**3. Statistical methodologies to evaluate the risk assessment and air pollution correlated diseases**

Extreme pollution episodes that took place in the period 1930–1960 have initiated the epide‐ miological studies for increase of health environmental concern. The association between air pollution and certain health variables was made clear by several studies [14, 15].

Thanks to the new approach, air pollution levels were reduced substantially, such that, for health effects assessment, now, longer monitoring plans are required. To this end, epidemiol‐ ogists began to use some statistical models. In the 1970s firstly were used the dynamic regression models, models in which the relationship between the dependent and explanatory variables were distributed over time rather than being expected to occur simultaneously. However, the problem of these types of models is that they assume that the dependent variable is distributed normally, but this is a condition extremely rare to achieve [16].

In the 1990s were used for the first time the linear models based on Poisson regression because the event counts more typically have a Poisson distribution. These models use the "time" and its transforms as variable [16].

Also Poisson regression is particularly useful only when cases rather than the entire population can be enumerated.

Nevertheless, Poisson regression poses the problem that, if any of these unmeasured variables follows a cyclical component of varying frequency, the parametric functions of time cannot be easily adapted. So, these limitations led to the development also of nonparametric Poisson regression [17], which is well adapted to the irregular cyclic components of unmeasured variables reducing any potential confounding.

One difficulty with this method is that the researcher must specify the number of degrees of freedom, with several discrepancies arising as the way to calculate this. Because inappropriate determination of the number of degrees of freedom is frequently a bias in the estimates of nonparametric Poisson designs, epidemiologists focused on the case-crossover (CCO) [18] design that purported to control time trends. The CCO design is characterized by the fact that each subject serves as his or her own control. This design was initially used to assess the effect of exposures measured at an individual level and was not applicable to exposures with a time trend, such as air pollution. A variant was developed to bypass this bias [19], the bidirectional CCO, characterized by having control time periods before and after the event. This design was already appropriate for ecologic-type exposures, such as air pollution, water pollution, etc. In addition, pollution values are not affected by the presence of prior morbidity and mortality events. In fact, in the CCO design in air pollution, pre- and post-event exposure values are independent of the hazard-period exposure; those that are post-event referent can be appro‐ priate. One advantage of CCO design respect to Poisson regression is its ability to assess potential effect modification at the individual level rather than at the group level [20]. As an alternative analytic methodology to Poisson regression, the CCO approach allows for direct modeling of interaction terms, rather than depending on multiple subgroup analyses [20].

The information provided by mosses, lichens, and higher plants on the deposition on the effects of air pollutants is an important complement to the data acquired with automatic systems. Distribution of tree species on a national or supranational lets you draw cheaply and in a short time maps of diffusion and deposition of persistent pollutants or the effects of tropospheric

Extreme pollution episodes that took place in the period 1930–1960 have initiated the epide‐ miological studies for increase of health environmental concern. The association between air

Thanks to the new approach, air pollution levels were reduced substantially, such that, for health effects assessment, now, longer monitoring plans are required. To this end, epidemiol‐ ogists began to use some statistical models. In the 1970s firstly were used the dynamic regression models, models in which the relationship between the dependent and explanatory variables were distributed over time rather than being expected to occur simultaneously. However, the problem of these types of models is that they assume that the dependent variable

In the 1990s were used for the first time the linear models based on Poisson regression because the event counts more typically have a Poisson distribution. These models use the "time" and

Also Poisson regression is particularly useful only when cases rather than the entire population

Nevertheless, Poisson regression poses the problem that, if any of these unmeasured variables follows a cyclical component of varying frequency, the parametric functions of time cannot be easily adapted. So, these limitations led to the development also of nonparametric Poisson regression [17], which is well adapted to the irregular cyclic components of unmeasured

One difficulty with this method is that the researcher must specify the number of degrees of freedom, with several discrepancies arising as the way to calculate this. Because inappropriate determination of the number of degrees of freedom is frequently a bias in the estimates of nonparametric Poisson designs, epidemiologists focused on the case-crossover (CCO) [18] design that purported to control time trends. The CCO design is characterized by the fact that each subject serves as his or her own control. This design was initially used to assess the effect of exposures measured at an individual level and was not applicable to exposures with a time trend, such as air pollution. A variant was developed to bypass this bias [19], the bidirectional CCO, characterized by having control time periods before and after the event. This design was already appropriate for ecologic-type exposures, such as air pollution, water pollution, etc. In addition, pollution values are not affected by the presence of prior morbidity and mortality

**3. Statistical methodologies to evaluate the risk assessment and air**

pollution and certain health variables was made clear by several studies [14, 15].

is distributed normally, but this is a condition extremely rare to achieve [16].

O3 and other phytotoxic pollutants [13].

426 Current Air Quality Issues

**pollution correlated diseases**

its transforms as variable [16].

variables reducing any potential confounding.

can be enumerated.

Besides the individual characteristics during the statistical analysis of epidemiological data, other variables must be considered, taking into account that the chemical characteristics of air pollutants mixture and PM may change over time and depending on the geographical location, emission sources, atmospheric chemistry, and weather conditions [21].

Interest in health effects of air pollution became more intense after two US cohort studies suggested that exposure to fine PM in the air was associated with life shortening [22,23].

Exposure to pollutants such as airborne PM and O3 has been associated with increases in mortality and hospital admissions mainly because of respiratory and cardiovascular disease, due to both acute and chronic exposure [24].

Health problems can include cancer, respiratory irritation, nervous system problems, and birth defects.

Air pollution, in the year 2012, was responsible for 3.7 million deaths, i.e., approximately 6.7% of the total deaths. In particular, air pollution is estimated to cause deaths from lung cancer (16%), COPD (11%), ischemic heart disease and stroke (>20%), and respiratory infection (13%). Furthermore, the IARC announced that it has classified outdoor air pollution as carcinogenic to humans (Group 1).

Typically, exposure to toxic components of air pollutant causes the well-established correlated diseases, respiratory and pulmonary ones, such as decreased lung function and increased incidence of chronic cough, bronchitis, chronic obstructive pulmonary disease, asthma, and conjunctivitis [25,26,27].

Recent clinical and epidemiological data suggest also that cardiovascular disease may be related to pollution [28,29] especially those associates with fine PM [30,31,32,33]. The PM's effects on the cardiovascular system seem to involve the activation of prothrombotic factors, leading to thrombosis, but destabilization of atherosclerotic plaques cannot also be excluded. In addition, there may be direct effects on the heart or indirectly on the nervous system.

Health effects have been seen at very low levels of exposure, and it is unclear whether a threshold concentration exists for PM and O3 below which no effects on health are likely.

It could be expected that the impact caused by a preventable risk factor would decline if the exposure to that risk factor could be reduced or removed. According to this approach, the proportional reduction in the number of health problems or deaths as a result of reducing the risk factor is known as the attributable fraction (AF) [34].

Public health agencies concerned with air quality perform risk assessments to determine the increased risk of illness from a specific human exposure to a toxic air pollutant.

The risk assessment approach outlined by the WHO in the Environmental Burden of Disease (EBD) series [35,36] includes the following steps:


The US Environmental Protection Agency (US-EPA) divided the risk assessment process into the following four steps:

	- **i.** Estimation of the maximum quantity of each pollutant emitted from the source.
	- **ii.** For each contaminant emitted, estimation of the resulting maximum annual average and (where applicable) maximum short-term average ambient air concentrations, using dispersion models, or air impact values based on disper‐ sion models.
	- **iii.** Estimation of the amount of contaminant taken in by a human receptor.

**e. Risk Characterization**: it is the final step in risk assessment where health risk is calculated and described thanks to data collected in the first three steps.

#### **• Carcinogens**

The risk assessment approach outlined by the WHO in the Environmental Burden of Disease

**1.** assessment of the air exposure of the population through data from Air model or moni‐ toring networks. A target concentration is also needed to determine the attributable

**3.** baseline data of incidence of the adverse health outcomes associated with air pollutants

**4.** concentration-response functions (CRFs) related to the incidence of adverse health effects. The US Environmental Protection Agency (US-EPA) divided the risk assessment process into

**a. Hazard Identification**: it allows determining the potential human health effects from exposure to a chemical. This is based on information provided by the scientific literature.

**b. Dose-Response Assessment:** it is the assessment of the relationship between a dose of chemical exposure and incidence or severity of the related adverse health effect. It takes into consideration the intensity and pattern of exposure and also age and lifestyle variables that may affect people's susceptibility. The dose-response relationship is evaluated differently for carcinogenic and non-carcinogenic pollutants. In fact, for carcinogens, it is assumed that there is a linear relationship between an increased dose of exposure and increase in cancer risk; this is expressed as slope factor (SF) and threshold is not accepted. For risks evaluation by inhalation of carcinogenic substances, US-EPA use the potency slopes to develop the unit risk factors (URFs). A URF is the upper-bound excess probability of contracting cancer as the result of a lifetime of exposure to a carcinogen at a concen‐ tration of 1 µg/m3 in air. For inhalation effects from non-carcinogens, dose-response data are used to develop reference concentrations (RfCs) for both long-term and short-term exposures. Unlike carcinogens, non-carcinogens are assumed to have thresholds dose, so the injury does not occur until exposure has exceeded a threshold limit. An RfC is derived from a no-observed adverse effect level (NOAEL) or lowest observed adverse effect level

(LOAEL) determined through human or animal exposure studies.

dose) of human exposure to a chemical in the environment.

**d.** There are three components to exposure assessment:

sion models.

**c. Exposure Assessment**: it determines the extent (intensity, frequency, and duration, or

**i.** Estimation of the maximum quantity of each pollutant emitted from the source. **ii.** For each contaminant emitted, estimation of the resulting maximum annual

**iii.** Estimation of the amount of contaminant taken in by a human receptor.

average and (where applicable) maximum short-term average ambient air concentrations, using dispersion models, or air impact values based on disper‐

(EBD) series [35,36] includes the following steps:

(like mortality rate); and

the following four steps:

428 Current Air Quality Issues

disease or the potential gains of a management plan; **2.** sufficient number of persons exposed to air pollutants; Human health risk estimates for inhalation of carcinogenic air pollutants are based on the following:

Cancer Risk = C x URF

where

C = maximum annual average ambient air concentration of a pollutant (µg/m3 )

URF = pollutant-specific inhalation unit risk factor (µg/m3 ) -1

For routes of exposure other than inhalation, risk is calculated by multiplying the estimated chemical dose (in mg/kg/day) by the chemical-specific oral slope factor (in (mg/kg/day)-1).

#### **• No carcinogens**

Human health risk estimates for inhalation of non-carcinogenic air pollutants are based on the following:

$$\text{Hazard Quotient} = \text{C/RfC}$$

where

C = maximum ambient air concentration, µg/m3

RfC = pollutant-specific reference concentration, µg/m3

The averaging time can be either annual, or a specific number of hours, depending on the basis of the reference dose [37]. For routes of exposure other than inhalation, the hazard quotient is calculated by ratio between the estimated chemical dose (in mg/kg/day) and the chemicalspecific reference dose (in mg/kg/day).

Hazard quotients can be summed (separately for inhalation and oral exposures and for different averaging times) to give a hazard index.
