**Section 1**

**Measurement and Methodology** 

**1** 

*Brazil* 

**Potential Risk: A New Approach** 

*Federal Institute of Education, Science and Technology of Bahia* 

Handerson J. Dourado Leite and Marcus V. Teixeira Navarro

Risk is a polysemic term that has been transformed throughout the historical process, but

The first rudimentary notion of what can be called risk, may have arisen, according to Covello and Munpower (1985), around 3200 BC in the valley between the Tigris and Euphrates Rivers, where lived a group called "*Asipu*". A major function of this group was to help people who needed to make difficult decisions. The "*Asipus*", when sought, identified the scale of the problem, the alternatives and the consequences of each alternative. Then, they drew up a table, marking the positive and negative points of each alternative to

With the great voyages in the fifteenth century it became necessary to evaluate the damage caused by the potential loss of ships. Emerges then the term risk, with connotations similar to what is meant today, but the understanding of its causes was related to accidents and, therefore, impossible to predict. The development of classical probability theory, in the midseventeenth century, to solve problems related to gambling, allowed the start of the process

Only from the nineteenth century, associated with the dominant thinking of the primacy of science and technique and propelled, among other factors, by the discoveries of Pasteur, emerged the association of risk with prevention, i.e., if the causes are known and quantified

The advent of modernity has produced and incorporated to the human way of life a variety of technologies and the risk became the distinguishing feature of this generated complexity. More and more, the sources of hazards1 were associated with daily social practices. In today's society, it is difficult to separate the manmade dangers of the "natural" dangers (Beck, 2003). A flood for example, that occurred as a completely spontaneous phenomenon, today can happen as a consequence of human action on nature. This new concept that the term risk assumes defies the human prediction capacity and rationality, because its causes are no longer accidental and the causes are not always known, or they are possible effects of

1 Hazards are "physical, chemical or biological agents or a set of conditions that present a source of

has always been associated to the idea of predicting an unwanted future event.

of quantifying the risks, but the causes were still credited to chance.

**1. Introduction** 

indicate the best decision.

one can predict the undesirable effects.

the technologies generated by man himself.

risk." (Kolluru, 1996. p. 3-41).

## **1**

### **Potential Risk: A New Approach**

 Handerson J. Dourado Leite and Marcus V. Teixeira Navarro *Federal Institute of Education, Science and Technology of Bahia Brazil* 

#### **1. Introduction**

Risk is a polysemic term that has been transformed throughout the historical process, but has always been associated to the idea of predicting an unwanted future event.

The first rudimentary notion of what can be called risk, may have arisen, according to Covello and Munpower (1985), around 3200 BC in the valley between the Tigris and Euphrates Rivers, where lived a group called "*Asipu*". A major function of this group was to help people who needed to make difficult decisions. The "*Asipus*", when sought, identified the scale of the problem, the alternatives and the consequences of each alternative. Then, they drew up a table, marking the positive and negative points of each alternative to indicate the best decision.

With the great voyages in the fifteenth century it became necessary to evaluate the damage caused by the potential loss of ships. Emerges then the term risk, with connotations similar to what is meant today, but the understanding of its causes was related to accidents and, therefore, impossible to predict. The development of classical probability theory, in the midseventeenth century, to solve problems related to gambling, allowed the start of the process of quantifying the risks, but the causes were still credited to chance.

Only from the nineteenth century, associated with the dominant thinking of the primacy of science and technique and propelled, among other factors, by the discoveries of Pasteur, emerged the association of risk with prevention, i.e., if the causes are known and quantified one can predict the undesirable effects.

The advent of modernity has produced and incorporated to the human way of life a variety of technologies and the risk became the distinguishing feature of this generated complexity. More and more, the sources of hazards1 were associated with daily social practices. In today's society, it is difficult to separate the manmade dangers of the "natural" dangers (Beck, 2003). A flood for example, that occurred as a completely spontaneous phenomenon, today can happen as a consequence of human action on nature. This new concept that the term risk assumes defies the human prediction capacity and rationality, because its causes are no longer accidental and the causes are not always known, or they are possible effects of the technologies generated by man himself.

<sup>1</sup> Hazards are "physical, chemical or biological agents or a set of conditions that present a source of risk." (Kolluru, 1996. p. 3-41).

Potential Risk: A New Approach 5

environmental, chemical agents and radioactive agents risks. In this context of large social movements, the need for State intervention was strengthened, in order to regulate the use of products potentially harmful to health and the environment (National Research Council,

The regulation of health risks is understood as a government interference in the market or in social processes, in order to control potentially damaging consequences to health (Hood; Rothstein; Baldwin, 2004). The model of the regulatory system, deployed in each country depends on political, economic and social conjunctures. Therefore, in the 1970s, while European countries exerted, initially, its regulatory power, by means of direct administration bodies of the State, the United States exercised this power, mainly, through

Currently, most European Union countries use the model of regulatory agencies (Lucchese, 2001). In Brazil, this role it's exercised in a hybrid way, because the National System of Sanitary Surveillance (*Sistema Nacional de Vigilância Sanitária* - SNVS) is composed of a regulatory agency in the federal sphere, the National Health Surveillance Agency (*Agência Nacional de Vigilância Sanitária* – ANVISA), but in most states and municipalities the regulation is exerted

The new technologies permeate the entire society and, therefore, influence and change the established social relations. These technologies are characterized by having intrinsic risks, by the possibility of adding new risks throughout their life cycle and by the incomplete scientific knowledge about the types of risks they generate and their interactions in different situations. Thus, the regulatory process occurs, in most cases, in situations of epistemic uncertainty, where risk factors are presented in a diffuse way, requiring from sanitary

As for the economic and social consequences related to the decisions of regulatory actions were amplified by the globalization process, as many decisions go beyond national borders and bring into play great interests. The first regulatory decisions showed that the process of definition and regulation of risk is an exercise of power, full of interests and political, economical, and social concepts, and can strongly influence the allocation of public and

Thus, the risk conceived as the probability of occurrence of an undesired event, calculated by specialists and presented to society as an absolute and neutral truth, began to be questioned. The conflicts of interest over the division of risk showed that it is not possible to separate the technical analysis about the risks from the decisions of who should be protected, from the costs and from the available alternatives, because the studies or risk

The fact that the calculation of risks undertaken by experts no longer represented the absolute truth and, also, the impossibility to eliminate the risks produced by the new technologies, because the benefits would also be suppressed, bring up new angles for the analysis of the phenomenon. Therefore, come into play other dimensions of risk as

surveillance the use of mutually complementary strategies of health protection.

private resources of a nation (Slovic, 2000, Fischhoff; Bostrum e Quadrel, 2005).

evaluations occur, necessarily, to subsidize decision-making.

acceptability, perception and confidence in the regulatory system.

1983, Lippmann; Cohen; Schlesinger, 2003, Omenn; Faustman, 2005)

independent and specialized agencies.

by direct administration.

**4. Other dimensions of risk** 

#### **2. Risk and probability**

The first report of a quantitative risk evaluation applied to health goes back to Laplace, in the late eighteenth century, which calculated the probability of death among people with and without vaccination for smallpox. With Pasteur's studies in the late nineteenth century, it was possible to use the tools of statistics to evaluate the factors related to communicable diseases, giving birth to the concept of epidemiological risk (Covello; Munpower, 1985, Czeresnia, 2004).

Epidemiological studies about contagious diseases have two very specific characteristics. The first refers to the object, which is only a source of damage. The second relates to the goals, which aim to determine the relationship between cause and effect, i.e., between exposure and disease. So, even with multifactorial determinants, it's an unidimensional evaluation. Therefore, in a evaluation between exposed and unexposed, the concept of risk approaches the definition of probability. However, when the objective includes the judgment about the severity of the injury or the comparison of different injuries in different exposures, the probability becomes one of the information that compose the concept of risk.

Therefore, the development of probability enabled the start of the process of quantifying risk. However, it's noteworthy that probability and risk are different concepts to most subjects. While the probability it's mathematically defined as the possibility or chance of a particular event occurs, and is represented by a number between 0 and 1 (Gelman; Nolan, 2004, Triola, 2005), the risk is associated with the probability of occurrence of an undesired event and its severity and cannot be represented by only one number.

If two events A and B have, respectively, 0.10 and 0.90 probability of occurring, the event B is classified as nine times more likely to occur than the event A. However, one can not say that the event B has a greater risk that the event A. For the concept of risk, is fundamental to know how much the event will be harmful. The evaluation of the probabilities of occurrence of the events A and B is done purely with mathematical analysis, while the risk assessment requires judgment of values. Thus, all observers will agree that the event B is more likely to happen than the event A, but not all should agree on which event represents a greater risk, knowing, or not, the damage.

As already explained, the notion of risk has been transformed throughout human history, it being understood nowadays as a theoretical elaboration that is historically constructed in order to mediate the relationship between man and the hazards, in order to minimize losses and maximize the benefits. Thus, it is not a greatness that is in nature to be measured, is not independent of the observer and his interests. It is formulated and evaluated within a political-economical-social context, having a multidimensional and multifactorial character (Fischhoff et al., 1983, Covello; Munpower, 1985, Beck, 2003, Hampel, 2006)

#### **3. The risk in the modern era**

The beginning of the twentieth century was marked by great scientific advances. The application of this knowledge produced new technologies such as X-rays, nuclear energy, asbestos and formaldehydes. The rapid use of these technologies as if they were only sources of benefits brought consequences to public health and to the environment, which only came to be perceived and understood by society, from the 70s of the last century. The disclosure of these risks led to pressures on governments, to control occupational,

The first report of a quantitative risk evaluation applied to health goes back to Laplace, in the late eighteenth century, which calculated the probability of death among people with and without vaccination for smallpox. With Pasteur's studies in the late nineteenth century, it was possible to use the tools of statistics to evaluate the factors related to communicable diseases, giving birth to the concept of epidemiological risk (Covello; Munpower, 1985, Czeresnia, 2004). Epidemiological studies about contagious diseases have two very specific characteristics. The first refers to the object, which is only a source of damage. The second relates to the goals, which aim to determine the relationship between cause and effect, i.e., between exposure and disease. So, even with multifactorial determinants, it's an unidimensional evaluation. Therefore, in a evaluation between exposed and unexposed, the concept of risk approaches the definition of probability. However, when the objective includes the judgment about the severity of the injury or the comparison of different injuries in different exposures, the probability becomes one of the information that compose the concept of risk. Therefore, the development of probability enabled the start of the process of quantifying risk. However, it's noteworthy that probability and risk are different concepts to most subjects. While the probability it's mathematically defined as the possibility or chance of a particular event occurs, and is represented by a number between 0 and 1 (Gelman; Nolan, 2004, Triola, 2005), the risk is associated with the probability of occurrence of an undesired

If two events A and B have, respectively, 0.10 and 0.90 probability of occurring, the event B is classified as nine times more likely to occur than the event A. However, one can not say that the event B has a greater risk that the event A. For the concept of risk, is fundamental to know how much the event will be harmful. The evaluation of the probabilities of occurrence of the events A and B is done purely with mathematical analysis, while the risk assessment requires judgment of values. Thus, all observers will agree that the event B is more likely to happen than the event A, but not all should agree on which event represents a greater risk,

As already explained, the notion of risk has been transformed throughout human history, it being understood nowadays as a theoretical elaboration that is historically constructed in order to mediate the relationship between man and the hazards, in order to minimize losses and maximize the benefits. Thus, it is not a greatness that is in nature to be measured, is not independent of the observer and his interests. It is formulated and evaluated within a political-economical-social context, having a multidimensional and multifactorial character

The beginning of the twentieth century was marked by great scientific advances. The application of this knowledge produced new technologies such as X-rays, nuclear energy, asbestos and formaldehydes. The rapid use of these technologies as if they were only sources of benefits brought consequences to public health and to the environment, which only came to be perceived and understood by society, from the 70s of the last century. The disclosure of these risks led to pressures on governments, to control occupational,

event and its severity and cannot be represented by only one number.

(Fischhoff et al., 1983, Covello; Munpower, 1985, Beck, 2003, Hampel, 2006)

**2. Risk and probability** 

knowing, or not, the damage.

**3. The risk in the modern era** 

environmental, chemical agents and radioactive agents risks. In this context of large social movements, the need for State intervention was strengthened, in order to regulate the use of products potentially harmful to health and the environment (National Research Council, 1983, Lippmann; Cohen; Schlesinger, 2003, Omenn; Faustman, 2005)

The regulation of health risks is understood as a government interference in the market or in social processes, in order to control potentially damaging consequences to health (Hood; Rothstein; Baldwin, 2004). The model of the regulatory system, deployed in each country depends on political, economic and social conjunctures. Therefore, in the 1970s, while European countries exerted, initially, its regulatory power, by means of direct administration bodies of the State, the United States exercised this power, mainly, through independent and specialized agencies.

Currently, most European Union countries use the model of regulatory agencies (Lucchese, 2001). In Brazil, this role it's exercised in a hybrid way, because the National System of Sanitary Surveillance (*Sistema Nacional de Vigilância Sanitária* - SNVS) is composed of a regulatory agency in the federal sphere, the National Health Surveillance Agency (*Agência Nacional de Vigilância Sanitária* – ANVISA), but in most states and municipalities the regulation is exerted by direct administration.

The new technologies permeate the entire society and, therefore, influence and change the established social relations. These technologies are characterized by having intrinsic risks, by the possibility of adding new risks throughout their life cycle and by the incomplete scientific knowledge about the types of risks they generate and their interactions in different situations. Thus, the regulatory process occurs, in most cases, in situations of epistemic uncertainty, where risk factors are presented in a diffuse way, requiring from sanitary surveillance the use of mutually complementary strategies of health protection.

As for the economic and social consequences related to the decisions of regulatory actions were amplified by the globalization process, as many decisions go beyond national borders and bring into play great interests. The first regulatory decisions showed that the process of definition and regulation of risk is an exercise of power, full of interests and political, economical, and social concepts, and can strongly influence the allocation of public and private resources of a nation (Slovic, 2000, Fischhoff; Bostrum e Quadrel, 2005).

Thus, the risk conceived as the probability of occurrence of an undesired event, calculated by specialists and presented to society as an absolute and neutral truth, began to be questioned. The conflicts of interest over the division of risk showed that it is not possible to separate the technical analysis about the risks from the decisions of who should be protected, from the costs and from the available alternatives, because the studies or risk evaluations occur, necessarily, to subsidize decision-making.

#### **4. Other dimensions of risk**

The fact that the calculation of risks undertaken by experts no longer represented the absolute truth and, also, the impossibility to eliminate the risks produced by the new technologies, because the benefits would also be suppressed, bring up new angles for the analysis of the phenomenon. Therefore, come into play other dimensions of risk as acceptability, perception and confidence in the regulatory system.

Potential Risk: A New Approach 7

In the center of the map is the information that characterizes the particularization of the model for the health surveillance: the object of study. Objects of action of health surveillance, herein referred to as technologies in health care, have three basic characteristics: they are of interest to health, produce benefits and have intrinsic risks. It is these characteristics that justify the action of health surveillance about the technologies for

In this triad, the risk is a feature that mobilizes a wide set of control strategies. As the risk is intrinsic to the object, it cannot be eliminated without eliminating the object, it can only be minimized. All technologies for health present some kind of risk and, if there is any that does not possess risks, it probably will not be object of action of the sanitary surveillance.

For possessing risks inherent in their nature, the technologies should be used in the

The diagram of the paradigm of risk, represented in Figure 1, is divided in half, pierced by social control and the object of study. The right side represents the field of risk assessment and the left side, the field of risk management. Risk assessment is the use of objective evidences to define the effects on health due to exposure of individuals or populations to hazardous materials or situations. Risk management refers to the process of integrating the results of risk assessment with social, economical and political issues, weighing the alternatives and selecting the most appropriate to the regulatory action (National Research

Risk assessment consists of three steps: identifying the source of damage, establishment of the dose x response and risk characterization. Risk identification is basically the answer to the question: which component of this health technology causes an adverse event? It is a question that can be answered based on causal, toxicological, and epidemiological evidence

In the second stage, two questions must be answered: how exposures occur? How is the relationship between exposure x effects (dose x response)? At this point, should be evaluated the conditions (intensity, frequency, duration, susceptibility and exposure period), in which the individuals or the populations are exposed. The second question should be answered with epidemiological, toxicological, experimental, and in vitro studies, using extrapolations or mathematical modeling, to establish the probability of occurrence

The last step is the characterization of the risk, in the classic sense. It is a moment of synthesis, when setting the damage likely to occur and its probability (P) the severity of the damage (D), the lifetime lost (T) and the vulnerabilities of exposure, as the intensity of exposure (I), the frequency of exposure (F), the duration of exposure (D), the exposed population (N), the populational groups (G) and the accessibility to the geographical

The risk assessment is a moment eminently technical and scientific, in which the theoretical models, the experimental procedures and the validation of the results are the elements of the performed studies (epidemiological, toxicological, in vitro and mathematical modeling, among others), so they can have rigor and scientific legitimacy. However, the evaluation

models are not independent of the observers and their objectives (Czeresnia, 2004).

or in vitro tests (National Research Council, 1983, Omenn; Faustman, 2005).

(National Research Council, 1983, Omenn; Faustman, 2005).

location of the population (L).

observance of the bioethical principle of the benefit (Costa, 2003, 2004)

health.

Council, 1983).

In beginning of the 1980, the U.S. Congress, realizing the need to structure a model of risk assessment that had wide acceptance, as well as standardizing the realization of studies in various areas, established a directive that designated the Food and Drug Administration (FDA) as responsible in coordinating a study for the harmonization. The FDA commissioned the National Academy of Sciences of the United States, which developed the project, whose results were of notorious and acknowledged importance, structuring the foundation for the paradigm of risk regulation (National Research Council, 1983, Omenn, Faustman, 2005).

This study, published in 1983 under the title *Risk assessment in the government: managing the process*, known internationally as the *Red Book*, establishes a process with seven stages: (1) Hazard identification, (2) dose x response assessment, (3) exposure assessment, (4) risk characterization; (5) Establishment of regulatory options, (6) Decision and implementation of the option of regulation, (7) Evaluation of the regulation. All steps occur with the participation of various actors, experts or not. The stages (1 to 4) are classified as risk assessment and are of technical and scientifically base. The other stages (5 to 7) are part of risk management, which, taking into account the information obtained in the first stage, evaluate and implement the best regulatory options, considering economical, political and social issues.

A diagram of the paradigm of risks applied to the area of health surveillance is represented in Figure 1.

Fig. 1. Diagram of the paradigm of risks applied to the area of health surveillance. Adapted Omenn and Faustman (2005, p. 1084)

In beginning of the 1980, the U.S. Congress, realizing the need to structure a model of risk assessment that had wide acceptance, as well as standardizing the realization of studies in various areas, established a directive that designated the Food and Drug Administration (FDA) as responsible in coordinating a study for the harmonization. The FDA commissioned the National Academy of Sciences of the United States, which developed the project, whose results were of notorious and acknowledged importance, structuring the foundation for the paradigm of risk regulation (National Research Council, 1983, Omenn, Faustman, 2005).

This study, published in 1983 under the title *Risk assessment in the government: managing the process*, known internationally as the *Red Book*, establishes a process with seven stages: (1) Hazard identification, (2) dose x response assessment, (3) exposure assessment, (4) risk characterization; (5) Establishment of regulatory options, (6) Decision and implementation of the option of regulation, (7) Evaluation of the regulation. All steps occur with the participation of various actors, experts or not. The stages (1 to 4) are classified as risk assessment and are of technical and scientifically base. The other stages (5 to 7) are part of risk management, which, taking into account the information obtained in the first stage, evaluate and implement the best regulatory options, considering economical, political and

A diagram of the paradigm of risks applied to the area of health surveillance is represented

Fig. 1. Diagram of the paradigm of risks applied to the area of health surveillance. Adapted

Omenn and Faustman (2005, p. 1084)

social issues.

in Figure 1.

In the center of the map is the information that characterizes the particularization of the model for the health surveillance: the object of study. Objects of action of health surveillance, herein referred to as technologies in health care, have three basic characteristics: they are of interest to health, produce benefits and have intrinsic risks. It is these characteristics that justify the action of health surveillance about the technologies for health.

In this triad, the risk is a feature that mobilizes a wide set of control strategies. As the risk is intrinsic to the object, it cannot be eliminated without eliminating the object, it can only be minimized. All technologies for health present some kind of risk and, if there is any that does not possess risks, it probably will not be object of action of the sanitary surveillance.

For possessing risks inherent in their nature, the technologies should be used in the observance of the bioethical principle of the benefit (Costa, 2003, 2004)

The diagram of the paradigm of risk, represented in Figure 1, is divided in half, pierced by social control and the object of study. The right side represents the field of risk assessment and the left side, the field of risk management. Risk assessment is the use of objective evidences to define the effects on health due to exposure of individuals or populations to hazardous materials or situations. Risk management refers to the process of integrating the results of risk assessment with social, economical and political issues, weighing the alternatives and selecting the most appropriate to the regulatory action (National Research Council, 1983).

Risk assessment consists of three steps: identifying the source of damage, establishment of the dose x response and risk characterization. Risk identification is basically the answer to the question: which component of this health technology causes an adverse event? It is a question that can be answered based on causal, toxicological, and epidemiological evidence or in vitro tests (National Research Council, 1983, Omenn; Faustman, 2005).

In the second stage, two questions must be answered: how exposures occur? How is the relationship between exposure x effects (dose x response)? At this point, should be evaluated the conditions (intensity, frequency, duration, susceptibility and exposure period), in which the individuals or the populations are exposed. The second question should be answered with epidemiological, toxicological, experimental, and in vitro studies, using extrapolations or mathematical modeling, to establish the probability of occurrence (National Research Council, 1983, Omenn; Faustman, 2005).

The last step is the characterization of the risk, in the classic sense. It is a moment of synthesis, when setting the damage likely to occur and its probability (P) the severity of the damage (D), the lifetime lost (T) and the vulnerabilities of exposure, as the intensity of exposure (I), the frequency of exposure (F), the duration of exposure (D), the exposed population (N), the populational groups (G) and the accessibility to the geographical location of the population (L).

The risk assessment is a moment eminently technical and scientific, in which the theoretical models, the experimental procedures and the validation of the results are the elements of the performed studies (epidemiological, toxicological, in vitro and mathematical modeling, among others), so they can have rigor and scientific legitimacy. However, the evaluation models are not independent of the observers and their objectives (Czeresnia, 2004).

Potential Risk: A New Approach 9

damages and its consequences, thus risk is characterized. In risk management, the forms of

The sanitary standards generally do not regulate the action of chemical, physical or biological substances, they regulate actions, procedures, products and equipments that must be used, so that the technologies for health may produce the maximum of benefit with the minimum of risk, considering the scientific, ethical, economical, political and social issues. The control actions are not related, necessarily, to the sources of risks. They may be related to conditions of the environment, of procedures, of human resources or of management of the own system of risk management. Since actions of health surveillance are focused, generally, on the control of risks and not on the risks itself, it becomes difficult the

The sanitary license, for example, is an operating concept that instrumentate the sanitary surveillance to control risk, but that is not directly related to any source of risk. A health service working without a sanitary license poses a risk to the system control, but may not represent a risk in the classical sense. One can not say what are the damages that may occur and in which probability. Even because the service can be fulfilling all technical and safety requirements. However, the absence of the license represents an unacceptable potential risk situation for the system control. Similar reasoning can be used to evaluate the equipment

The luminosity of the view box, used to view radiographic images, is another good example. The inadequate luminosity of the view box, despite not causing any direct harm to the patient, can hide radiological information and cause a misdiagnosis. In order to display the different tones of gray, in a radiography with optical density between 0.5 and 2.2, you need a view box with luminance between 2000 and 4000 nit2. So, what is the risk of using a view

There are so many variables involved that the question becomes difficult to answer. The possibility of error or loss of diagnostic information, for example, cannot be understood as a harm to the patient. The damage will be done when the decision making of the medical procedure, based on incorrect or incomplete diagnostic information, is made effective. Thus, one cannot determine the damage that will be caused and what are the probabilities of occurrence. One cannot say, even, that damage will occur. However, it is an unacceptable potentially hazardous situation, as is known to the minimum necessary light in a view box,

The potential risk concerns the possibility of an injury to health, without necessarily describing the injury and its probability of occurrence. It is an concept that expresses a value judgment about a potential exposure to a possible risk. It is as if it represents the risk of the

It is observed that the potential risk passes to present itself as a possibility of occurrence, or an expectation of the unexpected, therefore, it's related with possibility and not with probability. This difference is crucial to be able to clarify the proposed concept, after all, the probable is a category of the possible, that is, something is only probable if it's possible,

2 The unit of luminance in the International System is the cd/m2, known as nit.

control are identified, implemented and evaluated; thus control is characterized.

establishment of the cause-effect relationship.

box with a luminance of 500 nit?

to produce a reliable diagnosis condition.

risk.

registration, the professional certification, among others.

Risk assessment is not always possible to be performed quantitatively. In the case of the ionizing radiations, for example, the studied populations (Hiroshima and Nagasaki, Chernobyl and radiotherapy patients) were exposed to high doses, with high dose rates. Thus, it was necessary the use of the precautionary principle to postulate that, by extrapolation of the results of exposure at high doses, one must consider the linear relationship dose x response, without a threshold of exposure. Similar situations also occur in exposures to other physical and chemical elements, reflecting the complexity of the processes of risk assessment.

Based on information from the risk assessment, begins the process of management, conducted by the regulatory authority, also composed of three steps: establishment of regulatory options and decision making; implementation of control measures and risk communication and; assessment of the control actions.

In the first stage, are raised the possible actions that can minimize the risks, when the political-economical-cultural viability of each of the actions should be evaluated. Generally, there are several possibilities of regulation, when the best should be chosen. The best option is not, necessarily, the one with lowest risk or the one you want, it's the possible option in the evaluated context. The result of the value judgments will be the establishment of the limits of acceptability and of the control activities needed to keep the risks within these limits (National Research Council, 1983, Omenn; Faustman, 2005). In the case of the sanitary surveillance, this is the moment of development and publication of the standards for sanitary regulation.

The next step is the moment to inform society about the risks being regulated and the control measures being implemented. Parallel to the communication process, the regulatory authority should take the necessary measures, so that the control measures are effectively fulfilled by the regulated segment. An autonomous regulatory authority, with financial resources and skilled technicians, is a sine qua non condition for the implementation of the regulatory actions. However, the tradition of the institutions, of the regulated segment and of the society is essential so that risk control actions cease to be just rules and start to be practiced (National Research Council, 1983, Omenn; Faustman, 2005).

The last step is the evaluation of the entire process. It's the end of the first cycle and, perhaps, demands the beginning of a new cycle of risk assessment and management. To carry out the assessment, understood as a trial on a social practice or any of its components, in order to assist in decision-making, it is necessary to formulate strategies, select approaches, criteria, indicators and standards (Vieira Da Silva, 2005).

#### **5. The potential risk**

As seen so far, risk is a theoretical construct, historically grounded and, by the characteristics with which it presents itself in modern times, requires a regulatory system focused on protecting the health, due to the attributes that present the new technologies.

In the presented model of regulation of risks, the risk, in the classical sense, no longer has the central role, when passing from evaluation to management. In the process of risk management, the actions of health surveillance are focused, in general, on the control of risks and on the source of risks. In risk evaluation, the hazard is identified, related to the

Risk assessment is not always possible to be performed quantitatively. In the case of the ionizing radiations, for example, the studied populations (Hiroshima and Nagasaki, Chernobyl and radiotherapy patients) were exposed to high doses, with high dose rates. Thus, it was necessary the use of the precautionary principle to postulate that, by extrapolation of the results of exposure at high doses, one must consider the linear relationship dose x response, without a threshold of exposure. Similar situations also occur in exposures to other physical and chemical elements, reflecting the complexity of the

Based on information from the risk assessment, begins the process of management, conducted by the regulatory authority, also composed of three steps: establishment of regulatory options and decision making; implementation of control measures and risk

In the first stage, are raised the possible actions that can minimize the risks, when the political-economical-cultural viability of each of the actions should be evaluated. Generally, there are several possibilities of regulation, when the best should be chosen. The best option is not, necessarily, the one with lowest risk or the one you want, it's the possible option in the evaluated context. The result of the value judgments will be the establishment of the limits of acceptability and of the control activities needed to keep the risks within these limits (National Research Council, 1983, Omenn; Faustman, 2005). In the case of the sanitary surveillance, this is the moment of development and publication of the standards for

The next step is the moment to inform society about the risks being regulated and the control measures being implemented. Parallel to the communication process, the regulatory authority should take the necessary measures, so that the control measures are effectively fulfilled by the regulated segment. An autonomous regulatory authority, with financial resources and skilled technicians, is a sine qua non condition for the implementation of the regulatory actions. However, the tradition of the institutions, of the regulated segment and of the society is essential so that risk control actions cease to be just rules and start to be

The last step is the evaluation of the entire process. It's the end of the first cycle and, perhaps, demands the beginning of a new cycle of risk assessment and management. To carry out the assessment, understood as a trial on a social practice or any of its components, in order to assist in decision-making, it is necessary to formulate strategies, select

As seen so far, risk is a theoretical construct, historically grounded and, by the characteristics with which it presents itself in modern times, requires a regulatory system focused on protecting the health, due to the attributes that present the new technologies.

In the presented model of regulation of risks, the risk, in the classical sense, no longer has the central role, when passing from evaluation to management. In the process of risk management, the actions of health surveillance are focused, in general, on the control of risks and on the source of risks. In risk evaluation, the hazard is identified, related to the

practiced (National Research Council, 1983, Omenn; Faustman, 2005).

approaches, criteria, indicators and standards (Vieira Da Silva, 2005).

processes of risk assessment.

sanitary regulation.

**5. The potential risk** 

communication and; assessment of the control actions.

damages and its consequences, thus risk is characterized. In risk management, the forms of control are identified, implemented and evaluated; thus control is characterized.

The sanitary standards generally do not regulate the action of chemical, physical or biological substances, they regulate actions, procedures, products and equipments that must be used, so that the technologies for health may produce the maximum of benefit with the minimum of risk, considering the scientific, ethical, economical, political and social issues.

The control actions are not related, necessarily, to the sources of risks. They may be related to conditions of the environment, of procedures, of human resources or of management of the own system of risk management. Since actions of health surveillance are focused, generally, on the control of risks and not on the risks itself, it becomes difficult the establishment of the cause-effect relationship.

The sanitary license, for example, is an operating concept that instrumentate the sanitary surveillance to control risk, but that is not directly related to any source of risk. A health service working without a sanitary license poses a risk to the system control, but may not represent a risk in the classical sense. One can not say what are the damages that may occur and in which probability. Even because the service can be fulfilling all technical and safety requirements. However, the absence of the license represents an unacceptable potential risk situation for the system control. Similar reasoning can be used to evaluate the equipment registration, the professional certification, among others.

The luminosity of the view box, used to view radiographic images, is another good example. The inadequate luminosity of the view box, despite not causing any direct harm to the patient, can hide radiological information and cause a misdiagnosis. In order to display the different tones of gray, in a radiography with optical density between 0.5 and 2.2, you need a view box with luminance between 2000 and 4000 nit2. So, what is the risk of using a view box with a luminance of 500 nit?

There are so many variables involved that the question becomes difficult to answer. The possibility of error or loss of diagnostic information, for example, cannot be understood as a harm to the patient. The damage will be done when the decision making of the medical procedure, based on incorrect or incomplete diagnostic information, is made effective. Thus, one cannot determine the damage that will be caused and what are the probabilities of occurrence. One cannot say, even, that damage will occur. However, it is an unacceptable potentially hazardous situation, as is known to the minimum necessary light in a view box, to produce a reliable diagnosis condition.

The potential risk concerns the possibility of an injury to health, without necessarily describing the injury and its probability of occurrence. It is an concept that expresses a value judgment about a potential exposure to a possible risk. It is as if it represents the risk of the risk.

It is observed that the potential risk passes to present itself as a possibility of occurrence, or an expectation of the unexpected, therefore, it's related with possibility and not with probability. This difference is crucial to be able to clarify the proposed concept, after all, the probable is a category of the possible, that is, something is only probable if it's possible,

<sup>2</sup> The unit of luminance in the International System is the cd/m2, known as nit.

Potential Risk: A New Approach 11

The operationalization of the concept of potential risk has implications for the sanitary surveillance, because the quantification, classification and definition of acceptability levels of these risks will permit the monitoring and comparison of several objects under the control

A strategy to operationalize this concept is to establish a mathematical function that relates potential risk with risk control indicators. These control indicators are present in the rules, ie, are the characteristics associated with equipments, procedures, health services etc., that

The control indicators represent elements that, in most cases, you do not know the probability of generation of harmful effects, but, if outside of the pre-established parameters, there is a possibility that a harmful event may occur. Therefore, there is a causal relationship between indicators of control and potential risk, where both are inversely proportional, ie, the closer to the predetermined values are the control indicator, the lower

Having identified the causal relationship it's possible to establish mathematical formulations that describe the behavior of these relationships, through the traditional mathematical formalism or using new theoretical contributions to the theory of fuzzy sets which together with the theories of evidence and of possibility, constitute a new field of study that aims at the treatment of epistemic uncertainties within the possibilities, as will be shown below.

The theory of fuzzy sets, developed by Zadeh (1965), was born from the observation that in the real world certain objects or beings, such as the bacteria, are ambiguous as to which class they belong to, ie, have characteristics of animals and also vegetables. The observation of this ambiguity has led to the thought that there is no precision in the limits of a set and thus, it is possible to establish degrees of belonging of an element X, whatever, to a certain set. Taking as an example the bacteria, the number of animals characteristics that they exhibit allows us to establish a degree of belonging to the set of the animals, as well as, the amount of plant characteristics allows us to establish another degree of belonging to the set of the vegetables.This way, although they have a higher number of features of one kind or another, the bacterium does not cease to belong to both, though with different degrees of belonging. However, in the analysis of the ambiguities present in most of the everyday phenomena, is not always possible to quantify the characteristics of an element with precision to determine its degree of belonging. In most cases, these characteristics are presented in the form of uncertainties. To solve this problem, the modeling of the uncertainties uses the natural language (ordinary) and the membership functions express the possible values between 0

As in natural language are used variables or linguistic terms, also called inaccurate quantifiers , of common use in everyday life, but definers of many decisions, such as, "low," "high," "good," "very good"," tolerable "and so on. The membership functions consist of the association of each linguistic variable to a standard curve of possibilities (Shaw; Simões, 1999), which will define the membership degrees between 0 and 1, that the linguistic

**6. Strategy for operationalization of potential risk** 

of the sanitary surveillance, such as, the health services.

should be controlled within the pre-established parameters.

**6.1 A fuzzy logic system to evaluate potential risk** 

and 1, which each natural term may take. (Weber,2003).

variable may assume.

the potential risk and vice versa.

because if it's impossible, you cannot talk about probable or improbable. This condition of potential risk demonstrates its anteriority in relation to the classic risk. In the examples above, one can not calculate the probability of a damaging event for the lack of sanitary license or the low luminosity of the negatoscope, but, given what is known, there are chances that harmful events may occur due to these conditions.

Another important feature of the concept of potential risk refers to the temporal dimension of causal relationships. While the classic risk has its evaluation basis in occurred events, the potential risk has its causal evaluation foundations in the events that are occurring and the effects that may, or may not, occur in the future. Thus, allows work with the temporal dimension of risk facing the future or for a meta-reality and not for the past.

It is also possible to differentiate the potential risk from the classical risk according to the strategies used in the public health practices. These strategies can be divided into three great groups: health promotion in the restricted sense, health prevention (of risks or damages) and health protection.

In the practices of health promotion, strategies are aimed at capacity building and at raising awareness of the groups, so that they can take action to improve the quality of life and health, without being directed to a disease or injury whatsoever. They are actions of an educational nature which are not related to one or another specific risk factor (Almeida Filho, 2008). Thus, as their strategies do not involve specific risk factors, remains to discuss the concept of risk involving the two other strategies.

Regarding the preventive health strategy, the search for the determinants or the risk factors of a disease or of a specific aggravation on temporally and spatially defined individuals characterize their actions.In other words, are destined to act on these factors in order to reduce or eliminate new occurrences in the collective.It starts from "the assumption of recurrence of events in series, implying in an expectation of stability of the patterns of serial occurrence of the epidemiological facts" (Almeida Filho, 2000). As the action is given according to specific risk factors, ie, is related to the known behavior of the cause (risk factor) according to the probability of occurrence of the unwanted effect, the classical concept of risk seems to be the most appropriate.

On the other hand, health protection is intended to strengthen the individual defenses, therefore, is not always directed to known causes and specific risks, or relate to the referred events in series. They are used, in most cases, when there is an epistemic uncertainty, ie, when it's unknown or there is little information about the problem to be resolved or a decision to make. So, in the case of the health protection strategies, the central element in risk management is the potential risk that, despite not, necessarily, representing a defined relationship of cause and effect, can be quantified and classified into levels of acceptability, as will be discussed further, becoming an important operational concept of the sanitary surveillance.

However, the potential risk, as well as the classic risk, cannot be represented in most scientific fields by only a number. It should be understood and evaluated within a context and with limits of acceptability established by the technical and social determinants. Therefore, the evaluations made by regulatory authorities in the process of risk management have as indicators, in most cases, the tools of risk control and, as consequence, a measure of potential risk, which will indicate whether the control conditions are acceptable or not.

because if it's impossible, you cannot talk about probable or improbable. This condition of potential risk demonstrates its anteriority in relation to the classic risk. In the examples above, one can not calculate the probability of a damaging event for the lack of sanitary license or the low luminosity of the negatoscope, but, given what is known, there are

Another important feature of the concept of potential risk refers to the temporal dimension of causal relationships. While the classic risk has its evaluation basis in occurred events, the potential risk has its causal evaluation foundations in the events that are occurring and the effects that may, or may not, occur in the future. Thus, allows work with the temporal

It is also possible to differentiate the potential risk from the classical risk according to the strategies used in the public health practices. These strategies can be divided into three great groups: health promotion in the restricted sense, health prevention (of risks or damages)

In the practices of health promotion, strategies are aimed at capacity building and at raising awareness of the groups, so that they can take action to improve the quality of life and health, without being directed to a disease or injury whatsoever. They are actions of an educational nature which are not related to one or another specific risk factor (Almeida Filho, 2008). Thus, as their strategies do not involve specific risk factors, remains to discuss

Regarding the preventive health strategy, the search for the determinants or the risk factors of a disease or of a specific aggravation on temporally and spatially defined individuals characterize their actions.In other words, are destined to act on these factors in order to reduce or eliminate new occurrences in the collective.It starts from "the assumption of recurrence of events in series, implying in an expectation of stability of the patterns of serial occurrence of the epidemiological facts" (Almeida Filho, 2000). As the action is given according to specific risk factors, ie, is related to the known behavior of the cause (risk factor) according to the probability of occurrence of the unwanted effect, the classical

On the other hand, health protection is intended to strengthen the individual defenses, therefore, is not always directed to known causes and specific risks, or relate to the referred events in series. They are used, in most cases, when there is an epistemic uncertainty, ie, when it's unknown or there is little information about the problem to be resolved or a decision to make. So, in the case of the health protection strategies, the central element in risk management is the potential risk that, despite not, necessarily, representing a defined relationship of cause and effect, can be quantified and classified into levels of acceptability, as will be discussed further, becoming an important operational concept of the sanitary

However, the potential risk, as well as the classic risk, cannot be represented in most scientific fields by only a number. It should be understood and evaluated within a context and with limits of acceptability established by the technical and social determinants. Therefore, the evaluations made by regulatory authorities in the process of risk management have as indicators, in most cases, the tools of risk control and, as consequence, a measure of potential risk, which will indicate whether the control conditions are acceptable or not.

chances that harmful events may occur due to these conditions.

the concept of risk involving the two other strategies.

concept of risk seems to be the most appropriate.

and health protection.

surveillance.

dimension of risk facing the future or for a meta-reality and not for the past.

#### **6. Strategy for operationalization of potential risk**

The operationalization of the concept of potential risk has implications for the sanitary surveillance, because the quantification, classification and definition of acceptability levels of these risks will permit the monitoring and comparison of several objects under the control of the sanitary surveillance, such as, the health services.

A strategy to operationalize this concept is to establish a mathematical function that relates potential risk with risk control indicators. These control indicators are present in the rules, ie, are the characteristics associated with equipments, procedures, health services etc., that should be controlled within the pre-established parameters.

The control indicators represent elements that, in most cases, you do not know the probability of generation of harmful effects, but, if outside of the pre-established parameters, there is a possibility that a harmful event may occur. Therefore, there is a causal relationship between indicators of control and potential risk, where both are inversely proportional, ie, the closer to the predetermined values are the control indicator, the lower the potential risk and vice versa.

Having identified the causal relationship it's possible to establish mathematical formulations that describe the behavior of these relationships, through the traditional mathematical formalism or using new theoretical contributions to the theory of fuzzy sets which together with the theories of evidence and of possibility, constitute a new field of study that aims at the treatment of epistemic uncertainties within the possibilities, as will be shown below.

#### **6.1 A fuzzy logic system to evaluate potential risk**

The theory of fuzzy sets, developed by Zadeh (1965), was born from the observation that in the real world certain objects or beings, such as the bacteria, are ambiguous as to which class they belong to, ie, have characteristics of animals and also vegetables. The observation of this ambiguity has led to the thought that there is no precision in the limits of a set and thus, it is possible to establish degrees of belonging of an element X, whatever, to a certain set. Taking as an example the bacteria, the number of animals characteristics that they exhibit allows us to establish a degree of belonging to the set of the animals, as well as, the amount of plant characteristics allows us to establish another degree of belonging to the set of the vegetables.This way, although they have a higher number of features of one kind or another, the bacterium does not cease to belong to both, though with different degrees of belonging.

However, in the analysis of the ambiguities present in most of the everyday phenomena, is not always possible to quantify the characteristics of an element with precision to determine its degree of belonging. In most cases, these characteristics are presented in the form of uncertainties. To solve this problem, the modeling of the uncertainties uses the natural language (ordinary) and the membership functions express the possible values between 0 and 1, which each natural term may take. (Weber,2003).

As in natural language are used variables or linguistic terms, also called inaccurate quantifiers , of common use in everyday life, but definers of many decisions, such as, "low," "high," "good," "very good"," tolerable "and so on. The membership functions consist of the association of each linguistic variable to a standard curve of possibilities (Shaw; Simões, 1999), which will define the membership degrees between 0 and 1, that the linguistic variable may assume.

Potential Risk: A New Approach 13

Established the control points of the four possible causes, it is up to define which input variables of the FLS will be the results of the verification of the level of control of the set points. This level of control is called control indicator (CI) and shall be established by an observer, such as, a public health professional with expertise to make a subjective evaluation of each item, and may be defined, therefore, for a *fuzzy* linguistic variable, or inaccurate quantifier.

Defined the input and output variables of the FLS, it is necessary to establish the universe of discourse of each of them, ie, the variation range of the fuzzy linguistic variables of input and output. The universe of discourse limits the possible evaluations that the observer can present. As is the case of the input variables of the FLS, it is to check its adequacy, we will use the universe of discourse in terms of: Inadequate (IND), Shortly Adequate (SAD),

For the output variable of the FLS, since it is an indicator of potential risk, the universe of discourse adopted will be: Very Low (VL), Low (L), Medium (M), High (H) and Very High (VH). Note that in all cases the universe of discourse consists of 5 variables to allow good accuracy, since the greater the number of possibilities is, the better the accuracy of the evaluator. The next step will be to define the logical operations that must be made in the FLS so that, from the input variables, it can be obtained the potential risk indicator of biological contamination of the water for dialysis (PRI-BCW) in the output. Being the four input variables of the type, verification of the "level of adequacy", and as the output variable should represent an indicator of potential risk, two questions must be evaluated: which operations should be performed between the four input variables? and what is the

The operation between the input variables of the FLS should be held so that it is possible to obtain a single value, ie, a value that represents the level of control of all input variables (control indicator), ie, an indicator of aggregate control. Therefore, this must be one of the

The control indicators represent the level of control found by the observer and the 'potential risk' is the output of the FLS. Thus, the indicator of potential risk is inversely proportional to the control indicator, since the greater the observed control indicator, the lower the potential

To perform these operations, will be used *fuzzy* logic controllers. A *fuzzy* logic controller is a device that performs logical operations between *fuzzy* linguistic variables in its three stages: fuzzification, *fuzzy* inference and defuzzification. In this case, you need to build two types of

For each of the *fuzzy* controllers, it is necessary to develop the three steps referred above (fuzzification, *fuzzy* inference and defuzzification); therefore, it will be demonstrated, initially, the operation between the input variables of the FLS, known only as input controller. Each controller must perform only the operation between two input variables, so

Fuzzification means the process of transforming the possible existing information into *fuzzy* elements; consists in identifying the linguistic variables of input and output that you want to operate, defining the universe of discourse and the membership functions for each variable,

relationship between control indicator (CI) and potential risk indicator (PRI)?

risk and vice versa. So, this will be another operation to perform

there is no explosion of rules, as will be explained later.

*fuzzy* logic controllers, one for each type of operation you need to perform.

based on the experience and on the nature of the process being fuzzified.

Tolerable (TOL), Adequate (ADQ) and Very Adequate (VAD).

logical operations to be performed.

Zadeh (1965) developed operators for the *fuzzy* sets, enabling the establishment of relationships between them, being the most important the operations of maximum (max) and minimum (min), which can be easily understood if defined, respectively, as union and intersection in the classical set theory.

A fuzzy logic system (FLS), in a simplified manner, consists of performing logical operations with several fuzzy linguistic variables, in order to obtain a single value that represents the result of the performed operations.

To build an FLS, the first step consists of the definition of the input and output variables of the FLS, depending on the problem you want solved. When you want to, for example, know what is the potential risk indicator of biological contamination of the water for dialysis in the realization of the hemodialysis procedure; the output variable of the FLS may already be defined as the potential risk indicator of biological contamination of the water for dialysis (PRI-BCW)

To establish the input variables, the first question to be answered is: what are the possible causes to make water for dialysis potentially dangerous for biological contamination? Loosely, we can say that there are four causes: 1) Inadequacy of the drinking water treatment; 2) Inadequacy of the water treatment for dialysis; 3) Lack of knowledge or error of an employee who performs the procedure of water treatment for dialysis and 4) Inadequacy on the facilities of the water treatment plant.

The second question, in an attempt to define the input variables, is: how to handle each cause defined? Consulting the existing regulations for dialysis services in Brazil, you can display at least one control point for each defined cause, as described in Table 1.


Table 1. Relationship between possible causes and control points of the possibility of biological contamination of the water for dialysis.

Zadeh (1965) developed operators for the *fuzzy* sets, enabling the establishment of relationships between them, being the most important the operations of maximum (max) and minimum (min), which can be easily understood if defined, respectively, as union and

A fuzzy logic system (FLS), in a simplified manner, consists of performing logical operations with several fuzzy linguistic variables, in order to obtain a single value that represents the

To build an FLS, the first step consists of the definition of the input and output variables of the FLS, depending on the problem you want solved. When you want to, for example, know what is the potential risk indicator of biological contamination of the water for dialysis in the realization of the hemodialysis procedure; the output variable of the FLS may already be defined as the potential risk indicator of biological contamination of the water for dialysis

To establish the input variables, the first question to be answered is: what are the possible causes to make water for dialysis potentially dangerous for biological contamination? Loosely, we can say that there are four causes: 1) Inadequacy of the drinking water treatment; 2) Inadequacy of the water treatment for dialysis; 3) Lack of knowledge or error of an employee who performs the procedure of water treatment for dialysis and 4)

The second question, in an attempt to define the input variables, is: how to handle each cause defined? Consulting the existing regulations for dialysis services in Brazil, you can

518/2004 4

dialysis.

Table 1. Relationship between possible causes and control points of the possibility of

RDC nº 154/2004 5

Adequacy of the procedure for drinking water treatment, according to the Ordinance MS nº

Adequacy of the execution of the procedure of water treatment for dialysis, according to the

Adequacy of the capacity of an employee who performs the procedure of water treatment for

Adequacy of the constructive aspects and of the equipment used in the water treatment plant, according to the RDC nº 154/2004 5

display at least one control point for each defined cause, as described in Table 1.

intersection in the classical set theory.

result of the performed operations.

Inadequacy on the facilities of the water treatment plant.

Cause Control point

Inadequacy of the drinking water

Lack of knowledge or error of an employee who performs the procedure

of water treatment for dialysis.

Inadequacy of the water treatment for

Inadequacy on the facilities of the water

biological contamination of the water for dialysis.

(PRI-BCW)

treatment.

dialysis.

treatment plant.

Established the control points of the four possible causes, it is up to define which input variables of the FLS will be the results of the verification of the level of control of the set points. This level of control is called control indicator (CI) and shall be established by an observer, such as, a public health professional with expertise to make a subjective evaluation of each item, and may be defined, therefore, for a *fuzzy* linguistic variable, or inaccurate quantifier.

Defined the input and output variables of the FLS, it is necessary to establish the universe of discourse of each of them, ie, the variation range of the fuzzy linguistic variables of input and output. The universe of discourse limits the possible evaluations that the observer can present. As is the case of the input variables of the FLS, it is to check its adequacy, we will use the universe of discourse in terms of: Inadequate (IND), Shortly Adequate (SAD), Tolerable (TOL), Adequate (ADQ) and Very Adequate (VAD).

For the output variable of the FLS, since it is an indicator of potential risk, the universe of discourse adopted will be: Very Low (VL), Low (L), Medium (M), High (H) and Very High (VH). Note that in all cases the universe of discourse consists of 5 variables to allow good accuracy, since the greater the number of possibilities is, the better the accuracy of the evaluator.

The next step will be to define the logical operations that must be made in the FLS so that, from the input variables, it can be obtained the potential risk indicator of biological contamination of the water for dialysis (PRI-BCW) in the output. Being the four input variables of the type, verification of the "level of adequacy", and as the output variable should represent an indicator of potential risk, two questions must be evaluated: which operations should be performed between the four input variables? and what is the relationship between control indicator (CI) and potential risk indicator (PRI)?

The operation between the input variables of the FLS should be held so that it is possible to obtain a single value, ie, a value that represents the level of control of all input variables (control indicator), ie, an indicator of aggregate control. Therefore, this must be one of the logical operations to be performed.

The control indicators represent the level of control found by the observer and the 'potential risk' is the output of the FLS. Thus, the indicator of potential risk is inversely proportional to the control indicator, since the greater the observed control indicator, the lower the potential risk and vice versa. So, this will be another operation to perform

To perform these operations, will be used *fuzzy* logic controllers. A *fuzzy* logic controller is a device that performs logical operations between *fuzzy* linguistic variables in its three stages: fuzzification, *fuzzy* inference and defuzzification. In this case, you need to build two types of *fuzzy* logic controllers, one for each type of operation you need to perform.

For each of the *fuzzy* controllers, it is necessary to develop the three steps referred above (fuzzification, *fuzzy* inference and defuzzification); therefore, it will be demonstrated, initially, the operation between the input variables of the FLS, known only as input controller. Each controller must perform only the operation between two input variables, so there is no explosion of rules, as will be explained later.

Fuzzification means the process of transforming the possible existing information into *fuzzy* elements; consists in identifying the linguistic variables of input and output that you want to operate, defining the universe of discourse and the membership functions for each variable, based on the experience and on the nature of the process being fuzzified.

Potential Risk: A New Approach 15

It is also important to point out, in Figure 2, that each *fuzzy* linguistic variable was associated with a numerical value 0%; 25%, 50%, 75% and 100%, respectively. This fact can be identified, by observing that the top of the trapezoids corresponds to one of these values. So, if the point 0.5 is taken (50%), in the X-axis (blue dotted line), it will correspond to the center of the trapezoid for the *fuzzy* linguistic variables 'Tolerable', in the input and

We opted for the trapezoidal shape because it was recognized that in the observation made there is no accuracy of values; when reporting, for example, that a level of control is 'shortly adequate', this does not correspond exactly to 25% but to a range for that value. Now the option for the symmetry was made as it was considered that there are an equal number of chances of the observer to choose for any of the *fuzzy* linguistic variables that compose the

The importance of fuzzification can be understood, when we take a value, for example, 0.35 in the abscissa axis (red line), note that this value has a degree of membership greater than 50% for 'shortly adequate' and less than 50% for 'tolerable'. These membership differences

Completed the process of fuzzification, it is necessary to perform the *fuzzy* inference process. The *fuzzy* inference process consists in the processing of the *fuzzy* variables according to specific rules. There are basically two methods, the Mamdani model and the Takagi-Sugeno-Kang model (Shaw; Simões, 2005). Mamdani's method is the most used and recommended for the treatment with inaccurate information. It is based on the elaboration of rules of the 'IF' <condition>; 'THEN' <consequence> type, using the heuristic method. The rules are the knowledge bases, from which, an "inference machine" (*software* or *hardware*) acts and performs operations of minimum (intersection) between the input *fuzzy* linguistic variables of each rule, and of maximum (union) between the results obtained by the previous

A rule of the 'IF' <condition>; 'THEN' <consequence> type is a simple logic rule and it means that for a given situation, 'IF' a condition is met, even partially, 'THEN', a consequence will occur. For example, when one states that the potential risk is inversely proportional to the level of control, it is possible to say that 'IF' the level of control is high, 'THEN' the potential risk is low. When two variables (two conditions) are associated, using the Mamdani method, we use the operator 'AND' between the two variables to indicate that an operation will take place between them. This way, the rule is now stated as: 'IF' a condition is met, even partially 'AND' other condition is also met, even partially, 'THEN', some consequence will occur. This way, using the heuristic method was constructed the

It will be required the construction of twenty-five rules, because for two variables per controller and five fuzzy linguistic variables, one needs, therefore, twenty-five combinations

It should be noted, also, that in Table 2 the input variables were treated generically as 'Adequacy 1' and 'Adequacy 2', because it will be necessary to use more than one input controller, since there are four input variables. So, you can use the same set of rules for both

'Medium', in the output.

universe of discourse.

operation.

(52).

controllers.

will generate the sets that will be operationalized.

rules base for *fuzzy* logic controller input, shown in Table 2.

To perform the fuzzification of the input controller, some steps have been taken, as the identification of the input linguistic variables and the establishment of the universe of discourse (Inadequate – IND, Shortly Adequate – SAD, Tolerable – TOL, Adequate – ADQ and Very Adequate – VAD). However, it lacks defining the output variable and the universe of discourse for this controller, because, as has been identified, it will be required more than one logical operation between the *fuzzy* variables, the output of this input controller, will not, necessarily, be equal to the output variable of the FLS. Thus, considering that the objective of this controller is to aggregate the control indicators (CI) pointed by the observer and thinking about the future composition of the organization of the FLS, it was decided, in the example shown, to define the universe of discourse of the output variable as: Very Low (VL), Low (L), Medium (M), High (H) and Very High (VH).

The last step to accomplish the process of fuzzification is to define the membership function for each identified *fuzzy* linguistic variable. In this case, we took the function of trapezoidal and symmetrical shape for all the input controller's *fuzzy* linguistic variables, as can be seen in Figure 2.

A membership function defines the degree of belonging or membership of each *fuzzy* linguistic value, ie, it represents the curve of possibilities of the behavior of the *fuzzy* linguistic variable (Weber, 2003). Note that the membership functions are standard functions, ie, in its ordinate axis (Y) it only admits *fuzzy* values from '0 'to '1', ie, it goes from the not belonging (0%) to the total belonging (100%). In the abscissa axis (X) the values depend on the problem addressed; in this case, we used '0 'to '1', because those are variables that assume this behavior (potential risk and control indicator).

Fig. 2. Input controller's input and output membership functions

To perform the fuzzification of the input controller, some steps have been taken, as the identification of the input linguistic variables and the establishment of the universe of discourse (Inadequate – IND, Shortly Adequate – SAD, Tolerable – TOL, Adequate – ADQ and Very Adequate – VAD). However, it lacks defining the output variable and the universe of discourse for this controller, because, as has been identified, it will be required more than one logical operation between the *fuzzy* variables, the output of this input controller, will not, necessarily, be equal to the output variable of the FLS. Thus, considering that the objective of this controller is to aggregate the control indicators (CI) pointed by the observer and thinking about the future composition of the organization of the FLS, it was decided, in the example shown, to define the universe of discourse of the output variable as: Very Low

The last step to accomplish the process of fuzzification is to define the membership function for each identified *fuzzy* linguistic variable. In this case, we took the function of trapezoidal and symmetrical shape for all the input controller's *fuzzy* linguistic variables, as can be seen

A membership function defines the degree of belonging or membership of each *fuzzy* linguistic value, ie, it represents the curve of possibilities of the behavior of the *fuzzy* linguistic variable (Weber, 2003). Note that the membership functions are standard functions, ie, in its ordinate axis (Y) it only admits *fuzzy* values from '0 'to '1', ie, it goes from the not belonging (0%) to the total belonging (100%). In the abscissa axis (X) the values depend on the problem addressed; in this case, we used '0 'to '1', because those are variables

(VL), Low (L), Medium (M), High (H) and Very High (VH).

that assume this behavior (potential risk and control indicator).

Fig. 2. Input controller's input and output membership functions

in Figure 2.

It is also important to point out, in Figure 2, that each *fuzzy* linguistic variable was associated with a numerical value 0%; 25%, 50%, 75% and 100%, respectively. This fact can be identified, by observing that the top of the trapezoids corresponds to one of these values. So, if the point 0.5 is taken (50%), in the X-axis (blue dotted line), it will correspond to the center of the trapezoid for the *fuzzy* linguistic variables 'Tolerable', in the input and 'Medium', in the output.

We opted for the trapezoidal shape because it was recognized that in the observation made there is no accuracy of values; when reporting, for example, that a level of control is 'shortly adequate', this does not correspond exactly to 25% but to a range for that value. Now the option for the symmetry was made as it was considered that there are an equal number of chances of the observer to choose for any of the *fuzzy* linguistic variables that compose the universe of discourse.

The importance of fuzzification can be understood, when we take a value, for example, 0.35 in the abscissa axis (red line), note that this value has a degree of membership greater than 50% for 'shortly adequate' and less than 50% for 'tolerable'. These membership differences will generate the sets that will be operationalized.

Completed the process of fuzzification, it is necessary to perform the *fuzzy* inference process. The *fuzzy* inference process consists in the processing of the *fuzzy* variables according to specific rules. There are basically two methods, the Mamdani model and the Takagi-Sugeno-Kang model (Shaw; Simões, 2005). Mamdani's method is the most used and recommended for the treatment with inaccurate information. It is based on the elaboration of rules of the 'IF' <condition>; 'THEN' <consequence> type, using the heuristic method. The rules are the knowledge bases, from which, an "inference machine" (*software* or *hardware*) acts and performs operations of minimum (intersection) between the input *fuzzy* linguistic variables of each rule, and of maximum (union) between the results obtained by the previous operation.

A rule of the 'IF' <condition>; 'THEN' <consequence> type is a simple logic rule and it means that for a given situation, 'IF' a condition is met, even partially, 'THEN', a consequence will occur. For example, when one states that the potential risk is inversely proportional to the level of control, it is possible to say that 'IF' the level of control is high, 'THEN' the potential risk is low. When two variables (two conditions) are associated, using the Mamdani method, we use the operator 'AND' between the two variables to indicate that an operation will take place between them. This way, the rule is now stated as: 'IF' a condition is met, even partially 'AND' other condition is also met, even partially, 'THEN', some consequence will occur. This way, using the heuristic method was constructed the rules base for *fuzzy* logic controller input, shown in Table 2.

It will be required the construction of twenty-five rules, because for two variables per controller and five fuzzy linguistic variables, one needs, therefore, twenty-five combinations (52).

It should be noted, also, that in Table 2 the input variables were treated generically as 'Adequacy 1' and 'Adequacy 2', because it will be necessary to use more than one input controller, since there are four input variables. So, you can use the same set of rules for both controllers.

Potential Risk: A New Approach 17

(AWTD), a control indicator 'Adequate' (0.75). The red lines represent the values assigned to each input variable and the yellow forms, the set generated in each rule. Note that operations of minimum (intersection) are performed between the yellow sets of each rule, generating as results the blue sets. Among the blue sets, an operation of maximum (union) is performed, resulting in the set surrounded by a red line, representing the *fuzzy*

The defuzzification process is translated into the transformation of the *fuzzy* set resulting in a discrete value, seeking to define the value that best represents the distribution of possibilities present in the output variable. The three most used methods for defuzzification are the center of area (C-O-A), the center of maximum (C-O-M) and the mean of maximum (M-O-M). The C-O-A method calculates the centroid of the area obtained in the output, or the point that divides this area in half, after the max-min operations performed on *fuzzy* inference. The C-O-M method calculates a weighted average of the maximum values present in the exit area, which weights are the results of *fuzzy* inference, the area itself has no influence on the outcome. Finally, the M-O-M method, used in this work, calculates an average of the maximum values present in the

exit area, disregarding the format of this area, as shown in Figure 3.

Fig. 3. The steps of fuzzy inference and defuzzification for the input controller

result.


The 'inference machine' is a *software* or *hardware* that performs logic operations based on defined rules.

(Inadequate (IND), Shortly Adequate (SAD), Tolerable (TOL), Adequate (ADQ), Very Adequate (VAD), Very Low (VL), Low (L), Medium (M), High (H) and Very High (VH)).

Table 2. Rules 'IF'...'THEN' for *fuzzy* input controller

As shown in the example above, when it was shown the importance of fuzzification, when defining a control indicator for an input variable, it will be associated with a number that will produce different degrees of membership for each membership function and, at every point where it intercepts the membership function, it will generate *fuzzy* sets. In the *fuzzy* inference it is verified if there is a point of interception for all defined rules 'IF', 'THEN'. Among the sets generated in each variable and in each rule, it is performed an operation of minimum (intersection) that corresponds to the operator 'AND'. Among the resulting sets from the operation of minimum of every rule, it is performed an operation of maximum (union), coming to a set representing the results of the performed *fuzzy* operations.

In Figure 3, it is shown what happens in the process of *fuzzy* inference. It was assigned to the 'adequacy of the procedure for drinking water treatment' (ADWT) a control indicator 'Tolerable' (0.5) and to the 'adequacy of the procedure of water treatment for dialysis'

The 'inference machine' is a *software* or *hardware* that performs logic operations based on

Rules IF Condition AND Condition THEN Condition 1 'Adequacy 1' VAD 'Adequacy 2' VAD 'Control' VH 2 'Adequacy 1' VAD 'Adequacy 2' ADQ 'Control' VH 3 'Adequacy 1' VAD 'Adequacy 2' TOL 'Control' M 4 'Adequacy 1' VAD 'Adequacy 2' SAD 'Control' L 5 'Adequacy 1' VAD 'Adequacy 2' IND 'Control' L 6 'Adequacy 1' ADQ 'Adequacy 2' VAD 'Control' VH 7 'Adequacy 1' ADQ 'Adequacy 2' ADQ 'Control' H 8 'Adequacy 1' ADQ 'Adequacy 2' TOL 'Control' M 9 'Adequacy 1' ADQ 'Adequacy 2' SAD 'Control' L 10 'Adequacy 1' ADQ 'Adequacy 2' IND 'Control' L 11 'Adequacy 1' TOL 'Adequacy 2' VAD 'Control' M 12 'Adequacy 1' TOL 'Adequacy 2' ADQ 'Control' M 13 'Adequacy 1' TOL 'Adequacy 2' TOL 'Control' M 14 'Adequacy 1' TOL 'Adequacy 2' SAD 'Control' L 15 'Adequacy 1' TOL 'Adequacy 2' IND 'Control' VL 16 'Adequacy 1' SAD 'Adequacy 2' VAD 'Control' L 17 'Adequacy 1' SAD 'Adequacy 2' ADQ 'Control' L 18 'Adequacy 1' SAD 'Adequacy 2' TOL 'Control' L 19 'Adequacy 1' SAD 'Adequacy 2' SAD 'Control' VL 20 'Adequacy 1' SAD 'Adequacy 2' IND 'Control' VL 21 'Adequacy 1' IND 'Adequacy 2' VAD 'Control' L 22 'Adequacy 1' IND 'Adequacy 2' ADQ 'Control' L 23 'Adequacy 1' IND 'Adequacy 2' TOL 'Control' VL 24 'Adequacy 1' IND 'Adequacy 2' SAD 'Control' VL 25 'Adequacy 1' IND 'Adequacy 2' IND 'Control' VL (Inadequate (IND), Shortly Adequate (SAD), Tolerable (TOL), Adequate (ADQ), Very Adequate (VAD),

Very Low (VL), Low (L), Medium (M), High (H) and Very High (VH)).

As shown in the example above, when it was shown the importance of fuzzification, when defining a control indicator for an input variable, it will be associated with a number that will produce different degrees of membership for each membership function and, at every point where it intercepts the membership function, it will generate *fuzzy* sets. In the *fuzzy* inference it is verified if there is a point of interception for all defined rules 'IF', 'THEN'. Among the sets generated in each variable and in each rule, it is performed an operation of minimum (intersection) that corresponds to the operator 'AND'. Among the resulting sets from the operation of minimum of every rule, it is performed an operation of maximum

In Figure 3, it is shown what happens in the process of *fuzzy* inference. It was assigned to the 'adequacy of the procedure for drinking water treatment' (ADWT) a control indicator 'Tolerable' (0.5) and to the 'adequacy of the procedure of water treatment for dialysis'

(union), coming to a set representing the results of the performed *fuzzy* operations.

Table 2. Rules 'IF'...'THEN' for *fuzzy* input controller

defined rules.

(AWTD), a control indicator 'Adequate' (0.75). The red lines represent the values assigned to each input variable and the yellow forms, the set generated in each rule. Note that operations of minimum (intersection) are performed between the yellow sets of each rule, generating as results the blue sets. Among the blue sets, an operation of maximum (union) is performed, resulting in the set surrounded by a red line, representing the *fuzzy* result.

The defuzzification process is translated into the transformation of the *fuzzy* set resulting in a discrete value, seeking to define the value that best represents the distribution of possibilities present in the output variable. The three most used methods for defuzzification are the center of area (C-O-A), the center of maximum (C-O-M) and the mean of maximum (M-O-M). The C-O-A method calculates the centroid of the area obtained in the output, or the point that divides this area in half, after the max-min operations performed on *fuzzy* inference. The C-O-M method calculates a weighted average of the maximum values present in the exit area, which weights are the results of *fuzzy* inference, the area itself has no influence on the outcome. Finally, the M-O-M method, used in this work, calculates an average of the maximum values present in the exit area, disregarding the format of this area, as shown in Figure 3.

Fig. 3. The steps of fuzzy inference and defuzzification for the input controller

Potential Risk: A New Approach 19

dialysis (CI-AWTD), of the adequacy of the constructive aspects and of the equipment used in the water treatment plant (CI-ACEQ) and of the adequacy of the capacity of an employee who performs the procedure of water treatment for dialysis (CI-ACET), respectively, 'TOL', 'ADQ', 'SAD' e 'VAD', so, the PRI-BCW of this system will be considered high (H), ie, 0.75; indicating that there is a nonconformity at some point in the process under analysis. In this case, the inadequacy of the constructive aspects of the water treatment plant and / or of the

The formulation of the PRAM has been developed generalized so that it could be applied in

The PRAM was validated by evaluating potential risks in radiodiagnostic services in the State of Bahia, Brazil, enabling advance, in order to better understand the specific problems and the possibilities of action of the health surveillance system, as the regulatory authority,

The validation results showed that use of the PRAM model allowed going beyond simple situational description, indicating the possible explanatory factors of the health situation found. Some advantages of this approach are introduced, in comparison with other works that dealt with the theme. One of them concerns the graphical representation of the potential

This enables the regulatory system to classify and compare the evaluated procedures, so that you can plan and direct the actions for the services whose procedures are in unacceptable or

Another advantage relates to the possibility of applying the principle of optimization in the risk control system, enabling the continuous evolution of the system, evaluating the historical evolution of risk management. The monitoring of time evolution can show an advance or a retreat of the potential hazard, alerting the regulatory authority before the service moves to a range of higher degree of risk, allowing risks prevention actions, by

So, the Regulatory Authority has the possibility to act in preventing the risk and not just in control. The temporal evolution can be used easily, with computational aid, to monitor the

However, using PRAM to monitor the temporal evolution of the potential risks, as well as for comparison and risk assessment, should be carried out using the same rating scales and indicators of the same ranges of acceptability. Otherwise, the PRAM loses

The PRAM needs to consider important issues of risk governance. The first question refers to the range of variation. The PRAM needs to be represented by a mathematical formalism, whose values of the potential risk - PR are always within the same range of variation, regardless of the number of indicators, and there is no possibility of taking the

equipment used to perform the process.

in control of risks in radiodiagnostic.

anticipating and stopping a trend.

services individually or collectively.

comparability.

zero value.

risk of each procedure in each of the services.

tolerable level of potential risk , establishing priorities.

**6.2 O PRAM: Potential Risk Assessment Model** 

any area of risk governance and possibly, also outside it.

The second type of *fuzzy* logic controller to be built is called output controller. As can be seen in Figure 4, the input variables of this output controller will be equal to the output variables of the input controller and the output variables will be equal to the output variables of the FLS. The only difference will be the rule base 'IF'; 'THEN', but all other steps are identical to the input controller. The difference in the rule base exists, because the logical operation to be performed will be the conversion of the indicator of control for potential risk indicators that are inversely proportional. Thus, the rule base 'IF', 'THEN' was elaborated considering this criterion.

Finally, for the construction of the FLS the *fuzzy* logic controllers will be grouped so as to produce the desired information, as shown in Figure 4. To carry out the construction and operation of an FLS, the program MatLab can be used.

Fig. 4. Fuzzy logic system for indication of potential risk of biological contamination of the water for dialysis

Thus, as can be seen in Figure 4, when evaluating a service of dialysis a sanitary inspection team should consider the control indicators of the adequacy of the procedure for drinking water treatment (CI-ADWT), of the adequacy of the procedure of water treatment for

The second type of *fuzzy* logic controller to be built is called output controller. As can be seen in Figure 4, the input variables of this output controller will be equal to the output variables of the input controller and the output variables will be equal to the output variables of the FLS. The only difference will be the rule base 'IF'; 'THEN', but all other steps are identical to the input controller. The difference in the rule base exists, because the logical operation to be performed will be the conversion of the indicator of control for potential risk indicators that are inversely proportional. Thus, the rule base 'IF', 'THEN'

Finally, for the construction of the FLS the *fuzzy* logic controllers will be grouped so as to produce the desired information, as shown in Figure 4. To carry out the construction and

Fig. 4. Fuzzy logic system for indication of potential risk of biological contamination of the

Thus, as can be seen in Figure 4, when evaluating a service of dialysis a sanitary inspection team should consider the control indicators of the adequacy of the procedure for drinking water treatment (CI-ADWT), of the adequacy of the procedure of water treatment for

was elaborated considering this criterion.

water for dialysis

operation of an FLS, the program MatLab can be used.

dialysis (CI-AWTD), of the adequacy of the constructive aspects and of the equipment used in the water treatment plant (CI-ACEQ) and of the adequacy of the capacity of an employee who performs the procedure of water treatment for dialysis (CI-ACET), respectively, 'TOL', 'ADQ', 'SAD' e 'VAD', so, the PRI-BCW of this system will be considered high (H), ie, 0.75; indicating that there is a nonconformity at some point in the process under analysis. In this case, the inadequacy of the constructive aspects of the water treatment plant and / or of the equipment used to perform the process.

#### **6.2 O PRAM: Potential Risk Assessment Model**

The formulation of the PRAM has been developed generalized so that it could be applied in any area of risk governance and possibly, also outside it.

The PRAM was validated by evaluating potential risks in radiodiagnostic services in the State of Bahia, Brazil, enabling advance, in order to better understand the specific problems and the possibilities of action of the health surveillance system, as the regulatory authority, in control of risks in radiodiagnostic.

The validation results showed that use of the PRAM model allowed going beyond simple situational description, indicating the possible explanatory factors of the health situation found. Some advantages of this approach are introduced, in comparison with other works that dealt with the theme. One of them concerns the graphical representation of the potential risk of each procedure in each of the services.

This enables the regulatory system to classify and compare the evaluated procedures, so that you can plan and direct the actions for the services whose procedures are in unacceptable or tolerable level of potential risk , establishing priorities.

Another advantage relates to the possibility of applying the principle of optimization in the risk control system, enabling the continuous evolution of the system, evaluating the historical evolution of risk management. The monitoring of time evolution can show an advance or a retreat of the potential hazard, alerting the regulatory authority before the service moves to a range of higher degree of risk, allowing risks prevention actions, by anticipating and stopping a trend.

So, the Regulatory Authority has the possibility to act in preventing the risk and not just in control. The temporal evolution can be used easily, with computational aid, to monitor the services individually or collectively.

However, using PRAM to monitor the temporal evolution of the potential risks, as well as for comparison and risk assessment, should be carried out using the same rating scales and indicators of the same ranges of acceptability. Otherwise, the PRAM loses comparability.

The PRAM needs to consider important issues of risk governance. The first question refers to the range of variation. The PRAM needs to be represented by a mathematical formalism, whose values of the potential risk - PR are always within the same range of variation, regardless of the number of indicators, and there is no possibility of taking the zero value.

Potential Risk: A New Approach 21

maximum value of an indicator, ie, regardless of the number of indicators that is selected,

Once the non-critical indicators do not have the ability to, individually, represent the commitment of all the system potential risks control, cannot have its mean represented by a multiplicand. However, they also need to be represented by a mean, so that the representative value of the set is equal, at most, to the maximum value of one of its elements

Therefore, the best way to represent them is through an arithmetic mean. The non-critical indicators (INC) can be represented by a simple arithmetic average, because it can only be

1

*<sup>j</sup>*<sup>=</sup> = 

The function risk control (RC), which represents the result of the indicators of risks control,

Once more, we used the geometric mean, so that the risk control (RC) is in a range of

Taking the risk control (RC) as the independent variable, the function that best represent the relationship of cause and effect between risk control and potential risk is the exponential

PR (RC) - Potential risk function, which is dependent on the risk control function, will be referred to as PR; RC - Risk control, function that determines the potential risk and that, on

The shape of the exponential function, with a rapid decrease, represents a good model for critical phenomena, as is the case of the potential risk for health services. The complex relationship between the various factors that influence in the risk control exhibits a kind of not extensive sum, where the potential risk for an event, involving the junction between two factors, can be greater than the sum of the potential risk of the two factors

This type of behavior ends up generating a sudden increase of the potential risk, when adding many elements or some critics, being perfectly represented by the rapid decrease of

*M*

M

I

NC

variation known in advance and that depends only on the variation of IC and INC.

P( ) <sup>R</sup>

the other hand, is determined by the indicators of risk control.

j

NC

{NC ; NC ; NC ;...;NC III N <sup>123</sup> } (3)

( ) , *R C NC C NC CI I* = × (5)

*Rc RC e*<sup>−</sup> <sup>=</sup> (6)

(4)

the result will always be in the same range of variation.

and is within a known range of variation.

zero, if all control indicators are non-existent.

should be represented as the geometric mean, ie:

function, with the following form:

separately.

the exponential function.

The set of non-critical indicators is formed by the INC indicators.

The issue of the values being within the same range of variation allows the comparison and the establishment of limits of acceptability, while the not possibility of assuming the value zero is a condition of the problem, because the risks can be as small as possible, but will never be nulls.

The levels of acceptability should not have a direct border between the acceptable and the unacceptable. There should be a transition zone, where the condition of risk is tolerable in certain conditions or for some time. The levels of acceptability must permit its variation, for more or for less, allowing the application of the principle of optimization (Slovic, 2000).

On the other hand, the number of indicators should be opened, allowing the inclusion and exclusion of as many indicators as may be necessary. The indicators are classified, according to the level of potential risk they pose to the system.

The risk control indicators should be separated into two categories: critical indicators and non-critical indicators. Critical indicators are those that are associated, directly, to the unacceptable potential risk level. For its severity, they compromise the whole risk control system of the procedures. Therefore, report about critical situations, whose existence, regardless of the existence of any other, take the potential risk to the unacceptable levels.

The set of the non-critical indicators is formed by all the indicators that, individually, do not compromise, in a decisive way, the risk control system. The complete set of the non-critical indicators acts like a critical indicator, ie, if all non-critical indicators are null, the set of indicators will be null and thus, only then, will represent a critical commitment on the potential risks control system.

Once one can build as many risk indicators as needed or desired and the result must be within fixed limits, fundamental to the discussion and establishment of acceptability criteria of the potential risks, it was necessary to develop a mathematical formalism to represent the mean values of the sets of indicators (critical and noncritical) through a single value.

The set of critical indicators is formed by IC the indicators

$$\left\{ \mathbf{C}\_{\mathrm{I}\_{1}} \mathrel{\colon} \mathbf{C}\_{\mathrm{I}\_{2}} \mathrel{\colon} \mathbf{C}\_{\mathrm{I}3} \mathrel{\colon} \dots \textnormal{C}\_{\mathrm{IN}} \right\} \tag{1}$$

Since the critical indicators have the ability to compromise the entire potential risk control of the system, as well as they need to be represented by a mean, the most appropriate way is to represent them as a geometric mean. The geometric mean is the nth root of the product of N terms, representing a mean value of the product. Thus, to represent a mean of N terms, we have:

$$\overline{\mathbf{C}}\_{\mathrm{I}} = \sqrt[N]{\prod\_{i=1}^{N} \mathbf{C}\_{\mathrm{I}i}} \tag{2}$$

So, if any of the indicators has zero value, the value of IC will be zero, independent of the other indicators. On the other hand, the maximum value is, numerically, equal to the

The issue of the values being within the same range of variation allows the comparison and the establishment of limits of acceptability, while the not possibility of assuming the value zero is a condition of the problem, because the risks can be as small as possible, but will

The levels of acceptability should not have a direct border between the acceptable and the unacceptable. There should be a transition zone, where the condition of risk is tolerable in certain conditions or for some time. The levels of acceptability must permit its variation, for more or for less, allowing the application of the principle of optimization (Slovic,

On the other hand, the number of indicators should be opened, allowing the inclusion and exclusion of as many indicators as may be necessary. The indicators are classified, according

The risk control indicators should be separated into two categories: critical indicators and non-critical indicators. Critical indicators are those that are associated, directly, to the unacceptable potential risk level. For its severity, they compromise the whole risk control system of the procedures. Therefore, report about critical situations, whose existence, regardless of the existence of any other, take the potential risk to the

The set of the non-critical indicators is formed by all the indicators that, individually, do not compromise, in a decisive way, the risk control system. The complete set of the non-critical indicators acts like a critical indicator, ie, if all non-critical indicators are null, the set of indicators will be null and thus, only then, will represent a critical commitment on the

Once one can build as many risk indicators as needed or desired and the result must be within fixed limits, fundamental to the discussion and establishment of acceptability criteria of the potential risks, it was necessary to develop a mathematical formalism to represent the mean values of the sets of indicators (critical and noncritical) through a

Since the critical indicators have the ability to compromise the entire potential risk control of the system, as well as they need to be represented by a mean, the most appropriate way is to represent them as a geometric mean. The geometric mean is the nth root of the product of N terms, representing a mean value of the product. Thus, to represent a mean of N terms, we

> N <sup>N</sup> <sup>I</sup> Ii i 1 C C =

So, if any of the indicators has zero value, the value of IC will be zero, independent of the other indicators. On the other hand, the maximum value is, numerically, equal to the

{C ; C ; C ;...;C I I I3 IN 1 2 } (1)

<sup>=</sup> ∏ (2)

to the level of potential risk they pose to the system.

The set of critical indicators is formed by IC the indicators

never be nulls.

unacceptable levels.

single value.

have:

potential risks control system.

2000).

maximum value of an indicator, ie, regardless of the number of indicators that is selected, the result will always be in the same range of variation.

The set of non-critical indicators is formed by the INC indicators.

$$\left\{ \mathbf{NC}\_{\mathrm{I}\_{1}} \mathrel{\colon} \mathrm{NC}\_{\mathrm{I}\_{2}} \mathrel{\colon} \mathrm{NC}\_{\mathrm{I}\_{3}} \mathrel{\colon} \dots \mathrm{NC}\_{\mathrm{N}} \right\} \tag{3}$$

Once the non-critical indicators do not have the ability to, individually, represent the commitment of all the system potential risks control, cannot have its mean represented by a multiplicand. However, they also need to be represented by a mean, so that the representative value of the set is equal, at most, to the maximum value of one of its elements and is within a known range of variation.

Therefore, the best way to represent them is through an arithmetic mean. The non-critical indicators (INC) can be represented by a simple arithmetic average, because it can only be zero, if all control indicators are non-existent.

j 1 I NC NC M *M <sup>j</sup>*<sup>=</sup> = (4)

The function risk control (RC), which represents the result of the indicators of risks control, should be represented as the geometric mean, ie:

$$R\_{\mathbb{C}}\left(\mathbb{C}\_{I}, \mathrm{NC}\_{I}\right) = \sqrt{\mathbb{C} \times \mathrm{NC}} \tag{5}$$

Once more, we used the geometric mean, so that the risk control (RC) is in a range of variation known in advance and that depends only on the variation of IC and INC.

Taking the risk control (RC) as the independent variable, the function that best represent the relationship of cause and effect between risk control and potential risk is the exponential function, with the following form:

$$P\_R(R\_\mathbb{C}) = \mathcal{C}^{-R\_\varepsilon} \tag{6}$$

PR (RC) - Potential risk function, which is dependent on the risk control function, will be referred to as PR; RC - Risk control, function that determines the potential risk and that, on the other hand, is determined by the indicators of risk control.

The shape of the exponential function, with a rapid decrease, represents a good model for critical phenomena, as is the case of the potential risk for health services. The complex relationship between the various factors that influence in the risk control exhibits a kind of not extensive sum, where the potential risk for an event, involving the junction between two factors, can be greater than the sum of the potential risk of the two factors separately.

This type of behavior ends up generating a sudden increase of the potential risk, when adding many elements or some critics, being perfectly represented by the rapid decrease of the exponential function.

Potential Risk: A New Approach 23

The RC function can also be understood as the relationship between the macro and micro indicators of the service. The means IC and INC contain all the information service, so that they behave as if they were the micro systems states, that compose a given health service, determined by the individual indicators IC and INC. Through them, we can know the situation of the equipment, of the human resources or of the procedures, while RC reports a macro value, aggregated, indicating the situation of the total risk control service, but nothing about its components, specifically. Both, RC and IC or ICN, are of fundamental importance for the understanding of the risk control situation, depending on who is looking

As the potential risk (PR) cannot be understood only as a dimensionless number more information are needed to support a decision making. As a way to aggregate the dimension acceptability, the potential risk should be represented within an area of potential risk with

The idea of risks space was first proposed by Slovic et al. (1979), to perform a comparison of the perception of different types of risks and how experts and lay people perceive risks, by using psychometry to quantify the technologies, understood, in the broadest sense, such as

and what you want to analyze.

Fig. 5. Risk acceptance space of the PRAM

equipment, products, processes or practices.

their respective bands of acceptability, as shown in Figure 5.

Another important behavior of the exponential function, to represent the potential risk, is that it has a finite maximum value and the minimum value tends to zero, without necessarily assuming the zero value. The potential risk of a system cannot increase indefinitely, and cannot be zero. Its possibility of occurrence is finite and, for bigger and better that it is the risk control system, you cannot reach a situation of absence of potential risk.

The function proposed in this article, represented by equation (6), allows the potential risk to vary between the maximum value 1 and the minimum value that will be defined by the risk control indicator. The minimum value, will never be zero and, regardless of the number of indicators that it is used, the potential risk function will have fixed maximum and minimum values.

So, an important issue in this model is to establish the range of variation of the risk control indicators, as the maximum scale value defines the minimum value that the potential risk function (PR) can take and, consequently, its range of variation. It is worth noting that the potential risk assessments with this model can only be compared, if they use the same scale of variation of the risk control indicators.

The IC and INC indicators are evaluated, on a scale of zero to five, where zero represents nonexistent or inadequate risk control and five represents risk control excellent, with the following degrees: 0 – absent or inadequate; 1 – poorly; 2 – reasonable; 3 – good; 4 – great and 5 – excellent.

One should consider that the compliance with the rule is associated with the value 3. Thus, regardless of the number of critical and non-critical indicators, the risk control function (RC) will assume values, necessarily, between 0 and 5. Then, the maximum and minimum values of the potential risk (PR) will be:

$$P\_R(\mathbf{R}\_C = 0) = \mathbf{e}^{\star 0} = \mathbf{1},000\tag{7}$$

$$P\_R(\mathbb{R}\_C = \mathbb{S}) = \mathbf{e}^{\circ 5} = 0,00\nabla \tag{8}$$

When RC = 0, which means the absence of the set of non-critical risk controls or the absence of one of the critical risks controls, the potential risk will be PR (0) = 1, ie, there is a full potential risk situation. One can describe the possible potential damage; yet one can not specify a damage and its associated probability of occurrence. On the other hand, for greater that are the controls, the potential risk (PR) will never assume the zero value.

So, one can insert or remove as many risk control indicators as may be necessary, whether they are critical indicators or not, there will be no change in the variation of the function (0.007 ≤ PR ≤ 1.000).

The exponential function proves to be adequate to describe risk control systems, because it reflects well the concept of risks inherent to the technologies, ie, the risk can and should be minimized ever more, but can not be totally eliminated, because it is part of the technology itself. Ie, even if they have implemented all risk control mechanisms, it has a minimum potential risk value (intrinsic), which can not be eliminated, being that the benefits justify the use of this technology for health.

Another important behavior of the exponential function, to represent the potential risk, is that it has a finite maximum value and the minimum value tends to zero, without necessarily assuming the zero value. The potential risk of a system cannot increase indefinitely, and cannot be zero. Its possibility of occurrence is finite and, for bigger and better that it is the risk control system, you cannot reach a situation of absence of potential

The function proposed in this article, represented by equation (6), allows the potential risk to vary between the maximum value 1 and the minimum value that will be defined by the risk control indicator. The minimum value, will never be zero and, regardless of the number of indicators that it is used, the potential risk function will have fixed maximum and

So, an important issue in this model is to establish the range of variation of the risk control indicators, as the maximum scale value defines the minimum value that the potential risk function (PR) can take and, consequently, its range of variation. It is worth noting that the potential risk assessments with this model can only be compared, if they use the same scale

The IC and INC indicators are evaluated, on a scale of zero to five, where zero represents nonexistent or inadequate risk control and five represents risk control excellent, with the following degrees: 0 – absent or inadequate; 1 – poorly; 2 – reasonable; 3 – good; 4 – great

One should consider that the compliance with the rule is associated with the value 3. Thus, regardless of the number of critical and non-critical indicators, the risk control function (RC) will assume values, necessarily, between 0 and 5. Then, the maximum and minimum values

PR(RC=0) = e-0 = 1,000 (7)

 PR(RC=5) = e-5 = 0,007 (8) When RC = 0, which means the absence of the set of non-critical risk controls or the absence of one of the critical risks controls, the potential risk will be PR (0) = 1, ie, there is a full potential risk situation. One can describe the possible potential damage; yet one can not specify a damage and its associated probability of occurrence. On the other hand, for greater

So, one can insert or remove as many risk control indicators as may be necessary, whether they are critical indicators or not, there will be no change in the variation of the function

The exponential function proves to be adequate to describe risk control systems, because it reflects well the concept of risks inherent to the technologies, ie, the risk can and should be minimized ever more, but can not be totally eliminated, because it is part of the technology itself. Ie, even if they have implemented all risk control mechanisms, it has a minimum potential risk value (intrinsic), which can not be eliminated, being that the benefits justify

that are the controls, the potential risk (PR) will never assume the zero value.

risk.

minimum values.

and 5 – excellent.

(0.007 ≤ PR ≤ 1.000).

the use of this technology for health.

of variation of the risk control indicators.

of the potential risk (PR) will be:

The RC function can also be understood as the relationship between the macro and micro indicators of the service. The means IC and INC contain all the information service, so that they behave as if they were the micro systems states, that compose a given health service, determined by the individual indicators IC and INC. Through them, we can know the situation of the equipment, of the human resources or of the procedures, while RC reports a macro value, aggregated, indicating the situation of the total risk control service, but nothing about its components, specifically. Both, RC and IC or ICN, are of fundamental importance for the understanding of the risk control situation, depending on who is looking and what you want to analyze.

As the potential risk (PR) cannot be understood only as a dimensionless number more information are needed to support a decision making. As a way to aggregate the dimension acceptability, the potential risk should be represented within an area of potential risk with their respective bands of acceptability, as shown in Figure 5.

Fig. 5. Risk acceptance space of the PRAM

The idea of risks space was first proposed by Slovic et al. (1979), to perform a comparison of the perception of different types of risks and how experts and lay people perceive risks, by using psychometry to quantify the technologies, understood, in the broadest sense, such as equipment, products, processes or practices.

Potential Risk: A New Approach 25

\_\_\_\_\_\_. O conceito de saúde e a vigilância sanitária: notas para a compreensão

Costa, E. A. Vigilância sanitária: proteção e defesa da saúde. In: Rouquayrol, M.Z.; Almeida

Covello, V. T., Munpower, J. Risk analysis and risk management: an historical perspective*.*

Czeresnia, D. Ciência, Técnica e Cultura: relações entre riscos e práticas de saúde. *Cadernos* 

Fischhoff, B; Bostrum, A; Quadrel, M. J.Risk perception and communication*.* In: Detels,

Hampel, J. Different concepts of risk: a challenge for risk commnication*. International Journal* 

Hood, C.; Rothstein H.; Baldwin, R. *The government of risk: understanding risk regulation* 

IRGC/International Risk Governance Council. *White Paper on Nanotechnology.* Geneva,

Kolluru, R. Risk assessment and management: a unified approach. In: KOLLURU R.et al.

National Research Council (United States of America). *Risk assessment in the government: managing the process*. Washington DC: National Academy Press, 1983. Oberkampf, W. L.et al. Challenge problems: uncertainty in system response given uncertain parameters. *Reliability Engineering and System Safety*, v. 85, p. 11–19, 2004. Omenn, G. S.; Faustman, E. M. Risk assessment and risk manegement*.* In: DETELS, Roger

Shaw IS, Simões MG. *Controle e modelagem fuzzy*. São Paulo: Editora Edgard Blücher;

Lippmann, M.; Cohen, B. S.; Schlesinger, R. B. *Enviromental health science.* Oxford: 2003. Lucchese, G. Globalização e regulação sanitária: os rumos da vigilância sanitária no Brasil.

(Org.). *Risk assessment and management handbook*: for environmental, health and

2001. Tese (Doutorado em Saúde Pública) – Escola Nacional de Saúde Pública,

et al. *Oxford textbook of public health*. 4 th New York: Oxford University Press,

Roger et al. *Oxford Textbook of Public Health.* 4 th. New York: Oxford University

Filho, N. *Epidemiologia & saúde*. São Paulo: Medsi, 2003. p. 357-87. \_\_\_\_\_\_. *Vigilância sanitária: proteção e defesa da saúde.* São Paulo: Sobravime, 2004.

*de Saúde Pública,* São Paulo, *v.* 20, n. 2, p. 447-455, 2004.

Fischhoff, B et al. *Acceptable risk.* Cambridge: Cambridge University Press, 1983.

Gelman, A.; Nolan, D. *Teaching statistic a bag of tricks.* London: Oxford, 2004.

safety professionals. Boston: McGraw Hill, 1996. p. 3-41.

*regimes.* New York: Oxford Univerty Press, 2004.

de um conjunto organizado de práticas de saúde. In: Costa, E. A. (Org.). *Vigilância sanitária: desvendando o enigma*. Salvador: EDUFBA, 2008. p. 19-

Almeida Filho, N. *A ciência da saúde*. São Paulo: Hucitec, 2000.

Beck, U. World risck Society. Cambridge: Polity Press, 2003.

*Risk Analysis,* v. 5, n. 2, 1985.

*of Microbiology*, n. 296, p. 5-10, 2006.

Press, 2005. v.1.

Fiocruz, Rio de Janeiro.

Slovic, P. *The perception of risk.* London: Earthscan, 2000. Triola, M. F. *Introdução à estatística.* Rio de Janeiro: LTC, 2005.

2006.

2005.

1999.

**9. References** 

43.

As there is a possibility of more than one evaluation with the same value of potential risk, causing a point overlap in the spatial representation, you can add a pie chart, so that you can see the number of services / procedures evaluated.

The IRGC "International Risk Governance Council" in the "white paper nº2", of 2006, proposes a bidimensional graphical representation to classify the risk levels of the nanotechnologies, using a non-linear representation, ranges of acceptability and a undefined region between the lower limit of the curve and the X-axis. It is a qualitative representation without estimation of values, which is meant to represent the shape of risks behavior in nanotechnology and its acceptability (IRGC, 2006). The work points to the need for quantitative graphical representation, which seems to have bumped in the difficulty to mathematically formulate the model. This difficulty was surpassed with the presented formulation of potential risk.

#### **7. Conclusion**

The concept of potential risk regards the possibility of occurrence of a health problem, without necessarily describing the injury and its probability of occurrence. It is a concept that expresses the value judgment about potential exposure to a possible risk. It's like representing the risk of the risk.

An important aspect of the concept of potential risk refers to the temporal dimension of causal relationships. While the classical risk has its basis of evaluation in occurred events, the potential risk has its causal bases of evaluation in the events that are occurring and in the effects that may, or may not, occur in the future. Thus, allows working with the temporal dimension of risk facing the future or a meta-reality and not the past.

In the case of the inspections of the health regulatory authorities, the central element in risk management should be the potential risk that, although not representing, necessarily, a defined relation of cause and effect, can be quantified and classified into levels of acceptability, as discussed in the presented model.

However, the potential risk, as the classical risk, can not be represented, only, by a number. It should be understood and evaluated within a context and with limits of acceptability established by the technical and social determinants. Therefore, the evaluations made by regulatory authorities in the process of risk management as indicators have, in most cases, the tools of risk control and as a consequence, a measure of potential risk, which will indicate whether the control conditions are acceptable or not.

#### **8. Acknowledgments**

This publication was financed by the Federal Institute of Education, Science and Technology of Bahia

This study is part of INCT-Citecs funded by the National Intitutes of Science and Technology Programme (MCT-CNPq, Brazil). Contract no. 57386/2008-9.

The authors thank especially Drs. Gunter Drexler and Ediná Alves Costa.

#### **9. References**

24 Public Health – Methodology, Environmental and Systems Issues

As there is a possibility of more than one evaluation with the same value of potential risk, causing a point overlap in the spatial representation, you can add a pie chart, so that you

The IRGC "International Risk Governance Council" in the "white paper nº2", of 2006, proposes a bidimensional graphical representation to classify the risk levels of the nanotechnologies, using a non-linear representation, ranges of acceptability and a undefined region between the lower limit of the curve and the X-axis. It is a qualitative representation without estimation of values, which is meant to represent the shape of risks behavior in nanotechnology and its acceptability (IRGC, 2006). The work points to the need for quantitative graphical representation, which seems to have bumped in the difficulty to mathematically formulate the model. This difficulty was surpassed with the presented

The concept of potential risk regards the possibility of occurrence of a health problem, without necessarily describing the injury and its probability of occurrence. It is a concept that expresses the value judgment about potential exposure to a possible risk. It's like

An important aspect of the concept of potential risk refers to the temporal dimension of causal relationships. While the classical risk has its basis of evaluation in occurred events, the potential risk has its causal bases of evaluation in the events that are occurring and in the effects that may, or may not, occur in the future. Thus, allows working with the temporal

In the case of the inspections of the health regulatory authorities, the central element in risk management should be the potential risk that, although not representing, necessarily, a defined relation of cause and effect, can be quantified and classified into levels of

However, the potential risk, as the classical risk, can not be represented, only, by a number. It should be understood and evaluated within a context and with limits of acceptability established by the technical and social determinants. Therefore, the evaluations made by regulatory authorities in the process of risk management as indicators have, in most cases, the tools of risk control and as a consequence, a measure of potential risk, which will

This publication was financed by the Federal Institute of Education, Science and Technology

This study is part of INCT-Citecs funded by the National Intitutes of Science and

Technology Programme (MCT-CNPq, Brazil). Contract no. 57386/2008-9. The authors thank especially Drs. Gunter Drexler and Ediná Alves Costa.

dimension of risk facing the future or a meta-reality and not the past.

acceptability, as discussed in the presented model.

indicate whether the control conditions are acceptable or not.

can see the number of services / procedures evaluated.

formulation of potential risk.

representing the risk of the risk.

**8. Acknowledgments** 

of Bahia

**7. Conclusion** 

Almeida Filho, N. *A ciência da saúde*. São Paulo: Hucitec, 2000.


**2** 

**Child Mental Health Measurement:** 

**Reflections and Future Directions** 

*1University Medical Center Hamburg-Eppendorf 2Karolinska Institutet Stockholm University* 

*3Dublin City University* 

*1Germany 2Sweden 3Ireland* 

Veronika Ottova1, Anders Hjern2, Carsten-Hendrik Rasche1, Ulrike Ravens-Sieberer1 and the RICHE Project Group1,3

Over the course of the past decades, mental health has enjoyed increased interest, particularly in research on subjective health and well-being. In 2008, the EU has launched the European Pact for Mental Health and Well-being in which European Member States declared mental health as an important health issue and recognized it as their responsibility to undertake action. The Pact for Mental Health and Well-being recognizes youth and education as one of the top priority areas for action and sees prevention and reduction of mental disorders (i.e. mental ill-health) as one of the primary objectives (European

According to the World Health Organization's [WHO] definition, health is not "merely the absence of disease or infirmity", but "a state of complete physical, mental and social wellbeing" (WHO Constitution, 1946). Essential to this definition of health is that it has a positive slant (through the use of the term well-being) and stresses the equal importance of physical, mental and social health. Mental health can further be subdivided into two dimensions: Mental ill-health and positive mental health (Lehtinen et al., 2005). Positive mental health is a resource and is essential to subjective well-being (Lehtinen et al., 2005). Frequently, however, "mental health" is used when actually referring to "positive mental health" and as a consequence is also often (mis)understood as mental health problems or even as mental health diseases/disorders, and not in the positive sense. The persistence of the negative understanding of mental health is largely due to the fact that past and current epidemiological research largely was based on mental health problems and/or illness (Zubrick & Kovess-Masfety, 2005). Many instruments have been developed focusing on mental health problems,

 "[W]ith its awareness of human capital and education, [modern society] puts a new emphasis on children as the resource of the future, low fertility strengthens children's position as a scarce future resource" (Frønes 2007, p. 7). Upon the background of an

thus capturing non-positive outcomes rather than mental health, as such.

**1. Introduction** 

Commission & WHO, 2008).


### **Child Mental Health Measurement: Reflections and Future Directions**

Veronika Ottova1, Anders Hjern2, Carsten-Hendrik Rasche1, Ulrike Ravens-Sieberer1 and the RICHE Project Group1,3 *1University Medical Center Hamburg-Eppendorf 2Karolinska Institutet Stockholm University 3Dublin City University 1Germany 2Sweden 3Ireland* 

#### **1. Introduction**

26 Public Health – Methodology, Environmental and Systems Issues

Vieira Da Silva, L. M. Conceitos, abordagens e estratégias para a avaliação em saúde. In:

Weber L, Klain Pat. Aplicação da lógica *fuzzy* em software e hardware. Canoas: Editora

Edufba; Rio de Janeiro: Fiocruz, 2005, p. 15 – 39.

Zadeh L.A. *Fuzzy* sets. Information and Control 1965; 8:338-353

Ulbra; 2003

Hartz, Zulmira M. de A.; Vieira Da Silva, L. M. (Org.). *Avaliação em saúde: dos modelos teóricos à prática na avaliação de programas e sistemas de saúde*. Salvador:

> Over the course of the past decades, mental health has enjoyed increased interest, particularly in research on subjective health and well-being. In 2008, the EU has launched the European Pact for Mental Health and Well-being in which European Member States declared mental health as an important health issue and recognized it as their responsibility to undertake action. The Pact for Mental Health and Well-being recognizes youth and education as one of the top priority areas for action and sees prevention and reduction of mental disorders (i.e. mental ill-health) as one of the primary objectives (European Commission & WHO, 2008).

> According to the World Health Organization's [WHO] definition, health is not "merely the absence of disease or infirmity", but "a state of complete physical, mental and social wellbeing" (WHO Constitution, 1946). Essential to this definition of health is that it has a positive slant (through the use of the term well-being) and stresses the equal importance of physical, mental and social health. Mental health can further be subdivided into two dimensions: Mental ill-health and positive mental health (Lehtinen et al., 2005). Positive mental health is a resource and is essential to subjective well-being (Lehtinen et al., 2005). Frequently, however, "mental health" is used when actually referring to "positive mental health" and as a consequence is also often (mis)understood as mental health problems or even as mental health diseases/disorders, and not in the positive sense. The persistence of the negative understanding of mental health is largely due to the fact that past and current epidemiological research largely was based on mental health problems and/or illness (Zubrick & Kovess-Masfety, 2005). Many instruments have been developed focusing on mental health problems, thus capturing non-positive outcomes rather than mental health, as such.

> "[W]ith its awareness of human capital and education, [modern society] puts a new emphasis on children as the resource of the future, low fertility strengthens children's position as a scarce future resource" (Frønes 2007, p. 7). Upon the background of an

Child Mental Health Measurement: Reflections and Future Directions 29

Common terms often used in association with mental health are "emotional and behaviour problems", "Mental health problems", "Children's well-being", "Psychological health", "Health-related quality of life", "Behavioural problems", just to name a few (Ravens-Sieberer et al., 2008a). However, several of these are synonyms of "ill-mental health" and not terms that in any way describe positive mental health. According to the two continua model, mental health and mental ill-health (mental illness) are related but distinct dimensions (Westerhof & Keyes, 2010). Mental health is a positive phenomenon (Westerhof

As a matter of fact, the definition of positive mental health has a long history and goes back to the two traditions of well-being (hedonic well-being and eudaimonic well-being). According to Keyes (2002), good mental health consists of three components which are: emotional wellbeing (e.g. feelings of happiness and satisfaction with life), psychological well-being (e.g. positive individual functioning in terms of self-realization), and social well-being (e.g. positive societal functioning in terms of being of social value). He extends previous works of Ryff (1989) (six dimensions of psychological well-being) by adding five elements of social wellbeing which includes "optimal social functioning of inidividuals in terms of their social engagement and societal embeddedness" (Westerhof & Keyes, 2010, p. 111). According to Keyes, positive mental health consists of hedonic well-being and psychological and societal

In the following, when using the term well-being, we use the term subjective well-being which is based on self-reports of happiness and life satisfaction (Schwarz & Strack, 1999).

Today's understanding of mental health and well-being is the result of scientific research and political activities over the past decades. Building upon the original definition of health (WHO, 1948), the definition of mental health is specified as "a state of well-being in which the individual realizes his or her own abilities, can cope with the normal stresses of life, can work productively and fruitfully, and is able to make a contribution to his or her community" (WHO, 2005, p. 2). This contemporary definition contains also the positive view of mental health which is a precondition for well-being. In the eyes of Keyes (2006) the science on mental health and subjective well-being has now arrived at - after half a century starting with the work of Jahoda (1958) on positive mental health - its "third generation" of research which does not merely focus on the absence of illness but "also on the presence of

The concept of well-being first emerged in Greek philosophical writings. Already ancient civilizations considered health and well-being as one of their highest goods and values in life (Sigerist, 1941). However, the scientific interest in well-being did not begin until the 1950s, when the first indicators for quality of life were defined by social scientists to assess social change and to develop social policy (Land, 1975). The theories emerged during the recreation period after World War II where the "individual's perceptions and viewpoints, and the personal meaning and concerns about life" gained relevance in different scientific fields (Keyes, 2006, p. 2). Especially in philosophy (e.g. Phenomenology, cf. Husserl, 1913), sociology (e.g. Symbolic Interactionism, cf. Blumer 1962), and psychology (e.g. cognitive Psychology, cf.

Neisser, 1967), as well as in humanistic theories (cf. Rogers, 1951; Maslow, 1968).

elements of eudaimonic well-being (Keyes, 2005, 2007; Westerhof & Keyes, 2010).

**2.2 Historical development: Positive mental health and well-being** 

subjective well-being" (Keyes, 2006, p. 1).

& Keyes, 2010) and is to be distinguished from ill-mental health.

increasing prevalence of chronic disease and mental health problems – in adults and youth populations alike – research into mental health has become increasingly popular over the past years. The term "new morbidity" has been used to describe the changing morbidity pattern (from acute to chronic disease) and the rise of mental health problems (Palfrey et al., 2005). At the present, disabling mental health problems occur worldwide in 20% of children and adolescents (WHO, 2001). This is an alarming number, especially knowing that mental health problems can have a negative effect on the entire society, with consequences, such as loss of productivity and social functioning (Jané-Llopis & Braddick, 2008). The fact that children and adolescents are affected as well is particularly worrisome. We know for instance that the risk for mental health problems in childhood is higher if there is a lack of resources; in the long run this can have effects even later in life (Jané-Llopis & Braddick, 2008). Many adulthood mental health problems have their roots in childhood (WHO, 2005; Jané-Llopis & Braddick, 2008), and therefore monitoring of mental health in children is a promising strategy, particularly in times of profound societal changes (Mortimer & Larson, 2002). Early detection of problem areas is crucial, and therefore, it is essential that monitoring systems are established based on sound indicators.

It is important to stress that despite the above mentioned negative trends, the overall level of mental well-being in Europe is still high (Jané-Llopis & Braddick, 2008). And thus, it is worthwhile not to limit ourselves to only observing patterns of mental health problems, but to look at the positive side as well, in other words: how is the mental health situation in children and adolescents? How can it be measured adequately in this population group to enable identification (screening) of those with good mental health vs. those who are at risk for poor mental health?

The main objective of this chapter is to give the reader a better understanding and appreciation of child mental health measurement, its current state-of-the-art, and additionally, to generally raise attention to this important field of public health. Drawing upon the authors' expertise and involvement in child and adolescent mental health research, the chapter will briefly go into the history of positive mental health and well-being, including important concepts and definitions of mental health, well-being and indicators. The heart of the chapter will be on selected indicators of (positive and ill-) mental health and subjective well-being. Although surely not comprehensive in all regards, this chapter provides a solid background on this research field and the current state-of-the-art of child mental health measurement. A brief discussion with an outlook will close the chapter.

#### **2. Conceptualization of mental health in children and adolescents**

#### **2.1 Concepts of mental health and well-being**

Coming back to the WHO Definition from the introduction which defines health as "a state of complete physical, mental and social well-being and not merely the absence of disease or infirmity" (WHO, 2001, p. 1), the relationship between health and well-being becomes evident. There are two important ideas that emerge from this: first of all, we see that "mental health is an integral part of health, mental health is more than the absence of mental illness, and mental health is intimately connected with physical health and behaviour" (WHO, 2005, p. 2). From this perspective one can also see that "mental health is the foundation for well-being and effective functioning for an individual and for a community" (WHO, 2005, p. 2).

increasing prevalence of chronic disease and mental health problems – in adults and youth populations alike – research into mental health has become increasingly popular over the past years. The term "new morbidity" has been used to describe the changing morbidity pattern (from acute to chronic disease) and the rise of mental health problems (Palfrey et al., 2005). At the present, disabling mental health problems occur worldwide in 20% of children and adolescents (WHO, 2001). This is an alarming number, especially knowing that mental health problems can have a negative effect on the entire society, with consequences, such as loss of productivity and social functioning (Jané-Llopis & Braddick, 2008). The fact that children and adolescents are affected as well is particularly worrisome. We know for instance that the risk for mental health problems in childhood is higher if there is a lack of resources; in the long run this can have effects even later in life (Jané-Llopis & Braddick, 2008). Many adulthood mental health problems have their roots in childhood (WHO, 2005; Jané-Llopis & Braddick, 2008), and therefore monitoring of mental health in children is a promising strategy, particularly in times of profound societal changes (Mortimer & Larson, 2002). Early detection of problem areas is crucial, and therefore, it is essential that

It is important to stress that despite the above mentioned negative trends, the overall level of mental well-being in Europe is still high (Jané-Llopis & Braddick, 2008). And thus, it is worthwhile not to limit ourselves to only observing patterns of mental health problems, but to look at the positive side as well, in other words: how is the mental health situation in children and adolescents? How can it be measured adequately in this population group to enable identification (screening) of those with good mental health vs. those who are at risk

The main objective of this chapter is to give the reader a better understanding and appreciation of child mental health measurement, its current state-of-the-art, and additionally, to generally raise attention to this important field of public health. Drawing upon the authors' expertise and involvement in child and adolescent mental health research, the chapter will briefly go into the history of positive mental health and well-being, including important concepts and definitions of mental health, well-being and indicators. The heart of the chapter will be on selected indicators of (positive and ill-) mental health and subjective well-being. Although surely not comprehensive in all regards, this chapter provides a solid background on this research field and the current state-of-the-art of child mental health measurement. A brief discussion with an outlook will close the chapter.

Coming back to the WHO Definition from the introduction which defines health as "a state of complete physical, mental and social well-being and not merely the absence of disease or infirmity" (WHO, 2001, p. 1), the relationship between health and well-being becomes evident. There are two important ideas that emerge from this: first of all, we see that "mental health is an integral part of health, mental health is more than the absence of mental illness, and mental health is intimately connected with physical health and behaviour" (WHO, 2005, p. 2). From this perspective one can also see that "mental health is the foundation for well-being and

monitoring systems are established based on sound indicators.

**2. Conceptualization of mental health in children and adolescents** 

effective functioning for an individual and for a community" (WHO, 2005, p. 2).

**2.1 Concepts of mental health and well-being** 

for poor mental health?

Common terms often used in association with mental health are "emotional and behaviour problems", "Mental health problems", "Children's well-being", "Psychological health", "Health-related quality of life", "Behavioural problems", just to name a few (Ravens-Sieberer et al., 2008a). However, several of these are synonyms of "ill-mental health" and not terms that in any way describe positive mental health. According to the two continua model, mental health and mental ill-health (mental illness) are related but distinct dimensions (Westerhof & Keyes, 2010). Mental health is a positive phenomenon (Westerhof & Keyes, 2010) and is to be distinguished from ill-mental health.

As a matter of fact, the definition of positive mental health has a long history and goes back to the two traditions of well-being (hedonic well-being and eudaimonic well-being). According to Keyes (2002), good mental health consists of three components which are: emotional wellbeing (e.g. feelings of happiness and satisfaction with life), psychological well-being (e.g. positive individual functioning in terms of self-realization), and social well-being (e.g. positive societal functioning in terms of being of social value). He extends previous works of Ryff (1989) (six dimensions of psychological well-being) by adding five elements of social wellbeing which includes "optimal social functioning of inidividuals in terms of their social engagement and societal embeddedness" (Westerhof & Keyes, 2010, p. 111). According to Keyes, positive mental health consists of hedonic well-being and psychological and societal elements of eudaimonic well-being (Keyes, 2005, 2007; Westerhof & Keyes, 2010).

In the following, when using the term well-being, we use the term subjective well-being which is based on self-reports of happiness and life satisfaction (Schwarz & Strack, 1999).

#### **2.2 Historical development: Positive mental health and well-being**

Today's understanding of mental health and well-being is the result of scientific research and political activities over the past decades. Building upon the original definition of health (WHO, 1948), the definition of mental health is specified as "a state of well-being in which the individual realizes his or her own abilities, can cope with the normal stresses of life, can work productively and fruitfully, and is able to make a contribution to his or her community" (WHO, 2005, p. 2). This contemporary definition contains also the positive view of mental health which is a precondition for well-being. In the eyes of Keyes (2006) the science on mental health and subjective well-being has now arrived at - after half a century starting with the work of Jahoda (1958) on positive mental health - its "third generation" of research which does not merely focus on the absence of illness but "also on the presence of subjective well-being" (Keyes, 2006, p. 1).

The concept of well-being first emerged in Greek philosophical writings. Already ancient civilizations considered health and well-being as one of their highest goods and values in life (Sigerist, 1941). However, the scientific interest in well-being did not begin until the 1950s, when the first indicators for quality of life were defined by social scientists to assess social change and to develop social policy (Land, 1975). The theories emerged during the recreation period after World War II where the "individual's perceptions and viewpoints, and the personal meaning and concerns about life" gained relevance in different scientific fields (Keyes, 2006, p. 2). Especially in philosophy (e.g. Phenomenology, cf. Husserl, 1913), sociology (e.g. Symbolic Interactionism, cf. Blumer 1962), and psychology (e.g. cognitive Psychology, cf. Neisser, 1967), as well as in humanistic theories (cf. Rogers, 1951; Maslow, 1968).

Child Mental Health Measurement: Reflections and Future Directions 31

dynamic processes interacting with the environment (Ben-Arieh, 2008). These theoretical changes gave rise to new methodological perspectives. To better capture children's own living conditions, especially on terms like mental well-being or peer-relations, subjective

Efforts to synthesize data into national and international "state of the child" reports began during the 1970s. At the international level, UNICEF published in 1979 the State of the World's Children Report (United Nations International Children's Emergency Fund [UNICEF], 1979) – at that time only basic survival indicators were included (e.g. infant and child mortality). In the 1990s, significant developments were made in the reporting. For instance, the Census Bureau published the first comparable report at an international level including domains on family structure, economic status, health and education (Lippman, 2007). Of particular note is the work of an international group of child health experts on a project called "Measuring and Monitoring Child Well-being: Beyond Survival" (Ben-Arieh & Wintersburger, 1997) with the intention to create international indicators which measure quality of life from a child's perspective, including indicators which go beyond the traditionally used survival indicators, such as social connectedness, civic and personal life skills and children's subculture (Lippman, 2007). Until now, a varying set of indicators exist to measure different aspects of child wellbeing. Crucial factors for the development of mental health indicators were obtained in the child indicators movement: Indicators for negative or risk factor were complemented by indicators for protective factors. Also, the indicators shifted from "well-becoming" (e.g. preparing children to be productive and happy adults) to "well-being" (Ben-Arieh, 2008).

As this short historical overview shows, there have been groundbreaking developments on the understanding and conceptualization of mental well-being. In the next two sections we will take a closer look at mental health measurement in children and adolescents. We will begin by briefly highlighting the importance of indicators in health monitoring while also

The term indicator originates from the Latin word "indicator" which means "one who points out" or "indico" (=to point out). Indicators can cover anything from "indices, signs, and symptoms" to "calculated probabilities and systematic measurements" (Frønes 2007, p. 8), and include time and space. Bauer (1966) has referred to (social) indicators as "statistics, statistical series, and all other forms of evidence […] that enable us to assess where we stand and are going with respect to our values and goals" (p. 1). For policy makers, indicators provide valuable information on relevant public health issues, including their trend and direction of change (improvement or worsening) (Lippman, 2007). But also other groups, such as child advocacy groups, researchers, and media use them for various purposes (Ben-Arieh, 2008).

A good example of a widely-known and politically very influential programme is the OECD Programme for International Student Assessment [PISA]. PISA assesses to what extend knowledge and skills essential for participation in society have been acquired by 15-yearolds at the end of their compulsory education (www.pisa.oecd.org/). The PISA indicators of educational success and marginalization are "perhaps the most well-known example of highly elaborated comparative research indicators related to children" (Frønes, 2007, p. 7). The first PISA report in 2000 had a substantial national and international impact and PISA assessment continues to be an important strategy to benchmark improvements in education

pointing out the conceptual and methodological challenges.

**3. Indicators as tools for health monitoring** 

at international level.

reports and child-centred indicators became necessary (Ben-Arieh, 2008).

Sponsored by the Joint Commission on Mental Illness and Health in the United States, Marie Jahoda (1958) and Gurin et al. (1960) published two seminal reviews which Keyes categorize as the first research generation of subjective well-being (Keyes, 2006). Jahoda's review 'Current Concepts of Positive Mental Health' (1958) can be seen as pioneer work which shaped our current understanding and theories of positive mental health. Her work was later continued by other scientists like Carol Ryff (1989) who operationalized her theories on well-being (Keyes, 2006). In the first part of this important volume, Jahoda outlines the former understanding of mental health and emphasizes "that the absence of disease may constitute a necessary, but not a sufficient, criterion for mental health" (Jahoda, 1958, p. 15). Further, she investigates throughout a literature research six partly overlapping approaches to categorize positive mental health. They can be summarized as "Attitude of an individual towards his own self", "Self-actualization", "Integration", "Autonomy", "Perception of reality", "Environmental mastery". Yet, there is some criticism especially on the role of cultural influences affecting the understanding of mental health. Scientists, e.g. Murphy (1978), argued that western cultures are predominated by individualism and so other cultures which have a strong collectivistic viewpoint could have a different understanding of mental health. Therefore, cultural values have a strong influence on concepts of mental health (WHO, 2005). The second influential volume is an interview survey on approximately twenty-five hundred Americans conducted by Gurin, Veroff, and Feld (1960) covering the subjective dimension of mental health. Additionally, the volume "featured the hedonic stream of subjective well-being" (Keyes, 2006, p. 3) which together with the eudaimonic stream became more important and dominant in the "second generation" of research (Keyes, 2006). Hedonic well-being can be seen as a part of subjective well-being focusing predominantly on happiness and interest as well as satisfaction with life (Keyes, 2007). More generally contemplated is the existence of positive and the absence of negative affect (Deci & Ryan, 2008) and matches up with our everyday understanding of the word happiness (Waterman, 1993). In contrast, eudaimonia is a feeling of personal expressiveness, selfrealization and life satisfaction (Waterman, 1993; Deci & Ryan, 2008). The latter tradition has lost importance in recent well-being research, but contributes important aspects to the concept of well-being (Deci & Ryan, 2008). Three articles stimulated and leveraged the research during this time: Schwartz & Clore (1983) studied how current mood states can affect judgments of happiness and satisfaction with life; Diener (1984) reviewed the first generation of subjective well-being with a focus solely on the hedonic streaming; Ryff (1989) operationalized different aspects of well-being. While the well-being theories became more and more elaborated, various multidimensional scales were developed to measure different aspects of the concept. Early scales for adults included, e.g. the Bradburn Schedule (1969) and the General Well-Being Scale (GWBS) (1969). The first signs of child and adolescent well-being measurement can be found in the "social indicators movement" in the 1960s. Seminal work published by Campbell and Converse (1972) deals with the development of subjective indicators of the quality of life (e.g. aspiration, expectations, and life satisfaction) and Sheldon and Moore's (1968) volume "Indicators of Social Change" conceptualized "objective measures, reviewing available data, and recommending data needs that would enable descriptive reporting on the status of society across domains" (Lippman, 2007, p. 40; Aborn, 1985). In later years, theoretical, normative and methodical changes in science spurred and formed the development of child indicators. Of particular relevance were the United Nation's Convention on the Rights of the Child (CRC) which raised a normative framework for an integral view on children; the "new" sociology of childhood considered it as an independent stage in and of his self and child development theories became more

Sponsored by the Joint Commission on Mental Illness and Health in the United States, Marie Jahoda (1958) and Gurin et al. (1960) published two seminal reviews which Keyes categorize as the first research generation of subjective well-being (Keyes, 2006). Jahoda's review 'Current Concepts of Positive Mental Health' (1958) can be seen as pioneer work which shaped our current understanding and theories of positive mental health. Her work was later continued by other scientists like Carol Ryff (1989) who operationalized her theories on well-being (Keyes, 2006). In the first part of this important volume, Jahoda outlines the former understanding of mental health and emphasizes "that the absence of disease may constitute a necessary, but not a sufficient, criterion for mental health" (Jahoda, 1958, p. 15). Further, she investigates throughout a literature research six partly overlapping approaches to categorize positive mental health. They can be summarized as "Attitude of an individual towards his own self", "Self-actualization", "Integration", "Autonomy", "Perception of reality", "Environmental mastery". Yet, there is some criticism especially on the role of cultural influences affecting the understanding of mental health. Scientists, e.g. Murphy (1978), argued that western cultures are predominated by individualism and so other cultures which have a strong collectivistic viewpoint could have a different understanding of mental health. Therefore, cultural values have a strong influence on concepts of mental health (WHO, 2005). The second influential volume is an interview survey on approximately twenty-five hundred Americans conducted by Gurin, Veroff, and Feld (1960) covering the subjective dimension of mental health. Additionally, the volume "featured the hedonic stream of subjective well-being" (Keyes, 2006, p. 3) which together with the eudaimonic stream became more important and dominant in the "second generation" of research (Keyes, 2006). Hedonic well-being can be seen as a part of subjective well-being focusing predominantly on happiness and interest as well as satisfaction with life (Keyes, 2007). More generally contemplated is the existence of positive and the absence of negative affect (Deci & Ryan, 2008) and matches up with our everyday understanding of the word happiness (Waterman, 1993). In contrast, eudaimonia is a feeling of personal expressiveness, selfrealization and life satisfaction (Waterman, 1993; Deci & Ryan, 2008). The latter tradition has lost importance in recent well-being research, but contributes important aspects to the concept of well-being (Deci & Ryan, 2008). Three articles stimulated and leveraged the research during this time: Schwartz & Clore (1983) studied how current mood states can affect judgments of happiness and satisfaction with life; Diener (1984) reviewed the first generation of subjective well-being with a focus solely on the hedonic streaming; Ryff (1989) operationalized different aspects of well-being. While the well-being theories became more and more elaborated, various multidimensional scales were developed to measure different aspects of the concept. Early scales for adults included, e.g. the Bradburn Schedule (1969) and the General Well-Being Scale (GWBS) (1969). The first signs of child and adolescent well-being measurement can be found in the "social indicators movement" in the 1960s. Seminal work published by Campbell and Converse (1972) deals with the development of subjective indicators of the quality of life (e.g. aspiration, expectations, and life satisfaction) and Sheldon and Moore's (1968) volume "Indicators of Social Change" conceptualized "objective measures, reviewing available data, and recommending data needs that would enable descriptive reporting on the status of society across domains" (Lippman, 2007, p. 40; Aborn, 1985). In later years, theoretical, normative and methodical changes in science spurred and formed the development of child indicators. Of particular relevance were the United Nation's Convention on the Rights of the Child (CRC) which raised a normative framework for an integral view on children; the "new" sociology of childhood considered it as an independent stage in and of his self and child development theories became more dynamic processes interacting with the environment (Ben-Arieh, 2008). These theoretical changes gave rise to new methodological perspectives. To better capture children's own living conditions, especially on terms like mental well-being or peer-relations, subjective reports and child-centred indicators became necessary (Ben-Arieh, 2008).

Efforts to synthesize data into national and international "state of the child" reports began during the 1970s. At the international level, UNICEF published in 1979 the State of the World's Children Report (United Nations International Children's Emergency Fund [UNICEF], 1979) – at that time only basic survival indicators were included (e.g. infant and child mortality). In the 1990s, significant developments were made in the reporting. For instance, the Census Bureau published the first comparable report at an international level including domains on family structure, economic status, health and education (Lippman, 2007). Of particular note is the work of an international group of child health experts on a project called "Measuring and Monitoring Child Well-being: Beyond Survival" (Ben-Arieh & Wintersburger, 1997) with the intention to create international indicators which measure quality of life from a child's perspective, including indicators which go beyond the traditionally used survival indicators, such as social connectedness, civic and personal life skills and children's subculture (Lippman, 2007). Until now, a varying set of indicators exist to measure different aspects of child wellbeing. Crucial factors for the development of mental health indicators were obtained in the child indicators movement: Indicators for negative or risk factor were complemented by indicators for protective factors. Also, the indicators shifted from "well-becoming" (e.g. preparing children to be productive and happy adults) to "well-being" (Ben-Arieh, 2008).

As this short historical overview shows, there have been groundbreaking developments on the understanding and conceptualization of mental well-being. In the next two sections we will take a closer look at mental health measurement in children and adolescents. We will begin by briefly highlighting the importance of indicators in health monitoring while also pointing out the conceptual and methodological challenges.

#### **3. Indicators as tools for health monitoring**

The term indicator originates from the Latin word "indicator" which means "one who points out" or "indico" (=to point out). Indicators can cover anything from "indices, signs, and symptoms" to "calculated probabilities and systematic measurements" (Frønes 2007, p. 8), and include time and space. Bauer (1966) has referred to (social) indicators as "statistics, statistical series, and all other forms of evidence […] that enable us to assess where we stand and are going with respect to our values and goals" (p. 1). For policy makers, indicators provide valuable information on relevant public health issues, including their trend and direction of change (improvement or worsening) (Lippman, 2007). But also other groups, such as child advocacy groups, researchers, and media use them for various purposes (Ben-Arieh, 2008).

A good example of a widely-known and politically very influential programme is the OECD Programme for International Student Assessment [PISA]. PISA assesses to what extend knowledge and skills essential for participation in society have been acquired by 15-yearolds at the end of their compulsory education (www.pisa.oecd.org/). The PISA indicators of educational success and marginalization are "perhaps the most well-known example of highly elaborated comparative research indicators related to children" (Frønes, 2007, p. 7). The first PISA report in 2000 had a substantial national and international impact and PISA assessment continues to be an important strategy to benchmark improvements in education at international level.

Child Mental Health Measurement: Reflections and Future Directions 33

perspective of the child itself. Secondly, and this is perhaps even more important, the indicators are based on robust, scientifically valid measurement tools. Apart from having been frequently used in research studies in Europe (as well as internationally), the measures

When child well-being is of interest, the preferred method of assessment is via the child's own subjective perspective (Lippman et al., 2009). How children and adolescents reflect and perceive their world and life may differ quite substantially today from adult's reality (Bradshaw et al. 2006). Increasing their participation and asking for their insight and view is an indispensable component of present and future research. "Current attempts to measure children's well-being are problematic because they fail to incorporate an analysis of broader contextual structural and political factors" (Morrow & Mayall, 2010, p. 162). Subjective indicators reflecting the "voice of the child" need to be complemented by objective models on well-being and indicators (Frønes, 2007, p. 11). Furthermore, indicators should be based on measurement tools, which have undergone extensive piloting and ideally have been used previously in surveys. Measurement tools need to be age-, gender-, and culturally-sensitive and should also take the individual's

Many of the indicators which will be presented here originate from the KIDSCREEN survey ["Screening for and Promotion of Health-Related Quality of Life in Children and Adolescents - A European Public Health Perspective"] and the Health Behaviour in School-

The European KIDSCREEN project titled "Screening for and promotion of health-related well-being in children and adolescents: a European public health perspective (KIDSCREEN)" took place between 2001 and 2004 in 13 European countries (Austria, the Czech Republic, France, Germany, Greece, Hungary, Ireland, Poland, Spain, Sweden, Switzerland, the Netherlands and the United Kingdom) and had the aim to develop a standardised screening instrument for quality of life in children and adolescents. This instrument should be suitable for representative national and European health surveys and enable cross-cultural comparisons. The project, which also comprised data collection from large population-based samples in each of the participating countries, was part of the Quality of life and Management of Living Resources programme and was funded by the European Commission (EC) within the Fifth Framework Programme (EC Grant Number: QLG-CT-2000- 00751) (Ravens-Sieberer et al., 2001). The data collection targeted children between 8 and 18 years of age using both parents as well as children as information sources. The same kind of data collection tools (questionnaires) and the same assessment tools were used in all participating countries (Ravens-Sieberer et al., 2005). Data on physical health, mental health and socioeconomic status in children and adolescents in Europe was collected and the distribution of mental ill health and poor mental well-being estimated. Thee important instruments came out of the KIDSCREEN project: KIDSCREEN-52, KIDSCREEN-27 and KIDSCREEN-10. Single dimensions of these instruments and the global HRQoL score (KIDSCREEN-10) can be used as suitable indicators for quality of life resp. positive mental health. The KIDSCREEN-10 Mental Health Index assesses the child's perspective on his or her physical, mental and social well-being, identifies children at risk and suggests suitable early interventions. For this

also fulfil the scientific criteria for indicators (as proposed by the ECHI group).

socioeconomic background into account (Erhart et al., 2006).

aged Children [HBSC] Survey.

**4.1.1 The KIDSCREEN survey** 

The European Community Health Indicators Project (ECHI) is a similar effort, but has a different focus. Its aim is to lay the foundation for further development of health indicators targeting all population groups, not just school children. The initial projects on European Community Health Indicators (ECHI and ECHI-2) which were conducted between 1998 and 2005 developed ECHI indicator lists which formed the basis for the follow-up work in the ECHIM project. The ECHIM project is part of the European Health Strategy and builds upon the works of ECHI and ECHI-2. It has three main objectives (Kilpeläinen, Aromaa & the ECHIM project, 2008):


Within ECHI, an indicator was defined as a characteristic of an individual, population or environment which is subject to measurement (directly or indirectly) and can be used to describe one or more aspects of the health of an individual or population (quantity, quality and time). According to ECHI recommendations, indicators must fulfil the criteria of validity, sensitivity, comparability (Kramers, 2003).

Despite advances in indicator development through projects such as ECHI, the development of positive mental health indicators for children and young people is really only beginning (Maher & Waters, 2005). While we seek to gain a better understanding of the magnitude of mental health problems in children, we seem to oversee the importance of measurement tools and indicators to facilitate this process. Monitoring of both positive mental health and mental ill-health (i.e. mental health problems) is essential for human development (Zubrick & Kovess-Masfety, 2005). Unfortunately, mental health research in children and adolescents currently lacks well-established indicators. It is primarily "needs driven", focusing on "illness" rather than "wellness", and in consequence, aimed at physical rather than mental health (Zubrick & Kovess-Masfety, 2005). Furthermore, it is too focussed on distress, and mental health problems, such as delinquency, suicide, depression (Maher & Waters, 2005), rather than positive mental health.

Presently, existing indicators on health are available through organisations, such as the European Union [EU], the Organisation for Economic Co-Operation and Development [OECD], and the World Health Organisation [WHO]. The EU sustainable development indicators provide 120 indicators, the OECD social indicators have 34 indicators on employment, society, general health and social cohesion, and the EU social protection indicators comprise 11 primary social protection indicators (whereby none on mental health). In 2009, the Innocenti Research Centre of the UNICEF has published a working paper on "Positive indicators of child well-being: a conceptual framework, measures and methodological issues" outlining frameworks for further development of positive indicators of well-being of children as well as the challenges involved (Lippman et al., 2009).

#### **4. Child mental health measurement**

#### **4.1 Indicators of mental health**

As mentioned in the introduction of this chapter, we have limited ourselves to a narrow selection of indicators which we consider suitable for several reasons. First of all, all of the indicators presented here are based on tools/instruments assessing the subjective perspective of the child itself. Secondly, and this is perhaps even more important, the indicators are based on robust, scientifically valid measurement tools. Apart from having been frequently used in research studies in Europe (as well as internationally), the measures also fulfil the scientific criteria for indicators (as proposed by the ECHI group).

When child well-being is of interest, the preferred method of assessment is via the child's own subjective perspective (Lippman et al., 2009). How children and adolescents reflect and perceive their world and life may differ quite substantially today from adult's reality (Bradshaw et al. 2006). Increasing their participation and asking for their insight and view is an indispensable component of present and future research. "Current attempts to measure children's well-being are problematic because they fail to incorporate an analysis of broader contextual structural and political factors" (Morrow & Mayall, 2010, p. 162). Subjective indicators reflecting the "voice of the child" need to be complemented by objective models on well-being and indicators (Frønes, 2007, p. 11). Furthermore, indicators should be based on measurement tools, which have undergone extensive piloting and ideally have been used previously in surveys. Measurement tools need to be age-, gender-, and culturally-sensitive and should also take the individual's socioeconomic background into account (Erhart et al., 2006).

Many of the indicators which will be presented here originate from the KIDSCREEN survey ["Screening for and Promotion of Health-Related Quality of Life in Children and Adolescents - A European Public Health Perspective"] and the Health Behaviour in Schoolaged Children [HBSC] Survey.

#### **4.1.1 The KIDSCREEN survey**

32 Public Health – Methodology, Environmental and Systems Issues

The European Community Health Indicators Project (ECHI) is a similar effort, but has a different focus. Its aim is to lay the foundation for further development of health indicators targeting all population groups, not just school children. The initial projects on European Community Health Indicators (ECHI and ECHI-2) which were conducted between 1998 and 2005 developed ECHI indicator lists which formed the basis for the follow-up work in the ECHIM project. The ECHIM project is part of the European Health Strategy and builds upon the works of ECHI and ECHI-2. It has three main objectives (Kilpeläinen, Aromaa & the

Within ECHI, an indicator was defined as a characteristic of an individual, population or environment which is subject to measurement (directly or indirectly) and can be used to describe one or more aspects of the health of an individual or population (quantity, quality and time). According to ECHI recommendations, indicators must fulfil the criteria of

Despite advances in indicator development through projects such as ECHI, the development of positive mental health indicators for children and young people is really only beginning (Maher & Waters, 2005). While we seek to gain a better understanding of the magnitude of mental health problems in children, we seem to oversee the importance of measurement tools and indicators to facilitate this process. Monitoring of both positive mental health and mental ill-health (i.e. mental health problems) is essential for human development (Zubrick & Kovess-Masfety, 2005). Unfortunately, mental health research in children and adolescents currently lacks well-established indicators. It is primarily "needs driven", focusing on "illness" rather than "wellness", and in consequence, aimed at physical rather than mental health (Zubrick & Kovess-Masfety, 2005). Furthermore, it is too focussed on distress, and mental health problems, such as delinquency, suicide, depression (Maher & Waters, 2005),

Presently, existing indicators on health are available through organisations, such as the European Union [EU], the Organisation for Economic Co-Operation and Development [OECD], and the World Health Organisation [WHO]. The EU sustainable development indicators provide 120 indicators, the OECD social indicators have 34 indicators on employment, society, general health and social cohesion, and the EU social protection indicators comprise 11 primary social protection indicators (whereby none on mental health). In 2009, the Innocenti Research Centre of the UNICEF has published a working paper on "Positive indicators of child well-being: a conceptual framework, measures and methodological issues" outlining frameworks for further development of positive indicators

As mentioned in the introduction of this chapter, we have limited ourselves to a narrow selection of indicators which we consider suitable for several reasons. First of all, all of the indicators presented here are based on tools/instruments assessing the subjective

of well-being of children as well as the challenges involved (Lippman et al., 2009).

• to further develop health indicators (based on ECHI short list),

• to enable the establishment of a Health Monitoring System.

• to initiate implementation in the EU countries, and

validity, sensitivity, comparability (Kramers, 2003).

rather than positive mental health.

**4. Child mental health measurement** 

**4.1 Indicators of mental health** 

ECHIM project, 2008):

The European KIDSCREEN project titled "Screening for and promotion of health-related well-being in children and adolescents: a European public health perspective (KIDSCREEN)" took place between 2001 and 2004 in 13 European countries (Austria, the Czech Republic, France, Germany, Greece, Hungary, Ireland, Poland, Spain, Sweden, Switzerland, the Netherlands and the United Kingdom) and had the aim to develop a standardised screening instrument for quality of life in children and adolescents. This instrument should be suitable for representative national and European health surveys and enable cross-cultural comparisons. The project, which also comprised data collection from large population-based samples in each of the participating countries, was part of the Quality of life and Management of Living Resources programme and was funded by the European Commission (EC) within the Fifth Framework Programme (EC Grant Number: QLG-CT-2000- 00751) (Ravens-Sieberer et al., 2001). The data collection targeted children between 8 and 18 years of age using both parents as well as children as information sources. The same kind of data collection tools (questionnaires) and the same assessment tools were used in all participating countries (Ravens-Sieberer et al., 2005). Data on physical health, mental health and socioeconomic status in children and adolescents in Europe was collected and the distribution of mental ill health and poor mental well-being estimated. Thee important instruments came out of the KIDSCREEN project: KIDSCREEN-52, KIDSCREEN-27 and KIDSCREEN-10. Single dimensions of these instruments and the global HRQoL score (KIDSCREEN-10) can be used as suitable indicators for quality of life resp. positive mental health. The KIDSCREEN-10 Mental Health Index assesses the child's perspective on his or her physical, mental and social well-being, identifies children at risk and suggests suitable early interventions. For this

Child Mental Health Measurement: Reflections and Future Directions 35

Children at risk of poor quality of life are identified by coding of responses so that higher values indicate better quality of life. The KIDSCREEN-10 Mental Health Index was developed by means of a Rasch analysis which ensured that only those items which represented a global, unidimensional latent trait were included. The values on the individual items are summed up, Rasch person parameters (PP) are assigned to each possible sum score, and then the PP are transformed into values with a mean of 50 and standard deviation of approximately 10 (Ravens-Sieberer et al., 2006). A better differentiation between the children is made possible by the distribution of the Rasch scores that resemble the expected theoretical normal distribution. The index provides a good

Validation work on this instrument indicates that it is a valid and well-tested stable child centred self-report measure (indicator) for child and adolescent general quality of life and mental well-being status. It has good psychometric properties, with high reliability and Rasch-scale properties. The index provides a good discriminatory power and shows only few ceiling or floor effects. The strong internal consistency reliability (Cronbach's Alpha = .82) and test-retest reliability (r = .73) allow precise and stable measurements (Ravens-

The cut-off at T-value below 38 (which represents the lowest 10%) indicates lower quality of life resp. higher risk for poor mental health (Ravens-Sieberer et al., 2006). Comparisons with the European Community Health Indicators (ECHI) show that the Kidscreen-10 Index for children and adolescents corresponds well with the "General Quality of Life Indicator". The ECHI group proposes to use the Euroqol score from the Euroqol 5D instrument (Eurociss project) or

Since its development, the instrument has been employed in several EU funded European research projects (KIDSCREEN, DISABKIDS, MHADIE, SPARCLE), in the Flash Eurobarometer and in the PROMIS roadmap initiative of the US NIH (National Institutes of Health) to develop a patient reported outcome measurement information system. The instrument is also used in the Health Behaviour in School-aged Children (HBSC) study as an indicator of positive mental health and has been translated into a variety of languages.

Another positive mental health indicator is psychological well-being which refers to a child's or adolescents' positive emotions and perceptions, his/her satisfaction with life, covering various areas of his/her inner feelings and thus provides insight into an individual's mental health state. The psychological well-being dimension is one of ten dimensions of KIDSCREEN-52 and one of the five dimensions of KIDSCREEN-27 as shown in the figure below. In the latter, it also encompasses the Moods and Emotions and the Self-

alternatively the WHOQOL of the WHO (Kramers & the ECHI team, 2005) for adults.

• "Have you been able to do the things that you want to do in your free time?"

discriminatory power and shows only few ceiling or floor effects.

Sieberer et al., 2006; Ravens-Sieberer et al., 2010).

**4.2.2 Psychological well-being indicator** 

Perception scale of KIDSCREEN-52.

• "Have you had enough time for yourself?"

• "Have your parent(s) treated you fairly?" • "Have you had fun with your friends?" • "Have you got on well at school?" • "Have you been able to pay attention?"

reason it is particularly useful for identifying children with positive mental health. Section 4.4 of this chapter will present empirical results from the KIDSCREEN survey detailing the distribution of children with positive mental health in thirteen European countries.

Further information on the KIDSCREEN instruments is available at http://www.kidscreen.org.

#### **4.1.2 The Health Behaviour in School-aged Children (HBSC) survey**

The HBSC Study is a WHO-collaborative study dedicated to the study of adolescent health. The overall aim is to gain a better understanding of health behaviours, health, and wellbeing in children and adolescents at the age of 11, 13 and 15 years. HBSC is a cross-national study covering over 40 countries in Europe, North America and Israel. The design of the survey is cross-sectional and data collection is carried out every four years. The basis for each survey is a standardized research protocol which is renewed for each survey round. The survey is based on a questionnaire which consists of mandatory items (required from each country), and optional items which focus on topics of national interest. Mandatory items are part of the international file and enable cross-country comparisons. Data is collected in schools and the primary sampling unit is school class (or entire school in case this is not possible). Data is collected within a class period via questionnaire.

Further information on the HBSC Survey is available at: http://www.hbsc.org .

Collection of data on positive mental health is in line with the health definition of the WHO (Ravens-Sieberer et al., 2008a). Assessment of mental health is possible in one of two ways: positive mental health and negative (ill) mental health, and the HBSC and KIDSCREEN Surveys provide suitable instruments for both.

#### **4.2 Positive mental health indicators**

#### **4.2.1 Quality of life and positive mental health indicator**

One of the outputs of the European KIDSCREEN survey was the development of a screening tool for mental health. The KIDSCREEN-10 instrument is an index which assesses the child's perspective on his or her physical, mental and social well-being, thus enabling the identification of children at risk. As previously stated, the KIDSCREEN-10 Mental Health Index is a non-clinical measure of quality of life and positive mental health status and enables the assessment of school-aged children's general well-being. The index is especially sensitive for affective, cognitive, and psychovegetative, as well as psychosocial aspects of mental health.

The short instrument consists of ten items covering six aspects of quality of life (physical well-being, moods & emotions, autonomy, parent relation & home life, peers & social support, school environment). The short instrument consists of the following ten items and only takes a few minutes to complete:


reason it is particularly useful for identifying children with positive mental health. Section 4.4 of this chapter will present empirical results from the KIDSCREEN survey detailing the distribution of children with positive mental health in thirteen European countries.

The HBSC Study is a WHO-collaborative study dedicated to the study of adolescent health. The overall aim is to gain a better understanding of health behaviours, health, and wellbeing in children and adolescents at the age of 11, 13 and 15 years. HBSC is a cross-national study covering over 40 countries in Europe, North America and Israel. The design of the survey is cross-sectional and data collection is carried out every four years. The basis for each survey is a standardized research protocol which is renewed for each survey round. The survey is based on a questionnaire which consists of mandatory items (required from each country), and optional items which focus on topics of national interest. Mandatory items are part of the international file and enable cross-country comparisons. Data is collected in schools and the primary sampling unit is school class (or entire school in case

Collection of data on positive mental health is in line with the health definition of the WHO (Ravens-Sieberer et al., 2008a). Assessment of mental health is possible in one of two ways: positive mental health and negative (ill) mental health, and the HBSC and KIDSCREEN

One of the outputs of the European KIDSCREEN survey was the development of a screening tool for mental health. The KIDSCREEN-10 instrument is an index which assesses the child's perspective on his or her physical, mental and social well-being, thus enabling the identification of children at risk. As previously stated, the KIDSCREEN-10 Mental Health Index is a non-clinical measure of quality of life and positive mental health status and enables the assessment of school-aged children's general well-being. The index is especially sensitive for affective, cognitive, and psychovegetative, as well as psychosocial

The short instrument consists of ten items covering six aspects of quality of life (physical well-being, moods & emotions, autonomy, parent relation & home life, peers & social support, school environment). The short instrument consists of the following ten items and

Further information on the KIDSCREEN instruments is available at

**4.1.2 The Health Behaviour in School-aged Children (HBSC) survey** 

this is not possible). Data is collected within a class period via questionnaire. Further information on the HBSC Survey is available at: http://www.hbsc.org .

Surveys provide suitable instruments for both.

**4.2.1 Quality of life and positive mental health indicator** 

**4.2 Positive mental health indicators** 

only takes a few minutes to complete:

• "Have you felt fit and well?" • "Have you felt full of energy?"

aspects of mental health.

• "Have you felt sad?" • "Have you felt lonely?"

http://www.kidscreen.org.

• "Have you been able to pay attention?"

Children at risk of poor quality of life are identified by coding of responses so that higher values indicate better quality of life. The KIDSCREEN-10 Mental Health Index was developed by means of a Rasch analysis which ensured that only those items which represented a global, unidimensional latent trait were included. The values on the individual items are summed up, Rasch person parameters (PP) are assigned to each possible sum score, and then the PP are transformed into values with a mean of 50 and standard deviation of approximately 10 (Ravens-Sieberer et al., 2006). A better differentiation between the children is made possible by the distribution of the Rasch scores that resemble the expected theoretical normal distribution. The index provides a good discriminatory power and shows only few ceiling or floor effects.

Validation work on this instrument indicates that it is a valid and well-tested stable child centred self-report measure (indicator) for child and adolescent general quality of life and mental well-being status. It has good psychometric properties, with high reliability and Rasch-scale properties. The index provides a good discriminatory power and shows only few ceiling or floor effects. The strong internal consistency reliability (Cronbach's Alpha = .82) and test-retest reliability (r = .73) allow precise and stable measurements (Ravens-Sieberer et al., 2006; Ravens-Sieberer et al., 2010).

The cut-off at T-value below 38 (which represents the lowest 10%) indicates lower quality of life resp. higher risk for poor mental health (Ravens-Sieberer et al., 2006). Comparisons with the European Community Health Indicators (ECHI) show that the Kidscreen-10 Index for children and adolescents corresponds well with the "General Quality of Life Indicator". The ECHI group proposes to use the Euroqol score from the Euroqol 5D instrument (Eurociss project) or alternatively the WHOQOL of the WHO (Kramers & the ECHI team, 2005) for adults.

Since its development, the instrument has been employed in several EU funded European research projects (KIDSCREEN, DISABKIDS, MHADIE, SPARCLE), in the Flash Eurobarometer and in the PROMIS roadmap initiative of the US NIH (National Institutes of Health) to develop a patient reported outcome measurement information system. The instrument is also used in the Health Behaviour in School-aged Children (HBSC) study as an indicator of positive mental health and has been translated into a variety of languages.

#### **4.2.2 Psychological well-being indicator**

Another positive mental health indicator is psychological well-being which refers to a child's or adolescents' positive emotions and perceptions, his/her satisfaction with life, covering various areas of his/her inner feelings and thus provides insight into an individual's mental health state. The psychological well-being dimension is one of ten dimensions of KIDSCREEN-52 and one of the five dimensions of KIDSCREEN-27 as shown in the figure below. In the latter, it also encompasses the Moods and Emotions and the Self-Perception scale of KIDSCREEN-52.

Child Mental Health Measurement: Reflections and Future Directions 37

defined as the percent of the population below the cut-point of the energy-vitality scale from

The building blocks for good health are laid early in life, and therefore evaluation of health does not begin in adulthood but much earlier. Health is an important resource and poor health early in life can have long-term negative effects which may continue throughout adulthood (WHO, 2006). Being in good health – physically, emotionally and socially – helps young people deal productively with challenges in their development (Burt, 2002). In recent years, self-assessments of health have come more into use as they are based on an individual's perception and evaluation of his or her health. Focusing on the subjective perspective, self-rated health is usually founded on age-peer comparisons either consciously or unconsciously (Bjorner et al., 1996). It can be distinguished from more specific health constructs in that it captures an overall conception of health, rather than a summation across specific domains of health. Empirical studies have shown that self-reported health is an independent predictor of mortality (Idler & Benyamini, 1997). Benjamins et al. (2004) could also identify a relationship between self-reported health and cause-specific mortality, and moreover, also found gender effects for some causes of mortality. A gender effect in selfrated health was confirmed in a sample of children, whereby girls reported poorer health than boys (Cavallo et al., 2006). Another study on psychosocial, demographic, and healthrelated correlates of self-rated health showed that daily smoking, alcohol intoxication on at least one occasion, infrequent physical activity, and difficulty making friends were predictors of poor self-rated health (Kelleher et al., 2007). Seemingly, multiple independent correlates of adolescent self-rated health exist (Breidablik et al., 2009), whereby poor health

the SF-36 questionnaire (see ECHI long list, 2005).

**4.2.3 Self-rated health (subjective health indicator)** 

increases by age and throughout adolescence (Wade & Vingilis, 1999).

which enables more differentiation on the positive end of the scale.

situation of children (ECHI long list 2005, p. 50).

The single item question on health is a suitable indicator of subjective health. Individuals are asked to indicate how they perceive their general health on a Likert scale. The answer categories are either four or five scaled. The Health Behaviour in School-aged Children (HBSC) study uses the four point Likert scale with answer categories: "excellent", "good", "fair", "poor". Those with either "fair" or "poor" health respectively "excellent" and "good" health are then combined into subgroups of individuals with "poorer health" resp. "better health" (Currie et al., 2004). Other studies, such as e.g. the European KIDSCREEN survey (Ravens-Sieberer et al., 2006), use the five-scaled answer categories: "excellent", "very good", "good", "fair", "poor". The main difference is the extra answer category "very good"

According to ECHI recommendations, preference should be given to the five answer categories (Kramers & the ECHI team, 2005). In the ECHI report, the WHO recommended instrument is proposed which is based on a five response category item: "How is your health in general?" (Answer options are: very good/good/fair/bad/very bad). The ECHI report further proposes to set the cut-off for perceived health at the % (very) good/less than good/less than fair. As noted in the ECHI report, very little focus was placed on the specific

Although no specific validation has been done on the self-rated health item in HBSC, several studies support its validity. It has shown multiple independent health-related

Fig. 1. Dimensions of the KIDSCREEN-52, -27, -10 (http://www.kidscreen.org/)

The KIDSCREEN-52 is part of a family of health-related quality of life instruments which were developed in several stages, beginning with literature searches, expert consultations (Delphi method) and focus groups with children and adolescents (Herdman et al., 2002; Ravens-Sieberer et al., 2006, 2008b). Using this approach, relevant health-related quality of life (HRQoL) dimensions and items could be identified (Ravens-Sieberer et al., 2006, 2008b). Reduction of items gathered in focus groups were done following EUROHIS guidelines (Nosikov & Gudex, 2003). Following this, a procedure of forward-backward-forward translation and harmonization was applied, followed by a pilot study and an item reduction analysis, which finally yielded a questionnaire comprising 52 items (Ravens-Sieberer et al., 2006, 2008b).

The KIDSCREEN-52 Psychological Well-being dimension assesses the psychological wellbeing of the child, which covers positive emotions and life satisfaction, including the child's or adolescents' positive perceptions and emotions, and positive feelings, such as happiness, joy and cheerfulness (Ravens-Sieberer et al., 2006). A low score on this dimension implies no pleasure in life/high dissatisfaction with life while a high score indicates happiness, positive view of life, life satisfaction and cheerfulness. The cut-off is at T-value of 36.91 and identifies the lowest 10% (Ravens-Sieberer et al., 2006). This dimension of the KIDSCREEN-52 (for children and adolescents) is comparable to the psychological well-being dimension for adults which is used as an indicator for general mental health in the ECHI report, and is defined as the percent of the population below the cut-point of the energy-vitality scale from the SF-36 questionnaire (see ECHI long list, 2005).

#### **4.2.3 Self-rated health (subjective health indicator)**

36 Public Health – Methodology, Environmental and Systems Issues

**KIDSCREEN-52 KIDSCREEN-27 KIDSCREEN-10**

Physical Well-being

Psychological Well-being

Autonomy & Parent relation General HRQoL Index

Peers & Social Support

School Environment

The KIDSCREEN-52 is part of a family of health-related quality of life instruments which were developed in several stages, beginning with literature searches, expert consultations (Delphi method) and focus groups with children and adolescents (Herdman et al., 2002; Ravens-Sieberer et al., 2006, 2008b). Using this approach, relevant health-related quality of life (HRQoL) dimensions and items could be identified (Ravens-Sieberer et al., 2006, 2008b). Reduction of items gathered in focus groups were done following EUROHIS guidelines (Nosikov & Gudex, 2003). Following this, a procedure of forward-backward-forward translation and harmonization was applied, followed by a pilot study and an item reduction analysis, which finally yielded a

The KIDSCREEN-52 Psychological Well-being dimension assesses the psychological wellbeing of the child, which covers positive emotions and life satisfaction, including the child's or adolescents' positive perceptions and emotions, and positive feelings, such as happiness, joy and cheerfulness (Ravens-Sieberer et al., 2006). A low score on this dimension implies no pleasure in life/high dissatisfaction with life while a high score indicates happiness, positive view of life, life satisfaction and cheerfulness. The cut-off is at T-value of 36.91 and identifies the lowest 10% (Ravens-Sieberer et al., 2006). This dimension of the KIDSCREEN-52 (for children and adolescents) is comparable to the psychological well-being dimension for adults which is used as an indicator for general mental health in the ECHI report, and is

Fig. 1. Dimensions of the KIDSCREEN-52, -27, -10 (http://www.kidscreen.org/)

questionnaire comprising 52 items (Ravens-Sieberer et al., 2006, 2008b).

Physical Well-being

Psychological Well-being

> Moods & Emotions

Self Perception

Autonomy

Parent relation & Home Life

> Financial Resources

Peers & Social Support

School Environment

Bullying

The building blocks for good health are laid early in life, and therefore evaluation of health does not begin in adulthood but much earlier. Health is an important resource and poor health early in life can have long-term negative effects which may continue throughout adulthood (WHO, 2006). Being in good health – physically, emotionally and socially – helps young people deal productively with challenges in their development (Burt, 2002). In recent years, self-assessments of health have come more into use as they are based on an individual's perception and evaluation of his or her health. Focusing on the subjective perspective, self-rated health is usually founded on age-peer comparisons either consciously or unconsciously (Bjorner et al., 1996). It can be distinguished from more specific health constructs in that it captures an overall conception of health, rather than a summation across specific domains of health. Empirical studies have shown that self-reported health is an independent predictor of mortality (Idler & Benyamini, 1997). Benjamins et al. (2004) could also identify a relationship between self-reported health and cause-specific mortality, and moreover, also found gender effects for some causes of mortality. A gender effect in selfrated health was confirmed in a sample of children, whereby girls reported poorer health than boys (Cavallo et al., 2006). Another study on psychosocial, demographic, and healthrelated correlates of self-rated health showed that daily smoking, alcohol intoxication on at least one occasion, infrequent physical activity, and difficulty making friends were predictors of poor self-rated health (Kelleher et al., 2007). Seemingly, multiple independent correlates of adolescent self-rated health exist (Breidablik et al., 2009), whereby poor health increases by age and throughout adolescence (Wade & Vingilis, 1999).

The single item question on health is a suitable indicator of subjective health. Individuals are asked to indicate how they perceive their general health on a Likert scale. The answer categories are either four or five scaled. The Health Behaviour in School-aged Children (HBSC) study uses the four point Likert scale with answer categories: "excellent", "good", "fair", "poor". Those with either "fair" or "poor" health respectively "excellent" and "good" health are then combined into subgroups of individuals with "poorer health" resp. "better health" (Currie et al., 2004). Other studies, such as e.g. the European KIDSCREEN survey (Ravens-Sieberer et al., 2006), use the five-scaled answer categories: "excellent", "very good", "good", "fair", "poor". The main difference is the extra answer category "very good" which enables more differentiation on the positive end of the scale.

According to ECHI recommendations, preference should be given to the five answer categories (Kramers & the ECHI team, 2005). In the ECHI report, the WHO recommended instrument is proposed which is based on a five response category item: "How is your health in general?" (Answer options are: very good/good/fair/bad/very bad). The ECHI report further proposes to set the cut-off for perceived health at the % (very) good/less than good/less than fair. As noted in the ECHI report, very little focus was placed on the specific situation of children (ECHI long list 2005, p. 50).

Although no specific validation has been done on the self-rated health item in HBSC, several studies support its validity. It has shown multiple independent health-related

Child Mental Health Measurement: Reflections and Future Directions 39

In the past, assessment of mental health was for the most part aimed at assessing mental ill health, with the focus being placed on mental health disorders and -problems. This has the disadvantage that the information gathered only enables separation between individuals with (signs of) mental disorders and healthy individuals (without any signs of mental health problems). No information is available on individuals "in-between", in other words about the position of the individual on a mental health continuum (Ravens-Sieberer et al., 2008a). Moreover, earlier instruments for measuring mental health problems in children were based and validated on experiences with child psychiatric patients and were often developed as screening instruments for patients in care. KIDSCREEN instruments overcome this drawback as they have been developed to measure mental health in the general population

The KIDSCREEN-52 "Moods & Emotions" Dimension provides an important indicator of psychological distress which can be used to identify children with depressiveness, as well as those feeling lonely, sad, and unhappy (Ravens-Sieberer et al., 2006). This dimension of the KIDSCREEN-52 examines experiences of depressive moods and emotions, including stressful feelings, and how distressing these are to the individual. A low score indicates that the child or adolescent feels depressed, is unhappy and/or in bad mood. A high score in contrast, implies feeling good and being in a good mood. The cut-off identifying the lowest

The "moods & emotions" dimension of KIDSCREEN-52 for children and adolescents corresponds to the indicator of psychological distress for adults in the general mental health section as published in the ECHI indicator list. In the ECHI report, psychological distress is defined as the percent of the population below the cut-point of MHI-5 score from the SF-36 questionnaire (see ECHI report, long list: http://www.echim.org/docs/echi\_longlist.pdf). The KIDSCREEN instruments are robust and psychometrically sound instruments suitable for the assessment of the health-related quality of life and mental health in children and adolescents between 8 and 18 years of age. The internal consistency reliability (Cronbach's Alpha) for the individual dimensions show for the "Moods & Emotions" dimension a value of 0.86 and for the "Psychological Well-being" dimension a Cronbach's Alpha of 0.89 (Ravens-Sieberer et al.,

2008b; Ravens-Sieberer et al., 2006), both of which can be considered sufficiently high.

the ECHI report (Kramers & the ECHI team, 2005), and are thus suitable indicators.

Both, the "Moods & Emotions", as well as the "Psychological Well-being" dimension of KIDSCREEN-52, correspond well with the published indicators of general mental health in

The presence of subjective health complaints and the frequency of their occurrence can serve as a good approximation for the individual's physical well-being. Health complaints tend to cluster together (Alfven, 1993; Mikkelsson et al., 1997; Starfield et al., 1984; WHO, 2006) and in this way cause immense burden – not only on the individual, but also on the health care system. Within the international HBSC Study, the Symptom Checklist (HBSC-SCL) was developed to assess the various health complaints that might occur in children and adolescents. The

**4.3 Ill-mental health indicators** 

**4.3.1 Psychological distress indicator** 

and have been validated in large population studies.

10% is at a T-value of 37.76 (Ravens-Sieberer et al., 2006).

**4.3.2 Subjective health complaints index** 

correlates, including medical diagnosis, and health complaints. The self-rated health item shows a certain degree of stability across time, suggesting that these self-reports are not simply a fluctuating subjective impression. Cavallo et al. (2006) analyzed the item in terms of its feasibility and psychometric robustness using the HBSC 2001/2002 data from all countries involved. The results confirm the trend of an increasing perception of poor health with increasing age in the pre-adolescence phase and a higher risk for perceived poorer health in girls, (Cavallo et al., 2006). HBSC showed this was a consistent finding across a large number of countries in Europe and North America (see also Currie et al., 2008).

#### **4.2.4 Life satisfaction indicator**

Well-being is a multi-faceted concept (Diener, 1984; Wilkinson & Walford, 1998) and comprises the individual's own evaluation of life, i.e. life satisfaction. It was not until the early 1990s that determinants of life satisfaction were studied (Suldo et al., 2006). Unlike other concepts, life satisfaction is relatively stable over time (Pavot & Diener, 1993). It is associated with depression, anxiety, suicide, work disability, fatal accidents and all cause mortality in adults (Fiscella & Franks, 1997; Helliwell, 2007; Koivumaa-Honkanen et al., 2001; Koivumaa-Honkanen et al., 2002; Koivumaa-Honkanen et al., 2004a; Koivumaa-Honkanen et al., 2004b). During adolescence, life satisfaction is influenced by life experiences and relationships, especially within the family context (Edwards & Lopez, 2006; Gohm et al., 1998; Rask et al., 2003) and school (Samdal et al., 1998). Psychosocial resources and school satisfaction, especially perceptions of feeling treated fairly, feeling safe and perceiving teachers as supportive (Samdal et al., 1998), are linked with high life satisfaction. School-related resources and their impact on overall life satisfaction are a central issue as the acquirement of academic competence constitutes one of the developmental goals in adolescence (Hurrelmann & Lösel, 1990). Moreover, school creates a social environment for young people which can provide them with additional resources. At the certain time, some social factors, such as bullying, can pose a risk as they may be associated with low life satisfaction and low subjective health (Gobina et al., 2008).

The Cantril Ladder is a measure of life satisfaction which has been widely used in the Health Behaviour in School-aged Children (HBSC) study. The measure is also a suitable indicator for life satisfaction in children and adolescents (Cantril, 1965). The measure consists of a Visual Analogue Scale with 11 positions (0 through 10) where children can mark the position on the scale demonstrating how satisfied they are with their life: "Here is a picture of a ladder. The top of the ladder "10" is the best possible life for you and the bottom "0" is the worst possible life for you. In general, where on the ladder do you feel you stand at the moment? Tick the box next to the number that best describes where you stand." The Health Behaviour in School-aged Children (HBSC) study uses the cut-off at "6 or above" to identify children and adolescents with a positive level of life satisfaction (normal to high life satisfaction) (Currie et al., 2004).

The Cantril Ladder has not been subject to structured validation studies at the international level, but observed relationships with quality of life and with self-rated health are in the expected range, and support claims about its validity. Analyses using data from the HBSC study show that the item is associated with the general health item and the Symptom Checklist (HBSC-SCL) (Cavallo et al., 2006).

#### **4.3 Ill-mental health indicators**

38 Public Health – Methodology, Environmental and Systems Issues

correlates, including medical diagnosis, and health complaints. The self-rated health item shows a certain degree of stability across time, suggesting that these self-reports are not simply a fluctuating subjective impression. Cavallo et al. (2006) analyzed the item in terms of its feasibility and psychometric robustness using the HBSC 2001/2002 data from all countries involved. The results confirm the trend of an increasing perception of poor health with increasing age in the pre-adolescence phase and a higher risk for perceived poorer health in girls, (Cavallo et al., 2006). HBSC showed this was a consistent finding across a large number of countries in Europe and North America (see also Currie et al.,

Well-being is a multi-faceted concept (Diener, 1984; Wilkinson & Walford, 1998) and comprises the individual's own evaluation of life, i.e. life satisfaction. It was not until the early 1990s that determinants of life satisfaction were studied (Suldo et al., 2006). Unlike other concepts, life satisfaction is relatively stable over time (Pavot & Diener, 1993). It is associated with depression, anxiety, suicide, work disability, fatal accidents and all cause mortality in adults (Fiscella & Franks, 1997; Helliwell, 2007; Koivumaa-Honkanen et al., 2001; Koivumaa-Honkanen et al., 2002; Koivumaa-Honkanen et al., 2004a; Koivumaa-Honkanen et al., 2004b). During adolescence, life satisfaction is influenced by life experiences and relationships, especially within the family context (Edwards & Lopez, 2006; Gohm et al., 1998; Rask et al., 2003) and school (Samdal et al., 1998). Psychosocial resources and school satisfaction, especially perceptions of feeling treated fairly, feeling safe and perceiving teachers as supportive (Samdal et al., 1998), are linked with high life satisfaction. School-related resources and their impact on overall life satisfaction are a central issue as the acquirement of academic competence constitutes one of the developmental goals in adolescence (Hurrelmann & Lösel, 1990). Moreover, school creates a social environment for young people which can provide them with additional resources. At the certain time, some social factors, such as bullying, can pose a risk as they may be associated with low life

The Cantril Ladder is a measure of life satisfaction which has been widely used in the Health Behaviour in School-aged Children (HBSC) study. The measure is also a suitable indicator for life satisfaction in children and adolescents (Cantril, 1965). The measure consists of a Visual Analogue Scale with 11 positions (0 through 10) where children can mark the position on the scale demonstrating how satisfied they are with their life: "Here is a picture of a ladder. The top of the ladder "10" is the best possible life for you and the bottom "0" is the worst possible life for you. In general, where on the ladder do you feel you stand at the moment? Tick the box next to the number that best describes where you stand." The Health Behaviour in School-aged Children (HBSC) study uses the cut-off at "6 or above" to identify children and adolescents with a positive level of life satisfaction (normal

The Cantril Ladder has not been subject to structured validation studies at the international level, but observed relationships with quality of life and with self-rated health are in the expected range, and support claims about its validity. Analyses using data from the HBSC study show that the item is associated with the general health item and the Symptom

2008).

**4.2.4 Life satisfaction indicator** 

satisfaction and low subjective health (Gobina et al., 2008).

to high life satisfaction) (Currie et al., 2004).

Checklist (HBSC-SCL) (Cavallo et al., 2006).

#### **4.3.1 Psychological distress indicator**

In the past, assessment of mental health was for the most part aimed at assessing mental ill health, with the focus being placed on mental health disorders and -problems. This has the disadvantage that the information gathered only enables separation between individuals with (signs of) mental disorders and healthy individuals (without any signs of mental health problems). No information is available on individuals "in-between", in other words about the position of the individual on a mental health continuum (Ravens-Sieberer et al., 2008a). Moreover, earlier instruments for measuring mental health problems in children were based and validated on experiences with child psychiatric patients and were often developed as screening instruments for patients in care. KIDSCREEN instruments overcome this drawback as they have been developed to measure mental health in the general population and have been validated in large population studies.

The KIDSCREEN-52 "Moods & Emotions" Dimension provides an important indicator of psychological distress which can be used to identify children with depressiveness, as well as those feeling lonely, sad, and unhappy (Ravens-Sieberer et al., 2006). This dimension of the KIDSCREEN-52 examines experiences of depressive moods and emotions, including stressful feelings, and how distressing these are to the individual. A low score indicates that the child or adolescent feels depressed, is unhappy and/or in bad mood. A high score in contrast, implies feeling good and being in a good mood. The cut-off identifying the lowest 10% is at a T-value of 37.76 (Ravens-Sieberer et al., 2006).

The "moods & emotions" dimension of KIDSCREEN-52 for children and adolescents corresponds to the indicator of psychological distress for adults in the general mental health section as published in the ECHI indicator list. In the ECHI report, psychological distress is defined as the percent of the population below the cut-point of MHI-5 score from the SF-36 questionnaire (see ECHI report, long list: http://www.echim.org/docs/echi\_longlist.pdf).

The KIDSCREEN instruments are robust and psychometrically sound instruments suitable for the assessment of the health-related quality of life and mental health in children and adolescents between 8 and 18 years of age. The internal consistency reliability (Cronbach's Alpha) for the individual dimensions show for the "Moods & Emotions" dimension a value of 0.86 and for the "Psychological Well-being" dimension a Cronbach's Alpha of 0.89 (Ravens-Sieberer et al., 2008b; Ravens-Sieberer et al., 2006), both of which can be considered sufficiently high.

Both, the "Moods & Emotions", as well as the "Psychological Well-being" dimension of KIDSCREEN-52, correspond well with the published indicators of general mental health in the ECHI report (Kramers & the ECHI team, 2005), and are thus suitable indicators.

#### **4.3.2 Subjective health complaints index**

The presence of subjective health complaints and the frequency of their occurrence can serve as a good approximation for the individual's physical well-being. Health complaints tend to cluster together (Alfven, 1993; Mikkelsson et al., 1997; Starfield et al., 1984; WHO, 2006) and in this way cause immense burden – not only on the individual, but also on the health care system.

Within the international HBSC Study, the Symptom Checklist (HBSC-SCL) was developed to assess the various health complaints that might occur in children and adolescents. The

Child Mental Health Measurement: Reflections and Future Directions 41

**38 48 58**

European mean = 48

\*\* p<.01 \*\*\* p<.001

gender3,4

the low FAS group.

(according to Cohen 1988).

<sup>1</sup> Mean scores of the KIDSCREEN-10 are depicted

2 This figure was previously published in Ravens-Sieberer et al. (2008a).

4 This figure was previously published in Ravens-Sieberer et al. (2008a).

**Netherlands**

**Switzerland**

**Austria**

**Germany**

**Sweden**

**Spain**

Fig. 2. Positive mental health index (KIDSCREEN-10) across 13 European countries1,2

Table 1. Positive mental health (KIDSCREEN-10) in different countries according to

Next, the analysis of positive mental health by family affluence shows that higher family affluence, i.e. growing up a in a better-situated family, is generally associated with a higher level of positive mental health (i.e. above the European average). This is depicted in Figure 3 by the increasing line (with one or two exceptions) and also in the distribution of % of children in low, middle and high FAS group per country (results not shown). Countries with higher mental health score means are also those with the least number of adolescents in

<sup>3</sup> Effect size calculation was based on dividing the mean difference by the overall standard deviation

**Country Girls m(SD) Boys m(SD) Effect (d)**  Austria (n=878) 49.6 (8.9) 52.6 (9.4) 0.3\*\*\* Czech Republic (n=1016) 45.0 (7.1) 47.3 (7.8) 0.3\*\*\* France (n=622) 45.0 (8.4) 46.1 (8.0) n.s. Germany (n=1079) 49.3 (8.4) 51.0 (8.4) 0.2\*\*\* Greece (n=1146) 44.2 (7.6) 47.2 (8.0) 0.4\*\*\* Hungary (n=1839) 43.6 (7.6) 46.2 (8.9) 0.3\*\*\* Ireland (n=894) 45.5 (7.9) 48.1 (7.6) 0.3\*\*\* The Netherlands (n=1168) 50.2 (8.2) 53.6 (10.0) 0.4\*\*\* Poland (n=1120) 43.9 (7.9) 45.3 (7.3) 0.2\*\* Spain (n=522) 48.4 (9.6) 50.9 (8.7) 0.3\*\* Sweden (n=3097) 49.2 (10.0) 52.4 (10.0) 0.3\*\*\* Switzerland (n=1078) 49.6 (8.0) 52.6 (8.5) 0.4\*\*\* United Kingdom (n=883) 45.5 (8.3) 47.8 (8.5) 0.3\*\*\*

**Czech**

**Republic**

**KIDSCREEN-10 Index**

**Ireland**

**UK**

**France**

**Greece**

**Hungary**

**Poland**

HBSC-SCL has proven to be a suitable and effective screening tool for the assessment of physical well-being. The Checklist includes symptoms, such as headache, abdominal pain, backache, feeling low, irritability or bad mood, feeling nervous, sleeping difficulties and dizziness (Haugland et al., 2001). The advantage of the HBSC-SCL is that it is not limited to somatic symptoms, but also contains a number of psychological symptoms and hence constitutes an instrument suitable for detecting psychosomatic complaints.

The HBSC-SCL assesses the occurrence of health complaints in children and adolescents and is a useful indicator for identifying individuals at risk for impaired health. The HBSC-SCL asks about the occurrence of the following symptoms in the last 6 months: Headache, Stomach ache, Back ache, Feeling low, Irritability or bad temper, Feeling nervous, Difficulties in getting to sleep, Feeling dizzy. Ravens-Sieberer et al. (2008c) developed an international scoring system for the HBSC-SCL which enables a cross-cultural and intervalscaled assessment of subjective health complaints and which can further be used to identify individuals at a greater risk of health complaints. This uni-dimensional scoring algorithm is based on seven of the eight items. A score below 41 indicates a "higher risk" for health complaints (Ravens-Sieberer et al., 2008c).

A number of validation studies have been made on the HBSC-SCL (Haugland & Wold, 2001; Haugland et al., 2001). Qualitative semi-structured interviews with early adolescents revealed that adolescents perceive the symptoms to be aversive physical and psychological states that interfere with daily functional ability and well-being. Consistent accounts as to how the different symptoms were defined were given, suggesting that adolescents have a common frame of reference when they rate their frequency of symptoms (Haugland & Wold, 2001). Differences emerged in their lay perspectives on the causes of such symptoms. While some explanations were consistent with a stress-model of health complaints, others were associated with developmental processes, such as growing pain, or ergonomic factors, such as low quality of air in classrooms etc.

#### **4.4 Application of the mental health indicator (KIDSCREEN-10)**

As previously mentioned, the mental health index (KIDSCREEN-10) is a non-clinical measure of mental health status. It does not permit identification of groups with defined burden of mental health problems, but allows measurement along a continuum (Ravens-Sieberer et al. 2008a).

As stated previously, KIDSCREEN-10 is an indicator of quality of life and (positive) mental health and in the following, we will apply it on a European sample of adolescents from 13 countries. The overall mean score is 48 with a standard deviation of 10. The results in Figure 2 show that some countries fall above and some below the European mean. Countries towards the left side of the figure tend to show better positive mental health compared to the countries at the right end of the figure which fall below the European average. Additional results show that variation in mental health scores was generally lower in countries with lower positive mental health scores (results not shown).

To gain a better understanding of the distribution of mental health, we will now look at selected sociodemographic characteristics, such as gender and socioeconomic status (approximated by family affluence [FAS]). Table 1 below shows the distribution of positive mental health across the 13 countries by gender. Comparisons across gender groups show that boys report better mental health across all countries than girls. This difference is significant in all but one country, and the effect sizes are generally small.

Fig. 2. Positive mental health index (KIDSCREEN-10) across 13 European countries1,2


\*\* p<.01

40 Public Health – Methodology, Environmental and Systems Issues

HBSC-SCL has proven to be a suitable and effective screening tool for the assessment of physical well-being. The Checklist includes symptoms, such as headache, abdominal pain, backache, feeling low, irritability or bad mood, feeling nervous, sleeping difficulties and dizziness (Haugland et al., 2001). The advantage of the HBSC-SCL is that it is not limited to somatic symptoms, but also contains a number of psychological symptoms and hence

The HBSC-SCL assesses the occurrence of health complaints in children and adolescents and is a useful indicator for identifying individuals at risk for impaired health. The HBSC-SCL asks about the occurrence of the following symptoms in the last 6 months: Headache, Stomach ache, Back ache, Feeling low, Irritability or bad temper, Feeling nervous, Difficulties in getting to sleep, Feeling dizzy. Ravens-Sieberer et al. (2008c) developed an international scoring system for the HBSC-SCL which enables a cross-cultural and intervalscaled assessment of subjective health complaints and which can further be used to identify individuals at a greater risk of health complaints. This uni-dimensional scoring algorithm is based on seven of the eight items. A score below 41 indicates a "higher risk" for health

A number of validation studies have been made on the HBSC-SCL (Haugland & Wold, 2001; Haugland et al., 2001). Qualitative semi-structured interviews with early adolescents revealed that adolescents perceive the symptoms to be aversive physical and psychological states that interfere with daily functional ability and well-being. Consistent accounts as to how the different symptoms were defined were given, suggesting that adolescents have a common frame of reference when they rate their frequency of symptoms (Haugland & Wold, 2001). Differences emerged in their lay perspectives on the causes of such symptoms. While some explanations were consistent with a stress-model of health complaints, others were associated with developmental processes, such as growing pain, or ergonomic factors,

As previously mentioned, the mental health index (KIDSCREEN-10) is a non-clinical measure of mental health status. It does not permit identification of groups with defined burden of mental health problems, but allows measurement along a continuum (Ravens-Sieberer et al. 2008a).

As stated previously, KIDSCREEN-10 is an indicator of quality of life and (positive) mental health and in the following, we will apply it on a European sample of adolescents from 13 countries. The overall mean score is 48 with a standard deviation of 10. The results in Figure 2 show that some countries fall above and some below the European mean. Countries towards the left side of the figure tend to show better positive mental health compared to the countries at the right end of the figure which fall below the European average. Additional results show that variation in mental health scores was generally lower in

To gain a better understanding of the distribution of mental health, we will now look at selected sociodemographic characteristics, such as gender and socioeconomic status (approximated by family affluence [FAS]). Table 1 below shows the distribution of positive mental health across the 13 countries by gender. Comparisons across gender groups show that boys report better mental health across all countries than girls. This difference is

constitutes an instrument suitable for detecting psychosomatic complaints.

complaints (Ravens-Sieberer et al., 2008c).

such as low quality of air in classrooms etc.

**4.4 Application of the mental health indicator (KIDSCREEN-10)** 

countries with lower positive mental health scores (results not shown).

significant in all but one country, and the effect sizes are generally small.

\*\*\* p<.001

Table 1. Positive mental health (KIDSCREEN-10) in different countries according to gender3,4

Next, the analysis of positive mental health by family affluence shows that higher family affluence, i.e. growing up a in a better-situated family, is generally associated with a higher level of positive mental health (i.e. above the European average). This is depicted in Figure 3 by the increasing line (with one or two exceptions) and also in the distribution of % of children in low, middle and high FAS group per country (results not shown). Countries with higher mental health score means are also those with the least number of adolescents in the low FAS group.

<sup>1</sup> Mean scores of the KIDSCREEN-10 are depicted

<sup>2</sup> This figure was previously published in Ravens-Sieberer et al. (2008a).

<sup>3</sup> Effect size calculation was based on dividing the mean difference by the overall standard deviation (according to Cohen 1988).

<sup>4</sup> This figure was previously published in Ravens-Sieberer et al. (2008a).

Child Mental Health Measurement: Reflections and Future Directions 43

and indicators; to identify gaps in child health research as perceived by stakeholders; and lastly, to develop roadmaps for the future of child health research in Europe based on all these findings. One of the objectives of the RICHE project is to produce an inventory of measurement and indicators to facilitate the implementation of existing methods and at the same time to initiate new developments through exchange and networking. Based on the notion that development and implementation of sound indicators is essential for developing child health with the European Union (Rigby et al., 2003), it is important that high quality indicators are available for analyses and political decision making. Health measurement is

The review of research into child mental health measurement has revealed important advances, such as the development of quality of life and positive mental health indicators in the KIDSCREEN project, while on the other hand, also pointed out several shortcomings. The main shortcomings relate to the age-specification and the cultural adaptation of measurement tools. With regard to the age issue, our evaluation of measures of child mental health revealed that measurement tools generally target older children and adolescents, i.e. eight years and older. Many of the measurement tools come from the HBSC survey and all of these have originally been designed for 11-, 13-, and 15-year old children. Any application of these instruments on younger children would require further validation which to date has not been done. Although indicators based on KIDSCREEN measures are suitable for slightly younger children (beginning with age seven), they are not available for the very young children (0-6 year olds). For this young group, there is a clear gap on measurement tools, especially those enabling a valid and cross-cultural assessment of quality of life and well-being for the age group 0-3 years and for the age group 4-6 years. Generally, very young and young children are underrepresented in international data sources, and "a portrait of positive well-being among young children is not available, and in many cases, measures are lacking that are appropriate for their age" (Lippman et al., 2009, p. 24). This implies that many indicators are adolescent-focused and hence may point attention to matters relevant for adolescents which may be quite different from those that are essential

Another shortcoming of current research on indicators for child well-being lies in the cultural adaptation of the measurement tools. As mentioned above, all of the indicators presented here are based on measurement tools which have been developed within the European and North American context. In order to compare child well-being and quality of life in different cultural contexts outside of Europe (e.g. in Africa, Asia), cultural adaptations would need to be done with the instruments. Currently, this is a research challenge in this

To end on a positive note, it is important to acknowledge that there are already a number of programmes on mental health and well-being in children and adolescents underway. A good example of a promising strategy is Scotland's "National Programme for Improving Mental Health and Wellbeing" which was launched in April 2008 with the purpose to identify a core set of indicators on mental health to support the national action plan on mental health ("mental health profile for Scotland"; Parkinson, 2009). Building upon the experience from the establishment of a mental health indicator set for adults (Parkinson,

the core for developing prevention strategies in a life course perspective.

for children (Bradshaw et al., 2006).

**6. Outlook** 

field and needs to be addressed in the near future.

Fig. 3. Adolescent positive mental health (KIDSCREEN-10) and FAS in the participating countries (without Ireland and Sweden)5

The results that were presented here serve the purpose to exemplify the application of a robust measure of mental health in a European sample of children and adolescents. Results show the associations between the outcome (mental health) and various sociodemographic factors (age, gender, FAS) and in this way provide the basis for more comprehensive analyses of mental health status in children and adolescents in Europe.

#### **5. Closing comments**

The objective of this chapter was to give interested readers an insight into the state-of-the-art in child mental health measurement. Our aim was to show that progress has indeed been made in this vast field, and although we still do not have all the tools and information for a complete assessment of mental health in children and adolescents, we have been able to identify useful measures and important surveys at the European level which enable a good approximation. The complexity of the field has made it necessary for us to concentrate on a few indicators which in our view are good representatives of the respective constructs. The indicators and the results we have described in this chapter come from the HBSC and KIDSCREEN Studies and also reflect our insights from within the RICHE project.

The original idea for a publication on mental health measurement came up during the course of working in the RICHE project. RICHE stands for "Research into Child Health in Europe" and is an international project focusing on child health research in Europe. The project is funded within the EU 7th Framework Programme. RICHE embraces the full multidisciplinary diversity of European research and addresses its fragmentation by making the parts visible. This is done in part via a platform which provides the opportunity for open exchange (http://www.childhealthresearch.eu/). The aims of the project are: to provide an inventory of current research; to identify research into child health measurement, statistics,

<sup>5</sup> The Figure was previously published in Ravens-Sieberer et al. (2008a).

and indicators; to identify gaps in child health research as perceived by stakeholders; and lastly, to develop roadmaps for the future of child health research in Europe based on all these findings. One of the objectives of the RICHE project is to produce an inventory of measurement and indicators to facilitate the implementation of existing methods and at the same time to initiate new developments through exchange and networking. Based on the notion that development and implementation of sound indicators is essential for developing child health with the European Union (Rigby et al., 2003), it is important that high quality indicators are available for analyses and political decision making. Health measurement is the core for developing prevention strategies in a life course perspective.

The review of research into child mental health measurement has revealed important advances, such as the development of quality of life and positive mental health indicators in the KIDSCREEN project, while on the other hand, also pointed out several shortcomings. The main shortcomings relate to the age-specification and the cultural adaptation of measurement tools. With regard to the age issue, our evaluation of measures of child mental health revealed that measurement tools generally target older children and adolescents, i.e. eight years and older. Many of the measurement tools come from the HBSC survey and all of these have originally been designed for 11-, 13-, and 15-year old children. Any application of these instruments on younger children would require further validation which to date has not been done. Although indicators based on KIDSCREEN measures are suitable for slightly younger children (beginning with age seven), they are not available for the very young children (0-6 year olds). For this young group, there is a clear gap on measurement tools, especially those enabling a valid and cross-cultural assessment of quality of life and well-being for the age group 0-3 years and for the age group 4-6 years. Generally, very young and young children are underrepresented in international data sources, and "a portrait of positive well-being among young children is not available, and in many cases, measures are lacking that are appropriate for their age" (Lippman et al., 2009, p. 24). This implies that many indicators are adolescent-focused and hence may point attention to matters relevant for adolescents which may be quite different from those that are essential for children (Bradshaw et al., 2006).

Another shortcoming of current research on indicators for child well-being lies in the cultural adaptation of the measurement tools. As mentioned above, all of the indicators presented here are based on measurement tools which have been developed within the European and North American context. In order to compare child well-being and quality of life in different cultural contexts outside of Europe (e.g. in Africa, Asia), cultural adaptations would need to be done with the instruments. Currently, this is a research challenge in this field and needs to be addressed in the near future.

#### **6. Outlook**

42 Public Health – Methodology, Environmental and Systems Issues

Fig. 3. Adolescent positive mental health (KIDSCREEN-10) and FAS in the participating

analyses of mental health status in children and adolescents in Europe.

The results that were presented here serve the purpose to exemplify the application of a robust measure of mental health in a European sample of children and adolescents. Results show the associations between the outcome (mental health) and various sociodemographic factors (age, gender, FAS) and in this way provide the basis for more comprehensive

The objective of this chapter was to give interested readers an insight into the state-of-the-art in child mental health measurement. Our aim was to show that progress has indeed been made in this vast field, and although we still do not have all the tools and information for a complete assessment of mental health in children and adolescents, we have been able to identify useful measures and important surveys at the European level which enable a good approximation. The complexity of the field has made it necessary for us to concentrate on a few indicators which in our view are good representatives of the respective constructs. The indicators and the results we have described in this chapter come from the HBSC and

The original idea for a publication on mental health measurement came up during the course of working in the RICHE project. RICHE stands for "Research into Child Health in Europe" and is an international project focusing on child health research in Europe. The project is funded within the EU 7th Framework Programme. RICHE embraces the full multidisciplinary diversity of European research and addresses its fragmentation by making the parts visible. This is done in part via a platform which provides the opportunity for open exchange (http://www.childhealthresearch.eu/). The aims of the project are: to provide an inventory of current research; to identify research into child health measurement, statistics,

KIDSCREEN Studies and also reflect our insights from within the RICHE project.

5 The Figure was previously published in Ravens-Sieberer et al. (2008a).

countries (without Ireland and Sweden)5

**5. Closing comments** 

To end on a positive note, it is important to acknowledge that there are already a number of programmes on mental health and well-being in children and adolescents underway. A good example of a promising strategy is Scotland's "National Programme for Improving Mental Health and Wellbeing" which was launched in April 2008 with the purpose to identify a core set of indicators on mental health to support the national action plan on mental health ("mental health profile for Scotland"; Parkinson, 2009). Building upon the experience from the establishment of a mental health indicator set for adults (Parkinson,

Child Mental Health Measurement: Reflections and Future Directions 45

Prof Candace Currie, The University of Edinburgh, National Institute for Health and

Prof Angela Brand, Universiteit Maastricht, European Centre for Public Health Genomics

Aborn, M. (1985). Statistical legacies of the social indicators movement. *Paper presented at the annual meeting of the American Statistical Association*, Las Vegas, Nevada Alfven, G. (1993). The covariation of common psychosomatic symptoms among children

Ben-Arieh, A. & Wintersberger, H. (Eds.) (1997). Monitoring and Measuring the State of

Ben-Arieh, A. (2008). The Child Indicators Movement: Past, Present, and Future. *Child* 

Benjamins, M. R., Hummer, R. A., Eberstein, I. W. & Nam, C. B. (2004). Self-report health

Bjorner, J. P, Kristensen, T. O., Orth-Gomer, K., Tibblin, G., Sullivan, M. & Westerholm, P.

Breidablik, H. J., Meland, E. & Lydersen, S. (2009). Self-rated health during adolescence:

Blumer, H. (1962). Society as Symbolic Interaction, In: *Human Behavior and Social Process: An Interactionist Approach*, Arnold M. Rose, pp. 179-192, Houghton-Mifflin, Boston

Bradshaw, J., Hoelscher, P., Richardson, D. & UNICEF (2006). *Comparing Child Well-Being in* 

Burt, M.R. (2002). Reasons to invest in adolescents. *Journal of Adolescent Health*, 31, 2, pp. 136-

Cameron, E., Mathers, J. & Parry, J (2006). Health and well-being: questioning the use of

Campbell, A. & Converse, P.E. (Eds.) (1972). *The Human Meaning of Social Change*, Russell

Cavallo, F., Zambon, A., Borraccino, A., Ravens-Sieberer, U., Torsheim, T. & Lemma, P.

Cohen, J. (1988). *Statistical power analysis for the behavioral science*, Lawrence Erlbaum, New

Cantril, H. (1965). *The pattern of human concern*, Rutgers University Press, New York

*Quality of Life Research*, 15, 10, pp. 1577-1585, 1573-2649

from socio-economically differing residential areas: an epidemiological study. *Acta* 

Children – Beyond Survival. *Eurosocial Report No. 62*, European Centre for Social

and adult mortality risk: an analysis of cause-specific mortality. *Social Science &* 

(1996). *Self-rated health. A useful concept in research, prevention and clinical medicine,*

stability and predictors of change (Young-Hunt study, Norway). *European Journal of* 

*OECD Countries: Concepts and methods. Innocenti Working Paper No. 2006-03*, UNICEF

health concepts in public health policy and practice. *Critical Public Health*, 16, 4, pp.

(2006). Girls growing through adolescence have a higher risk of poor health.

Prof Ulrike Ravens-Sieberer, Universitätsklinikum Hamburg-Eppendorf

Bauer, R. A. (Ed.) (1966). *Social Indicators*, MIT Press, Cambridge

Forskningsradsnämnden, Ord & Form AB, Uppsala

Innocenti Research Centre, 1014-7837, Florence.

Bradburn, N. M. (1969). *The structure of psychological well-being*, Aldine, Chicago

Welfare Policy and Research, Vienna

*Indicators Research*, 1, pp. 3-16, 1874-8988

Dr Ales Bourek, Masarykova Univerzita

Prof Allan Colver, University of Newcastle upon Tyne

Prof Livia Popescu, Universitatea Babeş-Bolyai

*Paediatr*, 82, pp. 484-487

*Medicine*, 59, pp. 1297-1306

*Public Health*, 91, 1, pp. 73-78

152

York

347-354, 1469-3682

Sage Foundation, New York

Clinical Excellence

**8. References** 

2007), the same is now being done for the age group 18 years and younger. The goal is to "support and promote consistent and sustainable national monitoring of the state of mental health and associated contextual factors for children and young people in Scotland" (Parkinson, 2009, n.p.). Using different methods, such as a comprehensive review of literature on children's own views of mental health (Shucksmith et al., 2009), and wider consultations with researchers, policy makers and practitioners, as well as advisory groups will be used to inform the development of a framework on mental health and well-being. Direct consultations with children and young people on mental health indicators and the proposed framework will complement this information (Parkinson, 2009).

NHS Scotland is a good example of how research and policy making can work together to more forward on an important public health issue with high relevance not just in Scotland, but also at the European level. It highlights the importance of epidemiological data which delivers information relevant for development of policy making (Remschmidt & Belfer, 2005). HBSC and KIDSCREEN are good examples of this. Through regular, standardized data collection (such as through monitoring), health indicators can further help in the problem identification process as well as in its prioritization (Korkeila et al., 2006). By "screening" for certain (risk) groups or health problems, they are valuable tools for preventive action which requires early detection of hidden or manifest mental health problems (Erhart et al., 2009). In this sense, indicators are an important "bridge between health policy and scientific information" (Korkeila et al., 2006, p. 13).

#### **7. Acknowledgement**

The RICHE Project (Title: A platform and inventory for child health research in Europe) is financed by a grant from the European Commission within the EC Seventh Framework Programme (HEALTH-2009-3.3-5 European child health research platform).

Participants of the RICHE project are:

Prof Anthony Staines, Dublin City University (Coordinator) Dr Giorgio Tamburlini, Burlo Garofolo Ms Csilla Kaposvari, Egészség Monitor Prof Hein Raat, Erasmus Universitair Medisch Centrum Rotterdam Dr Else-Karin Groholt, Nasjonalt Folkehelseinstitutt Prof Margarida Gaspar de Matos, Faculdade de Motricidade Humana Dr Anne McCarthy, The Health Research Board Dr Mitch Blair, Imperial College of Science, Technology and Medicine Dr Matilde Leonardi, Fondazione IRCCS Istituto Neurologico Carlo Besta Dr Polonca Truden-Dobrin, Institute of Public Health of the Republic of Slovenia Dr Reli Mechtler, Johannes Kepler University Linz, Institute of Health System Research Prof Michael Rigby, Nordic School for Public Health Prof Anders Hjern, Nordic School for Public Health Dr Polanska Kinga, Nofer Institute of Occupational Medicine Mr Con Hennessy, Open Applications Consulting Limited Prof Geir Gunnlaugsson, Reykjavik University Prof Mika Gissler, National Institute for Health and Welfare Prof Toomas Veidebaum, National Institute for Health Development

Dr Ales Bourek, Masarykova Univerzita

Prof Candace Currie, The University of Edinburgh, National Institute for Health and Clinical Excellence

Prof Ulrike Ravens-Sieberer, Universitätsklinikum Hamburg-Eppendorf

Prof Allan Colver, University of Newcastle upon Tyne

Prof Livia Popescu, Universitatea Babeş-Bolyai

Prof Angela Brand, Universiteit Maastricht, European Centre for Public Health Genomics

#### **8. References**

44 Public Health – Methodology, Environmental and Systems Issues

2007), the same is now being done for the age group 18 years and younger. The goal is to "support and promote consistent and sustainable national monitoring of the state of mental health and associated contextual factors for children and young people in Scotland" (Parkinson, 2009, n.p.). Using different methods, such as a comprehensive review of literature on children's own views of mental health (Shucksmith et al., 2009), and wider consultations with researchers, policy makers and practitioners, as well as advisory groups will be used to inform the development of a framework on mental health and well-being. Direct consultations with children and young people on mental health indicators and the

NHS Scotland is a good example of how research and policy making can work together to more forward on an important public health issue with high relevance not just in Scotland, but also at the European level. It highlights the importance of epidemiological data which delivers information relevant for development of policy making (Remschmidt & Belfer, 2005). HBSC and KIDSCREEN are good examples of this. Through regular, standardized data collection (such as through monitoring), health indicators can further help in the problem identification process as well as in its prioritization (Korkeila et al., 2006). By "screening" for certain (risk) groups or health problems, they are valuable tools for preventive action which requires early detection of hidden or manifest mental health problems (Erhart et al., 2009). In this sense, indicators are an important "bridge between

The RICHE Project (Title: A platform and inventory for child health research in Europe) is financed by a grant from the European Commission within the EC Seventh Framework

proposed framework will complement this information (Parkinson, 2009).

health policy and scientific information" (Korkeila et al., 2006, p. 13).

Prof Anthony Staines, Dublin City University (Coordinator)

Dr Else-Karin Groholt, Nasjonalt Folkehelseinstitutt

Prof Michael Rigby, Nordic School for Public Health Prof Anders Hjern, Nordic School for Public Health

Prof Geir Gunnlaugsson, Reykjavik University

Dr Polanska Kinga, Nofer Institute of Occupational Medicine Mr Con Hennessy, Open Applications Consulting Limited

Prof Mika Gissler, National Institute for Health and Welfare

Prof Toomas Veidebaum, National Institute for Health Development

Dr Anne McCarthy, The Health Research Board

Prof Hein Raat, Erasmus Universitair Medisch Centrum Rotterdam

Prof Margarida Gaspar de Matos, Faculdade de Motricidade Humana

Dr Mitch Blair, Imperial College of Science, Technology and Medicine Dr Matilde Leonardi, Fondazione IRCCS Istituto Neurologico Carlo Besta Dr Polonca Truden-Dobrin, Institute of Public Health of the Republic of Slovenia Dr Reli Mechtler, Johannes Kepler University Linz, Institute of Health System Research

Programme (HEALTH-2009-3.3-5 European child health research platform).

**7. Acknowledgement** 

Participants of the RICHE project are:

Dr Giorgio Tamburlini, Burlo Garofolo Ms Csilla Kaposvari, Egészség Monitor


Child Mental Health Measurement: Reflections and Future Directions 47

Herdman, M., Rajmil, L., Ravens-Sieberer, U., Bullinger, M., Power, M., Alonso, J., the

adolescents: a Delphi study. *Acta Paediatrica*, 91, 12, pp. 1385–90, 1651-2227 Hurrelmann, K. & Lösel F. (1990). Basic issues and problem of health in adolescence, In:

Husserl, E. (1913). *Ideen zu einer reinen Phänomenologie und phänomenologischen Philosophie.* 

Idler, E.L. & Benyamini, Y. (1997). Self-related health and mortality: A review of twentyseven community studies. *Journal of Health and Behavior*, 38, pp. 21-37, 2150-6000

Jané-Llopis, E. & Braddick, F. (Eds). (2008). *Mental Health in Youth and Education. Consensus* 

Keyes, C. L. M. (2002). The mental health continuum: From languishing to flourishing in life.

Keyes, C. L. M. (2005). Mental illness and/or mental health? Investigating axioms of the

Keyes, C. L. M. (2006). Subjective Well-Being in Mental Health and Human Development

Keyes, C. L. M. (2007). Promoting and protecting mental health as flourishing: A

Kilpeläinen, K., Aromaa, A. & the ECHIM project (Eds.) (2008). *European Health Indicators:* 

Koivumaa-Honkanen, H., Honkanen, R., Viinamaki, H., Heikkila, K., Kaprio, J. &

Koivumaa-Honkanen, H.T., Honkanen, R., Koskenvuo, M., Viinamaki, H. & Kaprio, J.

Koivumaa-Honkanen, H. T., Kaprio, J., Honkanen, R., Viinamaki, H. & Koskenvuo, M.

Korkeila, J., Tuomi-Nikula, A., Wahlbeck, K., Lehtinen, V. & Lavikainen J. (2006). Proposal

Kramers, P.G.N. (2003). The ECHI project. Health indicators for the European Community.

Kramers, P.G.N. & the ECHI team (2005). *Public Health indicators for the European Union:* 

European Journal of Public Health, 13, pp. 101-106, 1464-360X

*Social Psychiatry and Psychiatric Epidemiology*, 39, pp. 994-999, 1433-9285 Koivumaa-Honkanen, H.T., Koskenvuo, M., Honkanen, R., Viinamaki, H., Heikkila, K. &

complete state model of health. *Journal of Consulting and Clinical Psychology*, 73, pp.

Research Worldwide: An Introduction. *Social Indicators Research*, 77, 1, pp. 1-10,

complementary strategy for improving national mental health. *American* 

*Development and Initial Implementation. Final Report of the ECHIM project*, National

Koskenvuo, M. (2001). Life satisfaction and suicide: A 20-year follow-up study.

(2002). Life satisfaction as a predictor of fatal injury in a 20-year follow-up. *Acta* 

(2004a). Life satisfaction and depression in a 15-year follow-up of healthy adults.

Kaprio, J. (2004b). Life dissatisfaction and subsequent work disability in an 11-year

for a harmonised set of mental health indicators, In: *Improving Mental Health Information in Europe*, Lavikainen, Fryers & Lehtinen, pp. 107-116, Stakes and

*Context, selection, definition. Final Report by the ECHI Project Phase II*, National

Jahoda, M. (1958). *Current Concepts of Positive Mental Health*, Basic Books, New York

*paper*, European Communities, Luxembourg, 978-92-79-09526-9

*Journal of Health and Social Behavior*, 43, pp. 207-222, 2150-6000

Berlin

Verlag, Halle

539–548, 0022-006X

*Psychologist*, 62, pp. 95–108, 0003-066X

Public Health Institute, 978-951-740-858-5, Helsinki

*American Journal of Psychiatry*, 158, pp. 433-439, 1535-7228

*Psychiatrica Scandinavica*, 105, pp. 444–450, 1600-0447

follow-up. *Psychological Medicine*, 34, pp. 221-228,

European Union, Helsinki

1573-0921

European KIDSREEN Group & DISABKIDS Group (2002). Expert consensus in the development of a European health-related quality of life measure for children and

*Health hazards in adolescence*, Hurrelmann & Lösel, pp. 1-21, Walter de Gruyter,

*Erstes Buch: Allgemeine Einführung in die reine Phänomenologie*, Max Niemeyer


Currie, C., Roberts, C., Morgan, A., Smith, R., Settertobulte, W., Samdal, O. & Rasmussen, V.

Currie, C., Nic Gabhainn, S. Godeau, E., Roberts, C. Smith, R. Currie, D., Pickett, W. Richter,

Deci, E. L. & Ryan, R. M. (2008). Hedonia, Eudaimonia, and Well-Being: An Introduction.

ECHI long list (2005). Annex 5, In: *The ECHI comprehensive indicators list (Long List)*,

<http://www.healthindicators.eu/healthindicators/object\_binary/o2701\_ECHI\_

Edwards, L. M. & Lopez, S. J. (2006). Perceived family support, acculturation, and life

Ehrlich, D. (1961). Americans View Their Mental Health (Review). *Archives of General* 

Erhart, M., Wille, N., Ravens-Sieberer, U. (2006). Die Messung der subjektiven Gesundheit:

Erhart, M. , Ottova, V., Gaspar, T., Jericek, H., Schnohr, C., Alikasifoglu, M., Morgan, A.,

Fiscella, K. & Franks, P. (1997). Does psychological distress contribute to racial and

Frønes, I. (2007). Theorizing indicators. On Indicators, Signs and Trends. *Social Indicators* 

Gobina, I., Zaborskis, A., Pudule, I., Kalnins, I. & Villerusa, A. (2008). Bullying and

Gohm, C., Oishi, S., Darlington, J. & Diener, E. (1998). Culture, parental conflict, parental

Gurin, G., Veroff, J. & Feld, S. (1960). *Americans View Their Mental Health*, Basic Books, New

Helliwell, J. F. (2007). Well-being and social capital: Does suicide pose a puzzle? *Social* 

*International Journal of Public Health*, 53, 5, pp. 272-276, 1661-8556

Diener, E. (1984). Subjective well-being. *Psychological Bulletin*, 95, pp. 542–575, 0033-2909 Dupuy, H. J. (1977). The General Well-being Schedule, In: *Measuring health: a guide to rating* 

Office for Europe, 92 890 1372 9, Copenhagen

*Journal of Happiness Studies*, 9, pp. 1-11, 1573-7780

*Counselling Psychology*, 53, pp. 279-287, 1939-2168

Sozialwissenschaften, 3531160842, Wiesbaden

*Research*, 83, pp. 5-23, pp. 1573-0921, 1573-0921

*the Family*, 60, pp. 319–334, 1741-3737

*Indicators Research*, 81, pp. 455-496, 1573-0921

*Psychiatry*, 5, 6, pp. 616-a-618, 0003-990X

Europe, Copenhagen

University Press, Oxford

longlist.pdf>

Brussels, June 2008

0277-9536

York

31.10.2011, Available from:

B. (Eds.) (2004). *Young People's Health in Context: international report from the HBSC 2001/02 survey, (Health Policy for Children and Adolescents, No.4)*, WHO Regional

M., Morgan, A. & Barnekow, V. (Eds.) (2008). *Inequalities in young people's health: HBSC international report from the 2005/2006 Survey*, WHO Regional Office for

*scales and questionnaire (2nd ed)*, McDowell & Newell, pp. 206-213, Oxford

satisfaction in Mexican American youth: A mixed-methods exploration. *Journal of* 

Stand der Forschung und Herausforderungen, In: *Gesundheitliche Ungleichheit. Grundlagen, Probleme, Perspektiven*, Richter, Hurrelmann, pp. 321-338, VS Verlag für

Ravens-Sieberer, U. & HBSC Positive Health Focus Group (2009). Measuring mental health and well-being of school-children in 15 European countries using the KIDSCREEN-10 Index. *International Journal of Public Health*, 54, 2, pp. 160-166 European Commission & WHO (2008). European Pact for Mental Health and Well-being,

*Proceeding of EU High-level Conference: Together for Mental Health and Well-Being*,

socioeconomic disparities in mortality? *Social Science & Medicine*, 45, pp. 1805–1809,

subjective health among adolescents at schools in Latvia and Lithuania.

marital status, and the subjective well-being of young adults. *Journal of Marriage and* 


Child Mental Health Measurement: Reflections and Future Directions 49

Ravens-Sieberer, U., Gosch, A., Abel, T., Auquier, P., Bellach, B.M., Bruil, J., Dür, W., Power,

Ravens-Sieberer, U., Gosch, A., Rajmil, L., Erhart, M., Bruil, J., Duer, W., Auquier, P., Power,

Ravens-Sieberer, U., et al. & the European KIDSCREEN Group. (2006). *The KIDSCREEN* 

Ravens-Sieberer, U., Wille, N., Erhart, M., Nickel, J. & Richter, M. (2008a). Socioeconomic

Ravens-Sieberer, U., Gosch, A., Rajmil, L., Erhart, M., Bruil, J., Power, M., Duer, W.,

Ravens-Sieberer, U., Erhart, M., Rajmil, L., Herdman, M., Auquier, P., Bruil, J., Power, M.,

Remschmidt, H. & Belfer, M. (2005). Mental health care for children and adolescents

Rigby, M. J., Köhler, L. I., Blair, M. E. & Metchler, R. (2003). Child Health Indicators for

Rogers, C. (1951). *Client-centered therapy: Its current practice, implications and theory*, Constable,

Ryff, C.D. (1989). Happiness is everything, or is it? Explorations on the meaning of

Samdal, O., Nutbeam, D., Wold, B. & Kannas, L. (1998). Achieving health and educational

worldwide: a review. *World Psychiatry*, 4, 3, pp. 147-153

*Social Psychology*, 45, pp. 513–523, 0022-3514

*Präventivmedizin*, 46, pp. 297-302, 1420-911X

Pabst Science Publisher, Lengerich

*Health*, 18, pp. 294-299, 1101-1262

10, pp. 1487-1500

46, 1464-360X

1081, 0022-3514

1-84119-840-4, London

92 890 4288 8, Copenhagen

353-364, 1473-7167

M., Rajmil, L., & European KIDSCREEN Group (2001). Quality of Life in children and adolescents – a European public health perspective. *Sozial- und* 

M., Abel, T., Czemy, L., Mazur, J., Czimbalmos, A., Tountas, Y., Hagquist, C., Kilroe, J. & Kidscreen Group Europe (2005). The KIDSCREEN-52 Quality of life measure for children and adolescents: development and first results from a European survey. *Expert Review in Pharmacoeconomics & Outcomes Research*, 5, pp.

*questionnaires—Quality of life questionnaires for children and adolescents—Handbook*,

inequalities in mental health among adolescents in Europe, In: *Social cohesion for mental well-being among adolescents*, pp. 26-42, WHO Regional Office for Europe, 978

Auquier, P., Cloetta, B., Czemy, L., Mazur, J., Czimbalmos, A., Tountas, Y., Hagquist, C., Kilroe, J., & KIDSCREEN Group (2008b). The KIDSCREEN-52 Quality of Life Measure for Children and Adolescents: Psychometric Results from a Cross-Cultural Survey in 13 European Countries. *Value in Health*, 11, 4, pp. 645-658 Ravens-Sieberer, U., Erhart, M., Torsheim, T., Hetland, J., Freeman, J., Danielson, M.,

Thomas, C. and The HBSC Positive Health Group (2008c). An international scoring system for self-reported health complaints in adolescents. *European Journal of Public* 

Duer, W., Abel, T., Czemy, L., Mazur, J., Czimbalmos, A., Tountas, Y., Hagquist, C., Kilroe, J. & European KIDSCREEN Group (2010). Reliability, construct and criterion validity of the KIDSCREEN-10 score: a short measure for children and adolescents' well-being and health-related quality of life. *Quality of Life Research*, 19,

Europe. A priority for a caring society. *European Journal of Public Health*, 1, 1, pp. 38-

psychological well-being. *Journal of Personality and Social Psychology*, 57, pp. 1069–

goals through schools: A study of the importance of school climate and students' satisfaction with school. *Health Education Research*, 13, 3, pp. 383-397, 1465-3648 Schwarz, N. & Clore G.L. (1983). Mood, misattribution, and judgments of well-being:

Informative and directive functions of affective states. *Journal of Personality and* 

Institute for Public Health and the Environment, Bilthoven, 31.10.2011, Available from: <

http://rivm.openrepository.com/rivm/bitstream/10029/7294/1/271558006.pdf>


Adult%20mental%20health%20indicators%20-%20final%20report.pdf


Lehtinen, V. Ozamiz, A., Underwood, L. & Weiss, M. (2005). The Intrinsic Value of Mental

Lippmann, L.H., Moore, K.A., McIntosh, H. (2009). *Positive indicators of child well-being: a* 

Maher, I. & Waters, E. (2005). Indicators of Positive Mental Health for Children, In:

Mikkelsson, M., Salminen, J. & Kautiainen, H. (1997). Non-specific muskuloskeletal pain in preadolescents: prevalence and 1-year persistence. *Pain*, 73, pp. 29-35, 0304-3959 Morrow, V. & Mayall, B. (2010). Measuring Children's Well-Being: Some Problems and

Mortimer, J. & Larson, R. (2002). *The Changing Adolescent Experience: Societal trends and the* 

Murphy, H.B.M. (1978). The meaning of symptom-checklist scores in mental health surveys: a testing of multiple hypotheses. *Social Science & Medicine*, 12, pp. 67-75, 0277-9536

Nosikov, A. & Gudex, C. (2003). EUROHIS: Developing Common Instruments for Health

Palfrey, J. S., Tonniges, T.F., Green, M. & Richmond, J. (2005). Introduction: Addressing the

Parkinson, J. (2007). *Establishing a core set of national, sustainable mental health indicators for* 

Pavot, W. G. & Diener, E. (1993). Review of the Satisfaction with Life Scale. *Psychological* 

Rask, K., Asted-Kurki, P., Paavilainen, E. & Laippala, P. (2003). Adolescent subjective well-

Adult%20mental%20health%20indicators%20-%20final%20report.pdf Parkinson, J. (2009). *Children and Young People's Mental Health Indicators: Background Briefing*,

Millennial Morbidity − The Context of Community Pediatrics. *Pediatrics*, 115,

*adults in Scotland: Final report*, Public Health Adviser and NHS Health Scotland,

Public Health Observatory and NHS Health Scotland, Edinburgh, 31.10.2011, Available from: <http://www.healthscotland.com/uploads/documents/9694- C&YP%20Mental%20Health%20Indicators%20Background%20Briefing%20-

being and family dynamics. *Scandinavian Journal of Caring Sciences*, 17, pp. 129–138,

*Transition to Adulthood*, Cambridge University Press, Cambridge

Neisser, U. (1967). *Cognitive Psychology*, Appleton-Century-Crofts, New York

Surveys. *Biomedical and Health Research*, 57, 1586033220

http://www.healthscotland.com/uploads/documents/5798-

Moodie, pp. 159-168, World Health Organization, 92 4 156294 3, Geneva Maslow, A. H. (1968). *Toward a psychology of being* (2th editon), John Wiley & Son,

Spilerman, pp. 5-36, Russell Sage Foundation, New York

History. *Social Indicators Research*, 83, pp. 39-53, 1573-0921

0471293091, New York

pp. 1121-1123, 1098-4275

Edinburgh, 31.10.2011, Available from:

%20May%202009%20Final.pdf>

1471-6712

*Assessment*, 5, pp. 164-172, 1040-3590

167, Springer, 978-1-4419-5920-1, New York

*2009-21*, UNICEF Innocenti Research Centre, 1014-7837, Florence

Institute for Public Health and the Environment, Bilthoven, 31.10.2011, Available from: < http://rivm.openrepository.com/rivm/bitstream/10029/7294/1/271558006.pdf> Land, K. C. (1975). Social indicators models: An overview, In: *Social Indicator Models*, Land &

Health, In: *Promoting Mental Health: Concepts, Emerging Evidence, Practice*, Herrman, Saxena & Moodie, pp. 46-57, World Health Organization, 92 4 156294 3, Geneva Lippman, L. H. (2007). Indicators and Indices of Child Well-Being: A Brief American

*conceptual framework, measures and methodological issues. Innocenti Working Paper No.* 

*Promoting Mental Health: Concepts, Emerging Evidence, Practice*, Herrman, Saxena &

Possibilities, In: *Health Assets in a Global Context*, Morgan, Davies & Ziglio, pp. 145-


**0**

**3**

*Brazil*

**Assessing the Outline Uncertainty**

<sup>1</sup>*Departament of Mathematics, Universidade Federal de Ouro Preto*

<sup>3</sup>*Departament of Statistics, Universidade Federal de Minas Gerais*

The spatial analysis of disease incidence is a fundamental tool in public health monitoring (Lawson et al., 1999). Suppose that a geographic study area is divided into administrative areas, with known populations at risk and observed cases of disease within a certain period of time. An interesting question is the possible existence of spatial anomalies in the study area: are there localized regions within the map for which the relative concentration of cases among the population at risk is significantly higher than would be expected if the cases were distributed at random? Such anomalies, known as *spatial clusters*, are inherently difficult to delineate, for several reasons (Cancado et al., 2010; Lawson, 2009). Due to the stochastic nature of the number of observed cases of disease, the uncertainty may be elevated in the disease rate estimation for aggregated area maps, especially for small population areas. Thus the most likely disease cluster produced by any given method for the detection and inference of spatial clusters (like SaTScan (Kulldorff, 1999) or any other irregularly shaped scan) is subject to a lot of variation. If it is found to be statistically significant, what could be said of the external areas adjacent to the cluster? Do we have enough information to exclude them from a health

A criterion was proposed (Goovaerts, 2006) to measure the uncertainty of each area being part of a possible localized anomaly in the map, finding error bounds for the delineation of spatial clusters in maps of areas with known populations and observed number of cases. A given map with the vector of real data (the number of observed cases for each area) was considered as just one of the possible realizations of the random variable vector with an unknown expected number of cases. In this methodology, *m* Monte Carlo replications were performed, considering that the simulated number of cases for each area is the realization of a random variable with average equal to the observed number of cases of the original map. Then the most likely cluster for each replicated map was detected. Finally, to each area *ai* it was assigned the number of simulations that *ai* was included in a most likely cluster. If an area belonged to the most likely cluster on all the *m* replications, it was colored as black;

**1. Introduction**

program of prevention?

**of Spatial Disease Clusters**

Fernando L. P. Oliveira1, André L. F. Cançado2, Luiz H. Duczmal3 and Anderson R. Duarte<sup>1</sup>

<sup>2</sup>*Departament of Statistics, Universidade de Brasília*


<http://www.healthscotland.com/uploads/documents/10772-


Views%20of%20C&YP%20on%20what%20impacts%20on%20their%20mental%20h ealth%20-%20Final%20report.pdf>

### **Assessing the Outline Uncertainty of Spatial Disease Clusters**

Fernando L. P. Oliveira1, André L. F. Cançado2,

Luiz H. Duczmal3 and Anderson R. Duarte<sup>1</sup> <sup>1</sup>*Departament of Mathematics, Universidade Federal de Ouro Preto* <sup>2</sup>*Departament of Statistics, Universidade de Brasília* <sup>3</sup>*Departament of Statistics, Universidade Federal de Minas Gerais Brazil*

#### **1. Introduction**

50 Public Health – Methodology, Environmental and Systems Issues

Schwarz, N. & Strack, F. (1999). Reports of Subjective Well-Being: Judgmental Processes and

Sheldon, E.B. & Moore, W.E. (Eds.) (1968). *Indicators of Social Change: Concepts and* 

Shucksmith, J., Spratt, J., Philip, K. & McNaughton, R. (2009). *A critical review of the literature* 

Starfield, B., Katz, H., Gabriel, A., Livingston, G., Benson, P., Hankin, J., Horn, S.,

Suldo, S. M., Riley, K. N. & Shaffer, E. J. (2006). Academic Correlates of Children

Wade, T. J. & Vingilis, E. (1999). The development of self-rated health during adolescence:

Waterman, A. S. (1993). Two Conceptions of Happiness: Contrasts of Personal

Westerhof, G., J. & Keyes, C. L. M. (2010). Mental Illness and Mental Health: The Two

Wilkinson, R.B. & Walford, W. (1998). The measurement of adolescent psychological health: One or two dimensions? *Journal of Youth and Adolescence*, 27, pp. 443-455, 1573-6601

World Health Organization (1946). Constitution of the World Health Organization, *Proceedings of International Health Conference*, New York, June-July 1964 World Health Organization (2001). *The world health report 2001 - Mental Health: New* 

World Health Organization (2005). *Promoting mental health: Concepts, emerging evidence,* 

World Health Organization (2006). *Addressing the socioeconomic determinants of healthy eating* 

Zubrick, S. R. & Kovess-Masfety, V. (2005). Indicators of Mental Health, In: *Promoting Mental* 

*habits and physical activity levels among adolescents: Report from the 2006 HBSC/WHO* 

*Health: Concepts, Emerging Evidence, Practice*, Herrman, Saxena & Moodie, pp. 146-

*Measurements*, Russell Sage Foundation, New York

Health Scotland, Edinburgh, 31.10.2011, Available from: <http://www.healthscotland.com/uploads/documents/10772-

Sigerist, H.E. (1941). *Medicine and Human Welfare*, Yale University Press, New Haven

UNICEF (1979). *The State of the World's Children*, Oxford University Press, New York

0871544237, New York

ealth%20-%20Final%20report.pdf>

*Journal of Medicine*, 310, pp. 824-829

UN General Assembly (1989). *Convention on the Rights of the Child*

*Revue Canadienne De Sante Publique*, 90, 2, pp. 90-94

*Understanding, New Hope*, WHO, 92 4 156201 3, Geneva

*Forum*, WHO Regional Office for Europe, Copenhagen

166, World Health Organization, 92 4 156294 3, Geneva

*practice*, WHO, 92 4 156294 3, Geneva

*Social Psychology*, 46, 4, pp. 678-691, 0022-3514

1461-7374

1573-3440, 1573-3440

Their Methodological Implications, In: *Well-Being: The Foundations of Hedonic Psychology*, Kahneman, Diener & Schwarz, pp. 61-84, Russell Sage Foundation, 7

*on children and young people's views of the factors that influence their mental health*, NHS

Views%20of%20C&YP%20on%20what%20impacts%20on%20their%20mental%20h

Steinwachs, D. (1984). Morbidity in childhood: a longitudinal view. *New England* 

Adolescents' Life Satisfaction. *School Psychology International*, 27, 5, pp. 567-582,

An exploration of inter- and intra-cohort effect. *Canadian Journal of Public Health-*

Expressiveness (Eudaimonia) and Hedonic Enjoyment. *Journal of Personality and* 

Continua Model Across the Lifespan. *Journal of Adult Development*, 17, pp. 110-119,

The spatial analysis of disease incidence is a fundamental tool in public health monitoring (Lawson et al., 1999). Suppose that a geographic study area is divided into administrative areas, with known populations at risk and observed cases of disease within a certain period of time. An interesting question is the possible existence of spatial anomalies in the study area: are there localized regions within the map for which the relative concentration of cases among the population at risk is significantly higher than would be expected if the cases were distributed at random? Such anomalies, known as *spatial clusters*, are inherently difficult to delineate, for several reasons (Cancado et al., 2010; Lawson, 2009). Due to the stochastic nature of the number of observed cases of disease, the uncertainty may be elevated in the disease rate estimation for aggregated area maps, especially for small population areas. Thus the most likely disease cluster produced by any given method for the detection and inference of spatial clusters (like SaTScan (Kulldorff, 1999) or any other irregularly shaped scan) is subject to a lot of variation. If it is found to be statistically significant, what could be said of the external areas adjacent to the cluster? Do we have enough information to exclude them from a health program of prevention?

A criterion was proposed (Goovaerts, 2006) to measure the uncertainty of each area being part of a possible localized anomaly in the map, finding error bounds for the delineation of spatial clusters in maps of areas with known populations and observed number of cases. A given map with the vector of real data (the number of observed cases for each area) was considered as just one of the possible realizations of the random variable vector with an unknown expected number of cases. In this methodology, *m* Monte Carlo replications were performed, considering that the simulated number of cases for each area is the realization of a random variable with average equal to the observed number of cases of the original map. Then the most likely cluster for each replicated map was detected. Finally, to each area *ai* it was assigned the number of simulations that *ai* was included in a most likely cluster. If an area belonged to the most likely cluster on all the *m* replications, it was colored as black;

**2. The intensity function**

*f*(*j*) = *LLR*(*j*), *j* = 1, . . . , *m*.

For each area *ai*, let:

for further details.

**3.1 Introduction**

formalize this procedure in the following.

Monte Carlo replication distributing randomly the *C* = ∑*<sup>K</sup>*

*<sup>q</sup>*(*ai*) = <sup>1</sup>

**3. Genetic algorithm for spatial cluster finding**

the number of simulated cases in the area *ai*, *i* = 1, . . . , *K*, where ∑*<sup>K</sup>*

In this section we define a criterion to measure the plausibility of each area being part of a possible localized anomaly in the map. Following Oliveira et al. (2011), instead of finding the most likely cluster in the original map with the observed number of cases for each area, we consider maps where the number of cases are replications of a vector of random variables, whose averages are defined based on the observed number of cases of the original map. We

Assessing the Outline Uncertainty of Spatial Disease Clusters 53

The original map has *ci* observed cases in the area *ai*, *i* = 1, . . . , *K*. Now we construct a

*a*1,..., *aK* according to a multinomial distribution where the probability associated to the area *ai* is *ci*/*C*. Let *V* = (*s*1,...,*sK*) the realization of the multinomial random vector where *si* is

finder algorithm (in our setting we use the circular scan or we use the elliptic scan) now finds the most likely cluster *MLC*<sup>1</sup> with likelihood ratio value *LLR*1. The Monte Carlo procedure above is repeated *m* times, generating a set of *m* likelihood ratio values {*LLR*1,..., *LLRm*} corresponding to the most likely clusters {*MLC*1,..., *MLCm*}. The likelihood ratio values are sorted in increasing order as {*LLR*(1),..., *LLR*(*m*)} for the corresponding most likely clusters found {*MLC*(1),..., *MLC*(*m*)}. We now define the *intensity f unction f* : {1, . . . , *m*} −→ **R** by

*<sup>m</sup>* arg max <sup>1</sup>≤*j*≤*m*,*ai*∈*MLC*(*j*)

If the area *ai* does not belong to any of the sets *MLC*(1),..., *MLC*(*m*) then we set *q*(*ai*) = 0. The value *q*(*ai*) represents the quantile of the highest likelihood ratio among the ranked values of the likelihood ratios of the most likely clusters found in the *m* Monte Carlo replications, which take into account the variability of the number of cases in each area. In this sense, the value *q*(*ai*) may be interpreted as the relative importance of the area *ai* as part of the anomaly of the map, where the value *f*(*ai*) represents the maximum likelihood ratio found for the most likely clusters which contain the area *ai*. This concept gives more information about the anomaly than the clear-cut division between cluster and non-cluster areas, as given by the usual process of finding the most likely cluster in the original map. See Oliveira et al. (2011)

Genetic algorithms (GA's) constitute an important class of optimization methods. Its importance comes from the fact that the GA's are robust algorithms, in the sense that they are able to treat a wide variety of problems. While some optimization methods require certain assumptions about the problem to be solved, without which these methods fail, the GA's do not require any assumption of continuity, convexity, differentiability and unimodality. In fact,

*f*(*j*), *i* = 1, . . . , *K*

*<sup>i</sup>*=<sup>1</sup> *ci* cases among the *K* areas

*<sup>i</sup>*=<sup>1</sup> *si* = *C*. The cluster

otherwise, if it never was part of a most likely cluster, then it was colored as white, with intermediate shades of gray in-between. A Bayesian variant along these lines, to detect and represent spatial clusters, was also proposed recently Neill (2011).

Another approach to represent the uncertainty in the delineation of spatial clusters appeared recently (Oliveira et al., 2011), employing a ranking based scheme known as *intensity function*. That procedure uses the circular spatial scan statistic (Kulldorff, 1999) to find the circularly shaped most likely cluster for each replicated map. The corresponding *m* likelihood values (obtained by means of the *m* Monte Carlo replications) are ranked. For each area *ai* , the maximum likelihood value, obtained among the most likely clusters containing the area *ai*, is determined. Finally, the intensity function associated to each areaŠs ranking of its respective likelihood value among the *m* obtained values is constructed. The latest procedure generally produce less biased results when compared with the two previous schemes.

However, the circular spatial scan has several limitations, which were discussed in the literature (Duczmal et al., 2006; Kulldorff et al., 2006). Particularly, the circular window is not suitable to make the correct delineation of irregularly shaped clusters because it either chooses a proper subset of the true cluster (underestimation) or chooses a large circle containing the cluster as a proper subset (overestimation). One important consequence is the reduction of the power of detection (Duczmal et al., 2006). In order to overcome this limitation, many algorithms were recently proposed to detect irregularly shaped clusters, replacing the circularly shaped window scheme for any strategy of finding irregularly shaped solutions. Usually, the only limitation in shape for those clusters is a connectivity requirement. In this work, we will analyze the utilization of irregularly shaped algorithms for the application of the intensity function (Oliveira et al., 2011), compared to the use of the simple circular scan, which was employed as the standard method. Due to the regular shape of the most likely cluster found, a question was left, at least in part unanswered: do all the areas inside the cluster have the same importance from a practitioner perspective? In this work is proposed an application of the intensity function for irregularly shaped algorithms, thus avoiding a potential problem inherent in the use of the circular spatial scan, which may described as the lack of resolution inside the circular cluster. As a consequence, it may be difficult or impossible to distinguish the relative importance of the areas inside the detected circular cluster. As we shall see, this problem does not occur when using irregularly shaped scans. Besides, the maximum allowed size for the most likely cluster has a large influence in the result of the cluster search (Chen J, 2008).

In this work novel results are presented, applying the multi-objective genetic algorithm scan (Duarte et al., 2010; Duczmal et al., 2008; 2007), adapted for the weighted non-connectivity penalty function (Cancado et al., 2010). Also, by allowing several different maximum sizes for the most likely cluster, the possible anomaly could be identified with greater precision. As will be demonstrated in the following sections, much better delineated cluster maps of the intensity function will be generated, as compared with the previous version using the simpler circular scan. As a consequence, the relative importance of individual regions composing the spatial anomalies may be assessed, and several interesting phenomena related to the geographical distribution of chronic and acute diseases may be visualized.

#### **2. The intensity function**

2 Will-be-set-by-IN-TECH

otherwise, if it never was part of a most likely cluster, then it was colored as white, with intermediate shades of gray in-between. A Bayesian variant along these lines, to detect and

Another approach to represent the uncertainty in the delineation of spatial clusters appeared recently (Oliveira et al., 2011), employing a ranking based scheme known as *intensity function*. That procedure uses the circular spatial scan statistic (Kulldorff, 1999) to find the circularly shaped most likely cluster for each replicated map. The corresponding *m* likelihood values (obtained by means of the *m* Monte Carlo replications) are ranked. For each area *ai* , the maximum likelihood value, obtained among the most likely clusters containing the area *ai*, is determined. Finally, the intensity function associated to each areaŠs ranking of its respective likelihood value among the *m* obtained values is constructed. The latest procedure generally

However, the circular spatial scan has several limitations, which were discussed in the literature (Duczmal et al., 2006; Kulldorff et al., 2006). Particularly, the circular window is not suitable to make the correct delineation of irregularly shaped clusters because it either chooses a proper subset of the true cluster (underestimation) or chooses a large circle containing the cluster as a proper subset (overestimation). One important consequence is the reduction of the power of detection (Duczmal et al., 2006). In order to overcome this limitation, many algorithms were recently proposed to detect irregularly shaped clusters, replacing the circularly shaped window scheme for any strategy of finding irregularly shaped solutions. Usually, the only limitation in shape for those clusters is a connectivity requirement. In this work, we will analyze the utilization of irregularly shaped algorithms for the application of the intensity function (Oliveira et al., 2011), compared to the use of the simple circular scan, which was employed as the standard method. Due to the regular shape of the most likely cluster found, a question was left, at least in part unanswered: do all the areas inside the cluster have the same importance from a practitioner perspective? In this work is proposed an application of the intensity function for irregularly shaped algorithms, thus avoiding a potential problem inherent in the use of the circular spatial scan, which may described as the lack of resolution inside the circular cluster. As a consequence, it may be difficult or impossible to distinguish the relative importance of the areas inside the detected circular cluster. As we shall see, this problem does not occur when using irregularly shaped scans. Besides, the maximum allowed size for the most likely cluster has a large influence in the result of the

In this work novel results are presented, applying the multi-objective genetic algorithm scan (Duarte et al., 2010; Duczmal et al., 2008; 2007), adapted for the weighted non-connectivity penalty function (Cancado et al., 2010). Also, by allowing several different maximum sizes for the most likely cluster, the possible anomaly could be identified with greater precision. As will be demonstrated in the following sections, much better delineated cluster maps of the intensity function will be generated, as compared with the previous version using the simpler circular scan. As a consequence, the relative importance of individual regions composing the spatial anomalies may be assessed, and several interesting phenomena related to the

geographical distribution of chronic and acute diseases may be visualized.

represent spatial clusters, was also proposed recently Neill (2011).

produce less biased results when compared with the two previous schemes.

cluster search (Chen J, 2008).

In this section we define a criterion to measure the plausibility of each area being part of a possible localized anomaly in the map. Following Oliveira et al. (2011), instead of finding the most likely cluster in the original map with the observed number of cases for each area, we consider maps where the number of cases are replications of a vector of random variables, whose averages are defined based on the observed number of cases of the original map. We formalize this procedure in the following.

The original map has *ci* observed cases in the area *ai*, *i* = 1, . . . , *K*. Now we construct a Monte Carlo replication distributing randomly the *C* = ∑*<sup>K</sup> <sup>i</sup>*=<sup>1</sup> *ci* cases among the *K* areas *a*1,..., *aK* according to a multinomial distribution where the probability associated to the area *ai* is *ci*/*C*. Let *V* = (*s*1,...,*sK*) the realization of the multinomial random vector where *si* is the number of simulated cases in the area *ai*, *i* = 1, . . . , *K*, where ∑*<sup>K</sup> <sup>i</sup>*=<sup>1</sup> *si* = *C*. The cluster finder algorithm (in our setting we use the circular scan or we use the elliptic scan) now finds the most likely cluster *MLC*<sup>1</sup> with likelihood ratio value *LLR*1. The Monte Carlo procedure above is repeated *m* times, generating a set of *m* likelihood ratio values {*LLR*1,..., *LLRm*} corresponding to the most likely clusters {*MLC*1,..., *MLCm*}. The likelihood ratio values are sorted in increasing order as {*LLR*(1),..., *LLR*(*m*)} for the corresponding most likely clusters found {*MLC*(1),..., *MLC*(*m*)}. We now define the *intensity f unction f* : {1, . . . , *m*} −→ **R** by *f*(*j*) = *LLR*(*j*), *j* = 1, . . . , *m*.

For each area *ai*, let:

$$q(a\_i) = \frac{1}{m} \quad \text{arg} \quad \max\_{1 \le j \le m, a\_i \in MLC\_{(j)}} f(j), i = 1, \dots, K$$

If the area *ai* does not belong to any of the sets *MLC*(1),..., *MLC*(*m*) then we set *q*(*ai*) = 0. The value *q*(*ai*) represents the quantile of the highest likelihood ratio among the ranked values of the likelihood ratios of the most likely clusters found in the *m* Monte Carlo replications, which take into account the variability of the number of cases in each area. In this sense, the value *q*(*ai*) may be interpreted as the relative importance of the area *ai* as part of the anomaly of the map, where the value *f*(*ai*) represents the maximum likelihood ratio found for the most likely clusters which contain the area *ai*. This concept gives more information about the anomaly than the clear-cut division between cluster and non-cluster areas, as given by the usual process of finding the most likely cluster in the original map. See Oliveira et al. (2011) for further details.

#### **3. Genetic algorithm for spatial cluster finding**

#### **3.1 Introduction**

Genetic algorithms (GA's) constitute an important class of optimization methods. Its importance comes from the fact that the GA's are robust algorithms, in the sense that they are able to treat a wide variety of problems. While some optimization methods require certain assumptions about the problem to be solved, without which these methods fail, the GA's do not require any assumption of continuity, convexity, differentiability and unimodality. In fact,

**3.2.1 Generating the initial population**

**3.2.2 The selection operator**

**3.2.3 The crossover operator**

gray).

**3.2.4 The mutation operator**

region, provided the cluster's connectivity.

grows until it reaches a maximum size set by the user.

is repeated *n* times, thus producing a set of *n* selected individuals.

operator that ensures that every generated solution is feasible.

The initial population is generated by a greedy procedure. Given a map with *n* regions we generate a population of *n* individuals, each of which is generated from one region of the map. So, starting with a region, the solution incorporates more regions, choosing at each step to aggregate, among all the regions that are neighbors of some region in the actual solution, the one that makes the *LLR* value to increase the most when added to the solution. The individual

Assessing the Outline Uncertainty of Spatial Disease Clusters 55

Each solution is evaluated by means of its *LLR* value and this is the adaptation indicator: higher *LLR*-valued individuals are more adapted. The selection operator will then give more chances to the more adapted individuals to generate offspring. This is done through a mechanism called binary tournment. For each tournment two individuals are chosen from the current population, each individual having the same probability of being chosed. Then we compare the two solutions and the one with higher *LLR* value is selected. This procedure

Now, selected individuals have the chance to trasmit their genetic information to new individuals by generating offspring. Crossover is applied to pairs of parents randomly chosen from the list of selected individuals. The offspring is generated in a way that the children inherit characteristis from both parents. In addition, it is well known that GA's particularly designed for a specific problem perform much better than multiple-purpose generic GA's. Thus, it is highly recommended that operators are designed so that they can take advantage of the intrinsic structure of the problem. For example, in our case we would discard any disconnected cluster candidate because it is an infeasible solution. While a generic crossover operator could, most of the time, generate infeasible clusters, we chose to use a crossover

The crossover operator described by Duczmal et al. (2007) presents all these features, being capable of efficiently generating feasible offspring having characteristics of both parents. The operator is implemented in sucha a way that it is only possible to perform a crossover between two parents if they share a nonempty intersection. Once this condition is verified, the offspring is generated. Figure 1 shows an example with two parents (*A* and *B*) and the generated offspring 1-5. Note that the offspring constitutes a "path" from one parent to another, with child 1 being more like parent *A*, while child 5 is almost like parent *B*. In the middle of the figure we can see parents inside the map with the two intersection regions (in

Each individual generated by the crossover process has a probability of suffering a mutation. Mutation consistis in introducing a random perturbation in the genetic code of the individual. In our case, the mutation consists of adding to or removing from the cluster a randomly chosen

the only assumption a GA requires is that the function to be optimized presents a "global trend" that can be captured or learned by the algorithm. Of course, not making any kind of assumption and, consequently, not using these characteristics in favor, GA'a tend to be computationally intensive, so its usage is justified for difficult problems.

When looking for a most likely cluster, one faces a challenging otimization problem: given a set *R* of *n* regions in a map, some of which are neighbors, find the connected subset *S* of *R* that assumes the highest *LLR* value. By "connected" we mean that, starting from any region in *S* there's always a path to any other region of *S* formed by a chain of neighbors, all of them inside *S*.

Solving this problem exactly means that we would have to look at all of the 2*<sup>n</sup>* subsets of *R*, test which ones are connected, evaluate their *LLR* values and pick up the most likely one. For maps with just a few dozens of regions this problem is already intractable. So we need another strategy to find such optimal solution. GA's showed to be a good alternative for the spatial cluster finding problem (Duczmal et al., 2008; 2007).

#### **3.2 The genetic algorithm**

The natural evolution of living beings can be compared to an optimization process. In fact, if individuals who are best adapted survive - in the sense of transmitting their genetic information - while less adapted individuals tend to disappear, it is expected that after a number of generations the population is composed of individuals who are generally better adapted than those of earlier generations. This is also the idea behind a genetic algorithm. It tries to simulate the mechanisms of random variation and selection of adaptive evolution. The mechanisms (or genetic operators) that form the basis of a genetic algorithm are:


In this context, an individual is a candidate-solution to the optimization problem and a population is a set of individuals. For the spatial cluster detection problem a solution - or individual - is a set of connected regions of the map (the candidate cluster). So, the population is a set of lists, each list being a set of regions that form the solution.

Starting with an initial population the GA forms a sequence of generations. At each iteration it applies the selection, crossover and mutation operators to the current population, generating a new population. The GA used in this work was primarily described in Duczmal et al. (2007) and its biobjective versions were used by Duczmal et al. (2008), Cancado et al. (2010) and Duarte et al. (2010).

#### **3.2.1 Generating the initial population**

4 Will-be-set-by-IN-TECH

the only assumption a GA requires is that the function to be optimized presents a "global trend" that can be captured or learned by the algorithm. Of course, not making any kind of assumption and, consequently, not using these characteristics in favor, GA'a tend to be

When looking for a most likely cluster, one faces a challenging otimization problem: given a set *R* of *n* regions in a map, some of which are neighbors, find the connected subset *S* of *R* that assumes the highest *LLR* value. By "connected" we mean that, starting from any region in *S* there's always a path to any other region of *S* formed by a chain of neighbors, all of them

Solving this problem exactly means that we would have to look at all of the 2*<sup>n</sup>* subsets of *R*, test which ones are connected, evaluate their *LLR* values and pick up the most likely one. For maps with just a few dozens of regions this problem is already intractable. So we need another strategy to find such optimal solution. GA's showed to be a good alternative for the

The natural evolution of living beings can be compared to an optimization process. In fact, if individuals who are best adapted survive - in the sense of transmitting their genetic information - while less adapted individuals tend to disappear, it is expected that after a number of generations the population is composed of individuals who are generally better adapted than those of earlier generations. This is also the idea behind a genetic algorithm. It tries to simulate the mechanisms of random variation and selection of adaptive evolution.

• crossover operator, which combines the information of two or more individuals - called

• mutation operator, which applies a random perturbation to the information of an

• selection operator, which defines the probability of an individual to transmit its genetic

In this context, an individual is a candidate-solution to the optimization problem and a population is a set of individuals. For the spatial cluster detection problem a solution - or individual - is a set of connected regions of the map (the candidate cluster). So, the population

Starting with an initial population the GA forms a sequence of generations. At each iteration it applies the selection, crossover and mutation operators to the current population, generating a new population. The GA used in this work was primarily described in Duczmal et al. (2007) and its biobjective versions were used by Duczmal et al. (2008), Cancado et al. (2010) and

The mechanisms (or genetic operators) that form the basis of a genetic algorithm are:

computationally intensive, so its usage is justified for difficult problems.

spatial cluster finding problem (Duczmal et al., 2008; 2007).

parents - generating new individuals - called children;

information (generate children) based on its adaptation level.

is a set of lists, each list being a set of regions that form the solution.

individual, generating a new one;

Duarte et al. (2010).

inside *S*.

**3.2 The genetic algorithm**

The initial population is generated by a greedy procedure. Given a map with *n* regions we generate a population of *n* individuals, each of which is generated from one region of the map. So, starting with a region, the solution incorporates more regions, choosing at each step to aggregate, among all the regions that are neighbors of some region in the actual solution, the one that makes the *LLR* value to increase the most when added to the solution. The individual grows until it reaches a maximum size set by the user.

#### **3.2.2 The selection operator**

Each solution is evaluated by means of its *LLR* value and this is the adaptation indicator: higher *LLR*-valued individuals are more adapted. The selection operator will then give more chances to the more adapted individuals to generate offspring. This is done through a mechanism called binary tournment. For each tournment two individuals are chosen from the current population, each individual having the same probability of being chosed. Then we compare the two solutions and the one with higher *LLR* value is selected. This procedure is repeated *n* times, thus producing a set of *n* selected individuals.

#### **3.2.3 The crossover operator**

Now, selected individuals have the chance to trasmit their genetic information to new individuals by generating offspring. Crossover is applied to pairs of parents randomly chosen from the list of selected individuals. The offspring is generated in a way that the children inherit characteristis from both parents. In addition, it is well known that GA's particularly designed for a specific problem perform much better than multiple-purpose generic GA's. Thus, it is highly recommended that operators are designed so that they can take advantage of the intrinsic structure of the problem. For example, in our case we would discard any disconnected cluster candidate because it is an infeasible solution. While a generic crossover operator could, most of the time, generate infeasible clusters, we chose to use a crossover operator that ensures that every generated solution is feasible.

The crossover operator described by Duczmal et al. (2007) presents all these features, being capable of efficiently generating feasible offspring having characteristics of both parents. The operator is implemented in sucha a way that it is only possible to perform a crossover between two parents if they share a nonempty intersection. Once this condition is verified, the offspring is generated. Figure 1 shows an example with two parents (*A* and *B*) and the generated offspring 1-5. Note that the offspring constitutes a "path" from one parent to another, with child 1 being more like parent *A*, while child 5 is almost like parent *B*. In the middle of the figure we can see parents inside the map with the two intersection regions (in gray).

#### **3.2.4 The mutation operator**

Each individual generated by the crossover process has a probability of suffering a mutation. Mutation consistis in introducing a random perturbation in the genetic code of the individual. In our case, the mutation consists of adding to or removing from the cluster a randomly chosen region, provided the cluster's connectivity.

f 1

Assessing the Outline Uncertainty of Spatial Disease Clusters 57

Fig. 2. A set of solutions in the objectives space: dominated solutions (×) and Pareto-set (•).

Once the most likely cluster is identified, we want to check its significace. This will allow the practitioner to verify if the cluster can be considered a disease outbreak or if the disease cases are randomly spreaded over the map. Since the distribution of LLR under *H*<sup>0</sup> is not known we must perform a Monte Carlo simulation. For the mono-objective case, the LLR value is calculated for the most likely cluster in each Monte Carlo replication under *H*<sup>0</sup> and the *p*-value is computed comparing the value of LLR for the observed data and the empirical

For the biobjective case, we consider the attainment function (da Fonseca et al., 2001; Fonseca et al., 2005), as also used by Cancado et al. (2010). A single run of the biobjective GA would produce a Pareto-set, defining two distinct regions in the objectives space: points that are dominated by the Pareto-set and points that are not dominated by it. Then, for inference purposes we can consider, for each point of the Pareto-set obtained for the observed data, the proportion times that the point is dominated by the Pareto-sets under *H*0. This is exactly the

One of the possible penalties that takes in account the cluster geometric shape is the called compactness geometric penalty function. This penalty function introduced in Duczmal et al.

f

*p*-value for that point.

**3.5 The geometric penalty function**

2

**3.4 Inference and the attainment function**

distribution obtained through the Monte Carlo procedure.

Fig. 1. A splitted vision of parents *A* and *B* (left), parents *A* and *B* inside the map (middle) and offspring (right).

#### **3.3 The biobjective genetic algorithm**

Many times one wants to find a solution that simultaneously optimizes two or more functionals. For example, a costumer may want to buy a car which is powerful and cheap. Of course, it is very unlikely that, say, the most powerful is also the cheaper car, because these two criteria are conflicting. Based on these two criteria, a whole set of cars can be of interest for this costumer: powerful (but expensive) cars and cheap (but underpowered) cars. Of course, a costumer (again, based on just these criteria) will reject cars that cost too much and are low powered.

Following the same reasoning, a biobjective GA was proposed (Cancado et al., 2010; Duarte et al., 2010; Duczmal et al., 2008) to deal with the problem of spatial cluster detection. Using the LLR as the only objective to be maximized would lead to geographically meaningless tree-shaped solutions and it is necessary to consider some shape regularity measure, such as geometric compactness (Duczmal et al., 2008) or topological corrections (Cancado et al., 2010; Yiannakoulias et al., 2007). This regularity measure works as a second objective to be maximized. As in the power/price car example, LLR and regularity are conflicting objectives, because high values of LLR are associated to very irregular clusters, while regular solutions tend to

Instead of an optimal solution, a biobjective maximization problem will lead, in general, to a set of optimal solutions, called the Pareto-set. This set is composed by all solutions that are not worse than any other solution in both objectives simulteanously. Such solution is called nondominated solution. Because GA's work with a population of candidate-solutions they can find the Pareto-set in one execution with almost the same effort spent by its mono-objective version. Figure 2 illustrates a set of solutions in the objectives space. Nondominated solutions are indicated by black dots.

6 Will-be-set-by-IN-TECH

Fig. 1. A splitted vision of parents *A* and *B* (left), parents *A* and *B* inside the map (middle)

Many times one wants to find a solution that simultaneously optimizes two or more functionals. For example, a costumer may want to buy a car which is powerful and cheap. Of course, it is very unlikely that, say, the most powerful is also the cheaper car, because these two criteria are conflicting. Based on these two criteria, a whole set of cars can be of interest for this costumer: powerful (but expensive) cars and cheap (but underpowered) cars. Of course, a costumer (again, based on just these criteria) will reject cars that cost too much and are low

Following the same reasoning, a biobjective GA was proposed (Cancado et al., 2010; Duarte et al., 2010; Duczmal et al., 2008) to deal with the problem of spatial cluster detection. Using the LLR as the only objective to be maximized would lead to geographically meaningless tree-shaped solutions and it is necessary to consider some shape regularity measure, such as geometric compactness (Duczmal et al., 2008) or topological corrections (Cancado et al., 2010; Yiannakoulias et al., 2007). This regularity measure works as a second objective to be maximized. As in the power/price car example, LLR and regularity are conflicting objectives, because high values of LLR are associated to very irregular clusters, while regular solutions

Instead of an optimal solution, a biobjective maximization problem will lead, in general, to a set of optimal solutions, called the Pareto-set. This set is composed by all solutions that are not worse than any other solution in both objectives simulteanously. Such solution is called nondominated solution. Because GA's work with a population of candidate-solutions they can find the Pareto-set in one execution with almost the same effort spent by its mono-objective version. Figure 2 illustrates a set of solutions in the objectives space.

and offspring (right).

powered.

tend to

**3.3 The biobjective genetic algorithm**

Nondominated solutions are indicated by black dots.

Fig. 2. A set of solutions in the objectives space: dominated solutions (×) and Pareto-set (•).

#### **3.4 Inference and the attainment function**

Once the most likely cluster is identified, we want to check its significace. This will allow the practitioner to verify if the cluster can be considered a disease outbreak or if the disease cases are randomly spreaded over the map. Since the distribution of LLR under *H*<sup>0</sup> is not known we must perform a Monte Carlo simulation. For the mono-objective case, the LLR value is calculated for the most likely cluster in each Monte Carlo replication under *H*<sup>0</sup> and the *p*-value is computed comparing the value of LLR for the observed data and the empirical distribution obtained through the Monte Carlo procedure.

For the biobjective case, we consider the attainment function (da Fonseca et al., 2001; Fonseca et al., 2005), as also used by Cancado et al. (2010). A single run of the biobjective GA would produce a Pareto-set, defining two distinct regions in the objectives space: points that are dominated by the Pareto-set and points that are not dominated by it. Then, for inference purposes we can consider, for each point of the Pareto-set obtained for the observed data, the proportion times that the point is dominated by the Pareto-sets under *H*0. This is exactly the *p*-value for that point.

#### **3.5 The geometric penalty function**

One of the possible penalties that takes in account the cluster geometric shape is the called compactness geometric penalty function. This penalty function introduced in Duczmal et al.

**4. Results and discussion**

treatment.

**4.1 Real data case studies**

Epidemiological surveillance is essential to monitoring possible changes in the geographical distribution pattern of both acute and chronic diseases. To illustrate the techniques presented in this chapter, four diseases (dengue fever, tuberculosis, diabetes and hypertension) are analyzed. Those four diseases are currently among the most serious health threats to the Brazilian population. Our studies were concentrated in the Minas Gerais state in southeast Brazil, with 853 municipalities and total population of 19,597,330 (2010 census). For each disease, only the specific population at risk at each municipality was considered. Population data was available at Instituto Brasileiro de Geografia e Estatística (www.ibge.gov.br), and disease data was obtained through DATASUS, the Brazilian Ministry of Health's central data system (www.datasus.gov.br). Dengue fever data was collected by SINAN/MS system from the Brazilian Ministry of Health (www.sinam.org.br). During the period 2007-2010, 349.005 cases were registered, and the population at risk was assumed to be the total population of the 2010 census. Tuberculosis disease cases, using SINAN/MS data, were considered for the 2001-2010 period for the following age groups (years): 15-19, 20-39 and 40-59, making a total of 41,824 cases for a population at risk of 12,892,744. Hypertension data was obtained through the Hiperdia program of Brazilian Ministry of Health from January 2002 to January 2011. Data was available to the following age groups (years): 50-59, 60-69, 70-79 and 80+, with a total population at risk of 4,365,352 individuals and 941,710 cases. Diabetes types 1 and 2 data were also obtained through the Hiperdia program from January 2002 to May 2011. The age groups were: 45-49, 50-54, 55-59, 60-64, 65-69, 70-74, 75-79 and 80+ years, with 28.039 cases. Diabetes mellitus and hypertension are considered chronic diseases and their control and treatment depend on the individuals behavior in relation to their lifestyle: healthy eating, physical activity, and weight control. These diseases are responsible for high rates of hospital expenses, so the investment in shares of health promotion and prevention is potentially very cost effective. The importance of dengue in our study lies in the fact that it is an infectious disease and even in regions with previous low incidence rates are subject to outbreaks. This disease is subject to major public health campaigns in Brazil. The report on the epidemiology of dengue published by the Secretariat of Health Surveillance in 2010 indicates Minas Gerais state as one of the critical states in need of stricter monitoring. Hypertension and diabetes are very common chronic diseases, and hypertension is a major public health problem in Brazil. Tuberculosis has become relevant to this study due to its high incidence, and its early diagnosis and effective treatment are of great importance to public health. The biggest challenge for public health professionals has been to promote action to encourage compliance and continuity of care, since many individuals do not join or do not follow the prescribed

Assessing the Outline Uncertainty of Spatial Disease Clusters 59

In what follows, we present the obtained sets of intensity function maps for dengue fever, tuberculosis, diabetes and hypertension in Minas Gerais state (Figures 3, 4, 5 and 6, respectively). North is up for all the maps. For each disease set we present six maps: (a) the quantiles of population at risk, (b) the quantiles of disease rate and the intensity function maps based on the genetic multi-objective algorithm for maximum clusters of sizes 10, 20, 30

(2006) aims to penalize zones in the map that have very irregular shape. The compactness geometric function *k*(*z*) of a zone *z* is given by the area of *z* divided by the area of a circle with the same perimeter as the convex hull of *z*. The compactness geometric function takes values between zero and one, the circle has the most compact shape (*k*(*z*) = 1). Compactness depends on the shape of the zone, but not on its size. The expression for *k*(*z*) is given by:

$$k(z) = \frac{4\pi A(z)}{H(z)^2} \tag{1}$$

where *A*(*z*) is the area of the zone *z* and *H*(*z*) the perimeter of the convex hull of *z*. The compactness penalyzed scan statistic is defined as *maxz*∈*Zk*(*z*).*LLR*(*z*).

#### **3.6 The non-connectivity penalty function**

Yiannakoulias et al. (2007) proposed a greedy algorithm to scan the set *Z* of all possible zones *z*. A new penalty function called non-connectivity was proposed. It was based on the ratio of the number of nodes *v*(*z*) to the number of edges *e*(*z*) of the subgraph associated with the zone *z*. The non-connectivity penalty was used as a multiplier for the *LLR*(*z*). The non-connectivity penalty function of a zone *z* is defined by:

$$mc(z) = \frac{e(z)}{\left[\Im\left(v(z) - 2\right)\right]}\tag{2}$$

the expression in the denominator represents the maximum number of edges of a planar graph given its number of vertices. The most penalized zones are the ones whith tree-like associated graphs, meaning that they have a small number of nodes compared with the number of edges. Although there is some similarity between the non-connectivity penalty to the geometric compactness penalty, there is an important difference: the non-connectivity penalty does not rely on the geometric shape of the candidate cluster, which could be an interesting feature when searching for real clusters which are highly irregularly shaped, but present good connectivity properties.

#### **3.7 Evaluation of the candidate solutions**

Differently from the previous procedure employing the circular scan, each run of the multiobjective genetic scan produces a set of several non-dominated solutions.

In the circular scan, the scan statistic value for the most likely cluster was assigned to each area of the solution cluster, and later the maximum value of this quantity was obtained for all the executions. However, in the multiobjective procedure, the scan statistic value will be assigned for each component area of each solution cluster of the non-dominated solution set. In the event that a given area belongs to more than one solution cluster, the largest scan statistic value is assigned to the area. The remaining of the process is identical to the usual procedure using the circular scan, obtaining the maximum value of this quantity for all the executions, and building the intensity function as usual.

#### **4. Results and discussion**

8 Will-be-set-by-IN-TECH

(2006) aims to penalize zones in the map that have very irregular shape. The compactness geometric function *k*(*z*) of a zone *z* is given by the area of *z* divided by the area of a circle with the same perimeter as the convex hull of *z*. The compactness geometric function takes values between zero and one, the circle has the most compact shape (*k*(*z*) = 1). Compactness depends on the shape of the zone, but not on its size. The expression for *k*(*z*) is given by:

*<sup>k</sup>*(*z*) = <sup>4</sup>*πA*(*z*)

where *A*(*z*) is the area of the zone *z* and *H*(*z*) the perimeter of the convex hull of *z*. The

Yiannakoulias et al. (2007) proposed a greedy algorithm to scan the set *Z* of all possible zones *z*. A new penalty function called non-connectivity was proposed. It was based on the ratio of the number of nodes *v*(*z*) to the number of edges *e*(*z*) of the subgraph associated with the zone *z*. The non-connectivity penalty was used as a multiplier for the *LLR*(*z*). The non-connectivity

*nc*(*z*) = *<sup>e</sup>*(*z*)

the expression in the denominator represents the maximum number of edges of a planar graph given its number of vertices. The most penalized zones are the ones whith tree-like associated graphs, meaning that they have a small number of nodes compared with the number of edges. Although there is some similarity between the non-connectivity penalty to the geometric compactness penalty, there is an important difference: the non-connectivity penalty does not rely on the geometric shape of the candidate cluster, which could be an interesting feature when searching for real clusters which are highly irregularly shaped, but

Differently from the previous procedure employing the circular scan, each run of the

In the circular scan, the scan statistic value for the most likely cluster was assigned to each area of the solution cluster, and later the maximum value of this quantity was obtained for all the executions. However, in the multiobjective procedure, the scan statistic value will be assigned for each component area of each solution cluster of the non-dominated solution set. In the event that a given area belongs to more than one solution cluster, the largest scan statistic value is assigned to the area. The remaining of the process is identical to the usual procedure using the circular scan, obtaining the maximum value of this quantity for all the executions,

multiobjective genetic scan produces a set of several non-dominated solutions.

compactness penalyzed scan statistic is defined as *maxz*∈*Zk*(*z*).*LLR*(*z*).

**3.6 The non-connectivity penalty function**

penalty function of a zone *z* is defined by:

present good connectivity properties.

**3.7 Evaluation of the candidate solutions**

and building the intensity function as usual.

*<sup>H</sup>*(*z*)<sup>2</sup> (1)

[<sup>3</sup> (*v*(*z*) <sup>−</sup> <sup>2</sup>)] (2)

Epidemiological surveillance is essential to monitoring possible changes in the geographical distribution pattern of both acute and chronic diseases. To illustrate the techniques presented in this chapter, four diseases (dengue fever, tuberculosis, diabetes and hypertension) are analyzed. Those four diseases are currently among the most serious health threats to the Brazilian population. Our studies were concentrated in the Minas Gerais state in southeast Brazil, with 853 municipalities and total population of 19,597,330 (2010 census). For each disease, only the specific population at risk at each municipality was considered. Population data was available at Instituto Brasileiro de Geografia e Estatística (www.ibge.gov.br), and disease data was obtained through DATASUS, the Brazilian Ministry of Health's central data system (www.datasus.gov.br). Dengue fever data was collected by SINAN/MS system from the Brazilian Ministry of Health (www.sinam.org.br). During the period 2007-2010, 349.005 cases were registered, and the population at risk was assumed to be the total population of the 2010 census. Tuberculosis disease cases, using SINAN/MS data, were considered for the 2001-2010 period for the following age groups (years): 15-19, 20-39 and 40-59, making a total of 41,824 cases for a population at risk of 12,892,744. Hypertension data was obtained through the Hiperdia program of Brazilian Ministry of Health from January 2002 to January 2011. Data was available to the following age groups (years): 50-59, 60-69, 70-79 and 80+, with a total population at risk of 4,365,352 individuals and 941,710 cases. Diabetes types 1 and 2 data were also obtained through the Hiperdia program from January 2002 to May 2011. The age groups were: 45-49, 50-54, 55-59, 60-64, 65-69, 70-74, 75-79 and 80+ years, with 28.039 cases.

Diabetes mellitus and hypertension are considered chronic diseases and their control and treatment depend on the individuals behavior in relation to their lifestyle: healthy eating, physical activity, and weight control. These diseases are responsible for high rates of hospital expenses, so the investment in shares of health promotion and prevention is potentially very cost effective. The importance of dengue in our study lies in the fact that it is an infectious disease and even in regions with previous low incidence rates are subject to outbreaks. This disease is subject to major public health campaigns in Brazil. The report on the epidemiology of dengue published by the Secretariat of Health Surveillance in 2010 indicates Minas Gerais state as one of the critical states in need of stricter monitoring. Hypertension and diabetes are very common chronic diseases, and hypertension is a major public health problem in Brazil. Tuberculosis has become relevant to this study due to its high incidence, and its early diagnosis and effective treatment are of great importance to public health. The biggest challenge for public health professionals has been to promote action to encourage compliance and continuity of care, since many individuals do not join or do not follow the prescribed treatment.

#### **4.1 Real data case studies**

In what follows, we present the obtained sets of intensity function maps for dengue fever, tuberculosis, diabetes and hypertension in Minas Gerais state (Figures 3, 4, 5 and 6, respectively). North is up for all the maps. For each disease set we present six maps: (a) the quantiles of population at risk, (b) the quantiles of disease rate and the intensity function maps based on the genetic multi-objective algorithm for maximum clusters of sizes 10, 20, 30

<=C10 C10 − C30 C30 − C70 C70 − C90 >= C90

=0 0 to 0.92 0.92 to 0.94 >0.94

=0 0 to 0.81 0.81 to 0.83 >0.83

40 (c, d, e and f respectively)

Intensity function quantile

Intensity function quantile

(a)

(c)

(e)

<=C10 C10 − C30 C30 − C70 C70 − C90 >= C90

Assessing the Outline Uncertainty of Spatial Disease Clusters 61

=0 0 to 0.85 0.85 to 0.88 >0.88

=0 0 to 0.69 0.69 to 0.76 >0.76

Fig. 3. Population at risk quantiles (a), dengue fever rates (b), and intensity function maps based on the genetic multi-objective algorithm for maximum clusters of sizes 10, 20, 30 and

Intensity function quantile

Intensity function quantile

(b)

(d)

(f)

and 40 (c, d, e and f respectively). The population at risk was different for each disease in our study.

As can be noted on all four disease sets, the probability that each area belongs to the "'true"' cluster decreases as the maximum cluster size increases from 10 to 40. For instance, in the dengue fever set, the dark brown areas have probability of belonging to the "'true cluster"' greater than 94%, 88%, 83% and 76%, as the maximum cluster size increases from 10, 20, 30 and 40, respectively. It means that the intensity function maps produced with the smaller maximum cluster sizes (10 and 20)indicate inner "'core"' regions within the "'true cluster"'. On the other hand, the intensity function maps produced with the larger maximum cluster sizes (30 and 40)indicate "'borderline"' regions with respect of the "'true cluster"'.

Another important feature is the complexity of the shapes displayed in the sequence of intensity function maps as the maximum cluster size increases.

#### **4.1.1 Dengue fever**

In Figure 3c, the maximum size 10 inner core region of dengue fever includes the municipalities arround the state capital Belo Horizonte urban area (population 4 million) in the central part of the state. The maximum sizes 20 and 30 intensity function maps (Figures 3d and 3e respectively) show the anomaly spreading northward following the São Francisco river basin, a region with elevated humidity and high mosquito incidence. Finally, the larger maximum size 40 anomaly (Figure 3f) spreads along the highway joining the cities of Ipatinga, Valadares and Teofilo Otoni in the eastern part of the state.

#### **4.1.2 Tuberculosis**

In Figure 4c, the maximum size 10 inner core region of tuberculosis includes the predominantly urban area of Belo Horizonte (in the central part of the state) and two weaker urban regions: (i) the highway joining the cities of Ipatinga, Valadares and Teofilo Otoni in the eastern part of the state, and (ii) the areas surrounding the city of Juiz de Fora, the second largest city of the state in the south. As the maximum cluster size increases (Figures 4d, 4e and 4f), the tuberculosis anomaly is reinforced to include the surrounding municipalities, and also the neighbors of the populous Montes Claros city in the northern part of the state.

#### **4.1.3 Diabetes**

In Figure 5c, the maximum size 10 inner core region of diabetes includes the southwest part of the state and the weaker urban region of Valadares city in the east. As the maximum cluster size increases (Figures 5d, 5e and 5f), the diabetes anomaly is reinforced to include the surrounding municipalities.

#### **4.1.4 Hypertension**

In Figure 6c, the maximum size 10 inner core region of hypertension includes several scattered regions in the center and mid southeast parts of the state. As the maximum cluster size increases (Figures 6d, 6e and 6f), the hypertension anomaly is reinforced to include the surrounding municipalities.

10 Will-be-set-by-IN-TECH

and 40 (c, d, e and f respectively). The population at risk was different for each disease in our

As can be noted on all four disease sets, the probability that each area belongs to the "'true"' cluster decreases as the maximum cluster size increases from 10 to 40. For instance, in the dengue fever set, the dark brown areas have probability of belonging to the "'true cluster"' greater than 94%, 88%, 83% and 76%, as the maximum cluster size increases from 10, 20, 30 and 40, respectively. It means that the intensity function maps produced with the smaller maximum cluster sizes (10 and 20)indicate inner "'core"' regions within the "'true cluster"'. On the other hand, the intensity function maps produced with the larger maximum cluster

Another important feature is the complexity of the shapes displayed in the sequence of

In Figure 3c, the maximum size 10 inner core region of dengue fever includes the municipalities arround the state capital Belo Horizonte urban area (population 4 million) in the central part of the state. The maximum sizes 20 and 30 intensity function maps (Figures 3d and 3e respectively) show the anomaly spreading northward following the São Francisco river basin, a region with elevated humidity and high mosquito incidence. Finally, the larger maximum size 40 anomaly (Figure 3f) spreads along the highway joining the cities of Ipatinga,

In Figure 4c, the maximum size 10 inner core region of tuberculosis includes the predominantly urban area of Belo Horizonte (in the central part of the state) and two weaker urban regions: (i) the highway joining the cities of Ipatinga, Valadares and Teofilo Otoni in the eastern part of the state, and (ii) the areas surrounding the city of Juiz de Fora, the second largest city of the state in the south. As the maximum cluster size increases (Figures 4d, 4e and 4f), the tuberculosis anomaly is reinforced to include the surrounding municipalities, and also the neighbors of the populous Montes Claros city in the northern part of the state.

In Figure 5c, the maximum size 10 inner core region of diabetes includes the southwest part of the state and the weaker urban region of Valadares city in the east. As the maximum cluster size increases (Figures 5d, 5e and 5f), the diabetes anomaly is reinforced to include

In Figure 6c, the maximum size 10 inner core region of hypertension includes several scattered regions in the center and mid southeast parts of the state. As the maximum cluster size increases (Figures 6d, 6e and 6f), the hypertension anomaly is reinforced to include the

sizes (30 and 40)indicate "'borderline"' regions with respect of the "'true cluster"'.

intensity function maps as the maximum cluster size increases.

Valadares and Teofilo Otoni in the eastern part of the state.

study.

**4.1.1 Dengue fever**

**4.1.2 Tuberculosis**

**4.1.3 Diabetes**

**4.1.4 Hypertension**

the surrounding municipalities.

surrounding municipalities.

(b)

(d)

Fig. 3. Population at risk quantiles (a), dengue fever rates (b), and intensity function maps based on the genetic multi-objective algorithm for maximum clusters of sizes 10, 20, 30 and 40 (c, d, e and f respectively)

(b)

<=C10 C10 − C30 C30 − C70 C70 − C90 >= C90

=0 0 to 0.86 0.86 to 0.88 >0.88

=0 0 to 0.74 0.74 to 0.83 >0.83

e and f respectively)

Intensity function quantile

Intensity function quantile

(a)

(c)

(e)

<=C10 C10 − C30 C30 − C70 C70 − C90 >= C90

Assessing the Outline Uncertainty of Spatial Disease Clusters 63

=0 0 to 0.78 0.78 to 0.86 >0.86

=0 0 to 0.77 0.77 to 0.79 >0.79

Fig. 5. Population at risk quantiles (a), diabetes rates (b), and intensity function maps based on the genetic multi-objective algorithm for maximum clusters of sizes 10, 20, 30 and 40 (c, d,

Intensity function quantile

Intensity function quantile

(b)

(d)

(f)

(c)

(d)

Fig. 4. Population at risk quantiles (a), tuberculosis rates (b), and intensity function maps based on the genetic multi-objective algorithm for maximum clusters of sizes 10, 20, 30 and 40 (c, d, e and f respectively)

12 Will-be-set-by-IN-TECH

<=C10 C10 − C30 C30 − C70 C70 − C90 >= C90

> =0 0 to 0.77 0.77 to 0.83 >0.83

=0 0 to 0.61 0.61 to 0.62 >0.62

Fig. 4. Population at risk quantiles (a), tuberculosis rates (b), and intensity function maps based on the genetic multi-objective algorithm for maximum clusters of sizes 10, 20, 30 and

Intensity function quantile

Intensity function quantile

(b)

(d)

(f)

<=C10 C10 − C30 C30 − C70 C70 − C90 >= C90

=0 0 to 0.89 0.89 to 0.93 >0.93

> =0 0 to 0.69 0.69 to 0.72 >0.72

40 (c, d, e and f respectively)

Intensity function quantile

Intensity function quantile

(a)

(c)

(e)

Fig. 5. Population at risk quantiles (a), diabetes rates (b), and intensity function maps based on the genetic multi-objective algorithm for maximum clusters of sizes 10, 20, 30 and 40 (c, d, e and f respectively)

**5. Conclusion**

Our methodology takes into account the variability in the observed number of disease cases on area aggregated maps to nonparametrically infer the uncertainty in the delineation of spatial clusters. A given real data map is regarded as just one possible realization of an unknown random variable vector with expected number of cases. The real data vector of the number of observed cases in each area is used to construct a new vector of expected values of random variables, considering the count of cases as the average of the random variables. This vector is now an estimate of the unknown random variable vector with expected number of cases. Our methodology performs *m* Monte Carlo replications based on this estimated vector of averages. The most likely cluster of each replicated map is detected and the *m* corresponding likelihood values obtained in the replications are ranked. For each area we determine the maximum likelihood value among the most likely clusters containing that area. Thus, we obtain the intensity function associated to each area's ranking of their respective likelihood value among the *m* values. The intensity of each area can be interpreted as the importance of that area in the delineation of the possibly existing anomaly on the map, considering only the initially given information of the observed number of cases. This procedure, based on the empirical distribution, takes into account the intrinsic variability of the observed number of cases, which generally is not considered directly in the existing algorithms used to detect spatial clusters. In our case studies we could see different situations with respect to the intrinsic variability of the existing spatial anomaly. When the most likely cluster is quite prominent, as in the diabetes case study, the intensity function is such that almost all areas associated with the most likely clusters found in the *m* replications coincides with those areas composing the most likely cluster detected for the original observed cases. In this situation the geographic anomaly is highly focused. However, in a different scenario, a disease map may present an intrinsically wide variability of data. Many areas near or adjacent to the most likely cluster have values of the intensity function close to the values corresponding to areas of the most likely cluster. In the case study of hypertension, this intrinsic variability produces a map with clearly unrelated areas, but with rather close probability ranking, indicating a situation of

Assessing the Outline Uncertainty of Spatial Disease Clusters 65

multiplicity of clusters, i. e., the most likely cluster is clearly poorly delineated.

"'borderline"' regions, with different levels of uncertainty.

provided by ordinary methods of cluster finding.

In this work we included two new features that extended the original ideas of the previous paper Oliveira et al. (2011). First, instead of the circular scan, we have used an irregularly shaped cluster finder based on a multiobjective genetic algorithm. It allowed a much better delineation of the complex shapes found in the real data clusters. As a consequence, several new phenomena could be distinguished in the spatial distribution of disease, which could not be observed with the simples spatial scan. The second modification was the sequential execution of runs with different sizes for the maximum allowed cluster to composing the intensity function maps. With this modified procedure, instead of only one map, it was obtained a sequence of intensity function maps: as the maximum cluster size increased, larger anomalies of lesser intensity were displayed. This allowed the identification of "'core"' and

The visualization tool developed in this work may serve as a support for the decision making process to prioritize areas of public health intervention, in a more precise manner than

(c)

Fig. 6. Population at risk quantiles (a), hypertension rates (b), and intensity function maps based on the genetic multi-objective algorithm for maximum clusters of sizes 10, 20, 30 and 40 (c, d, e and f respectively)

#### **5. Conclusion**

14 Will-be-set-by-IN-TECH

<=C10 C10 − C30 C30 − C70 C70 − C90 >= C90

> =0 0 to 0.76 0.76 to 0.79 >0.79

> > =0 0 to 0.60 0.60 to 0.66 >0.66

Fig. 6. Population at risk quantiles (a), hypertension rates (b), and intensity function maps based on the genetic multi-objective algorithm for maximum clusters of sizes 10, 20, 30 and

Intensity function quantile

Intensity function quantile

(b)

(d)

(f)

<=C10 C10 − C30 C30 − C70 C70 − C90 >= C90

=0 0 to 0.87 0.87 to 0.93 >0.93

> =0 0 to 0.7 0.7 to 0.73 >0.73

40 (c, d, e and f respectively)

Intensity function quantile

Intensity function quantile

(a)

(c)

(e)

Our methodology takes into account the variability in the observed number of disease cases on area aggregated maps to nonparametrically infer the uncertainty in the delineation of spatial clusters. A given real data map is regarded as just one possible realization of an unknown random variable vector with expected number of cases. The real data vector of the number of observed cases in each area is used to construct a new vector of expected values of random variables, considering the count of cases as the average of the random variables. This vector is now an estimate of the unknown random variable vector with expected number of cases. Our methodology performs *m* Monte Carlo replications based on this estimated vector of averages. The most likely cluster of each replicated map is detected and the *m* corresponding likelihood values obtained in the replications are ranked. For each area we determine the maximum likelihood value among the most likely clusters containing that area. Thus, we obtain the intensity function associated to each area's ranking of their respective likelihood value among the *m* values. The intensity of each area can be interpreted as the importance of that area in the delineation of the possibly existing anomaly on the map, considering only the initially given information of the observed number of cases. This procedure, based on the empirical distribution, takes into account the intrinsic variability of the observed number of cases, which generally is not considered directly in the existing algorithms used to detect spatial clusters.

In our case studies we could see different situations with respect to the intrinsic variability of the existing spatial anomaly. When the most likely cluster is quite prominent, as in the diabetes case study, the intensity function is such that almost all areas associated with the most likely clusters found in the *m* replications coincides with those areas composing the most likely cluster detected for the original observed cases. In this situation the geographic anomaly is highly focused. However, in a different scenario, a disease map may present an intrinsically wide variability of data. Many areas near or adjacent to the most likely cluster have values of the intensity function close to the values corresponding to areas of the most likely cluster. In the case study of hypertension, this intrinsic variability produces a map with clearly unrelated areas, but with rather close probability ranking, indicating a situation of multiplicity of clusters, i. e., the most likely cluster is clearly poorly delineated.

In this work we included two new features that extended the original ideas of the previous paper Oliveira et al. (2011). First, instead of the circular scan, we have used an irregularly shaped cluster finder based on a multiobjective genetic algorithm. It allowed a much better delineation of the complex shapes found in the real data clusters. As a consequence, several new phenomena could be distinguished in the spatial distribution of disease, which could not be observed with the simples spatial scan. The second modification was the sequential execution of runs with different sizes for the maximum allowed cluster to composing the intensity function maps. With this modified procedure, instead of only one map, it was obtained a sequence of intensity function maps: as the maximum cluster size increased, larger anomalies of lesser intensity were displayed. This allowed the identification of "'core"' and "'borderline"' regions, with different levels of uncertainty.

The visualization tool developed in this work may serve as a support for the decision making process to prioritize areas of public health intervention, in a more precise manner than provided by ordinary methods of cluster finding.

**4** 

*Iran* 

**Review of Ames Assay Studies** 

**of the Urine of Clinical Pathology and** 

**Other Occupations, such as Oncology** 

Majid Rezaei Basiri1,5, Mahmoud Ghazi-khansari2, Hasan Rezazadeh1,

Mohammad Ali Eghbal1,4\*, Iraj Aswadi-kermani3, H. Hamzeiy1, Hossein Babaei1, Ali Reza Mohajjel Naebi1 and Alireza Partoazar2 *1Department of Pharmacology and Toxicology in School of Medicine of Tabriz,* 

*5Students Researches Committee of Tabriz, Iran Medical Sciences University* 

This chapter we refer to mutagenicity activity in the urine samples of persons who are exposed to carcinogenic materials in their occupations, and so some studies evaluated mutagenicity determination in individuals who worked with and are exposed to active potential mutagenic materials. There are some mutagenic compounds present in workplaces such as among nursing personnel in oncology hospitals, farmers' fields, clinical pathology laboratories, clinical forensic laboratories and pharmacology investigation laboratories etc. Also, clinical forensic laboratory personnel use some dangerous solvents such as chloroform which is mixed with other solvents in solutions of preparations of tank thin layer chromatography. They are also exposed to some mutagenic compounds such as formaldehyde, benzene and some solvents and colour regents. We refer to some of the below mentioned compounds, such as benzene, formaldehyde, paraffin, colour regents, organochlorin, smear fixators and so on. The colour regent might be contained in heavy metal carcinogenic substances which were used for smear colouring in clinical pathology laboratories by technicians and clinical forensic laboratories. According to review of some studies, after the filling out of questionnaire forms by these individuals, urine samples were

**1. Introduction** 

 \*

Corresponding Author

**Forensic Laboratory Personnel and** 

**Hospitals and Nursing Personnel** 

*2Department of Pharmacology in School of Medicine of Tehran,* 

*3Shahid Ghazi Oncology Research Centre Departments in Tabriz,* 

 *Iran Medical Sciences University* 

*Iran Medical Sciences University* 

*Iran Medical Sciences University* 

*4Drug Applied Research Centre of Tabriz* 

#### **6. References**


### **Review of Ames Assay Studies of the Urine of Clinical Pathology and Forensic Laboratory Personnel and Other Occupations, such as Oncology Hospitals and Nursing Personnel**

Majid Rezaei Basiri1,5, Mahmoud Ghazi-khansari2, Hasan Rezazadeh1, Mohammad Ali Eghbal1,4\*, Iraj Aswadi-kermani3, H. Hamzeiy1, Hossein Babaei1, Ali Reza Mohajjel Naebi1 and Alireza Partoazar2 *1Department of Pharmacology and Toxicology in School of Medicine of Tabriz, Iran Medical Sciences University 2Department of Pharmacology in School of Medicine of Tehran, Iran Medical Sciences University 3Shahid Ghazi Oncology Research Centre Departments in Tabriz, Iran Medical Sciences University 4Drug Applied Research Centre of Tabriz 5Students Researches Committee of Tabriz, Iran Medical Sciences University Iran* 

#### **1. Introduction**

16 Will-be-set-by-IN-TECH

66 Public Health – Methodology, Environmental and Systems Issues

Cancado, A. L. F., Duarte, A. R., Duczmal, L., Ferreira, S. J., Fonseca, C. M. & Gontijo, E. C.

da Fonseca, V. G., Fonseca, C. M. & Hall, A. O. (2001). Inferential performance assessment of

Duarte, A. R., Duczmal, L. H., Ferreira, S. J. & Cancado, A. L. F. (2010). Internal cohesion and

Duczmal, L., Cancado, A. L. F. & Takahashi, R. H. C. (2008). Delineation of irregularly shaped

Duczmal, L., Cancado, A. L. F., Takahashi, R. H. C. & Bessegato, L. F. (2007). A genetic

Duczmal, L., Kulldorff, M. & Huang, L. (2006). Evaluation of spatial scan statistics

Fonseca, C. M., da Fonseca, V. G. & Paquete, L. (2005). Exploring the performance of stochastic

*Notes In Computer Science*, Vol. 3410, Springer-Verlag, Berlin, pp. 250–264. Goovaerts, P. (2006). Geostatistical analysis of disease data: visualization and propagation

Kulldorff, M. (1999). Spatial scan statistics: Models, calculations and applications, *in* J. Glaz

Kulldorff, M., Huang, L., Pickle, L. & Duczmal, L. (2006). An elliptic spatial scan statistic,

Lawson, A., Biggeri, A. & Bohning, D. (1999). *Disease mapping and risk assessment for public*

Neill, D. B. (2011). Fast bayesian scan statistics for multivariate event detection and

Oliveira, F. L. P., Duczmal, L., Cancado, A. L. F. & Tavares, R. (2011). Nonparametric

Yiannakoulias, N., Rosychuk, R. J. & Hodgson, J. (2007). Adaptations for finding irregularly shaped disease clusters, *International Journal of Health Geographics* 6(28).

intensity bounds for the delineation of spatial clusters, *International Journal of Health*

*Science*, Vol. 1993, Springer-Verlag, Berlin, pp. 213–225.

*Data Analysis* 52: 43–52. DOI:10.1016/j.csda.2007.01.016.

simulation, *International Journal of Health Geographics* 5(7).

D. M. (2010). Penalized likelihood and multi-objective spatial scans for the detection and inference of irregular clusters, *International Journal of Health Geographics* 55(9). Chen J, Roth RE, N. A. L. E. M. A. (2008). Geovisual analytics to enhance spatial scan statistic

interpretation: an analysis of u.s. cervical cancer mortality, *International Journal of*

stochastic optimisers and the attainment function, *Proceedings of the First International Conference on Evolutionary Multi-Criterion Optimization, Lecture Notes In Computer*

geometric shape of spatial clusters, *Environmental and Ecological Statistics* 17: 203–229.

disease clusters through multiobjective optimization, *Journal of Computational and*

algorithm for irregularly shaped spatial scan statistics, *Computational Statistics and*

for irregularly shaped clusters, *Journal of Computational and Graphical Statistics*

multiobjective optimisers with the second-order attainment function, *Proceedings of the Third International Conference on Evolutionary Multi-Criterion Optimization, Lecture*

of spatial uncertainty in cancer mortality risk using poisson kriging and p-field

& N. Balakrishnan (eds), *Scan Statistics and Applications*, Springer Netherlands,

**6. References**

*Health Geographics* 7(57).

*Graphical Statistics* 17(2): 243–262.

*Statistics in Medicine* 25: 3929–3943. Lawson, A. (2009). *Bayesian Disease mapping*, CRC Press.

*health*, John Wiley and Sons, New York.

visualization, *Statistics in Medicine* 30(28): 455–469.

15(2): 428–442.

pp. 303–322.

*Geographics* 1(10).

This chapter we refer to mutagenicity activity in the urine samples of persons who are exposed to carcinogenic materials in their occupations, and so some studies evaluated mutagenicity determination in individuals who worked with and are exposed to active potential mutagenic materials. There are some mutagenic compounds present in workplaces such as among nursing personnel in oncology hospitals, farmers' fields, clinical pathology laboratories, clinical forensic laboratories and pharmacology investigation laboratories etc. Also, clinical forensic laboratory personnel use some dangerous solvents such as chloroform which is mixed with other solvents in solutions of preparations of tank thin layer chromatography. They are also exposed to some mutagenic compounds such as formaldehyde, benzene and some solvents and colour regents. We refer to some of the below mentioned compounds, such as benzene, formaldehyde, paraffin, colour regents, organochlorin, smear fixators and so on. The colour regent might be contained in heavy metal carcinogenic substances which were used for smear colouring in clinical pathology laboratories by technicians and clinical forensic laboratories. According to review of some studies, after the filling out of questionnaire forms by these individuals, urine samples were

<sup>\*</sup> Corresponding Author

Review of Ames Assay Studies of the Urine of Clinical Pathology and Forensic

**4. Sampling of volunteers and urine extraction** 

[3,4,6,12,17]

**5. Ames assay** 

**6. Ames assay protocol** 

Laboratory Personnel and Other Occupations, such as Oncology Hospitals and Nursing Personnel 69

technique called Ames assay. The first experiments by professor Ames and his co-workers were conducted in the 1960s and 1970s and they were accepted by international agencies, including the American FDA. The test has its own specific conditions because it shows that mutagenicity activities with cost less and the samples of tested results would be expressed in a few days. As well as this, it can simultaneously test up some carcinogenic material over the same time in less than three days. It is held that it also tested for ease of mutagenicity activities which are available for review, of which there may be more in the future. However, in many work environments there are promutagenic and carcinogenic compounds that are present to which workers are exposed. The review of studies of more jobs and environments globally show that using this test for biological fluids – including the urine of personnel – proves potential mutagenic activities. It is also held that the assessment of them in all jobs as well as review by the study process including the preparation of urine samples, extraction of urine mutagenic compounds and the Ames test performed on the samples and biological extracted compounds have similarities processes. [1,2,3,6,12,13,17]

All of the urine samples were collected at the end of the week from the Iranian forensic organisation's laboratories and other relationship laboratories personnel and stored in a refrigerator, although some appropriate methods are accessible for the urine extracts' preparation, but in this review we almost see solid phase extraction methods. Extraction columns were filled with 1 g from one kind of C18 or XAD-2 resins between two layers of cotton, depending on studies' strategies. First, the resins were activated by Milli-Q water and methanol then the volunteer's clear urine was passed through a column by vacuum and the mutagenic substances were concentrated with V/V of methanol/Acetonitrile in tubes.

In these studies, mutagenicity was followed by the plates' incorporated procedures, as described by Mortelmans and Zeiger (2000), in overnight cultures of TA98 and TA100 salmonella typhymorium tester strains with and without the S9 mix fraction in the forensic organisation's personnel and just TA100 in the clinical laboratory technicians in Iranian laboratories. For this strain, the spontaneous background number of the revertant was approximately 20-50 for TA98 and 75-200 for TA100 salmonella typhymorium tester strains. In this study, Sodium azide was used as a positive control without using the S-9 mix buffer and the 2-amino anthracene as a positive control with the S-9 mix buffer. DMSO and distilled water were considered as negative controls and were used in this study. [1,2,3,6,12,13,17]

The first dose/response was determined for all urine extract samples. In our studies, the appropriate dose of urine extracts was 1/100 for exposed bacterial strains and 0.05ml of overnight bacterial culture (TA98 or TA100) was added in 2ml of melting top agar containing trace amounts of biotin and histidine then, and 0.05ml 1/100 of diluted urine extract with DMSO solvent was added in the melting top agar tube as well; after shaking the tube's contents were transferred in Glucose minimum agar plates which were incubated at 37C for 48 hours. The above mentioned protocols were repeated with the addition of 0.5ml

collected from candidates in all occupations at the end of the week and they were then held in a refrigerator for the urine extraction process. Special instruments – such as vacuum – and special materials such as MilliQ water and some solvents such as methanol and acetonitrile were used for urine solid phase extraction. Some relationship studies showed that urine extracts were prepared by resins of C18 and amberlite XAD-2 because they are particularly absorbent of the carcinogenic compounds of urine; the importance of the above mentioned resins and the sterilisation of urine extracts will be described in this chapter as well. The personnel who work for a long time in these fields are exposed to these potentially carcinogenic materials. According to the review studies, we will show the mutagenic activity evaluation of the urine of some professional personnel by Ames monitoring. After the preparation of the sterile urine extracts of these personnel, they were exposed to salmonella strains of bacteria for the determination of mutagenic activity. The function of salmonella typhimorium strains such as: TA98,TA100, and the overnight culture of these bacteria show the importance of suspension preparation in the Ames test ratio and all kinds of bacterial environment cultures and the holding of the above mentioned bacteria will be considered in this chapter too. All methods of Ames assay application in these work places will be explained in detail in this chapter. And finally, a risk-benefit assessment and the conditions of healthy work places and individuals' protection will be described as well at the end of this chapter.

#### **2. Mutagenic or carcinogenic material**

Some compounds are present in the occupations of clinical pathology and forensic laboratory personnel and other occupations, such as oncology hospitals. Others in this review are: hamatoxilene, eosin and some solvents and colour reagent. Benzene, formaldehyde, paraffin, colour reagents, organochlorin, smear fixators, antineoplastic drugs etc. So the personal of these places might absorb these materials in their work environment through the lungs, the skin, orally etc. These compounds are metabolised in their liver by the P450 enzyme to activate potential mutagenic compounds. Figure.1 shows some these compounds. [5,6,7,8,9,10,12,13,14,15,17,18,19,20,21]

Fig. 1. Mutagenic or Carcinogenic material in laboratory, oxide metabolites of these compounds cause genotoxic effects.

#### **3. Samples and Ames test history**

For the mutagenicity activity evaluation of biological fluids in different job environments in previous decades, numerous studies of the founder of mutagenicity have a review up technique called Ames assay. The first experiments by professor Ames and his co-workers were conducted in the 1960s and 1970s and they were accepted by international agencies, including the American FDA. The test has its own specific conditions because it shows that mutagenicity activities with cost less and the samples of tested results would be expressed in a few days. As well as this, it can simultaneously test up some carcinogenic material over the same time in less than three days. It is held that it also tested for ease of mutagenicity activities which are available for review, of which there may be more in the future. However, in many work environments there are promutagenic and carcinogenic compounds that are present to which workers are exposed. The review of studies of more jobs and environments globally show that using this test for biological fluids – including the urine of personnel – proves potential mutagenic activities. It is also held that the assessment of them in all jobs as well as review by the study process including the preparation of urine samples, extraction of urine mutagenic compounds and the Ames test performed on the samples and biological extracted compounds have similarities processes. [1,2,3,6,12,13,17]

### **4. Sampling of volunteers and urine extraction**

All of the urine samples were collected at the end of the week from the Iranian forensic organisation's laboratories and other relationship laboratories personnel and stored in a refrigerator, although some appropriate methods are accessible for the urine extracts' preparation, but in this review we almost see solid phase extraction methods. Extraction columns were filled with 1 g from one kind of C18 or XAD-2 resins between two layers of cotton, depending on studies' strategies. First, the resins were activated by Milli-Q water and methanol then the volunteer's clear urine was passed through a column by vacuum and the mutagenic substances were concentrated with V/V of methanol/Acetonitrile in tubes. [3,4,6,12,17]

#### **5. Ames assay**

68 Public Health – Methodology, Environmental and Systems Issues

collected from candidates in all occupations at the end of the week and they were then held in a refrigerator for the urine extraction process. Special instruments – such as vacuum – and special materials such as MilliQ water and some solvents such as methanol and acetonitrile were used for urine solid phase extraction. Some relationship studies showed that urine extracts were prepared by resins of C18 and amberlite XAD-2 because they are particularly absorbent of the carcinogenic compounds of urine; the importance of the above mentioned resins and the sterilisation of urine extracts will be described in this chapter as well. The personnel who work for a long time in these fields are exposed to these potentially carcinogenic materials. According to the review studies, we will show the mutagenic activity evaluation of the urine of some professional personnel by Ames monitoring. After the preparation of the sterile urine extracts of these personnel, they were exposed to salmonella strains of bacteria for the determination of mutagenic activity. The function of salmonella typhimorium strains such as: TA98,TA100, and the overnight culture of these bacteria show the importance of suspension preparation in the Ames test ratio and all kinds of bacterial environment cultures and the holding of the above mentioned bacteria will be considered in this chapter too. All methods of Ames assay application in these work places will be explained in detail in this chapter. And finally, a risk-benefit assessment and the conditions of healthy work places and individuals' protection

Some compounds are present in the occupations of clinical pathology and forensic laboratory personnel and other occupations, such as oncology hospitals. Others in this review are: hamatoxilene, eosin and some solvents and colour reagent. Benzene, formaldehyde, paraffin, colour reagents, organochlorin, smear fixators, antineoplastic drugs etc. So the personal of these places might absorb these materials in their work environment through the lungs, the skin, orally etc. These compounds are metabolised in their liver by the P450 enzyme to activate potential mutagenic compounds. Figure.1 shows some these

Fig. 1. Mutagenic or Carcinogenic material in laboratory, oxide metabolites of these

For the mutagenicity activity evaluation of biological fluids in different job environments in previous decades, numerous studies of the founder of mutagenicity have a review up

will be described as well at the end of this chapter.

compounds. [5,6,7,8,9,10,12,13,14,15,17,18,19,20,21]

compounds cause genotoxic effects.

**3. Samples and Ames test history** 

**2. Mutagenic or carcinogenic material** 

In these studies, mutagenicity was followed by the plates' incorporated procedures, as described by Mortelmans and Zeiger (2000), in overnight cultures of TA98 and TA100 salmonella typhymorium tester strains with and without the S9 mix fraction in the forensic organisation's personnel and just TA100 in the clinical laboratory technicians in Iranian laboratories. For this strain, the spontaneous background number of the revertant was approximately 20-50 for TA98 and 75-200 for TA100 salmonella typhymorium tester strains. In this study, Sodium azide was used as a positive control without using the S-9 mix buffer and the 2-amino anthracene as a positive control with the S-9 mix buffer. DMSO and distilled water were considered as negative controls and were used in this study. [1,2,3,6,12,13,17]

#### **6. Ames assay protocol**

The first dose/response was determined for all urine extract samples. In our studies, the appropriate dose of urine extracts was 1/100 for exposed bacterial strains and 0.05ml of overnight bacterial culture (TA98 or TA100) was added in 2ml of melting top agar containing trace amounts of biotin and histidine then, and 0.05ml 1/100 of diluted urine extract with DMSO solvent was added in the melting top agar tube as well; after shaking the tube's contents were transferred in Glucose minimum agar plates which were incubated at 37C for 48 hours. The above mentioned protocols were repeated with the addition of 0.5ml

Review of Ames Assay Studies of the Urine of Clinical Pathology and Forensic

positive control plates with the S-9mix buffer.

S-9 mix buffer. Rezai-basiri et al., 2008.

Laboratory Personnel and Other Occupations, such as Oncology Hospitals and Nursing Personnel 71

persons' urine sample colony counts, so the ratio of all the samples was more than 2 and significant for the description of mutagenic compounds in all occupations. According to figure.3, in these studies we used a 0.05ml DMSO solvent and diluted water for negative control plates with and without using the S-9mix buffer, and we also used sodium azide for positive control plates without using the S-9mix buffer and 2-amino anthracene for

Fig. 3a. Diluted water (D.W) for the negative control colony count plate without using the

Fig. 3b. Sodium azide (5µg/plate) for the positive control colony count plate (and TA100 Salmonella typhymorium tester strain) without using the S-9 mix buffer. Rezai-basiri et al., 2008.

of the S-9mix buffer with the contents of the rat liver p450 enzyme. After the incubation of all plates with and without S-9mix buffer, the colony counts of all the plates were reported for data collection. [1,2,3,4,6,12,13,17]

#### **7. Ames assay on urine samples of hospitals nursings**

Most of the anti-neoplastic drugs are cytotoxic as well as nephrotoxic and cause DNA damage. For example, some of them are: doxorubicin, bleomycin, vinblastine, dacarbazine, methotrexate, fluorouracil , prednisone , epirubicin , irinotecan , leucoverin, prednisone, 6-mercaptopurine, procarbazine, lomustine, cisplatin (platinum), etoposide, 6-thioguanine, dexamethasone. According to figure.2, these drugs (such as cycloposphamide, mechlortamine, melphalan, chlorambucil) are nitrogen mustard group and so they have genotoxic effects because they will link to DNA, and they have carcinogenic effects on individuals who are exposed to them in their occupations. They involve a rapid proliferation of normal tissues (bone marrow, hair follicles and the intestinal epithelium). Therefore, most hospital nursing personnel are chronically exposed to anti-neoplastic drugs, especially during the course of giving therapeutic doses to cancer patients. Some studies have shown mutagenic activity in the urine of these nursing staff through an Ames assay. [3,4,6,7,8,9,11,12,13,17]

Fig. 2. Some antineoplastic drugs with nitrogen mustard groups.

#### **8. Studies' results**

The data was collected from all the occupational personnel and their negative group control urine extracts were exposed to salmonella bacteria tester strains, such as TA98 and TA100 in these review studies; then, according to the below mentioned results and data from the Iranian clinical pathology and medicinal forensic laboratories, they show the colony counts of all the tester strains of salmonella typhymorium bacteria (such as TA 98 and TA100) as being more than 400 for the positive samples of personnel. On the other hand, the colony count of these individuals' urine samples in the bacterial culture shows that these personnel excrete mutagenic compounds in their urine. Each ratio is determined from individuals' urine samples colony counts per negative group of control

of the S-9mix buffer with the contents of the rat liver p450 enzyme. After the incubation of all plates with and without S-9mix buffer, the colony counts of all the plates were reported

Most of the anti-neoplastic drugs are cytotoxic as well as nephrotoxic and cause DNA damage. For example, some of them are: doxorubicin, bleomycin, vinblastine, dacarbazine, methotrexate, fluorouracil , prednisone , epirubicin , irinotecan , leucoverin, prednisone, 6-mercaptopurine, procarbazine, lomustine, cisplatin (platinum), etoposide, 6-thioguanine, dexamethasone. According to figure.2, these drugs (such as cycloposphamide, mechlortamine, melphalan, chlorambucil) are nitrogen mustard group and so they have genotoxic effects because they will link to DNA, and they have carcinogenic effects on individuals who are exposed to them in their occupations. They involve a rapid proliferation of normal tissues (bone marrow, hair follicles and the intestinal epithelium). Therefore, most hospital nursing personnel are chronically exposed to anti-neoplastic drugs, especially during the course of giving therapeutic doses to cancer patients. Some studies have shown mutagenic activity in the urine of these nursing staff through an Ames assay.

for data collection. [1,2,3,4,6,12,13,17]

[3,4,6,7,8,9,11,12,13,17]

**8. Studies' results** 

**7. Ames assay on urine samples of hospitals nursings** 

Fig. 2. Some antineoplastic drugs with nitrogen mustard groups.

The data was collected from all the occupational personnel and their negative group control urine extracts were exposed to salmonella bacteria tester strains, such as TA98 and TA100 in these review studies; then, according to the below mentioned results and data from the Iranian clinical pathology and medicinal forensic laboratories, they show the colony counts of all the tester strains of salmonella typhymorium bacteria (such as TA 98 and TA100) as being more than 400 for the positive samples of personnel. On the other hand, the colony count of these individuals' urine samples in the bacterial culture shows that these personnel excrete mutagenic compounds in their urine. Each ratio is determined from individuals' urine samples colony counts per negative group of control persons' urine sample colony counts, so the ratio of all the samples was more than 2 and significant for the description of mutagenic compounds in all occupations. According to figure.3, in these studies we used a 0.05ml DMSO solvent and diluted water for negative control plates with and without using the S-9mix buffer, and we also used sodium azide for positive control plates without using the S-9mix buffer and 2-amino anthracene for positive control plates with the S-9mix buffer.

Fig. 3a. Diluted water (D.W) for the negative control colony count plate without using the S-9 mix buffer. Rezai-basiri et al., 2008.

Fig. 3b. Sodium azide (5µg/plate) for the positive control colony count plate (and TA100 Salmonella typhymorium tester strain) without using the S-9 mix buffer. Rezai-basiri et al., 2008.

Review of Ames Assay Studies of the Urine of Clinical Pathology and Forensic

bacteria strains both with and without the S-9mix (cell of rat liver).

\*\*=Positive cases in the personnel of the medicinal forensic organisation.

bacteria strains both with and without the S-9mix (cell of rat liver).

**laboratory technicians** 

technicians. Rezai-basiri et al., 2008.

Laboratory Personnel and Other Occupations, such as Oncology Hospitals and Nursing Personnel 73

\*\*=The samples belonged to the anatomy personnel of the medicinal forensic organisation. Partoazr, et al., 2009. Table 2. Positive and doubtful cases. The results of the mutagenicity assay in the urine of medicinal forensic laboratory personnel with the use of TA100 salmonella typhymorium

 with+S-9mix without-S-9mix without-S-9mix with+S-9mix 1\*\* 550 55 20 2.2 2\* 340 30 15 1.2 3\* 41 31 1.86 1.24

Samples Colony count TA100 Ratio of TA100 with+S-9mix without-S-9mix without-S-9mix with+S-9mix 1\* 95 95 1 1.26 2\* 95 78 1 1.04 3\*\* 85 74 0.89 0.98

\*=The samples belonged to the pathology personnel of the medicinal forensic organisation.

Samples Colony count TA98 Ratio of TA98

\*=Doubtful cases in the personnel of the medicinal forensic organisation. Partoazr, et al., 2009. Table 3. Positive and doubtful cases. The results of the mutagenicity assay in the urine of medicinal forensic laboratory personnel with the use of TA98 salmonella typhymorium

**11. Evaluation of mutagenic compound in urine of clinical pathology** 

At least four clinical pathology laboratories witnessed the evaluation of the mutagenicity of the urine of technicians by the Ames assay. That the ratio would be more than two was significant for the potential of the mutagenic activity of the urine of technician.

Fig. 4. The plates of positive cases of the TA100 salmonella tester strain cultures at 37C for 48hours for the Ames assay on the urine extracts of the clinical pathology laboratory

Fig. 3c. DMSO for the positive control colony count plate (and TA100 Salmonella typhymorium tester strain) with using the S-9 mix buffer. Rezai-basiri et al., 2008.

#### **9. Statistical analysis methods**

In the Ames assay on the Iranian forensic organisation's laboratory personnel's urines with two bacterial tester strains of TA98, TA100 for colony counts in exposure to urine extract samples, and positive or negative samples the Anova two-way statistical method was used.


#### **10. Evaluation of mutagenic compounds the in urine of medicinal forensic laboratory personnel**

Partoazr, et al., 2009.

Table 1. Results of the control group of the mutagenicity assay in the urine of medicinal forensic laboratory personnel with the use of TA98 and TA100 bacteria strains both with and without the S-9mix.


\*=The samples belonged to the pathology personnel of the medicinal forensic organisation. \*\*=The samples belonged to the anatomy personnel of the medicinal forensic organisation. Partoazr, et al., 2009.

Table 2. Positive and doubtful cases. The results of the mutagenicity assay in the urine of medicinal forensic laboratory personnel with the use of TA100 salmonella typhymorium bacteria strains both with and without the S-9mix (cell of rat liver).


\*\*=Positive cases in the personnel of the medicinal forensic organisation.

72 Public Health – Methodology, Environmental and Systems Issues

Fig. 3c. DMSO for the positive control colony count plate (and TA100 Salmonella typhymorium tester strain) with using the S-9 mix buffer. Rezai-basiri et al., 2008.

In the Ames assay on the Iranian forensic organisation's laboratory personnel's urines with two bacterial tester strains of TA98, TA100 for colony counts in exposure to urine extract samples, and positive or negative samples the Anova two-way statistical method was used.

**10. Evaluation of mutagenic compounds the in urine of medicinal forensic** 

Control group Colony count TA100 Colony countTA98 without-S-9mix with+S-9mix without-S-9mix with+S-9mix Male 86 65 26 25 Male 106 72 23 27 Male 100 77 27 21 Male 85 75 19 24 Male 88 68 22 25 Male 98 65 21 14 Male 95 85 23 29 Female 91 85 20 20 Female 102 86 24 28 Female 98 76 18 24 Average 94.9 75.4 22.2 24.7

Table 1. Results of the control group of the mutagenicity assay in the urine of medicinal forensic laboratory personnel with the use of TA98 and TA100 bacteria strains both with and

**9. Statistical analysis methods** 

**laboratory personnel** 

Partoazr, et al., 2009.

without the S-9mix.

\*=Doubtful cases in the personnel of the medicinal forensic organisation. Partoazr, et al., 2009.

Table 3. Positive and doubtful cases. The results of the mutagenicity assay in the urine of medicinal forensic laboratory personnel with the use of TA98 salmonella typhymorium bacteria strains both with and without the S-9mix (cell of rat liver).

#### **11. Evaluation of mutagenic compound in urine of clinical pathology laboratory technicians**

At least four clinical pathology laboratories witnessed the evaluation of the mutagenicity of the urine of technicians by the Ames assay. That the ratio would be more than two was significant for the potential of the mutagenic activity of the urine of technician.

Fig. 4. The plates of positive cases of the TA100 salmonella tester strain cultures at 37C for 48hours for the Ames assay on the urine extracts of the clinical pathology laboratory technicians. Rezai-basiri et al., 2008.

Review of Ames Assay Studies of the Urine of Clinical Pathology and Forensic

database, CRS Press, Boca Raton, FL, 1997, 687-729.

method, Mutat Res, 2002; 520(1-2): 199-205.

9(6): 269-72.

113(2): 103-16.

38(11): 1693-5.

2010, 7(2): 192-196.

June 2009, 67(3): 184-189.

Biochem Suppl, 1996, 25: 92-8.

Pharm, 1981, 38(11): 1686-93.

691-9.

Laboratory Personnel and Other Occupations, such as Oncology Hospitals and Nursing Personnel 75

[2] Zeiger E, Genotoxicity Database, Hand book of carcinogenic potency and genotoxicity

[3] Andre V, Deslandes D, Henry-Amar M, Gauduchon P, Biomonitoring of urine

[4] Hyde PM, Evaluation of drug extraction procedures from urine, J Anal Toxicol, 1985,

[5] Aeschbacher HU, Finot PA, Wolleb U, Interaction of histidine- containing test substances

[6] Andre V, Lebailly P, Pottier D, Deslandes E, De Meo M, Henry-Amar M, Gauduchon P,

[8] Harrison BR, Developing guidelines for working with antineoplastic drugs, Am J Hosp

[9] Zimmerman PF, Larsen RK, Barkley EW, Gallelli JF, Recommendations for the safe

[10] Anderson RW, Puckett WH Jr, Dana WJ, Nguyen TV, Theiss JC, Matney, TS, Risk of handling injectable antineoplastic agents, Am J Hosp Pharm,1982;39(11):1881-7. [11] Rezaei-Basiri M, Ghazi-khansari M, Faghih A, Sadeghi M, Lotfalizadeh N, Eghbal M,

[12] Rezaei-Basiri M, Samini M, Ghazi- khansari M, Rezayat M, Sahebgharani M and

Technicians, Journal of Pharmacology and Toxicology, 2008, 3 (3): 230-235. [13] Partoazar A, Ghazi -Khansari M, Abedi MH, Kaviani M, Norashrafeddin SM, Rezaei-

[14] Shelef LA and Chin B, Effect of Phenolic Antioxidants on the Mutagenicity of Aflatoxin B1, Applied and Environmental Microbiology, Dec. 1980: 1039-1043. [15] Cerná M, Pastorková A, Bacterial urinary mutagenicity test for monitoring of exposure

[16] De Flora S, Camoirano A, Bagnasco M, Bennicelli C, van Zandwijk N, Wigbout G, Qian

[17] Ames B, Methods for detecting Carcinogens and mutagens with the salmonella/ mammalian micro some mutagenicity test, Mutation Res, 1975, 31: 347-364.

to genotoxic compounds: a review, Sep. 2002, 10(3): 124-9.

mutagenicity with the Ames test: improvement of the extraction/concentration

and extraction methods with the Ames mutagenicity test, Mutation Res, 1983;

Urine mutagenicity of farmers occupationally exposed during a 1-day use of chlorothalonil and insecticides, Int Arch Occup Environ Health, 2003, 76(1): 55-62. [7] Ahlborg G Jr, Bergstrom B, Hogstedt C, Einisto P, Sorsa M, Urinary screening for

potentially genotoxic exposures in a chemical industry, Br J Ind Med, 1985, 42(10):

handling of injectable antineoplastic drug products, Am J Hosp Pharm, 1981,

Mohajell-Nayebi A, Rezazadeh H, Arshad Zadeh M, Screening of Morphine & Codeine in Urine of Opioid Abusers by Rapid and TLC Analysis, Eur J Gen Med,

Partoazar A, Monitoring Ames Assay on Urine of Clinical Pathology Laboratories

Basiri M,Talebi M, Determining urine sample mutagenicity ratio using Ames test: Tehran forensic medicine laboratory personnel, Tehran University Medical Journal,

GS, Zhu YR, Kensler TW, Smokers and urinary genotoxins: implications for selection of cohorts and modulation of endpoints in chemoprevention trials, J Cell

The number of personnel who had greater than two ratios was five people in this study. According to the below mentioned, we see some significant results showing mutagenic excretions in the urine of clinical pathology laboratory technicians:

The ratios without the S-9mix for two of the technicians were 2.01 and 2.0, and the ratio with the S-9mix for these individuals was 2.05 and 2.01. [3,6,12,13,17]

#### **12. Conclusion and recommendation**

In this review study, we observed mutagenic activity in the urine extracts of some of the technicians of clinical pathology laboratories and the forensic organisation laboratories. According to these studies, in order to reduce mutagenicity in technicians it was suggested to all of them to use masks, gloves and work under laminar flu and avoid drinking in the laboratories. Considering the contamination of all personnel with mutagenic substances such as colour reagents contained in carcinogenic heavy metals, formaldehyde, benzene, hamatoxilen-Eosin and so on in these laboratories, the observation of the principles of health conditions in clinical pathology laboratories and forensic organisation laboratories is recommended. According to this review study, it is held that the long duration of working and exposure to mutagenic and carcinogenic substances in high risk conditions leads to the excreting of mutagenic compounds in the urine of these individuals and so they should decrease their time spent working over the week. The some studies showed which personnel has used antioxidant compounds, such as thiol group drugs; they had decreased the mutagenic activity in their urine. Also these results were shown to smokers who have used thiol group drugs such as acetyl cystein, had low mutagenic activity in urine samoles. As such, it considers the consumption of vitamins with antioxidant effects which are useful to individuals who are exposed to mutagenic compounds in their occupations. [3,6,12,13,16].

#### **13. Acknowledgments**

We are grateful to the below mentioned departments and the research centres of universities and organisation laboratories as well as to the volunteer personnel for their participation in these studies and their professors for the writing of this article:

Department of Pharmacology in the School of Medicine of Tehran/Iran Medical Sciences University.

Department of Pharmacology and Toxicology in the School of Medicine of Tabriz/Iran Medical Sciences University.

Laboratories of Forensic Organization of Tehran/Iran.

Shahid Ghazi Oncology Research Centre Department in Tabriz/Iran Medical Sciences University.

Drug Applied Research Centre of Tabriz/Iran.

Students Researches Committee of Tabriz/Iran Medical Sciences University.

#### **14. References**

[1] Mortelmans K, Zeiger E, the Ames Salmonella/Microsome Mutagenicity Assay; Mutat Research, 2000, 29-60.

The number of personnel who had greater than two ratios was five people in this study. According to the below mentioned, we see some significant results showing mutagenic

The ratios without the S-9mix for two of the technicians were 2.01 and 2.0, and the ratio with

In this review study, we observed mutagenic activity in the urine extracts of some of the technicians of clinical pathology laboratories and the forensic organisation laboratories. According to these studies, in order to reduce mutagenicity in technicians it was suggested to all of them to use masks, gloves and work under laminar flu and avoid drinking in the laboratories. Considering the contamination of all personnel with mutagenic substances such as colour reagents contained in carcinogenic heavy metals, formaldehyde, benzene, hamatoxilen-Eosin and so on in these laboratories, the observation of the principles of health conditions in clinical pathology laboratories and forensic organisation laboratories is recommended. According to this review study, it is held that the long duration of working and exposure to mutagenic and carcinogenic substances in high risk conditions leads to the excreting of mutagenic compounds in the urine of these individuals and so they should decrease their time spent working over the week. The some studies showed which personnel has used antioxidant compounds, such as thiol group drugs; they had decreased the mutagenic activity in their urine. Also these results were shown to smokers who have used thiol group drugs such as acetyl cystein, had low mutagenic activity in urine samoles. As such, it considers the consumption of vitamins with antioxidant effects which are useful to individuals who are exposed to mutagenic compounds in their occupations. [3,6,12,13,16].

We are grateful to the below mentioned departments and the research centres of universities and organisation laboratories as well as to the volunteer personnel for their participation in

Department of Pharmacology in the School of Medicine of Tehran/Iran Medical Sciences

Department of Pharmacology and Toxicology in the School of Medicine of Tabriz/Iran

Shahid Ghazi Oncology Research Centre Department in Tabriz/Iran Medical Sciences

[1] Mortelmans K, Zeiger E, the Ames Salmonella/Microsome Mutagenicity Assay; Mutat

Students Researches Committee of Tabriz/Iran Medical Sciences University.

excretions in the urine of clinical pathology laboratory technicians:

the S-9mix for these individuals was 2.05 and 2.01. [3,6,12,13,17]

these studies and their professors for the writing of this article:

Laboratories of Forensic Organization of Tehran/Iran.

Drug Applied Research Centre of Tabriz/Iran.

Research, 2000, 29-60.

**12. Conclusion and recommendation** 

**13. Acknowledgments** 

Medical Sciences University.

University.

University.

**14. References** 


**5**

*UK* 

**Old Obstacles on New Horizons:**

**Gene X Environment Discoveries in** 

Conrad Iyegbe, Gemma Modinos and Margarita Rivera Sanchez

Genetics and Social Sciences are divergent disciplines for whom it is customary to compete to explain the greater part of Schizophrenia risk 1. These days, a convincing case can be made for the prospective public health value of either discipline 2,3. However the practical implementation of such knowledge continues to prove challenging for either field alone: From a genetic perspective, progress was traditionally hindered by the inconsistent nature of discoveries made in the pre-GWAS (Genome-wide Association Study) era. It is now held back by the fact that the heritability attributed to this disorder remains largely impermeable

Socio-environmental research, on the other hand, has not progressed to the point of being able to pinpoint the precise origins of the high attributable risk fractions repeatedly

However ongoing progress on two fronts is fuelling hopes that a successful marriage of the two fields will benefit both the rate and the integrity of new discoveries, so that clinical interventions can eventually be targeted to patient sub-groups on the basis of their

i. Firstly, the credibility of Schizophrenia genetics is benefiting from a recent upswing in the generation of verifiable new findings. This has led to a palpable mood change

ii. Secondly, it is anticipated that social science research will benefit from an unprecedented program of investment that will stimulate the emergence of newer methodologies designed to improve the resolution with which social risk factors are

There are high hopes that the formal integration of these two fields will help to invigorate the search for tailored clinical interventions, whether they be therapeutic or prophylactic in nature. Thus it seems an opportune time to consider the potential obstacles that lie ahead for Schizophrenia research in the newly revitalised era of translational research. We do this by

**1. Introduction** 

encountered 2.

measured 5,6.

to GWAS and other genomic approaches.

combined genetic and environmental risk profile:

within the psychiatric genetics community 4.

**The Challenge of Implementing**

**Schizophrenia Research** 

*Institute of Psychiatry, Kings College London,* 


### **Old Obstacles on New Horizons: The Challenge of Implementing Gene X Environment Discoveries in Schizophrenia Research**

Conrad Iyegbe, Gemma Modinos and Margarita Rivera Sanchez *Institute of Psychiatry, Kings College London, UK* 

#### **1. Introduction**

76 Public Health – Methodology, Environmental and Systems Issues

[18] Bartczak AW, Sangaiah R, Ball LM, Warren SH1 and Gold A, Synthesis and bacterial

[19] Varella SD, Rampazo RA, Varanda EA. Urinary mutagenicity in chemical laboratory

[20] Siwińska E, Mielzyńska D, Kapka L. Association between urinary 1-hydroxypyrene and genotoxic effects in coke oven workers. Occup Environ Med, 2004, 61: e10. [21] Chamberlain PL, Brynes SD. The regulatory status of xylazine for use in food-producing

[22] Goodson-Gregg N and De Stasio EA Reinventing the Ames Test as a Quantitative Lab that Connects Classical and Molecular Genetics. Genetics , 2009,181: 21-30. [23] Ming-Fang Wu, Fu-Chuo Peng, Yung-Liang Chen et al, Evaluation of Genotoxicity of

[24] Meltem Boyacoğlu, Özlem Çakal Arslan, Hatice Parlak, Muhammet Ali Karaaslan,

[25] Alva Biran , Rami Pedahzur, Sebastian Buchinger, Georg Reifferscheid ,and Shimshon

[26] Jadwiga Marczewska, Ewa Karwicka et al, Assessment of Cytotoxic and Genotoxic

Antrodia cinnamomea in the Ames Test and the In Vitro Chromosomal Aberration

Mutagenicity of Nonylphenol and Octylphenol Using *Salmonella* Mutation Assay,

Belkin, Genetically Engineered Bacteria for Genotoxicity Assessment, Hdb Env

Activity of Alcohol Extract of Polyscias Filicifolia Shoot, Leaf, Cell Biomass of Suspension Culture and Saponin Fraction, Acta Poloniae Pharmaceutica,2011, 703-

workers exposed to solvents. J Occup Health, 2008, 50: 415-22.

animals in the United States. J Vet Pharmacol Ther, 1998, 21: 322-9.

E.U. Journal of Fisheries & Aquatic Sciences, 2007, 299–302.

benzanthracene, Mutagenesis, 2(2), 1987: 101-105.

Test, in vivo,2011, 419-424.

Chem ,2009,161–186.

710.

mutagenicity of the cyclopenta oxides of the four cyclopenta-fused of isomers of

Genetics and Social Sciences are divergent disciplines for whom it is customary to compete to explain the greater part of Schizophrenia risk 1. These days, a convincing case can be made for the prospective public health value of either discipline 2,3. However the practical implementation of such knowledge continues to prove challenging for either field alone: From a genetic perspective, progress was traditionally hindered by the inconsistent nature of discoveries made in the pre-GWAS (Genome-wide Association Study) era. It is now held back by the fact that the heritability attributed to this disorder remains largely impermeable to GWAS and other genomic approaches.

Socio-environmental research, on the other hand, has not progressed to the point of being able to pinpoint the precise origins of the high attributable risk fractions repeatedly encountered 2.

However ongoing progress on two fronts is fuelling hopes that a successful marriage of the two fields will benefit both the rate and the integrity of new discoveries, so that clinical interventions can eventually be targeted to patient sub-groups on the basis of their combined genetic and environmental risk profile:


There are high hopes that the formal integration of these two fields will help to invigorate the search for tailored clinical interventions, whether they be therapeutic or prophylactic in nature. Thus it seems an opportune time to consider the potential obstacles that lie ahead for Schizophrenia research in the newly revitalised era of translational research. We do this by

Old Obstacles on New Horizons:



**3. Multifactorial risk factors for Schizophrenia** 

Social Context - Neighbourhood

Migration

Birth defects/Obstetric

Cannabis Use Henquet et al, 2008 20

Childhood Trauma Morgan and Fisher, 2007 24 -

**Familial** Advancing Paternal age Miller et al, 2010 25 Miller et al 2011 26

Seasonal birth - -



**Social** Social Discrimination-

**Economic** Developed vs Developing

include;

childhood 15.

**Neurodevelopme ntal** 

The Challenge of Implementing Gene X Environment Discoveries in Schizophrenia Research 79

This ambiguity probably explains why heritability estimates for Schizophrenia have historically been so variable; the value of each estimate is affected by parameters defined by the population under study, and also the degree to which characteristics such as gender, age, and environment exposure profile have been averaged-over 11. Therefore it comes as no surprise that genetic epidemiology research in Psychiatry is fast becoming preoccupied with redefining heritability itself 11. Some principle areas of interest emerging from such work

Twin studies and other methods impose a fixed-point approximation of heritability. But this fails to adequately capture the inherent mobility of heritability over time. Evidence for this drift comes from longitudinal studies of both Substance Misuse and Depression. These demonstrate a tendency for heritability to increase across the developmental period between adolescence and adulthood 12, and also with later stages of decline 13. These studies show that the initiation of cannabis use is predominantly an environmental phenomenon, although genetic influences become increasingly important as the level of usage progresses towards substance abuse and drug dependency 14. In extreme scenarios within polygenic disorders, heritability may reach a higher level during earlier neuro-developmental stages. Such cases tend to result in earlier onset. For Schizophrenia, the earliest cases are known to occur during

Table 1 lists some of the important exposures known to affect the risk of Schizophrenia. The main origins are social, socio-economic and neuro-developmental. As well as being very common many of these risk factors are associated with large effects. Odds ratios (ORs), reflect the odds of exposure to a risk factor in cases relative to controls (expressed as a fold-difference).

**Table 1. Environmental Risk factors for Schizophrenia - a non-exhaustive list Context Environmental Risk Factor Recent Review Recent meta-analysis**  Urban-Rural dwelling March et al, 2008 16 -

effects - -

Cantor-Graae and Selten,

Discrimination - Allardyce et al, 2005 17

Seasonal birth Davies et al, 2003 27 Davies et al, 2003 27

Country - Saha et al, 2005 25 (25)

complications - -

Vitamin D - -

**Other** Gender - Aleman et al, 2003 23 (23) McGrath et al, 2004 24 (24)

Socio-Economic status Cohen et al, 2008 (22) -

2005 18 DeAlberto et al, 2010 19

Arsenault et al, 2002 21; Henquet et al, 2005 22; Moore et al, 2007 23

taking a fresh look through a retrospective lens, at the historical stumbling blocks for the GxE field. We discuss some of the new opportunities (horizons) at the disposal of GxE researchers designed to circumvent them. Some of these hail from recent advances in biobanking, meanwhile new bioinformatic initiatives are helping to transform electronic clinical databases into similarly powerful research tools.

We also highlight the potential pitfalls of an over-regulated clinical trial environment and the detrimental consequences this may eventually have on the pipeline for new drugs. Currently there are fears that an over-burdensome European regulatory legislature is responsible for the recent efflux of companies away from the European clinical trial market. This may create an unwanted bottleneck (or worse still, a precipice) within the new and fully-functional formal framework designed to shepherd only the most robust GxE discoveries into the clinic. We begin this chapter with a brief review of some important concepts central to a discussion on Gene-Environment inter-dependency.

#### **2. The enigma surrounding heritability**

Heritability is defined as the proportion of phenotypic variance due to genetic variance. The concept of Schizophrenia as a heritable disorder was once considered to be controversial, though this is no longer the case. From a scientific perspective it is well worth knowing beforehand that a phenotype of potential interest is heritable enough to merit the effort of dissecting genetically. Thus, establishing that this is the case, is a prerequisite first step in genetic research.

Formal estimates of heritability can be obtained through a number of different methods. The archetypal approach uses twins 7. Twin studies suggest that susceptibility to Schizophrenia is predominantly a genetic phenomenon that accounts for 65-80% of overall risk 7,8. But that upper estimate is likely to understate the true importance of the environment. Even highly penetrant genetic risk factors (such as a syndromic deletion on chromosome 22q11), are not always sufficient to elicit Schizophrenia on their own 9. This is confirmed by the fact that pathogenic genetic anomalies are often harboured by asymptomatic controls, as well as cases 10. This suggests that the underlying risk conferred is heavily mitigated by the environment and other background genetic modifiers of main effects.

Heritability studies estimate that the environmental contribution to Schizophrenia is between 15-35% of the phenotypic variance. The issue of which science explains the greater part of risk is contentious; social science research bases its own claims of dominance on larger explained effects, and also recent calculations which suggest that the burden of cases occurring in the general population could be averted through social interventions 2. In truth, methodological biases in both fields mean attempting to delineate between the effects of genes and environment is a somewhat arbitrary exercise. This is because classical approaches to heritability estimation do not automatically factor-in the dependency which may occur between genes and environment. Meanwhile, one all-important confounder not accounted for by the social risk liability models of Kirkbride et al 2, is a family history of psychiatric disorder, (a proxy for genetic influence). It is important to keep in mind that these methodological limitations mean that a disorder caused by GxE will be attributed to Genes in a twin study and Environment in an epidemiological study.

This ambiguity probably explains why heritability estimates for Schizophrenia have historically been so variable; the value of each estimate is affected by parameters defined by the population under study, and also the degree to which characteristics such as gender, age, and environment exposure profile have been averaged-over 11. Therefore it comes as no surprise that genetic epidemiology research in Psychiatry is fast becoming preoccupied with redefining heritability itself 11. Some principle areas of interest emerging from such work include;


78 Public Health – Methodology, Environmental and Systems Issues

taking a fresh look through a retrospective lens, at the historical stumbling blocks for the GxE field. We discuss some of the new opportunities (horizons) at the disposal of GxE researchers designed to circumvent them. Some of these hail from recent advances in biobanking, meanwhile new bioinformatic initiatives are helping to transform electronic

We also highlight the potential pitfalls of an over-regulated clinical trial environment and the detrimental consequences this may eventually have on the pipeline for new drugs. Currently there are fears that an over-burdensome European regulatory legislature is responsible for the recent efflux of companies away from the European clinical trial market. This may create an unwanted bottleneck (or worse still, a precipice) within the new and fully-functional formal framework designed to shepherd only the most robust GxE discoveries into the clinic. We begin this chapter with a brief review of some important

Heritability is defined as the proportion of phenotypic variance due to genetic variance. The concept of Schizophrenia as a heritable disorder was once considered to be controversial, though this is no longer the case. From a scientific perspective it is well worth knowing beforehand that a phenotype of potential interest is heritable enough to merit the effort of dissecting genetically. Thus, establishing that this is the case, is a prerequisite first step in

Formal estimates of heritability can be obtained through a number of different methods. The archetypal approach uses twins 7. Twin studies suggest that susceptibility to Schizophrenia is predominantly a genetic phenomenon that accounts for 65-80% of overall risk 7,8. But that upper estimate is likely to understate the true importance of the environment. Even highly penetrant genetic risk factors (such as a syndromic deletion on chromosome 22q11), are not always sufficient to elicit Schizophrenia on their own 9. This is confirmed by the fact that pathogenic genetic anomalies are often harboured by asymptomatic controls, as well as cases 10. This suggests that the underlying risk conferred is heavily mitigated by the

Heritability studies estimate that the environmental contribution to Schizophrenia is between 15-35% of the phenotypic variance. The issue of which science explains the greater part of risk is contentious; social science research bases its own claims of dominance on larger explained effects, and also recent calculations which suggest that the burden of cases occurring in the general population could be averted through social interventions 2. In truth, methodological biases in both fields mean attempting to delineate between the effects of genes and environment is a somewhat arbitrary exercise. This is because classical approaches to heritability estimation do not automatically factor-in the dependency which may occur between genes and environment. Meanwhile, one all-important confounder not accounted for by the social risk liability models of Kirkbride et al 2, is a family history of psychiatric disorder, (a proxy for genetic influence). It is important to keep in mind that these methodological limitations mean that a disorder caused by GxE will be attributed to

clinical databases into similarly powerful research tools.

**2. The enigma surrounding heritability** 

genetic research.

concepts central to a discussion on Gene-Environment inter-dependency.

environment and other background genetic modifiers of main effects.

Genes in a twin study and Environment in an epidemiological study.


Twin studies and other methods impose a fixed-point approximation of heritability. But this fails to adequately capture the inherent mobility of heritability over time. Evidence for this drift comes from longitudinal studies of both Substance Misuse and Depression. These demonstrate a tendency for heritability to increase across the developmental period between adolescence and adulthood 12, and also with later stages of decline 13. These studies show that the initiation of cannabis use is predominantly an environmental phenomenon, although genetic influences become increasingly important as the level of usage progresses towards substance abuse and drug dependency 14. In extreme scenarios within polygenic disorders, heritability may reach a higher level during earlier neuro-developmental stages. Such cases tend to result in earlier onset. For Schizophrenia, the earliest cases are known to occur during childhood 15.

#### **3. Multifactorial risk factors for Schizophrenia**

Table 1 lists some of the important exposures known to affect the risk of Schizophrenia. The main origins are social, socio-economic and neuro-developmental. As well as being very common many of these risk factors are associated with large effects. Odds ratios (ORs), reflect the odds of exposure to a risk factor in cases relative to controls (expressed as a fold-difference).



Old Obstacles on New Horizons:

7q36.3

15q13.2

16p11.2

16p13.11

22q11.21

Xp22.3 &

Xp22.32 &

15

16

22

X

significance.

Schizophrenia 65.

The Challenge of Implementing Gene X Environment Discoveries in Schizophrenia Research 81

7 7q11.23-q21.3 *PCLO* piccolo (presynaptic cytomatrix protein) Strongly suggestive <sup>60</sup>

8 8p12 Significant <sup>38</sup>

duplication *VIPR2* vasoactive intestinal peptide receptor 2 Too rare to compute 28,32,53,61,62

8p21-p12 *NRG1* neuregulin 1 Strongly suggestive <sup>49</sup> 8q21 *MMP16* matrix metallopeptidase 16 Significant <sup>4</sup> 8p23.2 *CSMD1* CUB and Sushi multiple domains 1 Significant <sup>4</sup>

deletion Too rare to compute 28,32,34

duplication Too rare to compute 28,32,35

duplication Too rare to compute 28,30,31,34

17 17q12 deletion Too rare to compute 32,34 18 18q21.1 *TCF4* transcription factor 4 Significant 41,57

deletion Too rare to compute

Table 2. CNVs (Copy Number Variants) are sub-microscopic deletions and duplications of DNA (typically greater than 100kb in size). SNP (Single Nucleotide Polymorphism) refers to a single subnit (base) change in the DNA sequence. *CACNA1C, ZNF804A, NRGN, MHC* and *PBRM1*, all overlap with Bipolar Disorder. \*Genome-wide significant = P<5x10-8; Strong significance is defined as a *P* value of between 5x10-4 and 5x10-8. \*\*Notes the pre-existence of this gene as a commonly-researched candidate in Schizophrenia research prior to GWAS. Note within table 2 the high occurence of findings validated by more than one study. This is particularly obvious for CNVs,

The effects of Environmental risk factors are on a par with those of the structural variants 9, catalogued in Table 2, but the latter occur much too infrequently to explain the fact that Schizophrenia is common mental disorder, affecting 1% of the global population. In fact, the molecular modalities identified so far for Schizophrenia (namely copy number and common variation) currently account for no more than 3% of the total phenotypic variance of

The discrepancy between theoretical and observed heritability estimates has led many to speculate on possible reasons why the 'missing' component is so elusive 66. The possibilities span a wide array of plausible theories, most of which are based on the premise that the

**threshold Reference** 

28,31,32,34,36, 54

RC3) Significant 41,57

affinity) Strongly suggestive <sup>64</sup>

alpha Strongly suggestive <sup>64</sup>

type, alpha 1C subunit Suggestive <sup>63</sup>

**Chromosome Gene/Region Symbol Full Gene Name GWAS Significance** 

11 11q24.2 *NRGN* neurogranin (protein kinase C substrate,

12 12p13.3 *CACNA1C* calcium channel, voltage-dependent, L

Yp13.3 *IL3RA* interleukin 3 receptor, alpha (low

Yp11.3 *CSF2RA* colony stimulating factor 2 receptor,

but is also evident for SNP variants, including those not reaching overall

additive component of heritability is probably exaggerated. eg 67.

The typical effect range of the risk factors shown in table 1 typically range from 1.5 to 11. In contrast, common genetic risk factors for Schizophrenia are much smaller, typically with Odds ratios of between 1.1 – 1.4. See table 2 for a summary of genetic risk factors for Schizophrenia deriving from large-scale (genome-wide) genetic studies.

**Table 2. Genetic Risk factors for Schizophrenia**


The typical effect range of the risk factors shown in table 1 typically range from 1.5 to 11. In contrast, common genetic risk factors for Schizophrenia are much smaller, typically with Odds ratios of between 1.1 – 1.4. See table 2 for a summary of genetic risk factors for

> deletion Too rare to compute 28-36 1q21.1 *BCL9* B-cell CLL/lymphoma 9 Strongly suggestive <sup>37</sup>

deletion *NRXN1* neurexin 1 Too rare to compute 28,31-35,42-44

 1q24 Significant <sup>38</sup> 1q32.2 *PLXNA2* plexin A2 Strongly suggestive <sup>39</sup> 2 2p16.1 *VRK2* vaccinia related kinase 2 Significant 40,41

2q32.1 *ZNF804A* zinc finger protein 804A Strongly suggestive 46-48

 3p21.1 *PBRM1* Polybromo 1 Strongly suggestive <sup>51</sup> 3 3q21-q23 *RELN* reelin Strongly suggestive <sup>52</sup>

5 5q14.1 *CMYA5* cardiomyopathy associated 5 Strongly suggestive <sup>55</sup>

 6p21.3 *NOTCH4* notch 4 Strongly suggestive 45,57 6p22.1 MHC region Significant 49,57,58

6q23.2 *AHI1* Abelson helper integration site 1 Strongly suggestive <sup>59</sup>

3q39 deletion Too rare to compute

3q21-q23 *RBP1* retinol binding protein 1, cellular Strongly suggestive <sup>53</sup>

6p22.1 *NKAPL* NFKB activating protein-like Significant <sup>56</sup>

(intergenic) neuromedin B receptor Significant <sup>50</sup>

miRNA transcript) Significant <sup>4</sup>

RNA transcript) (prostate-specific transcript 1 Significant <sup>4</sup>

(intergenic) Holliday junction recognition protein Significant <sup>50</sup>

(intergenic) twist homolog 2 (Drosophila) Significant <sup>50</sup>

member 1 Strongly suggestive <sup>45</sup>

member 3 Significant <sup>50</sup>

protein 1 Significant <sup>50</sup>

domains 4 Significant <sup>56</sup>

1 Significant 56,57

28,32,34,35,43, 54

oncogene homolog 4 (avian) Strongly suggestive <sup>49</sup>

repeat and PH domain 1 Strongly suggestive <sup>49</sup>

**threshold Reference** 

Schizophrenia deriving from large-scale (genome-wide) genetic studies.

**Chromosome Gene/Region Symbol Full Gene Name GWAS Significance** 

**Table 2. Genetic Risk factors for Schizophrenia**

*MIR137* (intron 3 of

PCGEM1(non-coding

*UGT1A1-HJURP*

*AK573765-TWIST2*

2q34-q35 *ACSL3-KCNE4* acyl-CoA synthetase long-chain family

2q37 *CENTG2/AGAP1* ArfGAP with GTPase domain, ankyrin

2q37.3 *LRRFIP1* leucine rich repeat (in FLII) interacting

6 6p21 *ZKSCAN4* zinc finger with KRAB and SCAN

6p22.1 *PGBD1* piggyBac transposable element derived

*LOC645434-NMBR*

2p22.2 *SULT6B1* sulfotransferase family, cytosolic, 6B,

2q33.3-q34 *ERBB4* v-erb-a erythroblastic leukemia viral

1

1q21.1

1p21.3

2p16.3

2q32.3

2q37.1

2q37.3

6q21-qter


Table 2. CNVs (Copy Number Variants) are sub-microscopic deletions and duplications of DNA (typically greater than 100kb in size). SNP (Single Nucleotide Polymorphism) refers to a single subnit (base) change in the DNA sequence. *CACNA1C, ZNF804A, NRGN, MHC* and *PBRM1*, all overlap with Bipolar Disorder. \*Genome-wide significant = P<5x10-8; Strong significance is defined as a *P* value of between 5x10-4 and 5x10-8. \*\*Notes the pre-existence of this gene as a commonly-researched candidate in Schizophrenia research prior to GWAS. Note within table 2 the high occurence of findings validated by more than one study. This is particularly obvious for CNVs, but is also evident for SNP variants, including those not reaching overall significance.

The effects of Environmental risk factors are on a par with those of the structural variants 9, catalogued in Table 2, but the latter occur much too infrequently to explain the fact that Schizophrenia is common mental disorder, affecting 1% of the global population. In fact, the molecular modalities identified so far for Schizophrenia (namely copy number and common variation) currently account for no more than 3% of the total phenotypic variance of Schizophrenia 65.

The discrepancy between theoretical and observed heritability estimates has led many to speculate on possible reasons why the 'missing' component is so elusive 66. The possibilities span a wide array of plausible theories, most of which are based on the premise that the additive component of heritability is probably exaggerated. eg 67.

Old Obstacles on New Horizons:

**4.2 G-E interaction (GxE)** 

Schizophrenia genes in the initiation of cannabis use.

infection 79, but seemingly not of obstetric complications 80.

**5. Candidate gene studies of gene-environment interaction** 

interactions themselves are proving difficult to individually identify.

isolation of shared environmental influences.

derived from the general population 81.

experimental evidence (ie. by meta-analysis).

The Challenge of Implementing Gene X Environment Discoveries in Schizophrenia Research 83

no difference between these two groups and thus does not find support a role for

Interaction is a more solidly supported mechanism of G-E dependency in Schizophrenia, whose influence clearly extends to cannabis use. For example, early studies have suggested that familial (presumably genetic) influences on SZ risk also augment the psychotogenetic effects of this drug 76. Another study finds that the same level of familial liability is reached among cases of cannabis-induced psychosis, as that found among Schizophrenia patients; a strong indication that the enhanced responsiveness to cannabis in these hospitalised users is enabled by Schizophrenia genes 77. Cannabis use can thus be said to advance the genetic risk of Schizophrenia onset. The same can be said of urbanicity 78 and prenatal exposure to

One drawback of the *familial liability* study design is that genetic and environmental effects cannot so easily be discerned within the construct of 'familiality', which is inferred as being predominantly genetic in origin, but which also incorporates an element of shared environmental risk. The adoption study design therefore, is a convenient way to disentangle the components of this construct, by allowing the genetic component to be assessed in

The adoption study design has been widely used for this purpose in Schizophrenia research. A recent exemplar for the approach investigated psychosis in 13,000 entrants on the Swedish National Adoption Register. Using an empirical approach, the study confirmed the relevance of early life parental employment status, parental separation and housing status to underlying Schizophrenia risk. Importantly this occurred both dependently and independently of underlying genetic liability. The synergism between genes and environment was many times greater than either additive or multiplicative risk thresholds, indicating a strong interaction. The findings were later validated in 26,000 individuals

Currently, Gene-Environment interaction is one of a few areas of genetic research in which the candidate-gene design has had the upper hand over the more systematic approach represented by Genome-wide Association studies (GWAS). A favoured approach uses biological plausibility to guide the formulation of coherent hypotheses 82. This strategy has several high profile discoveries to its credit. Table 3 Lists the GxE studies performed to date in psychosis and summarises the individual outcome of each. Heterogeneity among hypotheses and methodological approaches precludes a more formal assessment of current

Universal acknowledgement of the GxE concept in Schizphrenia alone 78-81,83-86 tends to suggest that its pervasiveness across psychiatry should be on a par with the rest of nature. However the paradox of GxE in Psychiatry is that though generally acknowledged, the

#### **4. The fundamental models of gene-environment dependency**

Nowadays, it is possible to examine the theory that heritability has been overstated, by testing the significance of the difference in heritability between exposure and non-exposure twin models 11. This is an appropriate way to empirically test the dependency between genes and environment.

The risk factors shown in table 1, while very common, are mitigated by the genetic make-up of the individual, such that the overall effect on risk is relatively small. At a population level this means that only a small proportion of those encountering these exposures will ever go on to develop clinical symptoms. This example of inter-dependency is known as Gene-Environment interaction. Cumulatively it may have a large impact on psychosis risk at the population level.

#### **4.1 G-E correlation (rGE)**

Analytically, GxE is difficult to distinguish from gene-environment correlation (rGE), a phenomena whereby exposure to exogenous risk factors is encoded within the DNA of the individual. rGE represents the social manifestation of one's genetic heritage, and its influence on subsequent lifestyle choices. If not properly accounted for, rGE can quietly confound the apparent interaction between genes and the environment.

There are many behavioural examples of this phenomenon in the psychiatric literature (reviewed in 68). For instance, genes can have an indirect influence on adolescent substance misuse, through a mechanism in which genes drive the selection of friends who facilitate this behaviour. In this example, peer-group choice can be redefined as a lifestyle trait with a strong genetic component 69. An equally compelling case can be made for rGE in Depression, as there is an indication that genetic susceptibility to Depression may also partly reflect a person's tendency to experience stressful experiences, such as interpersonal and romantic difficulties 70.

The evidence used to discuss G-E dependency in the context of Schizophrenia is drawn almost exclusively from the cannabis literature, as it is one of the most commonly investigated risk factors in GxE research. Its popularity probably reflects the relative ease with which data on this exposure may be obtained and verified, with good sensitivity and specificity. This makes it comparatively easy to derive a fairly accurate profile of exposure using retrospective assessments 71,72.

While there is little in the way of direct experimental evidence to support the occurrence of rGE in Schizophrenia, it would be surprising if Schizophrenia were shown to be completely devoid of the phenomenon, given its demonstration in other areas of behavioural research 68. Only one study has purported to show evidence of the rGE mechanism in Schizophrenia 73. Meanwhile the evidence that contradicts this finding has withstood the many different experimental designs applied to re-address the same question eg. 21,74,75. The most recent of these used a case-control design 75, and also included a comparison of lifetime cannabis consumption between the siblings of cases (who have a higher genetic propensity for Schizophrenia) and healthy controls. It found no difference between these two groups and thus does not find support a role for Schizophrenia genes in the initiation of cannabis use.

#### **4.2 G-E interaction (GxE)**

82 Public Health – Methodology, Environmental and Systems Issues

Nowadays, it is possible to examine the theory that heritability has been overstated, by testing the significance of the difference in heritability between exposure and non-exposure twin models 11. This is an appropriate way to empirically test the dependency between

The risk factors shown in table 1, while very common, are mitigated by the genetic make-up of the individual, such that the overall effect on risk is relatively small. At a population level this means that only a small proportion of those encountering these exposures will ever go on to develop clinical symptoms. This example of inter-dependency is known as Gene-Environment interaction. Cumulatively it may have a large impact on psychosis risk at the

Analytically, GxE is difficult to distinguish from gene-environment correlation (rGE), a phenomena whereby exposure to exogenous risk factors is encoded within the DNA of the individual. rGE represents the social manifestation of one's genetic heritage, and its influence on subsequent lifestyle choices. If not properly accounted for, rGE can quietly

There are many behavioural examples of this phenomenon in the psychiatric literature (reviewed in 68). For instance, genes can have an indirect influence on adolescent substance misuse, through a mechanism in which genes drive the selection of friends who facilitate this behaviour. In this example, peer-group choice can be redefined as a lifestyle trait with a strong genetic component 69. An equally compelling case can be made for rGE in Depression, as there is an indication that genetic susceptibility to Depression may also partly reflect a person's tendency to experience stressful experiences, such as interpersonal

The evidence used to discuss G-E dependency in the context of Schizophrenia is drawn almost exclusively from the cannabis literature, as it is one of the most commonly investigated risk factors in GxE research. Its popularity probably reflects the relative ease with which data on this exposure may be obtained and verified, with good sensitivity and specificity. This makes it comparatively easy to derive a fairly accurate profile of exposure

While there is little in the way of direct experimental evidence to support the occurrence of rGE in Schizophrenia, it would be surprising if Schizophrenia were shown to be completely devoid of the phenomenon, given its demonstration in other areas of behavioural research 68. Only one study has purported to show evidence of the rGE mechanism in Schizophrenia 73. Meanwhile the evidence that contradicts this finding has withstood the many different experimental designs applied to re-address the same question eg. 21,74,75. The most recent of these used a case-control design 75, and also included a comparison of lifetime cannabis consumption between the siblings of cases (who have a higher genetic propensity for Schizophrenia) and healthy controls. It found

confound the apparent interaction between genes and the environment.

**4. The fundamental models of gene-environment dependency** 

genes and environment.

population level.

**4.1 G-E correlation (rGE)** 

and romantic difficulties 70.

using retrospective assessments 71,72.

Interaction is a more solidly supported mechanism of G-E dependency in Schizophrenia, whose influence clearly extends to cannabis use. For example, early studies have suggested that familial (presumably genetic) influences on SZ risk also augment the psychotogenetic effects of this drug 76. Another study finds that the same level of familial liability is reached among cases of cannabis-induced psychosis, as that found among Schizophrenia patients; a strong indication that the enhanced responsiveness to cannabis in these hospitalised users is enabled by Schizophrenia genes 77. Cannabis use can thus be said to advance the genetic risk of Schizophrenia onset. The same can be said of urbanicity 78 and prenatal exposure to infection 79, but seemingly not of obstetric complications 80.

One drawback of the *familial liability* study design is that genetic and environmental effects cannot so easily be discerned within the construct of 'familiality', which is inferred as being predominantly genetic in origin, but which also incorporates an element of shared environmental risk. The adoption study design therefore, is a convenient way to disentangle the components of this construct, by allowing the genetic component to be assessed in isolation of shared environmental influences.

The adoption study design has been widely used for this purpose in Schizophrenia research. A recent exemplar for the approach investigated psychosis in 13,000 entrants on the Swedish National Adoption Register. Using an empirical approach, the study confirmed the relevance of early life parental employment status, parental separation and housing status to underlying Schizophrenia risk. Importantly this occurred both dependently and independently of underlying genetic liability. The synergism between genes and environment was many times greater than either additive or multiplicative risk thresholds, indicating a strong interaction. The findings were later validated in 26,000 individuals derived from the general population 81.

#### **5. Candidate gene studies of gene-environment interaction**

Currently, Gene-Environment interaction is one of a few areas of genetic research in which the candidate-gene design has had the upper hand over the more systematic approach represented by Genome-wide Association studies (GWAS). A favoured approach uses biological plausibility to guide the formulation of coherent hypotheses 82. This strategy has several high profile discoveries to its credit. Table 3 Lists the GxE studies performed to date in psychosis and summarises the individual outcome of each. Heterogeneity among hypotheses and methodological approaches precludes a more formal assessment of current experimental evidence (ie. by meta-analysis).

Universal acknowledgement of the GxE concept in Schizphrenia alone 78-81,83-86 tends to suggest that its pervasiveness across psychiatry should be on a par with the rest of nature. However the paradox of GxE in Psychiatry is that though generally acknowledged, the interactions themselves are proving difficult to individually identify.

Old Obstacles on New Horizons:

98 broadly defined SZ, 79 narrowly defined SZ, 86 siblings

742 SZ, 884 HC

954 UPAD, BPAD, and SZ 395 HC

110 SZ, 493 HC

54 SZ, 53 HC

116 SZ spectrum disorders, 134 HC

Kéri et al 2009 200 SZ *NRG1* Psychosocial

60 SZ + HLA-DR1, 307 SZ no HLA-DR1

Husted et al 2010

Muntjeswerff et al 2011

Chotai et al 2003

Tochigi et al 2002

Narita et al 2000

Haukvik et al 2010

Nicodemus et al 2008

**Author Sample size Candidate G Candidate E Outcome** 

*MTHFR 677*

*TPH, 5- HTTLPR* and *DRD4*

*HLA-A24* and A26

32 SNPs in *BDNF, DTNBP1, GRM3* and *NRG1*

*AKT1, BDNF, CAPON, CHRNA7, COMT, DTNBP1, GAD1, GRM3, NOTCH4, NRG1, PRODH, RGS4, TNFalpha* 

*NOS1AP* Childhood

C>T Winter birth SZ

Seasonality of birth

Seasonality of birth

Obstetric Complications (OCs)

Obstetric Complications (OCs)

*HLA-DR1* Seasonality of birth

trauma SZ

The Challenge of Implementing Gene X Environment Discoveries in Schizophrenia Research 85

Season of birth variations in UPAD, BPAD and

Association between HLA-A and birth-season in

Association between HLA-DR1and winter birth in SZ

Hippocampal volume

SZ

SZ

SZ

stress Unusual thoughts

**Variable Results Statistics** 

Narrowly defined SZ more likely to have a history of early trauma than their unaffected family members (similar results after controlling for NOS1AP).

No winter period x MTHFR 677- T/T x SZ interaction.


No association between winter birth (December-March) and A24/A26 SZ

HLA-DR1 associated with winter births in patients.


Interactions between serious OCs and: - AKT1 rs1130233. - BDNF rs2049046 and rs76882600. - DTNBP1 rs875462 - GRM3 rs7808623 - No GxE interaction in


volume. -No significant interaction with SZ

controls.

OR=4.17; 95% CI=1.52,

p=0.744; OR=0.90

11.44

p=0.05

p=0.01

p=0.01

p=0.6/0.4; 2(1)=0.4/0.7

p=0.003; 2(1)=8.64

pdiagnosis×OCs=0.25

p=0.031; OR=3.97 p=0.019/0.015; OR=12.45 p=0.031; OR=9.49 p=0.061; OR=3.39

pdiagnosis×OCs×hemisphere=0.77

p<0.0001; F(2,197)=17.98

p<0.0001

p=0.5

#### **Table 3. Studies investigating interactions between candidate susceptibility genes and candidate environmental pathogens in relation to psychosis.**


#### Old Obstacles on New Horizons: The Challenge of Implementing Gene X Environment Discoveries in Schizophrenia Research 85

84 Public Health – Methodology, Environmental and Systems Issues

Self-reported psychotic experiences at age

genes Cannabis Psychotic disorder Interaction with AKT1

Brain Volume and Neurocognition in

(Age of onset)

cannabis use

Psychotic symptoms (hallucinations, delusions) in daily life (ESM)

Tobacco Psychotic disorder

D-9-THC-induced psychotic experiences

disorder

Positive and negative psychoticlike experiences

SZ

16

**Variable Results Statistics** 



No association cannabis use-COMT genotypes

American/Caucasians)

Cannabis increased hallucinations in Val/Val carriers with high levels of psychometric psychosis

No interaction with CNR1 or COMT genotypes.

Condition x Val/Val x Psychometric psychosis interaction.

Cannabis x Val/Val x Schizophreniform disorder interaction.

BDNF (Met/-) x childhood abuse x positive psychotic-like experiences interaction.

(African-

liability.

No interaction p=0.304-0.981

rs2494732 only in cases p=0.007

OR=0.83-1.10

p<0.05

p≤0.05

p≤0.05

mean z=−1.78

p=0.420; 2(1)=0.65

p=0.023; 2(1)=5.15

p<0.001; =0.78

p>0.05; OR=0.83-0.98

p=0.003; 2(1)=8.86

p=0.025; OR=10.9

p=0.004; =0.27, SE=0.10

p=0.23/0.49; 2(2)=2.9/1.4

**Table 3. Studies investigating interactions between candidate susceptibility genes and candidate environmental pathogens in relation to psychosis.** 

**Author Sample size Candidate G Candidate E Outcome** 

152 SNPs in 42 candidate

12 tag SNPs in *CB1/CNR* gene

*COMT*

*COMT* 

*CNR1, CHRNA7, COMT*  Val(158)Met

*COMT* 

*COMT* 

*BDNF*  Val(66)Met Cannabis

Val(66)Met Cannabis Psychotic disorder

Val(158)Met Cannabis Adolescent

Val(158)Met Cannabis (ESM)

Val(158)Met Cannabis

Cannabis,

Val(158)Met Cannabis Schizophreniform

Childhood abuse and neglect

Val(158)Met Cannabis

*COMT* 

Zammit et al 2011

vanWinkel et al 2011

Ho et al 2011 235 SZ

Decoster et al

Kantrowitz et al 2009

Henquet et al 2009

Zammit et al 2007

Henquet et al 2006

Caspi et al 2005

Alemany et al 2011

2011 585 SZ *BDNF* 

92 SZ (33 Caucasian, 46 African-American)

31 psychotic disorder, 25 HC

750 SZ, 688 HC

803 HC general population

533 HC general population

30 psychotic disorder, 12 relatives, 32 HC

2630 HC general population

810 SZ, 740 siblings, 419 HC


Old Obstacles on New Horizons:

**6. Methodological constraints in GxE research** 

high importance (eg. Caspi vs Zammit) 92,93.


with which environmental exposures are measured.

include:


**6.1 Measurement error** 


The Challenge of Implementing Gene X Environment Discoveries in Schizophrenia Research 87

There are lingering doubts about the experimental validity of many GxE findings reported in the literature. This is in contrast to the renewed sense of optimism about genetic association, a method which focuses on direct gene effects rather than the consequences of their interactions. Genetic association studies now have GWAS to look to as a methodological reference point 4,38,40,41,50,56,89, and it is out of the technological infrastructure supporting GWAS, that two competing theories (they may not be mutually exclusive) regarding the genetic architecture of Schizophrenia have emerged: The Common Disease-Common Variant and Multiple Rare Variant hypotheses 4,9,10,58. The latter of these will be explored in great detail through new sequencing initiatives already underway for

A combination of meticulous study design, unprecedented sample sizes and good governance over methodological practice 90, now mean that replication is no longer the rarity it once was for genetic association studies (see table 3). This is in part due to the fact that genetic association studies are becoming more methodologically homogeneous; many of the rigorous methodological practices and standards routinely implemented in GWAS research (internally validated findings, population stratification, etc) have also been widely

In contrast, the diversity of methodologies and standards used in GxE research has remained stubbornly heterogeneous to date; the multitude of study designs used to followup new discoveries, has seen only varying levels of success 91. Longitudinal studies sit very high within the complex methodological hierarchy of epidemiological designs, but even they are failing to provide the swift resolutions hoped for, to ongoing research questions of

Several recurring factors limit the success rate of replication attempts in GxE research. These

Arguably the most replicated GxE finding in Psychiatry belongs to the field of Depression, and involves the short allele of the Serotonin transporter gene (*5HTTLPR*) and Stressful Life Events (SLE), which interactively augment the risk of Depression 94. Reviews that have delved into the matter of how consistently the finding can be reproduced, have noted that there is an inverse relationship between sample sizes and the associated likelihood of replication. This appears to be due to the larger degree of measurement error (associated with exposure) inherent to large studies 95. Small studies, which have fewer resources, shun large-scale recruitment, but place greater importance instead on maximising the accuracy

The next section is dedicated to exploring each of these aspects in greater detail.

adopted by studies whose scope does not extend beyond individual candidate genes.

Schizophrenia (for example, the UK10K study: www.uk10k.org/goals.html).


Table 3. AKT1 = Serine-threonine protein kinase; *BDNF* = Brain-derived neurotrophic factor; *CB1* = Cannabinoid receptor type 1; *CAPON* = Carboxyl-terminal PDZ ligand of neuronal nitric oxide; *CHRNA7* = Neuronal acetylcholine receptor subunit alpha-7; *COMT* = Catechol-O-methyltransferase; *DTNBP1* = dystrobrevin-binding protein 1; ESM = Experience Sampling Method. *GAD1* = Glutamate decarboxylase 1; *GRM3* = Metabotropic glutamate receptor 3; HC = Healthy Controls; *HLA* = Human Leukocyte Antigen; *MTHFR* = Methylenetetrahydrofolatereductase; *NOTCH4* = Neurogenic locus notch homolog protein 4; *NOS1AP* = Carboxyl-terminal PDZ ligand of neuronal nitric oxide synthase protein; *NRG1* = Neuregulin; *PRODH* = Prolinedehydrogenase; *RGS4* = Regulator of G protein signaling 4; SZ = Patients with Schizophrenia; *TNF-alpha* = Tumor necrosis factor.

The particulate nature of the molecular element to GxE interaction means that genetically pre-ordained outcomes could in theory be averted. The extent to which this is true depends on the effect sizes involved and whether the proposed intervention can be made in timely fashion. This has positive implications for public health. For example in some circumstances, it may be preferable to eliminate the environmental risk component altogether, rather than attempt the more tedious task of targeting genetic risk groups for a given intervention. Phenotype expression is normally suppressed as risk-inducing environmental exposures become scarce; this correlates with a decline in heritability and impacts on the number of diagnosed cases. The contextual nature of heritability can be exploited through use of the 'exposure only' study design 87, which facilitates the detection of environmentally-sensitive genetic variation. This approach is particularly powerful at the genome-wide level. The success of such strategies is determined by the extent of GxE contribution to the heritability of a given disorder.

A good illustration of the relationship between exposure and heritability comes from a US study that compared interstate influences on the heritability of teenage nicotine use. It was found that heavier state control of tobacco availability, through a combination of higher taxation, lower advertising and controlled vending machine supply, resulted in lower levels of detectable genetic influence on daily smoking 88. The high incidence of Schizophrenia could benefit from interventions in several areas of public policy. First and foremost would be those policies that made it more difficult to acquire cannabis, as this could reduce rates of Schizophrenia within genetically-prone sub-populations.

#### **6. Methodological constraints in GxE research**

86 Public Health – Methodology, Environmental and Systems Issues

Feelings of paranoia in daily life (ESM)

experiences (ESM)

Table 3. AKT1 = Serine-threonine protein kinase; *BDNF* = Brain-derived neurotrophic factor; *CB1* = Cannabinoid receptor type 1; *CAPON* = Carboxyl-terminal PDZ ligand of neuronal nitric oxide; *CHRNA7* = Neuronal acetylcholine receptor subunit alpha-7; *COMT* = Catechol-

Methylenetetrahydrofolatereductase; *NOTCH4* = Neurogenic locus notch homolog protein 4; *NOS1AP* = Carboxyl-terminal PDZ ligand of neuronal nitric oxide synthase protein; *NRG1* = Neuregulin; *PRODH* = Prolinedehydrogenase; *RGS4* = Regulator of G protein signaling 4; SZ = Patients with Schizophrenia; *TNF-alpha* = Tumor necrosis factor.

The particulate nature of the molecular element to GxE interaction means that genetically pre-ordained outcomes could in theory be averted. The extent to which this is true depends on the effect sizes involved and whether the proposed intervention can be made in timely fashion. This has positive implications for public health. For example in some circumstances, it may be preferable to eliminate the environmental risk component altogether, rather than attempt the more tedious task of targeting genetic risk groups for a given intervention. Phenotype expression is normally suppressed as risk-inducing environmental exposures become scarce; this correlates with a decline in heritability and impacts on the number of diagnosed cases. The contextual nature of heritability can be exploited through use of the 'exposure only' study design 87, which facilitates the detection of environmentally-sensitive genetic variation. This approach is particularly powerful at the genome-wide level. The success of such strategies is determined by the extent of GxE contribution to the heritability

A good illustration of the relationship between exposure and heritability comes from a US study that compared interstate influences on the heritability of teenage nicotine use. It was found that heavier state control of tobacco availability, through a combination of higher taxation, lower advertising and controlled vending machine supply, resulted in lower levels of detectable genetic influence on daily smoking 88. The high incidence of Schizophrenia could benefit from interventions in several areas of public policy. First and foremost would be those policies that made it more difficult to acquire cannabis, as this could reduce rates of

Schizophrenia within genetically-prone sub-populations.

O-methyltransferase; *DTNBP1* = dystrobrevin-binding protein 1; ESM = Experience Sampling Method. *GAD1* = Glutamate decarboxylase 1; *GRM3* = Metabotropic glutamate

receptor 3; HC = Healthy Controls; *HLA* = Human Leukocyte Antigen; *MTHFR* =

**Variable Results Statistics** 



delusions. -No interaction in controls.

p=0.002; =0.05 p=0.10; =0.02 p<0.001; =0.04 p=0.33; =0.05

p=<0.001; =0.77

p=0.01; 2=12.4 p=0.20; 2=3.3

**Author Sample size Candidate G Candidate E Outcome** 

*COMT*  Val(158)Met *BDNF*  Val(66)Met

*COMT* 

Stress (ESM)

Val(158)Met Stress (ESM) Psychotic

Simons et al 2009

vanWinkel et al 2008

of a given disorder.

579 young adult female twins (general population)

31 psychotic disorder + cannabis, 25 HC + cannabis

There are lingering doubts about the experimental validity of many GxE findings reported in the literature. This is in contrast to the renewed sense of optimism about genetic association, a method which focuses on direct gene effects rather than the consequences of their interactions. Genetic association studies now have GWAS to look to as a methodological reference point 4,38,40,41,50,56,89, and it is out of the technological infrastructure supporting GWAS, that two competing theories (they may not be mutually exclusive) regarding the genetic architecture of Schizophrenia have emerged: The Common Disease-Common Variant and Multiple Rare Variant hypotheses 4,9,10,58. The latter of these will be explored in great detail through new sequencing initiatives already underway for Schizophrenia (for example, the UK10K study: www.uk10k.org/goals.html).

A combination of meticulous study design, unprecedented sample sizes and good governance over methodological practice 90, now mean that replication is no longer the rarity it once was for genetic association studies (see table 3). This is in part due to the fact that genetic association studies are becoming more methodologically homogeneous; many of the rigorous methodological practices and standards routinely implemented in GWAS research (internally validated findings, population stratification, etc) have also been widely adopted by studies whose scope does not extend beyond individual candidate genes.

In contrast, the diversity of methodologies and standards used in GxE research has remained stubbornly heterogeneous to date; the multitude of study designs used to followup new discoveries, has seen only varying levels of success 91. Longitudinal studies sit very high within the complex methodological hierarchy of epidemiological designs, but even they are failing to provide the swift resolutions hoped for, to ongoing research questions of high importance (eg. Caspi vs Zammit) 92,93.

Several recurring factors limit the success rate of replication attempts in GxE research. These include:


The next section is dedicated to exploring each of these aspects in greater detail.

#### **6.1 Measurement error**

Arguably the most replicated GxE finding in Psychiatry belongs to the field of Depression, and involves the short allele of the Serotonin transporter gene (*5HTTLPR*) and Stressful Life Events (SLE), which interactively augment the risk of Depression 94. Reviews that have delved into the matter of how consistently the finding can be reproduced, have noted that there is an inverse relationship between sample sizes and the associated likelihood of replication. This appears to be due to the larger degree of measurement error (associated with exposure) inherent to large studies 95. Small studies, which have fewer resources, shun large-scale recruitment, but place greater importance instead on maximising the accuracy with which environmental exposures are measured.

Old Obstacles on New Horizons:

http://www.uk10k.org/).

disclosure of this information.

**6.4 Sample size** 

curse' 105.

credible 84.

strategy).

The Challenge of Implementing Gene X Environment Discoveries in Schizophrenia Research 89

assumptions 101 that may not necessarily hold true for Schizophrenia 85. A statistical

The difference between these two definitions (of Biological versus Statistical Interaction) can be problematic, as there remains plenty of scope for conflict between the two. In some cases discrepancies between the two may be artefactual. For example, the logarithmic transformation inherent to the multiplicative model can cause *bona fide* interactions to disappear, or else induce them spuriously 86, (an important caveat of this

These issues have fuelled a debate about the more appropriate way to scale interaction effects eg.85,86. Some of the rhetoric surrounding this issue is seemingly prejudicial to the question of whether GxE research can make a positive contribution towards Schizophrenia's translational goals 85. A key step to obtaining a definitive answer to this question will be the introduction of more systematic approaches to GxE discovery. The model for the type of approach needed is epitomised by GWAS 102,103. In the future this will be further complemented by the genome sequencing projects now underway in Schizophrenia 104 (also see details of the UK MRC's cross-disorder sequencing initiative, the UK10K study;

The tendency to overstate initial effect sizes results in a phenomenon known as 'winners'

Sample sizes in a replication study must accordingly be adjusted (upwards) to compensate for this associated loss of power. This is a practice embraced by the genetic association field of late, due to a combination of good governance 90 and a 'trickle down' of good research

A similar level of rigour is lacking from GxE research. This is worrying on two counts. Firstly, because from the outset, the power of an interaction analysis is typically lower than it is for a study of main effects 106, and secondly, the GxE field has tended to avoid facing such issues head on. This is typified by a reluctance among researchers to divulge vital information regarding statistical power in many instances 84. Such *faux pas* are propagated by the willingness of reviewers to accept such work, without enforcing appropriate

A recent critical appraisal shines a spotlight on the immediate shortcomings of GxE research in the psychiatric field 84. Its findings are still being digested by the psychiatric research community 107. A main accusation again centres on underpowering, (described by its authors to have skewed a decade's worth of research). Face value interpretation of their calculations suggests that average effect sizes would have to be 10 times larger than those normally found in GWAS, for the small sample sizes used to be even remotely

The problem of underpowering was found to have a bi-directional relationship with publication bias, (the tendency to only report trends that support a given hypothesis). The

etiquette from GWAS, through to mainstream (candidate gene) genetic research.

**7. 10 years of GxE research in psychiatry – A post-assessment review** 

interaction may still have great predictive value nonetheless.

Studies in which the SLE represents a single specific source of adversity, tend to extend the interaction trend, (even if they do not strictly reach the criteria of a 'replication' study) 91. This reaffirms the statistical importance of maintaining measurement error at low levels 96,97. Opportunistic replication studies, typically performed using cohorts not primarily intended to address the original research question, tend to be more detrimental to replication efforts, as even variables with the same name can reflect either subtly, or grossly different constructs.

#### **6.2 The distribution of genotypes and exposure**

The issue of replication is further complicated by the fact that, depending on the frequency of the exposure, the same GxE construct may range from having:


As genotypic frequency has a similar influence on interaction detection, it is only recommendable to attempt the reproduction of an interaction in samples where exposure and allelic frequencies compare with the original study. Additionally, power to detect interactions is optimal only when both minor allele frequencies and exposure rates are at the 50% level. Idealised distributions such as these are unlikely to occur under normal recruitment conditions, although they can be ensured by the use of selective sampling 98. Deviation away from these two statistical optima may, along with other methodological deficiencies, compromise the overall power of a GxE study.

#### **6.3 Effect sizes**

Biological interactions need not give any statistical clues to their existence. This is demonstrated by the example of Phenylketonuria (PKU), (a syndrome that gives rise to neurodevelopmental and psychiatric symptomatologies). PKU results from a combination of allelic deficiency in the gene encoding the phenylalanine hydroxylase enzyme, and dietary exposure to phenylalanine. In this case, any statistical trace of this biological interaction is obfuscated by the ubiquitous nature of phenylalanine in the human diet.

A typical GxE analysis requires large samples to facilitate the detection of targeted effects. A wider debate surrounds how these interactions should be scaled. In order to determine the presence of an interaction, a product term is added to the regression model. In linear regression, the regression coefficient of the product term defines interaction as departure from additivity, whereas an interaction using logistic regression indicates a departure from multiplicativity 99. An additive model is thought to best approximate the concept of biological interaction 100, though this view is heavily contentious. Meanwhile multiplicative effects, though more difficult to interpret, generally allude to larger effects on risk, and so are still predictively useful.

Biological validity remains a panacea for all GxE research, as the concept of a purely biological interaction is easy to understand and design interventions around (assuming the consequences of the interaction is large enough to merit this course of action). In contrast, inferring a mechanistic relationship out of a statistical effect, relies on conditions and assumptions 101 that may not necessarily hold true for Schizophrenia 85. A statistical interaction may still have great predictive value nonetheless.

The difference between these two definitions (of Biological versus Statistical Interaction) can be problematic, as there remains plenty of scope for conflict between the two. In some cases discrepancies between the two may be artefactual. For example, the logarithmic transformation inherent to the multiplicative model can cause *bona fide* interactions to disappear, or else induce them spuriously 86, (an important caveat of this strategy).

These issues have fuelled a debate about the more appropriate way to scale interaction effects eg.85,86. Some of the rhetoric surrounding this issue is seemingly prejudicial to the question of whether GxE research can make a positive contribution towards Schizophrenia's translational goals 85. A key step to obtaining a definitive answer to this question will be the introduction of more systematic approaches to GxE discovery. The model for the type of approach needed is epitomised by GWAS 102,103. In the future this will be further complemented by the genome sequencing projects now underway in Schizophrenia 104 (also see details of the UK MRC's cross-disorder sequencing initiative, the UK10K study; http://www.uk10k.org/).

#### **6.4 Sample size**

88 Public Health – Methodology, Environmental and Systems Issues

Studies in which the SLE represents a single specific source of adversity, tend to extend the interaction trend, (even if they do not strictly reach the criteria of a 'replication' study) 91. This reaffirms the statistical importance of maintaining measurement error at low levels 96,97. Opportunistic replication studies, typically performed using cohorts not primarily intended to address the original research question, tend to be more detrimental to replication efforts, as even variables with the same name can reflect either subtly, or grossly different

The issue of replication is further complicated by the fact that, depending on the frequency

As genotypic frequency has a similar influence on interaction detection, it is only recommendable to attempt the reproduction of an interaction in samples where exposure and allelic frequencies compare with the original study. Additionally, power to detect interactions is optimal only when both minor allele frequencies and exposure rates are at the 50% level. Idealised distributions such as these are unlikely to occur under normal recruitment conditions, although they can be ensured by the use of selective sampling 98. Deviation away from these two statistical optima may, along with other methodological

Biological interactions need not give any statistical clues to their existence. This is demonstrated by the example of Phenylketonuria (PKU), (a syndrome that gives rise to neurodevelopmental and psychiatric symptomatologies). PKU results from a combination of allelic deficiency in the gene encoding the phenylalanine hydroxylase enzyme, and dietary exposure to phenylalanine. In this case, any statistical trace of this biological interaction is

A typical GxE analysis requires large samples to facilitate the detection of targeted effects. A wider debate surrounds how these interactions should be scaled. In order to determine the presence of an interaction, a product term is added to the regression model. In linear regression, the regression coefficient of the product term defines interaction as departure from additivity, whereas an interaction using logistic regression indicates a departure from multiplicativity 99. An additive model is thought to best approximate the concept of biological interaction 100, though this view is heavily contentious. Meanwhile multiplicative effects, though more difficult to

interpret, generally allude to larger effects on risk, and so are still predictively useful.

Biological validity remains a panacea for all GxE research, as the concept of a purely biological interaction is easy to understand and design interventions around (assuming the consequences of the interaction is large enough to merit this course of action). In contrast, inferring a mechanistic relationship out of a statistical effect, relies on conditions and

constructs.

**6.3 Effect sizes** 

**6.2 The distribution of genotypes and exposure** 

i. no effect when the exposure is low,

iii. a main effect when the exposure is high 91.

of the exposure, the same GxE construct may range from having:

ii. statistical interaction when the exposure is moderate, or

deficiencies, compromise the overall power of a GxE study.

obfuscated by the ubiquitous nature of phenylalanine in the human diet.

The tendency to overstate initial effect sizes results in a phenomenon known as 'winners' curse' 105.

Sample sizes in a replication study must accordingly be adjusted (upwards) to compensate for this associated loss of power. This is a practice embraced by the genetic association field of late, due to a combination of good governance 90 and a 'trickle down' of good research etiquette from GWAS, through to mainstream (candidate gene) genetic research.

A similar level of rigour is lacking from GxE research. This is worrying on two counts. Firstly, because from the outset, the power of an interaction analysis is typically lower than it is for a study of main effects 106, and secondly, the GxE field has tended to avoid facing such issues head on. This is typified by a reluctance among researchers to divulge vital information regarding statistical power in many instances 84. Such *faux pas* are propagated by the willingness of reviewers to accept such work, without enforcing appropriate disclosure of this information.

#### **7. 10 years of GxE research in psychiatry – A post-assessment review**

A recent critical appraisal shines a spotlight on the immediate shortcomings of GxE research in the psychiatric field 84. Its findings are still being digested by the psychiatric research community 107. A main accusation again centres on underpowering, (described by its authors to have skewed a decade's worth of research). Face value interpretation of their calculations suggests that average effect sizes would have to be 10 times larger than those normally found in GWAS, for the small sample sizes used to be even remotely credible 84.

The problem of underpowering was found to have a bi-directional relationship with publication bias, (the tendency to only report trends that support a given hypothesis). The

Old Obstacles on New Horizons:

nature.

domain.

Genetics community 115.

only study.

The Challenge of Implementing Gene X Environment Discoveries in Schizophrenia Research 91

through a combination of biological theory and functional evidence 82. Given our rudimentary understanding of the complexity encoded at the genomic level, it is perhaps not so surprising that the doctrine of 'biological plausibility' is often questioned. Additional scepticism is reserved for the notion that the molecular dissection of psychiatric phenotypes can be formularised 82. This is a pertinent point, given that GWAS has shown us that the underlying biological basis of many complex and Mendelian traits is largely abstract in

Advocates of the biological plausibility doctrine can rightly point to the robust experimental and analytical settings in which several of these discoveries have been made 93,94,108. However detractors often cite the peculiarly low level of GWAS support for traditional Schizophrenia candidate gene favourites, (all of which are 'plausible' in one way or

The apparent discord between candidate-gene and GWAS findings is typical for most of Psychiatry, with very few exceptions 111 (convergent GWAS and candidate-gene findings in Schizophrenia are noted in table 2). If anything, GWAS has diverted attention towards less-obvious genomic points of interest, many of which lie within the non-coding

Thus the non-coding genome has proved to be a rich source of pathogenic variation; approximately 90% of all GWAS findings (across disorders) originate from there. But for now, the jury is still out regarding the possible contribution of first-generation candidate genes to the risk, pathology and outcome of Schizophrenia. The delay in implementing GxEWAS studies of Schizophrenia means that the relevance of historical genetic candidates to the GxE paradigm remains untested in modern-day genome-wide protocols. It is still premature therefore, to exclude a possible wider role for some of these genes in the

GxEWAS studies are steadily becoming entrenched in the literature. A number of neurodevelopmental and neurological phenotypes have already been investigated. These highlight interactions ranging from the effect of coffee-drinking on Parkinson's Disease, to the effect of adverse intrauterine environments on brain growth 112-114. As this innovative branch of genomics is yet to take off in Schizophrenia, the current crop of GxE findings both in table 3 and in other areas of Psychiatry, are still yet to face the same acid test used to put the previous generation of association candidates on trial 109,110. GxEWAS is currently one of many longer-term aspirations for policymakers in the Psychiatric

Several alternatives to standard Case-Control analysis methods will be at the disposal of the community by the time this occurs. Bayesian Case-control approaches already feature among them 116. However the Case-only model is currently considered to be the most effective (in terms of power and efficiency) methodology for this branch of research 117,118. The one proviso of the approach is that genes and exposure must be independent in the population from which cases are drawn 117,119. This condition can be tested directly, by repeating the GxE analytical procedure on controls, and appropriately filtering out signals (that cross the designated threshold of significance) from the case-

another), 109,110 to suggest the perils of a religious fixation on biological dogma 84.

aetiological or pathological course of Schizophrenia.

authors' report outlines an interesting chain of events, initiated by the instinctive preference among journal editors for novel findings. This distortion of the literature is sustained by additional biases that favour the publication of corroborating evidence, at which point statistical considerations such as power and study design are less rigorously enforced 84. Leniency in areas such as sample size and study design has long been self-evident in GxE research 91,95 but can, for the first time, be quantified; studies which have failed to replicate an existing discovery are, on average, 6 times larger than studies that did manage to replicate. This suggests that the sample-size threshold required for a negative finding to be published is 6x higher than that of a positive study 84.

One non-intuitive factor that such appraisals have failed to acknowledge is that samples characterised by a low *n* may also be those most immune from measurement error 91. For the 5HTTLPR x SLE interaction alone, low measurement error has been qualitatively shown to be the single most important determinant of a successful replication 91,95. Simulations of measurement error by Wong et al help to qualify this point 96. They suggest that an increase in correlation with true values of 'E' from .4 to .7 can equate to as much as a 20-fold gain in sample size. It is apparent therefore, that any review of the field must take into account the fact that the problem of a small sample can, to an extent, be overcome by maximising the precision of environmental measures. These days purposefully-designed tools (eg. http://www.hsph.harvard.edu/faculty/peter-kraft/software/ or the ESPRESSO power calculator at http://www.p3gobservatory.org/powercalculator.htm ) allow one to factor-in the variable precision of exposure measurement to estimations of power.

But in its defence, the Duncan-Keller assessment (a systematic assessment of 103 studies over a 10-year period) extends way beyond the Serotonin transporter. Therefore the critique is a formulation which applies to the field as a whole. Its take home message suggests that replication studies in Psychiatry currently only rarely achieve what they purport to, to a satisfactory standard.

This message is resounding, and also provides a convenient narrative for the poor progress made in bringing new findings to the clinic. At present it is largely explained by the shortage of high quality evidence entering the translational pipeline.

The crystallisation of lessons learned over the past 10 years 84 should be capitalised upon to make this a watershed moment for the application of GxE methodology in Psychiatry. However the type of cultural revolution needed can only be prompted by:


#### **8. New horizons in GxE research**

#### **8.1 GxEWAS: The systematically tractable meets the biologically plausible**

The archetypal approach to identifying potential GxE candidates avoids the statistical pitfalls of multiple testing, and is instead guided towards appropriate candidate regions nature.

90 Public Health – Methodology, Environmental and Systems Issues

authors' report outlines an interesting chain of events, initiated by the instinctive preference among journal editors for novel findings. This distortion of the literature is sustained by additional biases that favour the publication of corroborating evidence, at which point statistical considerations such as power and study design are less rigorously enforced 84. Leniency in areas such as sample size and study design has long been self-evident in GxE research 91,95 but can, for the first time, be quantified; studies which have failed to replicate an existing discovery are, on average, 6 times larger than studies that did manage to replicate. This suggests that the sample-size threshold required for a negative finding to be

One non-intuitive factor that such appraisals have failed to acknowledge is that samples characterised by a low *n* may also be those most immune from measurement error 91. For the 5HTTLPR x SLE interaction alone, low measurement error has been qualitatively shown to be the single most important determinant of a successful replication 91,95. Simulations of measurement error by Wong et al help to qualify this point 96. They suggest that an increase in correlation with true values of 'E' from .4 to .7 can equate to as much as a 20-fold gain in sample size. It is apparent therefore, that any review of the field must take into account the fact that the problem of a small sample can, to an extent, be overcome by maximising the precision of environmental measures. These days purposefully-designed tools (eg. http://www.hsph.harvard.edu/faculty/peter-kraft/software/ or the ESPRESSO power calculator at http://www.p3gobservatory.org/powercalculator.htm ) allow one to factor-in

But in its defence, the Duncan-Keller assessment (a systematic assessment of 103 studies over a 10-year period) extends way beyond the Serotonin transporter. Therefore the critique is a formulation which applies to the field as a whole. Its take home message suggests that replication studies in Psychiatry currently only rarely achieve what they purport to, to a

This message is resounding, and also provides a convenient narrative for the poor progress made in bringing new findings to the clinic. At present it is largely explained by the

The crystallisation of lessons learned over the past 10 years 84 should be capitalised upon to make this a watershed moment for the application of GxE methodology in Psychiatry.

i. An all-encompassing redefinition of what constitutes methodological good practice in GxE research 107 (this could be achieved by developing something equivalent to the STREGA (*ST*rengthening the *RE*porting of *G*enetic *A*ssociations) principles, specifically

ii. A consensus between journal editors, reviewers and researchers that these guidelines

The archetypal approach to identifying potential GxE candidates avoids the statistical pitfalls of multiple testing, and is instead guided towards appropriate candidate regions

the variable precision of exposure measurement to estimations of power.

shortage of high quality evidence entering the translational pipeline.

However the type of cultural revolution needed can only be prompted by:

**8.1 GxEWAS: The systematically tractable meets the biologically plausible** 

published is 6x higher than that of a positive study 84.

satisfactory standard.

for the GxE research.

should be adhered to.

**8. New horizons in GxE research** 

through a combination of biological theory and functional evidence 82. Given our rudimentary understanding of the complexity encoded at the genomic level, it is perhaps not so surprising that the doctrine of 'biological plausibility' is often questioned. Additional scepticism is reserved for the notion that the molecular dissection of psychiatric phenotypes can be formularised 82. This is a pertinent point, given that GWAS has shown us that the underlying biological basis of many complex and Mendelian traits is largely abstract in

Advocates of the biological plausibility doctrine can rightly point to the robust experimental and analytical settings in which several of these discoveries have been made 93,94,108. However detractors often cite the peculiarly low level of GWAS support for traditional Schizophrenia candidate gene favourites, (all of which are 'plausible' in one way or another), 109,110 to suggest the perils of a religious fixation on biological dogma 84.

The apparent discord between candidate-gene and GWAS findings is typical for most of Psychiatry, with very few exceptions 111 (convergent GWAS and candidate-gene findings in Schizophrenia are noted in table 2). If anything, GWAS has diverted attention towards less-obvious genomic points of interest, many of which lie within the non-coding domain.

Thus the non-coding genome has proved to be a rich source of pathogenic variation; approximately 90% of all GWAS findings (across disorders) originate from there. But for now, the jury is still out regarding the possible contribution of first-generation candidate genes to the risk, pathology and outcome of Schizophrenia. The delay in implementing GxEWAS studies of Schizophrenia means that the relevance of historical genetic candidates to the GxE paradigm remains untested in modern-day genome-wide protocols. It is still premature therefore, to exclude a possible wider role for some of these genes in the aetiological or pathological course of Schizophrenia.

GxEWAS studies are steadily becoming entrenched in the literature. A number of neurodevelopmental and neurological phenotypes have already been investigated. These highlight interactions ranging from the effect of coffee-drinking on Parkinson's Disease, to the effect of adverse intrauterine environments on brain growth 112-114. As this innovative branch of genomics is yet to take off in Schizophrenia, the current crop of GxE findings both in table 3 and in other areas of Psychiatry, are still yet to face the same acid test used to put the previous generation of association candidates on trial 109,110. GxEWAS is currently one of many longer-term aspirations for policymakers in the Psychiatric Genetics community 115.

Several alternatives to standard Case-Control analysis methods will be at the disposal of the community by the time this occurs. Bayesian Case-control approaches already feature among them 116. However the Case-only model is currently considered to be the most effective (in terms of power and efficiency) methodology for this branch of research 117,118. The one proviso of the approach is that genes and exposure must be independent in the population from which cases are drawn 117,119. This condition can be tested directly, by repeating the GxE analytical procedure on controls, and appropriately filtering out signals (that cross the designated threshold of significance) from the caseonly study.

Old Obstacles on New Horizons:

validating discoveries made elsewhere.

The primary functions of a biobank include:



and type 2 Diabetes 65.

data.

often elusive.

The Challenge of Implementing Gene X Environment Discoveries in Schizophrenia Research 93

Observations by Caspi 91, Uher and McGuffin 95, Vineis 97 and Wong 96 collectively highlight the challenge of balancing sample size and measurement error for optimal statistical benefit. It is in this respect that the Dunedin study (to which a disproportionate number of GxE discoveries belong) enjoys an unparalleled advantage over many of the cohorts that have since revisited the original *5HTTLPR* finding. The study combines the higher accuracy of exposure measurement often found in smaller studies, with a large sample size that is so

A large number of replication studies do not share this same rare-but-optimal combination of properties 91,96,97. It is this variability which may be incapacitating to the field as a whole. Such problems can be addressed by applying greater epidemiological rigour to the collection, storage and power of genetic datasets. The rapid proliferation of biobanks in biomedical research is accompanied by the expectation that this will directly improve the quality of translational research, (and not just for Schizophrenia). Biobanks provide a means to satisfy the growing demand for high quality population data, thus they will be a key driver of genetic discovery in the future. They will also be an essential resource for

Of course genetics is just one of many important biological areas that can be served by such resources. This is why the rapid proliferation of biobanks is vital, even for the many nonpsychiatric traits that have, to a large degree, already profited from GWAS. This includes traits such as Age-related Macular Degeneration, Prostate Cancer, Coronary Heart disease

A recurrent concern among commentators in the GxE field is the increased scope for measurement error in these heterogeneously-assembled datasets 91. Additional problems may occur due to the fact that geneticists, epidemiologists, biologists and biostatisticians, often use different vocabularies 123. Extrapolating these issues to the large number of biobanks in existence around the globe suggests that there is a need for overall governance to maximise data harmonisation. A large number of international bodies have been created for this purpose, many with overlapping functions. For example in Europe, PHOEBE (Promoting Harmonisation Of Epidemiological Biobanks in Europe), ENGAGE (European Network of Genomic and Genetic Epidemiology) P3G (Public Population Project in Genomics), are three independent organisations that provide a continent-wide consensus on procedures ranging from collection, storage and format of biological samples and associated

Perhaps this overlap is needed to counteract the organisational absence of other major institutions from this exercise. Regulatory bodies such as the European Medicines Agency (EMA) and The Food and Drug Administration (FDA) were at some point considered, but ultimately deemed too inherently conservative to oversee such a task 124. Top-down

**8.2 Strategies for data harmonisation and how this will help** 

Post-genomic technological advances, namely the advent of micro-array technology, have led to huge increases in the scale at which genetic variation can be sampled from a genome by a single study. The abundance of this data can propel the formulation of *post-hoc* hypotheses based on biological plausibility. Useful resources that can help to inform the decision-making process include tools such as the UCSC and ENSEMBL genome browsers (http://genome.ucsc.edu/cgi-bin/hgGateway and http://www.ensembl.org/index.html). These contain a wealth of information highlighting the organisation, structure and function of the genome. Other specialist resources provide a dense functional annotation of regions that border GWAS hits (http://jjwanglab.org:8080/gwasdb/) 120.

One area in which Schizophrenia genetic research has been slow (compared to other fields such as Alzheimer's Disease), is its readiness to combine genetics with other flavours of system biology that can now be feasibly explored. This multi-level approach could provide insights about fundamental bio-mechanic processes that lie at the heart of gene-environment interaction.

One potential class of mediaries are known as Quantitative Trait Loci (QTLs). These are regulatory variants associated with control of gene-expression (eQTLs), protein levels (pQTLs) and gene activation status (methQTLs).

The ever-decreasing cost of implementing these system-based biological approaches continues to increase their accessibility. Meanwhile, whole-genome sequencing provides the means to increase both the resolution of regulatory variants across the genome, and the fuel for further biological hypotheses.

A key objective within the universal objectives of personalised medicine (to which the field of Psychiatric Genetics is also subscribed) is to enhance both the visibility and efficiency with which promising new evidence is vetted and then turned into new diagnostics and treatments. Crucially however, neither a purely biological, nor a purely systematic approach, (such as GxEWAS) can secure these goals alone. This is due to two main reasons:


The many lines of derivative research resulting from GWAS in Schizophrenia collectively demonstrate how both systematic and biological candidate approaches can work in tandem 55,103,115,121,122. Thus, an emphasis on *post-hoc* explorations of candidate pathways, genes and variants may be the best bet for turning a cursory screen of the genome (such as GxEWAS) into something that is potentially much more substantive. This kind of combinatorial approach, which marries systematic and hypothesis-led discovery through data-mining, may one day reveal (and explain) the true pervasiveness of GxE in Schizophrenia.

#### **8.2 Strategies for data harmonisation and how this will help**

92 Public Health – Methodology, Environmental and Systems Issues

Post-genomic technological advances, namely the advent of micro-array technology, have led to huge increases in the scale at which genetic variation can be sampled from a genome by a single study. The abundance of this data can propel the formulation of *post-hoc* hypotheses based on biological plausibility. Useful resources that can help to inform the decision-making process include tools such as the UCSC and ENSEMBL genome browsers (http://genome.ucsc.edu/cgi-bin/hgGateway and http://www.ensembl.org/index.html). These contain a wealth of information highlighting the organisation, structure and function of the genome. Other specialist resources provide a dense functional annotation of regions

One area in which Schizophrenia genetic research has been slow (compared to other fields such as Alzheimer's Disease), is its readiness to combine genetics with other flavours of system biology that can now be feasibly explored. This multi-level approach could provide insights about fundamental bio-mechanic processes that lie at the heart of gene-environment

One potential class of mediaries are known as Quantitative Trait Loci (QTLs). These are regulatory variants associated with control of gene-expression (eQTLs), protein levels

The ever-decreasing cost of implementing these system-based biological approaches continues to increase their accessibility. Meanwhile, whole-genome sequencing provides the means to increase both the resolution of regulatory variants across the genome, and the fuel

A key objective within the universal objectives of personalised medicine (to which the field of Psychiatric Genetics is also subscribed) is to enhance both the visibility and efficiency with which promising new evidence is vetted and then turned into new diagnostics and treatments. Crucially however, neither a purely biological, nor a purely systematic approach, (such as GxEWAS) can secure these goals alone. This is due to two



The many lines of derivative research resulting from GWAS in Schizophrenia collectively demonstrate how both systematic and biological candidate approaches can work in tandem 55,103,115,121,122. Thus, an emphasis on *post-hoc* explorations of candidate pathways, genes and variants may be the best bet for turning a cursory screen of the genome (such as GxEWAS) into something that is potentially much more substantive. This kind of combinatorial approach, which marries systematic and hypothesis-led discovery through data-mining, may one day reveal (and explain) the true pervasiveness of GxE in

mechanisms to assume we can become routinely successful at this.

that border GWAS hits (http://jjwanglab.org:8080/gwasdb/) 120.

(pQTLs) and gene activation status (methQTLs).

up being repeated, on a yet grander scale 91.

for further biological hypotheses.

interaction.

main reasons:

Schizophrenia.

Observations by Caspi 91, Uher and McGuffin 95, Vineis 97 and Wong 96 collectively highlight the challenge of balancing sample size and measurement error for optimal statistical benefit. It is in this respect that the Dunedin study (to which a disproportionate number of GxE discoveries belong) enjoys an unparalleled advantage over many of the cohorts that have since revisited the original *5HTTLPR* finding. The study combines the higher accuracy of exposure measurement often found in smaller studies, with a large sample size that is so often elusive.

A large number of replication studies do not share this same rare-but-optimal combination of properties 91,96,97. It is this variability which may be incapacitating to the field as a whole.

Such problems can be addressed by applying greater epidemiological rigour to the collection, storage and power of genetic datasets. The rapid proliferation of biobanks in biomedical research is accompanied by the expectation that this will directly improve the quality of translational research, (and not just for Schizophrenia). Biobanks provide a means to satisfy the growing demand for high quality population data, thus they will be a key driver of genetic discovery in the future. They will also be an essential resource for validating discoveries made elsewhere.

Of course genetics is just one of many important biological areas that can be served by such resources. This is why the rapid proliferation of biobanks is vital, even for the many nonpsychiatric traits that have, to a large degree, already profited from GWAS. This includes traits such as Age-related Macular Degeneration, Prostate Cancer, Coronary Heart disease and type 2 Diabetes 65.

The primary functions of a biobank include:


A recurrent concern among commentators in the GxE field is the increased scope for measurement error in these heterogeneously-assembled datasets 91. Additional problems may occur due to the fact that geneticists, epidemiologists, biologists and biostatisticians, often use different vocabularies 123. Extrapolating these issues to the large number of biobanks in existence around the globe suggests that there is a need for overall governance to maximise data harmonisation. A large number of international bodies have been created for this purpose, many with overlapping functions. For example in Europe, PHOEBE (Promoting Harmonisation Of Epidemiological Biobanks in Europe), ENGAGE (European Network of Genomic and Genetic Epidemiology) P3G (Public Population Project in Genomics), are three independent organisations that provide a continent-wide consensus on procedures ranging from collection, storage and format of biological samples and associated data.

Perhaps this overlap is needed to counteract the organisational absence of other major institutions from this exercise. Regulatory bodies such as the European Medicines Agency (EMA) and The Food and Drug Administration (FDA) were at some point considered, but ultimately deemed too inherently conservative to oversee such a task 124. Top-down

Old Obstacles on New Horizons:

and diet).

previously reached.

**9.1 Background** 

**9. New horizons in pharmacogenomic research** 

complex for breakdown (see table 4).

sustain the high level of psychosis in the general population.

The Challenge of Implementing Gene X Environment Discoveries in Schizophrenia Research 95

The true potential of the EMR model will become more apparent only when highdimensional genetic and molecular profiling becomes economically feasible and clinically routine. This will make it practically possible to integrate a whole manner of clinical data into diagnostic/prognostic genetic research. But such times are almost already upon us, thus we do not have far to search, to find examples of how an integrative approach may work in practice. One such model is that of Ashley and colleagues 132, who recently reported a far-reaching genomic health assessment of a patient showing a strong familial indication of Coronary Heart Disease and Sudden Death Syndrome. A graphical account of the relative genetic liability for other disorders (Coronary Artery Disease, Obesity, Osteoarthritis and Type 2 Diabetes) depicts the genetic relationship between these disorders and several conditional environmental risk exposures, (stress, smoking, exercise

For Schizophrenia, a more precise account of the relationship with environmental risk factors could be achieved with the help of a new generation of instruments (questionnaires) and devices that will enable their measurement to be conducted with greater sensitivity than ever. Many examples of these have been devised for a large multi-centre study: The European network study of Gene-Environment Interaction (EUGEI) 5. Of particular relevance is a work package entitled 'Functional Enviromics', which aims to take the elucidation of socio-environmental risk factors for Schizophrenia to a level of resolution not

One consequence of GxE interaction is that any undesired outcomes can be averted through interventions targeted at the level of the individual, or the population, through changes in wider socio-economic policy. Primary avenues of social intervention for Schizophrenia would include redressing social inequalities 2, as well as challenging permissive attitudes to the use of illegal psychotogenic substances which, in tandem with other risk factors, help to

Meanwhile, molecular strategies for moderating or ameliorating the detrimental consequences of GxE interaction, fall within an area of personalised medicine known as Pharmacogenomics. This discipline is concerned with devising optimal therapeutic treatments for genetic sub-groups of patients. A competing goal is to minimise the risk of ill effects resulting from such treatments. Large inter-individual variability in both drug response and side-effects are the main foundation for this branch of research 133. Much of this variability can be traced to genetic variation within key liver enzymes (the cytochrome P450 complex). It is the fate of all antipsychotic drugs to be channelled to this biological

Of all the enzymes known to have a role in the metabolism of antipsychotic drugs, *CYP2D6*  has been the most extensively characterised. This is not a great surprise, given that the protein product of this gene catalyses the breakdown of up to 25% of all pharmacological compounds. Current knowledge about functional variation within this gene alone is enough

implementation of new and emerging international standards and protocols for data collection, sample acquisition, etc is managed by national biobanking initiatives, such as the UK Biobank. Policies may then be channelled down to a set of regional hubs such as the National Institute for Health Research's Biomedical Research Centres (UK). It is encouraging that Schizophrenia research is now beginning to derive the benefits of biobank-based research 125-127.

#### **8.3 A note on methods for research synthesis**

Such initiatives inevitably generate an abundance of data. A critical mass of high quality data is usually the trigger for the synthesisation of this evidence to begin. This typically uses meta-analysis, whose conventional format uses the null hypothesis (a construct of frequentist statistical theory) as its reference point. However the rationale for this becomes increasingly questionable as new evidence is added to an existing literature 128,129. A Bayesian approach (ie. one that would allow the posterior probability of a hypothesis to be derived from prior knowledge, after taking into account new data), would allow any uncertainty about a hypothesis, to be acknowledged in an adaptive way.

The conspicuous absence of Bayesian methods from the science of data synthesis was only recently lamented, by key stakeholders involved in the process of evaluating new drugs for the UK's National Health Service 128. Such messages may yet help to expedite the uptake of these methods, although there is already evidence of their adoption in clinical trial research 128. These methods could widen the net used to gather new evidence, by allowing the incorporation of data from *in vivo* and cellular studies into the evaluation process. Thus Bayesian methodologies could provide an important means of channelling a wide range of functional evidence into synthesised data 130, as well as providing an alternative set of rules for assessing the validity of a hypothesis.

#### **8.4 The future of clinical databases in psychiatric GxE research**

It will soon be much easier to harvest the valuable clinical data derived out of even routine patient contact with clinical services, given that a switch-over to electronic medical records (EMRs) is now underway. The integrative blueprint for the new digital clinical age would allow a comprehensive (clinical, molecular and environmental risk profile) to be compiled for each patient. The front-end portal for this is as a personal record that follows the individual around as they move between different mental health institutions. Back-end access to such data is possible (for research purposes), but necessarily anonymised. The information itself can be processed in a way that allows even the interrogation of unstructured data (eg. clinical notes) to now be formularised (eg. see 131). The huge potential of EMRs represents great scope for integrative research. It is anticipated that such resources will:


The true potential of the EMR model will become more apparent only when highdimensional genetic and molecular profiling becomes economically feasible and clinically routine. This will make it practically possible to integrate a whole manner of clinical data into diagnostic/prognostic genetic research. But such times are almost already upon us, thus we do not have far to search, to find examples of how an integrative approach may work in practice. One such model is that of Ashley and colleagues 132, who recently reported a far-reaching genomic health assessment of a patient showing a strong familial indication of Coronary Heart Disease and Sudden Death Syndrome. A graphical account of the relative genetic liability for other disorders (Coronary Artery Disease, Obesity, Osteoarthritis and Type 2 Diabetes) depicts the genetic relationship between these disorders and several conditional environmental risk exposures, (stress, smoking, exercise and diet).

For Schizophrenia, a more precise account of the relationship with environmental risk factors could be achieved with the help of a new generation of instruments (questionnaires) and devices that will enable their measurement to be conducted with greater sensitivity than ever. Many examples of these have been devised for a large multi-centre study: The European network study of Gene-Environment Interaction (EUGEI) 5. Of particular relevance is a work package entitled 'Functional Enviromics', which aims to take the elucidation of socio-environmental risk factors for Schizophrenia to a level of resolution not previously reached.

#### **9. New horizons in pharmacogenomic research**

#### **9.1 Background**

94 Public Health – Methodology, Environmental and Systems Issues

implementation of new and emerging international standards and protocols for data collection, sample acquisition, etc is managed by national biobanking initiatives, such as the UK Biobank. Policies may then be channelled down to a set of regional hubs such as the National Institute for Health Research's Biomedical Research Centres (UK). It is encouraging that Schizophrenia

Such initiatives inevitably generate an abundance of data. A critical mass of high quality data is usually the trigger for the synthesisation of this evidence to begin. This typically uses meta-analysis, whose conventional format uses the null hypothesis (a construct of frequentist statistical theory) as its reference point. However the rationale for this becomes increasingly questionable as new evidence is added to an existing literature 128,129. A Bayesian approach (ie. one that would allow the posterior probability of a hypothesis to be derived from prior knowledge, after taking into account new data), would allow any

The conspicuous absence of Bayesian methods from the science of data synthesis was only recently lamented, by key stakeholders involved in the process of evaluating new drugs for the UK's National Health Service 128. Such messages may yet help to expedite the uptake of these methods, although there is already evidence of their adoption in clinical trial research 128. These methods could widen the net used to gather new evidence, by allowing the incorporation of data from *in vivo* and cellular studies into the evaluation process. Thus Bayesian methodologies could provide an important means of channelling a wide range of functional evidence into synthesised data 130, as well as providing an alternative set of rules

It will soon be much easier to harvest the valuable clinical data derived out of even routine patient contact with clinical services, given that a switch-over to electronic medical records (EMRs) is now underway. The integrative blueprint for the new digital clinical age would allow a comprehensive (clinical, molecular and environmental risk profile) to be compiled for each patient. The front-end portal for this is as a personal record that follows the individual around as they move between different mental health institutions. Back-end access to such data is possible (for research purposes), but necessarily anonymised. The information itself can be processed in a way that allows even the interrogation of unstructured data (eg. clinical notes) to now be formularised (eg. see 131). The huge potential of EMRs represents great scope for integrative research. It is anticipated that such resources




research is now beginning to derive the benefits of biobank-based research 125-127.

uncertainty about a hypothesis, to be acknowledged in an adaptive way.

**8.4 The future of clinical databases in psychiatric GxE research** 

**8.3 A note on methods for research synthesis** 

for assessing the validity of a hypothesis.

*silico* (bioinformatically-oriented) studies.

will:

prognoses.

One consequence of GxE interaction is that any undesired outcomes can be averted through interventions targeted at the level of the individual, or the population, through changes in wider socio-economic policy. Primary avenues of social intervention for Schizophrenia would include redressing social inequalities 2, as well as challenging permissive attitudes to the use of illegal psychotogenic substances which, in tandem with other risk factors, help to sustain the high level of psychosis in the general population.

Meanwhile, molecular strategies for moderating or ameliorating the detrimental consequences of GxE interaction, fall within an area of personalised medicine known as Pharmacogenomics. This discipline is concerned with devising optimal therapeutic treatments for genetic sub-groups of patients. A competing goal is to minimise the risk of ill effects resulting from such treatments. Large inter-individual variability in both drug response and side-effects are the main foundation for this branch of research 133. Much of this variability can be traced to genetic variation within key liver enzymes (the cytochrome P450 complex). It is the fate of all antipsychotic drugs to be channelled to this biological complex for breakdown (see table 4).

Of all the enzymes known to have a role in the metabolism of antipsychotic drugs, *CYP2D6*  has been the most extensively characterised. This is not a great surprise, given that the protein product of this gene catalyses the breakdown of up to 25% of all pharmacological compounds. Current knowledge about functional variation within this gene alone is enough

Old Obstacles on New Horizons:

**Phase 1: Discovery and Clinical validity** 

**Phase 2: Clinical Utility to Clinical guidelines** 

**Phase 3: Implementation in Clinical practice** 

(Adapted from references 138,139)

**9.3 Regulation and decision-making** 

observed in a recent Danish study 142.

(FDA) in the US.

The Challenge of Implementing Gene X Environment Discoveries in Schizophrenia Research 97

**Translation Research Phase Example Study Approach to overcoming phase** 

Phases I and II clinical trials; observational studies

Phase III clinical trials; observational studies; evidence synthesis and guidelines development

Dissemination and implementation research; Phase IV clinical trials

Outcomes research; Population monitoring; Phase IV clinical trials

eg. Reliable series of associations between a SNP and drug response

Does SNP improve drug response and what is its predictive accuracy?

Explore data regarding the uptake of the SNP test in clinical settings - explore potential barriers

outcome in the population?

Although this framework has been developed to support emerging new pharmacogenomic technologies, devices and treatments, its generic nature means it provides a model that is also extrapolable across genetic research (including GxE). The model adopts the ACCE (Analytical validity, Clinical Validity, Clinical Utility) and ELSI (Ethical, legal, social issues) criteria to ensure a rigorously vetted transition between phases 139. The solid foundation provided by the framework will help to ensure that promising findings do not become 'lost in translation' 140, a problem that has characterised the last 60 years of drug development. This issue still continues to affect the industry acutely: It takes an average of 17 years for just 14% of new scientific discoveries to enter day-to-day clinical practice 137, while the cost per

Regulatory governance fulfils several objectives, the most important of which is to ensure that patients and research subjects are protected from any undesired consequences ('adverse events') of new drugs intended for the market. GxE discoveries that make it into clinical evaluation phases fall under the jurisdiction of various geographical regulatory institutions such as the European Medicines Agency (EMA) in Europe, the Medicines and Healthcare products Regulatory Agency (MHRA) in the UK, and the Food and Drug Administration

Adherence to the process of regulation is essential for ensuring a smooth progression through the translation scheme outlined in table 5. For instance, failing to procure accreditation for genetic tests and therapies from decision-making bodies such as the EMA and the FDA tends to adversely affect the uptake of these innovations in other global regions. This may partly explain the poor uptake of CYP2D6 and CYP2C19 genetic tests

However, over-zealous regulation can itself create obstacles, particularly if perceived to be of no discernible benefit to patients 143. This has potentially been the case in Europe, where the much-criticised 2001 European Union Clinical Trial Directive has caused the cost of

Table 5. Table 4 shows the 4 phases of clinical translation and the critical approaches required to negotiate each one. Though initially designed to provide a translational model for pharmacogenetic research, it can also be applied in the context of GxE research.

successful drug exceeds \$1billion, after adjusting for all the failures 141.

**Table 5. The 4 phases of clinical translation** 

**Phase 4: Public Health Impact** Does SNP improve clinical

to explain inter-individual differences in drug efficacy. For example, 2% of Caucasians and 25% of East Africans who express multiple functional CYP2D6 alleles, (ultra-rapid metabolisers) can be phenotypically distinguished on account of having the poorest levels of response to specific treatments 134,135. Unfortunately however, current assessments of the clinical utility of pharmacogenetic testing in Schizophrenia, suggest that a heavy reliance on CYP2D6 genotyping is currently not the most beneficial way to formulate pescribing guidelines regarding the use of antipsychotic drugs 134. A similar study of CYP2D6 (looking at Selective Serotonin Re-uptake Inhibitor treatments in Depression), recently came to a similar conclusion 136.


#### **Table 4. Commonly used antipsychotics metabolised by CYP enzymes**

Table 4. (Adapted from reference 134)

#### **9.2 A generalisable translation framework for GxE discovery**

Poor performance of novel findings across different formulations of synthesised data represents an obvious barrier to clinical translation. But even if this obstacle can be overcome, a further series of hurdles may replace it. A clear framework now exists to prompt and signpost the long path between discovery and clinical application 137. Implementation of the framework is marshalled by the Human Genome Epidemiology Network (HUGENET), a global collaboration of individuals and organisations whose remit is to assess the impact of genomic variation on population health. According to HUGENET, the pathway to clinical translation can be divided into four key stages (see table 5).


#### **Table 5. The 4 phases of clinical translation**

96 Public Health – Methodology, Environmental and Systems Issues

to explain inter-individual differences in drug efficacy. For example, 2% of Caucasians and 25% of East Africans who express multiple functional CYP2D6 alleles, (ultra-rapid metabolisers) can be phenotypically distinguished on account of having the poorest levels of response to specific treatments 134,135. Unfortunately however, current assessments of the clinical utility of pharmacogenetic testing in Schizophrenia, suggest that a heavy reliance on CYP2D6 genotyping is currently not the most beneficial way to formulate pescribing guidelines regarding the use of antipsychotic drugs 134. A similar study of CYP2D6 (looking at Selective Serotonin Re-uptake Inhibitor treatments in Depression), recently came to a

**Table 4. Commonly used antipsychotics metabolised by CYP enzymes** 

**Enzyme Typical Antipsychotics Atypical Antipsychotics**  *CYP2D6* **Primary metabolism Primary metabolism**  Chlorpromazine Risperidone

Perphenazine **Secondary Metabolism**  Thioridazine Olanzapine Quetiapine

*CYP1A2* **Primary metabolism Primary metabolism**  Chlorpromazine Clozapine Perphenazine Olanzapine

*CYP3A4* **Primary metabolism Primary metabolism**  Haloperidol Quetiapine Ziprasidone **Secondary Metabolism**  Clozapine Olanzapine Risperidone

Poor performance of novel findings across different formulations of synthesised data represents an obvious barrier to clinical translation. But even if this obstacle can be overcome, a further series of hurdles may replace it. A clear framework now exists to prompt and signpost the long path between discovery and clinical application 137. Implementation of the framework is marshalled by the Human Genome Epidemiology Network (HUGENET), a global collaboration of individuals and organisations whose remit is to assess the impact of genomic variation on population health. According to HUGENET, the pathway to clinical translation can be divided into four key stages (see

similar conclusion 136.

table 5).

Haloperidol

Thioridazine

Table 4. (Adapted from reference 134)

**Secondary Metabolism**  Haloperidol Perphenazine

**9.2 A generalisable translation framework for GxE discovery** 

**Secondary Metabolism**  Zuclopenthixol

Table 5. Table 4 shows the 4 phases of clinical translation and the critical approaches required to negotiate each one. Though initially designed to provide a translational model for pharmacogenetic research, it can also be applied in the context of GxE research. (Adapted from references 138,139)

Although this framework has been developed to support emerging new pharmacogenomic technologies, devices and treatments, its generic nature means it provides a model that is also extrapolable across genetic research (including GxE). The model adopts the ACCE (Analytical validity, Clinical Validity, Clinical Utility) and ELSI (Ethical, legal, social issues) criteria to ensure a rigorously vetted transition between phases 139. The solid foundation provided by the framework will help to ensure that promising findings do not become 'lost in translation' 140, a problem that has characterised the last 60 years of drug development. This issue still continues to affect the industry acutely: It takes an average of 17 years for just 14% of new scientific discoveries to enter day-to-day clinical practice 137, while the cost per successful drug exceeds \$1billion, after adjusting for all the failures 141.

#### **9.3 Regulation and decision-making**

Regulatory governance fulfils several objectives, the most important of which is to ensure that patients and research subjects are protected from any undesired consequences ('adverse events') of new drugs intended for the market. GxE discoveries that make it into clinical evaluation phases fall under the jurisdiction of various geographical regulatory institutions such as the European Medicines Agency (EMA) in Europe, the Medicines and Healthcare products Regulatory Agency (MHRA) in the UK, and the Food and Drug Administration (FDA) in the US.

Adherence to the process of regulation is essential for ensuring a smooth progression through the translation scheme outlined in table 5. For instance, failing to procure accreditation for genetic tests and therapies from decision-making bodies such as the EMA and the FDA tends to adversely affect the uptake of these innovations in other global regions. This may partly explain the poor uptake of CYP2D6 and CYP2C19 genetic tests observed in a recent Danish study 142.

However, over-zealous regulation can itself create obstacles, particularly if perceived to be of no discernible benefit to patients 143. This has potentially been the case in Europe, where the much-criticised 2001 European Union Clinical Trial Directive has caused the cost of

Old Obstacles on New Horizons:

Nature 454, 824 (2008).

Nat Genet 43, 969-976 (2011).

Hum Mol Genet 19, 3477-3481 (2010).

Twin Res Hum Genet 10, 423-433 (2007).

dependence. Addiction 101, 801-812 (2006).

review. Am J Psychiatry 162, 12-24 (2005).

Schizophr Bull 32, 592-598 (2006).

Psychiatry (2011).

339 (2010).

(2004).

Biotechnol.(2012)

1082 (2008).

**11. References** 

The Challenge of Implementing Gene X Environment Discoveries in Schizophrenia Research 99

[1] McGrath, J.J. & Selten, J.P. Mental health: don't overlook environment and its risk factors.

[2] Kirkbride, J., et al. Translating the epidemiology of psychosis into public mental health: evidence, challenges and future prospects. J Public Ment Health 9, 4-14 (2010). [3] Rujescu, D., Genius, J., Benninghoff, J. & Giegling, I. Current progress in the genetic

[4] Ripke, S., et al. Genome-wide association study identifies five new schizophrenia loci.

[5] van Os, J., Rutten, B.P. & Poulton, R. Gene-environment interactions in schizophrenia:

[6] Weis, B.K., et al. Personalized exposure assessment: promising approaches for human environmental health research. Environ Health Perspect 113, 840-848 (2005). [7] Cardno, A.G., et al. Heritability estimates for psychotic disorders: the Maudsley twin

[8] Lichtenstein, P., et al. Common genetic determinants of schizophrenia and bipolar

[9] Vassos, E., et al. Penetrance for copy number variants associated with schizophrenia.

[10] Grozeva, D., et al. Independent estimation of the frequency of rare CNVs in the UK population confirms their role in schizophrenia. Schizophr Res (2011). [11] Dick, D.M., Riley, B. & Kendler, K.S. Nature and nurture in neuropsychiatric genetics:

[12] Bergen, S.E., Gardner, C.O. & Kendler, K.S. Age-related changes in heritability of

[13] Pagan, J.L., et al. Genetic and environmental influences on stages of alcohol use across adolescence and into young adulthood. Behav Genet 36, 483-497 (2006). [14] Agrawal, A. & Lynskey, M.T. The genetic epidemiology of cannabis use, abuse and

[15] Scherr, M., et al. Environmental risk factors and their impact on the age of onset of

[17] Allardyce, J. & Boydell, J. Review: the wider social environment and schizophrenia.

[18] Cantor-Graae, E. & Selten, J.P. Schizophrenia and migration: a meta-analysis and

[19] Dealberto, M.J. Ethnic origin and increased risk for schizophrenia in immigrants to

[20] Henquet, C., Di Forti, M., Morrison, P., Kuepper, R. & Murray, R.M. Gene-environment interplay between cannabis and psychosis. Schizophr Bull 34, 1111-1121 (2008). [21] Arseneault, L., Cannon, M., Witton, J. & Murray, R.M. Causal association between

psychosis series. Arch Gen Psychiatry 56, 162-168 (1999).

where do we stand? Dialogues Clin Neurosci 12, 7-23 (2010).

[16] March, D., et al. Psychosis and place. Epidemiol Rev 30, 84-100 (2008).

research of schizophrenia: relevance for drug discovery? Curr Pharm

review of epidemiological findings and future directions. Schizophr Bull 34, 1066-

disorder in Swedish families: a population-based study. Lancet 373, 234-239 (2009).

behavioral phenotypes over adolescence and young adulthood: a meta-analysis.

schizophrenia: Comparing familial to non-familial schizophrenia. Nord J

countries of recent and longstanding immigration. Acta Psychiatr Scand 121, 325-

cannabis and psychosis: examination of the evidence. Br J Psychiatry 184, 110-117

running clinical trials to spiral. Other knock-on effects attributed to the legislation include a 30% decline in the numbers of participants agreeing to take part in trials across Europe, over the last few years 144. As clinical trials are an integral component within any translation scheme, such problems threaten to create a fatal bottle-neck in the pipeline, for discoveries that might otherwise have made it through the process relatively unscathed.

An overhaul of regulatory governance at national level has been proposed to circumvent this problem. In the UK, it is being done in conjunction with The National Institute for Health and Clinical Excellence (NICE), an organisation primarily responsible for assessing the cost-effectiveness, on behalf of the National Health Service (NHS), of providing new therapies and treatments. However the change of UK government means it is not even clear that there is a timetable for putting such proposals into practice 143.

As just hinted at, all novel genetic disoveries (including GxE interactions) that have safely negotiated the rigours of the validation stages shown in table 5, must still run the gauntlet of proving their overall cost-effectivenesss, before they can progress beyond validity, into utility. But new technology and treatments can only be considered to be cost-effective if their health benefits can be shown to outweigh the opportunity costs of services or treatments that they may displace 145. When viewed in the context of the many benefits that personalised health care will bring, the additional expenses inherent to many new genomic technologies, are unlikely to present much of a barrier to widespread uptake.

#### **10. Conclusion**

Lessons of the past decade of GxE research in psychiatry (and more specifically, Schizophrenia) mean that the focus of the next should be to ensure that effort and resources already spent, or else earmarked for future investment, do not go wasted. In order to ensure this a course of greater methodological rigour should be pursued.

It would be advantageous to complement this with the encouraging array of new specialist tools, methodologies and infrastructures available, some of which are highlighted in this article. A combination of falling economic costs and increasing accessibility make this proposition the most practical and logical way forward. In the category of 'methodologies' we additionally include innovations that enable the epigenomes, transcriptomes and proteomes of Schizophrenic patients to be characterised in high-dimension. Each of these domains reflects a different dynamic (and environmentally-responsive) element within a broader biological scheme. But each also remains curiously under-represented in mainstream GxE research today. This is despite evidence to suggest they may serve a functional purpose as biomarkers of environmentally-induced pathogenesis, susceptibility, illness progression and treatment outcome 146-152. Despite these documented examples, each discipline also faces thematic questions about how to achieve methodological best practice, given their various respective constraints 147,153,154.

Thus the current outlook would suggest that no single biological domain will have a monopoly on the clinical insights that may yet emerge out of future studies that may link genes, environment and Schizophrenia. The option to harness the various biological domains collectively, with genetics as the focal point, is promising, but currently underresourced 155-157. But this type of expansive approach is additionally attractive and may propel us towards fulfilling the unrealised clinical ambitions of GxE research.

#### **11. References**

98 Public Health – Methodology, Environmental and Systems Issues

running clinical trials to spiral. Other knock-on effects attributed to the legislation include a 30% decline in the numbers of participants agreeing to take part in trials across Europe, over the last few years 144. As clinical trials are an integral component within any translation scheme, such problems threaten to create a fatal bottle-neck in the pipeline, for discoveries

An overhaul of regulatory governance at national level has been proposed to circumvent this problem. In the UK, it is being done in conjunction with The National Institute for Health and Clinical Excellence (NICE), an organisation primarily responsible for assessing the cost-effectiveness, on behalf of the National Health Service (NHS), of providing new therapies and treatments. However the change of UK government means it is not even clear

As just hinted at, all novel genetic disoveries (including GxE interactions) that have safely negotiated the rigours of the validation stages shown in table 5, must still run the gauntlet of proving their overall cost-effectivenesss, before they can progress beyond validity, into utility. But new technology and treatments can only be considered to be cost-effective if their health benefits can be shown to outweigh the opportunity costs of services or treatments that they may displace 145. When viewed in the context of the many benefits that personalised health care will bring, the additional expenses inherent to many new genomic

Lessons of the past decade of GxE research in psychiatry (and more specifically, Schizophrenia) mean that the focus of the next should be to ensure that effort and resources already spent, or else earmarked for future investment, do not go wasted. In order to ensure

It would be advantageous to complement this with the encouraging array of new specialist tools, methodologies and infrastructures available, some of which are highlighted in this article. A combination of falling economic costs and increasing accessibility make this proposition the most practical and logical way forward. In the category of 'methodologies' we additionally include innovations that enable the epigenomes, transcriptomes and proteomes of Schizophrenic patients to be characterised in high-dimension. Each of these domains reflects a different dynamic (and environmentally-responsive) element within a broader biological scheme. But each also remains curiously under-represented in mainstream GxE research today. This is despite evidence to suggest they may serve a functional purpose as biomarkers of environmentally-induced pathogenesis, susceptibility, illness progression and treatment outcome 146-152. Despite these documented examples, each discipline also faces thematic questions about how to achieve methodological best practice,

Thus the current outlook would suggest that no single biological domain will have a monopoly on the clinical insights that may yet emerge out of future studies that may link genes, environment and Schizophrenia. The option to harness the various biological domains collectively, with genetics as the focal point, is promising, but currently underresourced 155-157. But this type of expansive approach is additionally attractive and may

propel us towards fulfilling the unrealised clinical ambitions of GxE research.

that might otherwise have made it through the process relatively unscathed.

technologies, are unlikely to present much of a barrier to widespread uptake.

that there is a timetable for putting such proposals into practice 143.

this a course of greater methodological rigour should be pursued.

given their various respective constraints 147,153,154.

**10. Conclusion** 


Old Obstacles on New Horizons:

441 (2011).

(2008).

The Challenge of Implementing Gene X Environment Discoveries in Schizophrenia Research 101

[45] Ikeda, M., et al. Genome-wide association study of schizophrenia in a Japanese

[46] O'Donovan, M.C., et al. Identification of loci associated with schizophrenia by genome-

[47] Riley, B., et al. Replication of association between schizophrenia and ZNF804A in the

[48] Williams, H.J., et al. Fine mapping of ZNF804A and genome-wide significant evidence

[49] Shi, J., et al. Common variants on chromosome 6p22.1 are associated with

[50] Alkelai, A., et al. Identification of new schizophrenia susceptibility loci in an ethnically homogeneous, family-based, Arab-Israeli sample. FASEB J 25, 4011-4023 (2011). [51] Williams, H.J., et al. Most genome-wide significant susceptibility loci for schizophrenia

[52] Shifman, S., et al. Genome-wide association identifies a common variant in the reelin

[53] Kirov, G., et al. A genome-wide association study in 574 schizophrenia trios using DNA

[54] Mulle, J.G., et al. Microdeletions of 3q29 confer high risk for schizophrenia. Am J Hum

[55] Chen, X., et al. GWA study data mining and independent replication identify

[56] Yue, W.H., et al. Genome-wide association study identifies a susceptibility locus for schizophrenia in Han Chinese at 11p11.2. Nat Genet 43, 1228-1231 (2011) . [57] Stefansson, H., et al. Common variants conferring risk of schizophrenia. Nature 460,

[58] Purcell, S.M., et al. Common polygenic variation contributes to risk of schizophrenia

[59] Ingason, A., et al. A large replication study and meta-analysis in European samples

[60] Athanasiu, L., et al. Gene variants associated with schizophrenia in a Norwegian

[61] Shi, Y.Y., et al. A study of rare structural variants in schizophrenia patients and normal controls from Chinese Han population. Mol Psychiatry 13, 911-913 (2008). [62] Vacic, V., et al. Duplications of the neuropeptide receptor gene VIPR2 confer significant

[63] Green, E.K., et al. The bipolar disorder risk allele at CACNA1C also confers risk of

[64] Lencz, T., et al. Converging evidence for a pseudoautosomal cytokine receptor gene

Irish Case-Control Study of Schizophrenia sample. Mol Psychiatry 15, 29-37 (2010).

for its involvement in schizophrenia and bipolar disorder. Mol Psychiatry 16, 429-

and bipolar disorder reported to date cross-traditional diagnostic boundaries. Hum

gene that increases the risk of schizophrenia only in women. PLoS Genet 4, e28

cardiomyopathy-associated 5 (CMYA5) as a risk gene for schizophrenia. Mol

provides further support for association of AHI1 markers with schizophrenia. Hum

genome-wide study are replicated in a large European cohort. J Psychiatr Res 44,

recurrent major depression and of schizophrenia. Mol Psychiatry 15, 1016-1022

wide association and follow-up. Nat Genet 40, 1053-1055 (2008).

population. Biol Psychiatry 69, 472-478 (2010).

schizophrenia. Nature 460, 753-757 (2009).

pooling. Mol Psychiatry 14, 796-803 (2009).

and bipolar disorder. Nature 460, 748-752 (2009).

risk for schizophrenia. Nature 471, 499-503 (2011)

locus in schizophrenia. Mol Psychiatry 12, 572-580 (2007).

Mol Genet 20, 387-391 (2011) .

Genet 87, 229-236 (2010).

744-747 (2009).

748-753 (2010)

(2010).

Psychiatry 16, 1117-1129 (2011).

Mol Genet 19, 1379-1386 (2010)


[22] Henquet, C., Murray, R., Linszen, D. & van Os, J. The environment and schizophrenia:

[23] Moore, T.H., et al. Cannabis use and risk of psychotic or affective mental health

[24] Morgan, C. & Fisher, H. Environment and schizophrenia: environmental factors in schizophrenia: childhood trauma--a critical review. Schizophr Bull 33, 3-10 (2007). [25] Miller, B., et al. Paternal age and mortality in nonaffective psychosis. Schizophr Res 121,

[26] Miller, B., et al. Meta-analysis of paternal age and schizophrenia risk in male versus

[27] Davies, G., Welham, J., Chant, D., Torrey, E.F. & McGrath, J. A systematic review and

[28] Rare chromosomal deletions and duplications increase risk of schizophrenia. Nature

[29] Glessner, J.T., et al. Strong synaptic transmission impact by copy number variations in

[30] Ikeda, M., et al. Copy number variation in schizophrenia in the Japanese population.

[31] Kirov, G., et al. Support for the involvement of large copy number variants in the pathogenesis of schizophrenia. Hum Mol Genet 18, 1497-1503 (2009). [32] Levinson, D.F., et al. Copy number variants in schizophrenia: confirmation of five

[33] Need, A.C., et al. A genome-wide investigation of SNPs and CNVs in schizophrenia.

[34] Stefansson, H., et al. Large recurrent microdeletions associated with schizophrenia.

[35] Walsh, T., et al. Rare structural variants disrupt multiple genes in neurodevelopmental

[36] Xu, B., et al. Strong association of de novo copy number mutations with sporadic

[37] Li, J., et al. Common variants in the BCL9 gene conferring risk of schizophrenia. Arch

[38] Shi, Y., et al. Common variants on 8p12 and 1q24.2 confer risk of schizophrenia. Nat

[39] Mah, S., et al. Identification of the semaphorin receptor PLXNA2 as a candidate for

[40] Rietschel, M., et al. Association between genetic variation in a region on chromosome 11 and schizophrenia in large samples from Europe. Mol Psychiatry (2011). [41] Steinberg, S., et al. Common variants at VRK2 and TCF4 conferring risk of

[42] Kirov, G., et al. Comparative genome hybridization suggests a role for NRXN1 and

[43] Magri, C., et al. New copy number variations in schizophrenia. PLoS One 5, e13422

[44] Vrijenhoek, T., et al. Recurrent CNVs disrupt three candidate genes in schizophrenia

susceptibility to schizophrenia. Mol Psychiatry 11, 471-478 (2006).

schizophrenia. Proc Natl Acad Sci U S A 107, 10584-10589 (2010).

meta-analysis of Northern Hemisphere season of birth studies in schizophrenia.

previous findings and new evidence for 3q29 microdeletions and VIPR2

the role of cannabis use. Schizophr Bull 31, 608-612 (2005).

outcomes: a systematic review. Lancet 370, 319-328 (2007).

female offspring. Schizophr Bull 37, 1039-1047 (2011).

duplications. Am J Psychiatry 168, 302-316 (2011).

pathways in schizophrenia. Science 320, 539-543 (2008).

schizophrenia. Hum Mol Genet 20, 4076-4081 (2011).

patients. Am J Hum Genet 83, 504-510 (2008).

APBA2 in schizophrenia. Hum Mol Genet 17, 458-465 (2008).

schizophrenia. Nat Genet 40, 880-885 (2008).

Schizophr Bull 29, 587-593 (2003).

Biol Psychiatry 67, 283-286 (2010).

PLoS Genet 5, e1000373 (2009).

Gen Psychiatry 68, 232-240 (2011).

Nature 455, 232-236 (2008).

Genet 43, 1224-1227 (2011).

(2010).

218-226 (2010).

455, 237-241 (2008).


Old Obstacles on New Horizons:

1049(2011)

J Psychiatry 197, 170-171 (2010)

Health 99, 480-486 (2009).

1117-1127 (2005).

998 (2004).

The Challenge of Implementing Gene X Environment Discoveries in Schizophrenia Research 103

[84] Duncan, L.E. & Keller, M.C. A critical review of the first 10 years of candidate gene-by-

[85] Zammit, S., Lewis, G., Dalman, C. & Allebeck, P. Examining interactions between risk

[86] Kendler, K.S. & Gardner, C.O. Interpretation of interactions: guide for the perplexed. Br

[87] Kotb, M., et al. An immunogenetic and molecular basis for differences in outcomes of invasive group A streptococcal infections. Nat Med 8, 1398-1404 (2002). [88] Boardman, J.D. State-level moderation of genetic tendencies to smoke. Am J Public

[89] Carrera, N., et al. Association study of nonsynonymous single nucleotide polymorphisms in schizophrenia. Biol Psychiatry 71, 169-177 (2012) [90] Little, J., et al. STrengthening the REporting of Genetic Association Studies (STREGA):

[91] Caspi, A., Hariri, A.R., Holmes, A., Uher, R. & Moffitt, T.E. Genetic sensitivity to the

studying complex diseases and traits. Am J Psychiatry 167, 509-527 (2011) [92] Zammit, S., Owen, M.J., Evans, J., Heron, J. & Lewis, G. Cannabis, COMT and psychotic

[93] Caspi, A., et al. Moderation of the effect of adolescent-onset cannabis use on adult

[94] Caspi, A., et al. Influence of life stress on depression: moderation by a polymorphism in

[95] Uher, R. & McGuffin, P. The moderation by the serotonin transporter gene of

[96] Wong, M.Y., Day, N.E., Luan, J.A. & Wareham, N.J. Estimation of magnitude in gene-

[97] Vineis, P. A self-fulfilling prophecy: are we underestimating the role of the environment in gene-environment interaction research? Int J Epidemiol 33, 945-946 (2004). [98] Boks, M.P., et al. Investigating gene environment interaction in complex diseases:

[99] Knol, M.J., van der Tweel, I., Grobbee, D.E., Numans, M.E. & Geerlings, M.I. Estimating

[102] Engelman, C.D., et al. Detecting gene-environment interactions in genome-wide

[103] Thomas, D. Gene--environment-wide association studies: emerging approaches. Nat

[100] Darroch, J. Biologic synergism and parallelism. Am J Epidemiol 145, 661-668 (1997). [101] VanderWeele, T.J., Hernandez-Diaz, S. & Hernan, M.A. Case-only gene-environment

methodological analysis. Mol Psychiatry 13, 131-146 (2008).

regression model. Int J Epidemiol 36, 1111-1118 (2007).

association data. Genet Epidemiol 33 Suppl 1, S68-73 (2009).

environment: the case of the serotonin transporter gene and its implications for

psychosis by a functional polymorphism in the catechol-O-methyltransferase gene: longitudinal evidence of a gene X environment interaction. Biol Psychiatry 57,

environmental adversity in the aetiology of mental illness: review and

environment interactions in the presence of measurement error. Stat Med 23, 987-

increasing power by selective sampling for environmental exposure. Int J

interaction on an additive scale between continuous determinants in a logistic

interaction studies: when does association imply mechanistic interaction? Genet

an extension of the STROBE statement. PLoS Med 6, e22 (2009).

experiences. Br J Psychiatry 199, 380-385 (2011)

the 5-HTT gene. Science 301, 386-389 (2003).

Epidemiol 36, 1363-1369 (2007).

Epidemiol 34, 327-334 (2010)

Rev Genet 11, 259-272 (2010)

factors for psychosis. Br J Psychiatry 197, 207-211(2010)

environment interaction research in psychiatry. Am J Psychiatry 168, 1041-


[65] So, H.C., Gui, A.H., Cherny, S.S. & Sham, P.C. Evaluating the heritability explained by

[66] Eichler, E.E., et al. Missing heritability and strategies for finding the underlying causes

[67] Zuk, O., Hechter, E., Sunyaev, S.R. & Lander, E.S. The mystery of missing heritability: Genetic interactions create phantom heritability. Proc Natl Acad Sci U S A (2012) [68] Jaffee, S.R. & Price, T.S. Genotype-environment correlations: implications for

[69] Dick, D.M., et al. Gender differences in friends' influences on adolescent drinking: a genetic epidemiological study. Alcohol Clin Exp Res 31, 2012-2019 (2007). [70] Kendler, K.S. & Karkowski-Shuman, L. Stressful life events and genetic liability to major

[71] Di Forti, M., et al. High-potency cannabis and the risk of psychosis. Br J Psychiatry 195,

[72] Barkus, E.J., Stirling, J., Hopkins, R.S. & Lewis, S. Cannabis-induced psychosis-like

[73] Ferdinand, R.F., et al. Cannabis use predicts future psychotic symptoms, and vice versa.

[74] Fergusson, D.M., Horwood, L.J. & Ridder, E.M. Tests of causal linkages between cannabis use and psychotic symptoms. Addiction 100, 354-366 (2005). [75] Veling, W., Mackenbach, J.P., van Os, J. & Hoek, H.W. Cannabis use and genetic

[76] McGuire, P.K., et al. Morbid risk of schizophrenia for relatives of patients with

[77] Arendt, M., Mortensen, P.B., Rosenberg, R., Pedersen, C.B. & Waltoft, B.L. Familial

[80] Margari, F., et al. Familial liability, obstetric complications and childhood development

[81] Wicks, S., Hjern, A. & Dalman, C. Social risk or genetic liability for psychosis? A study

[82] Moffitt, T.E., Caspi, A. & Rutter, M. Strategy for investigating interactions between

[83] Tienari, P., et al. Genetic boundaries of the schizophrenia spectrum: evidence from the

cannabis-associated psychosis. Schizophr Res 15, 277-281 (1995).

causation of schizophrenia. Am J Psychiatry 166, 1025-1030 (2009).

of complex disease. Nat Rev Genet 11, 446-450 (2010)

illness. Psychiatry 7, 496-499 (2008).

Addiction 100, 612-618 (2005).

35, 310-317 (2011)

547 (1997).

(2006).

(2008).

11, 60 (2011)

(2005).

(2003).

1240-1246 (2010)

488-491 (2009).

known susceptibility variants: a survey of ten complex diseases. Genet Epidemiol

determining the relationship between environmental exposures and psychiatric

depression: genetic control of exposure to the environment? Psychol Med 27, 539-

experiences are associated with high schizotypy. Psychopathology 39, 175-178

predisposition for schizophrenia: a case-control study. Psychol Med 38, 1251-1256

predisposition for psychiatric disorder: comparison of subjects treated for cannabisinduced psychosis and schizophrenia. Arch Gen Psychiatry 65, 1269-1274 (2008). [78] van Os, J., Hanssen, M., Bak, M., Bijl, R.V. & Vollebergh, W. Do urbanicity and familial liability coparticipate in causing psychosis? Am J Psychiatry 160, 477-482 (2003). [79] Clarke, M.C., Tanskanen, A., Huttunen, M., Whittaker, J.C. & Cannon, M. Evidence for

an interaction between familial liability and prenatal exposure to infection in the

abnormalities in early onset schizophrenia: a case control study. BMC Psychiatry

of children born in Sweden and reared by adoptive parents. Am J Psychiatry 167,

measured genes and measured environments. Arch Gen Psychiatry 62, 473-481

Finnish Adoptive Family Study of Schizophrenia. Am J Psychiatry 160, 1587-1594


Old Obstacles on New Horizons:

Epidemiol 35, 887-898 (2011).

(Wellcome Trust, 2006).

257-263 (2010).

Hetil 151, 1403-1408 (2010).

Psychiatry 21, 318-321 (2008).

BMC Psychiatry 9, 51 (2009).

Sci U S A 90, 11825-11829 (1993).

biomarkers. Clin Med 9, 68-73 (2009).

Engl J Med 349, 868-874 (2003).

reuptake inhibitors. Genet Med 9, 826-835 (2007).

1525-1535 (2010).

265-269 (2001).

1-14 (2011) .

(2011).

interventions. Lancet 372, 2152-2161 (2008).

interactions. Eur J Pharmacol 668 Suppl 1, S108-116.

The Challenge of Implementing Gene X Environment Discoveries in Schizophrenia Research 105

[123] Zuvich, R.L., et al. Pitfalls of merging GWAS data: lessons learned in the eMERGE

[124] WellcomeTrust. Translating the potential of human population genetics research to

[125] Inczedy-Farkas, G., et al. [SCHIZOBANK - The Hungarian national schizophrenia

[126] McGrath, J.J., et al. Neonatal vitamin D status and risk of schizophrenia: a population-

[127] Mortensen, P.B., et al. A Danish National Birth Cohort study of maternal HSV-2

[128] Rawlins, M. De testimonio: on the evidence for decisions about the use of therapeutic

[129] Uher, R. Forum: The case for gene-environment interactions in psychiatry. Curr Opin

[130] Inselman, A.L., et al. Assessment of research models for testing gene-environment

[131] Stewart, R., et al. The South London and Maudsley NHS Foundation Trust Biomedical

[132] Ashley, E.A., et al. Clinical assessment incorporating a personal genome. Lancet 375,

[133] Wilson, J.F., et al. Population genetic structure of variable drug response. Nat Genet 29,

[134] Fleeman, N., et al. Cytochrome P450 testing for prescribing antipsychotics in adults

[135] Johansson, I., et al. Inherited amplification of an active gene in the cytochrome P450

[136] Thakur, M., et al. Review of evidence for genetic testing for CYP450 polymorphisms in

[137] Khoury, M.J., et al. The continuum of translation research in genomic medicine: how

[138] Pirmohamed, M. Acceptance of biomarker-based tests for application in clinical practice: criteria and obstacles. Clin Pharmacol Ther 88, 862-866 (2010). [139] Zimmern, R.L. Testing challenges: evaluation of novel diagnostics and molecular

[140] Lenfant, C. Shattuck lecture--clinical research to clinical practice--lost in translation? N

[141] Collins, F.S. Reengineering translational science: the time is right. Sci Transl Med 3,

[142] Jurgens, G., et al. Utility and adoption of CYP2D6 and CYP2C19 genotyping and its translation into psychiatric clinical practice. Acta Psychiatr Scand (2011).

health care and disease prevention? Genet Med 9, 665-674 (2007).

based case-control study. Arch Gen Psychiatry 67, 889-894 (2010).

network and quality control procedures to maintain high data quality. Genet

improve the quality of health of the EU citizen. in From Biobanks to biomarkers

biobank and its role in schizophrenia research and in personalized medicine]. Orv

antibodies as a risk factor for schizophrenia in their offspring. Schizophr Res 122,

Research Centre (SLAM BRC) case register: development and descriptive data.

with schizophrenia: systematic review and meta-analyses. Pharmacogenomics J 11,

CYP2D locus as a cause of ultrarapid metabolism of debrisoquine. Proc Natl Acad

management of patients with nonpsychotic depression with selective serotonin

can we accelerate the appropriate integration of human genome discoveries into


[104] Bickeboller, H., Houwing-Duistermaat, J.J., Wang, X. & Yan, X. Dealing with high

[106] Kraft, P. & Hunter, D. Integrating epidemiology and genetic association: the challenge

[107] Hewitt, J.K. Editorial Policy on Candidate Gene Association and Candidate Gene-by-Environment Interaction Studies of Complex Traits. Behav Genet 42, 1-2 (2012) [108] Caspi, A., et al. Role of genotype in the cycle of violence in maltreated children. Science

[109] Collins, A.L., Kim, Y., Sklar, P., O'Donovan, M.C. & Sullivan, P.F. Hypothesis-driven

[110] Sanders, A.R., et al. No significant association of 14 candidate genes with

[111] Lasky-Su, J., et al. Genome-wide association scan of quantitative traits for attention

[112] Hamza, T.H., et al. Genome-wide gene-environment study identifies glutamate

[113] Paus, T., et al. KCTD8 Gene and Brain Growth in Adverse Intrauterine Environment: A

[114] Tan, A., et al. A genome-wide association and gene-environment interaction study for

[115] A framework for interpreting genome-wide association studies of psychiatric

[116] Mukherjee, B., Ahn, J., Gruber, S.B., Ghosh, M. & Chatterjee, N. Case-control studies of

[117] Mukherjee, B., Ahn, J., Gruber, S.B. & Chatterjee, N. Testing gene-environment

[118] Pierce, B.L. & Ahsan, H. Case-only genome-wide interaction study of disease risk,

[119] Piegorsch, W.W., Weinberg, C.R. & Taylor, J.A. Non-hierarchical logistic models and

[120] Li, M.J., et al. GWASdb: a database for human genetic variants identified by genome-

[121] Chen, J., et al. Two non-synonymous markers in PTPN21, identified by genome-wide

[122] Havik, B., et al. The complement control-related genes CSMD1 and CSMD2 associate

wide association studies. Nucleic Acids Res 40, D1047-1054 (2012).

649-651; discussion 657-648 (2008).

Psychol Med 42, 607-616 (2012)

genetics. Am J Psychiatry 165, 497-506 (2008).

Genome-wide Association Study. Cereb Cortex (2011)

coffee. PLoS Genet 7, e1002237 (2011)

disorders. Mol Psychiatry 14, 10-17 (2009).

studies. Stat Med 13, 153-162 (1994).

Schizophr Res 131, 43-51(2011).

comparisons. Am J Epidemiol 175, 177-190 (2012) .

to schizophrenia. Biol Psychiatry 70, 35-42 (2011).

prognosis and treatment. Genet Epidemiol 34, 7-15 (2010).

(2005).

(2012)

(2010).

297, 851-854 (2002).

dimensionality for the identification of common and rare variants as main effects and for gene-environment interaction. Genet Epidemiol 35 Suppl 1, S35-40 (2011) [105] Kraft, P. Curses--winner's and otherwise--in genetic epidemiology. Epidemiology 19,

of gene-environment interaction. Philos Trans R Soc Lond B Biol Sci 360, 1609-1616

candidate genes for schizophrenia compared to genome-wide association results.

schizophrenia in a large European ancestry sample: implications for psychiatric

deficit hyperactivity disorder identifies novel associations and confirms candidate gene associations. Am J Med Genet B Neuropsychiatr Genet 147B, 1345-1354 (2008).

receptor gene GRIN2A as a Parkinson's disease modifier gene via interaction with

serum triglycerides levels in a healthy Chinese male population. Hum Mol Genet

gene-environment interaction: Bayesian design and analysis. Biometrics 66, 934-948

interaction in large-scale case-control association studies: possible choices and

case-only designs for assessing susceptibility in population-based case-control

association study data-mining and replication, are associated with schizophrenia.


**Section 2** 

**Environmental and Nutritional Issues** 

