**3. Problems with the 'Swiss cheese' model: why are hospitals different from other industries?**

There are three fundamental problems with the application of the 'Swiss Cheese' model to adverse events in hospitals. First, in the hospital, the distinction between individual and organisational accidents is not clear. The entire premise of the 'Swiss Cheese' model was the investigation of causation factors of large industrial accidents as opposed to individual accidents. In the hospital, we do not have large-scale accidents but, instead, multiple little accidents or adverse events daily, if not hourly, and in almost every setting. The study on the causation of adverse events in hospitals overwhelmingly points to failures at the sharp end of care delivery to the patient by frontline staff. Analysis of the causative factors associated with the adverse events in The Quality in Australian Health Care Study found that cognitive failure was a factor in 57% of these adverse events [49]. In this analysis, cognitive failure included such errors as failure to synthesise, decide and act on available information; failure to request or arrange an investigation, procedure or consultation; lack of care or attention; failure to attend; misapplication of, or failure to apply, a rule, or use of a bad or inadequate rule [49]. In a two-hospital study from the United Kingdom that looked at 100 sequential admissions to the intensive care unit (ICU) from ward areas, it was found that 54 had sub-optimal care on the ward prior to transfer [50]. This group of patients had a mortality rate of 56%. Some of the sub-optimal treatment factors included failure to seek advice, lack of knowledge, failure to appreciate clinical urgency and lack of supervision [50].

The adoption of the Reason 'Swiss Cheese' model for organisational accidents has led the whole Quality and Safety industry, and in particular hospitals, looking for system solutions to what can be explained by individual competency and micro-environment cultural issues at the patient interface. In particular, a major rationale of Reason's philosophy is to avoid individual accountability for errors and the culture of blame and shame. Nearly 20 years ago, Reason himself noted the folly of this approach in medicine when he stated, '*It is curious that such a bastion of discretionary action as medicine should be moving towards a 'Feed Forward' mode of control when many other hitherto rule dominated domains – notably railways and oil exploration and production – are shifting towards performance-based controls and away from prescriptive ones'* [42]. When Reason talks about human contribution to organisational accidents, he describes two schemas of control [42]. A 'Feed Forward' control system is one where human performance is determined by rules and procedures as determined by an organisational standards and objectives (**Figure 2**). In this schema, occasional accidents and incidents are analysed and then fed back into either an alteration of an existing rule or a procedure or the creation of a new one. At the other end of the control spectrum, there is the model where organisational output is largely determined by individual human performance (**Figure 3**). The basis for this model is that, in the first instance, the humans are generally highly trained and that performance is controlled by continual performance reinforcement against a known or a standard comparator. The best example of this, in hospitals, is specialist medical practice. To even start specialist training, there have been many years of training and experience (medical school, house officer jobs and pre-specialty registrar placements) followed by a period of mentoring and in essence apprenticeship to learn the specialty to the known standard of the comparator, the standard of practice as maintained by the specialty colleges. Taking these two schemas, one can immediately see the trouble with health care in hospitals. It is a large industry with community and political expectations that are more congruent with the 'Feed Forward' schema, but yet with most of the actual clinical activity being undertaken by the 'Human Performance' schema.

with the assumption that the health-care professionals at the patient end are competent and will be compliant. The shift to looking for hospital-wide problems has come at the cost of avoiding the issue of individual professional accountability and associated issues, most notably the education and certification of health-care professionals. In Australia and the United Kingdom, several studies indicate that the medical undergraduate syllabus does not provide graduates with the basic knowledge, skills and judgement to manage acute life-threatening emergencies [51–53]. These studies identified deficiencies in cognitive abilities, procedural skills and communication. Despite this, undergraduate and postgraduate curricula have been

Adverse Events in Hospitals: "Swiss Cheese" Versus the "Hierarchal Referral Model of Care and…

http://dx.doi.org/10.5772/intechopen.75380

41

**Figure 3.** The reason feedback process control system [37] (with kind permission from Ashgate Publishing).

The second fundamental problem with the 'Swiss Cheese' model and the Palmieri variation of this are that they are overly simplistic and do not take into account the complexity of the patient and the hospital system. When a patient enters a hospital system, they enter a system where they will be exposed to a variety of hazards which, in turn, have numerous defences in place to prevent an adverse patient outcome. Operations, anaesthesia, medical interventions and procedures, drugs and fluids and even oxygen therapy constitute the hazards. Most defences in health care are reliant on the competence of the health-care professional and as such are 'soft'. 'Hard' defences are those that are impossible to overcome, for example in anaesthesia where the administration of hypoxic gas mixtures is physically prevented. The soft defences, in health care, include treatment policies and procedures, manual alarm systems, and ad hoc hierarchical and lateral human checking systems. Soft defences are very reliant on the training and education that health-care workers receive and the culture of compliance. Superimposed on these layers of hazards and defences that confront a patient, there are the latent conditions that exist, most obviously within the patient, but more insidiously within the hospital as an organisation. A patient's past medical history, family history, social

slow to embrace a patient safety culture [54–56].

Thus, what we have seen in the construction of hospital adverse event defences is an over-reliance on the administrative blunt end of the organisation, in terms of policy and procedures,

**Figure 2.** The reason 'feedforward' process control system [37] (with kind permission from Ashgate Publishing).

Adverse Events in Hospitals: "Swiss Cheese" Versus the "Hierarchal Referral Model of Care and… http://dx.doi.org/10.5772/intechopen.75380 41

sub-optimal treatment factors included failure to seek advice, lack of knowledge, failure to

The adoption of the Reason 'Swiss Cheese' model for organisational accidents has led the whole Quality and Safety industry, and in particular hospitals, looking for system solutions to what can be explained by individual competency and micro-environment cultural issues at the patient interface. In particular, a major rationale of Reason's philosophy is to avoid individual accountability for errors and the culture of blame and shame. Nearly 20 years ago, Reason himself noted the folly of this approach in medicine when he stated, '*It is curious that such a bastion of discretionary action as medicine should be moving towards a 'Feed Forward' mode of control when many other hitherto rule dominated domains – notably railways and oil exploration and production – are shifting towards performance-based controls and away from prescriptive ones'* [42]. When Reason talks about human contribution to organisational accidents, he describes two schemas of control [42]. A 'Feed Forward' control system is one where human performance is determined by rules and procedures as determined by an organisational standards and objectives (**Figure 2**). In this schema, occasional accidents and incidents are analysed and then fed back into either an alteration of an existing rule or a procedure or the creation of a new one. At the other end of the control spectrum, there is the model where organisational output is largely determined by individual human performance (**Figure 3**). The basis for this model is that, in the first instance, the humans are generally highly trained and that performance is controlled by continual performance reinforcement against a known or a standard comparator. The best example of this, in hospitals, is specialist medical practice. To even start specialist training, there have been many years of training and experience (medical school, house officer jobs and pre-specialty registrar placements) followed by a period of mentoring and in essence apprenticeship to learn the specialty to the known standard of the comparator, the standard of practice as maintained by the specialty colleges. Taking these two schemas, one can immediately see the trouble with health care in hospitals. It is a large industry with community and political expectations that are more congruent with the 'Feed Forward' schema, but yet with most of the actual clinical activity being undertaken by the 'Human Performance' schema.

Thus, what we have seen in the construction of hospital adverse event defences is an over-reliance on the administrative blunt end of the organisation, in terms of policy and procedures,

**Figure 2.** The reason 'feedforward' process control system [37] (with kind permission from Ashgate Publishing).

appreciate clinical urgency and lack of supervision [50].

40 Vignettes in Patient Safety - Volume 3

**Figure 3.** The reason feedback process control system [37] (with kind permission from Ashgate Publishing).

with the assumption that the health-care professionals at the patient end are competent and will be compliant. The shift to looking for hospital-wide problems has come at the cost of avoiding the issue of individual professional accountability and associated issues, most notably the education and certification of health-care professionals. In Australia and the United Kingdom, several studies indicate that the medical undergraduate syllabus does not provide graduates with the basic knowledge, skills and judgement to manage acute life-threatening emergencies [51–53]. These studies identified deficiencies in cognitive abilities, procedural skills and communication. Despite this, undergraduate and postgraduate curricula have been slow to embrace a patient safety culture [54–56].

The second fundamental problem with the 'Swiss Cheese' model and the Palmieri variation of this are that they are overly simplistic and do not take into account the complexity of the patient and the hospital system. When a patient enters a hospital system, they enter a system where they will be exposed to a variety of hazards which, in turn, have numerous defences in place to prevent an adverse patient outcome. Operations, anaesthesia, medical interventions and procedures, drugs and fluids and even oxygen therapy constitute the hazards. Most defences in health care are reliant on the competence of the health-care professional and as such are 'soft'. 'Hard' defences are those that are impossible to overcome, for example in anaesthesia where the administration of hypoxic gas mixtures is physically prevented. The soft defences, in health care, include treatment policies and procedures, manual alarm systems, and ad hoc hierarchical and lateral human checking systems. Soft defences are very reliant on the training and education that health-care workers receive and the culture of compliance. Superimposed on these layers of hazards and defences that confront a patient, there are the latent conditions that exist, most obviously within the patient, but more insidiously within the hospital as an organisation. A patient's past medical history, family history, social history, associated co-morbidities, drug regimen and allergies largely constitute their latent conditions. These conditions and their relation to the current presenting complaint that brings the patient into the hospital system are territory that individual health-care workers are usually extremely well trained in and familiar with. Hospital latent conditions are not so explicit, particularly to the patient or the frontline health-care worker. They are made up of a complex matrix of production and cultural imperatives such as the financial operating environment, political and societal imperatives, medico-legal and insurance concerns, compliance issues imposed by various regulatory bodies (often with associated financial incentives or disincentives) and workforce and work-practice issues. Thus, in the hospital system, unlike any other industry, we have a high degree of ever-changing complexity, complex patients and a complex system where adverse events are essentially prevented by a whole host of predominantly soft defences [57]. The 'Swiss Cheese' model is a static model with fixed defences in terms of the layers and the size of holes in each layer. This translates well into most industries, but in health care, the complexity is dynamic and ever changing, the number of holes and layers change with every patient and each and every different health-care professional.

occurs in a traditional hierarchal referral model of care that by its very nature is often either unresponsive or slowly responsive and where the exhaustive policy and procedures are often ignored.

Adverse Events in Hospitals: "Swiss Cheese" Versus the "Hierarchal Referral Model of Care and…

http://dx.doi.org/10.5772/intechopen.75380

43

In the hospital, the CFC usually starts with the most junior level of the 'traditional hierarchical referral model of care', at the bedside with the interaction between the junior nurse and the patient (**Figure 4**). With a clinical abnormality, be it an observation, a wrong drug order or a procedural failure, the junior nurse must make a decision as to the significance of the abnormality and the importance of reporting it to a more senior team member, either a senior nurse or the most available (usually junior) doctor. However, that decision to escalate the issue can be influenced in the workplace culture that exists in the particular micro-environment of that bedside and that ward at that time [58]. If the concern or abnormality is escalated, it is to the next person in the care hierarchy of the team looking after that patient. This is often the junior doctor who then needs to attend, assess and then also make a decision about whether or not to escalate the issue to the next person in the hierarchy. This is important because, for the most, the junior doctor does not have the skills or emotional intelligence to appropriately manage many of these clinical abnormalities [51–54]. If the issue is escalated, it is often to a middle-grade doctor, one who is often a specialist in training and who as such may be difficult to find. Unlike their juniors, usually this grade of doctor does have the technical and clinical abilities to deal with the particular issue. However, they are often over-committed with clinics, operating theatre, but more importantly often see themselves more like the consultants they aspire to be rather than a junior doctor having to deal with patient problems on

**Figure 4.** Clinical futile cycles [38, 39].

The third problem with the 'Swiss Cheese' model is that adverse events in hospitals occur so insidiously that they become normalised into the operating behaviour and practice of the organisation. This is distinct from large-scale industrial accidents, where the impact of the event has a high degree of face validity, primarily due to the immediacy and scale of the event. Therefore, in terms of numbers, patient adverse events may constitute a crisis. However, to the individual practitioner or even hospital, these events may not appear to be a problem. On the whole, such events are infrequent and occur over a long time frame. For example, The Quality in Australian Health Care Study looked at a random sample of 14,179 admissions to 28 hospitals in two states of Australia in 1992 and documented 112 deaths (0.79%) and 109 cases where the adverse event caused greater than 50% disability (0.77%) [14]. Seventy per cent of the deaths and 58% of the cases of significant disability were considered to have had a high degree of preventability [49]. Thus, for the individual clinicians, treating departments and units, and even the 28 study hospitals themselves, their actual experience of these outcomes over the year would be minimal (one or two cases) [14].

The 'Swiss Cheese' model gives a poor explanation of the multitude of insidious individual accidents that occur in hospitals and is too simplistic for the complexity of most patients and the complex matrix of health care that is provided in a hospital. Most importantly, the focus on system issues whilst valid and important has detracted from what is really needed: focussed attention on clinical competence and accountability at the patient interface.

#### **4. CFC and the traditional hierarchical referral model of care**

The term 'Futile Cycle' is a term used in cell biology and biochemistry to explain the conversion of one substance to another and back to the original substance by two always on enzymatic pathways. However, despite the enzymatic activity and energy utilisation, there is no net output or gain from this energy-consuming and active process. This is exactly what we see with hospital patient adverse events and in particular the deteriorating patient, a lot of clinical activity, none of which effectively alters the trajectory of the patient towards the adverse event. The clinical activity occurs in a traditional hierarchal referral model of care that by its very nature is often either unresponsive or slowly responsive and where the exhaustive policy and procedures are often ignored.

In the hospital, the CFC usually starts with the most junior level of the 'traditional hierarchical referral model of care', at the bedside with the interaction between the junior nurse and the patient (**Figure 4**). With a clinical abnormality, be it an observation, a wrong drug order or a procedural failure, the junior nurse must make a decision as to the significance of the abnormality and the importance of reporting it to a more senior team member, either a senior nurse or the most available (usually junior) doctor. However, that decision to escalate the issue can be influenced in the workplace culture that exists in the particular micro-environment of that bedside and that ward at that time [58]. If the concern or abnormality is escalated, it is to the next person in the care hierarchy of the team looking after that patient. This is often the junior doctor who then needs to attend, assess and then also make a decision about whether or not to escalate the issue to the next person in the hierarchy. This is important because, for the most, the junior doctor does not have the skills or emotional intelligence to appropriately manage many of these clinical abnormalities [51–54]. If the issue is escalated, it is often to a middle-grade doctor, one who is often a specialist in training and who as such may be difficult to find. Unlike their juniors, usually this grade of doctor does have the technical and clinical abilities to deal with the particular issue. However, they are often over-committed with clinics, operating theatre, but more importantly often see themselves more like the consultants they aspire to be rather than a junior doctor having to deal with patient problems on

history, associated co-morbidities, drug regimen and allergies largely constitute their latent conditions. These conditions and their relation to the current presenting complaint that brings the patient into the hospital system are territory that individual health-care workers are usually extremely well trained in and familiar with. Hospital latent conditions are not so explicit, particularly to the patient or the frontline health-care worker. They are made up of a complex matrix of production and cultural imperatives such as the financial operating environment, political and societal imperatives, medico-legal and insurance concerns, compliance issues imposed by various regulatory bodies (often with associated financial incentives or disincentives) and workforce and work-practice issues. Thus, in the hospital system, unlike any other industry, we have a high degree of ever-changing complexity, complex patients and a complex system where adverse events are essentially prevented by a whole host of predominantly soft defences [57]. The 'Swiss Cheese' model is a static model with fixed defences in terms of the layers and the size of holes in each layer. This translates well into most industries, but in health care, the complexity is dynamic and ever changing, the number of holes and layers

42 Vignettes in Patient Safety - Volume 3

change with every patient and each and every different health-care professional.

comes over the year would be minimal (one or two cases) [14].

The third problem with the 'Swiss Cheese' model is that adverse events in hospitals occur so insidiously that they become normalised into the operating behaviour and practice of the organisation. This is distinct from large-scale industrial accidents, where the impact of the event has a high degree of face validity, primarily due to the immediacy and scale of the event. Therefore, in terms of numbers, patient adverse events may constitute a crisis. However, to the individual practitioner or even hospital, these events may not appear to be a problem. On the whole, such events are infrequent and occur over a long time frame. For example, The Quality in Australian Health Care Study looked at a random sample of 14,179 admissions to 28 hospitals in two states of Australia in 1992 and documented 112 deaths (0.79%) and 109 cases where the adverse event caused greater than 50% disability (0.77%) [14]. Seventy per cent of the deaths and 58% of the cases of significant disability were considered to have had a high degree of preventability [49]. Thus, for the individual clinicians, treating departments and units, and even the 28 study hospitals themselves, their actual experience of these out-

The 'Swiss Cheese' model gives a poor explanation of the multitude of insidious individual accidents that occur in hospitals and is too simplistic for the complexity of most patients and the complex matrix of health care that is provided in a hospital. Most importantly, the focus on system issues whilst valid and important has detracted from what is really needed:

The term 'Futile Cycle' is a term used in cell biology and biochemistry to explain the conversion of one substance to another and back to the original substance by two always on enzymatic pathways. However, despite the enzymatic activity and energy utilisation, there is no net output or gain from this energy-consuming and active process. This is exactly what we see with hospital patient adverse events and in particular the deteriorating patient, a lot of clinical activity, none of which effectively alters the trajectory of the patient towards the adverse event. The clinical activity

focussed attention on clinical competence and accountability at the patient interface.

**4. CFC and the traditional hierarchical referral model of care**

**Figure 4.** Clinical futile cycles [38, 39].

the ward. In addition, this grade of doctor is diagnosis–focused and often we see them giving instructions to their juniors (usually appropriately) to organise specialised investigations and other speciality consultations. There is nothing wrong with this, except for the fact that it is time-consuming [59].

and subsequent education takes no account of the fact that as soon as these people graduate,

Adverse Events in Hospitals: "Swiss Cheese" Versus the "Hierarchal Referral Model of Care and…

http://dx.doi.org/10.5772/intechopen.75380

45

The clinical care we deliver (and receive) is a function of the education and capability of our students who will eventually be our doctors and ultimately clinical leaders and decision-makers. What we teach and practise best is the point-of-care medicine and clinical interventions. Therefore, it is no surprise that what we examine and what students focus on are specific point-of-care clinical assessments and interventions [74]. This is best represented by the objective, structured, clinical, examination system (OSCE) that is now a widespread and common form of summative assessment [75]. In the OSCE, candidates undertake clinical assessment tasks at a number of specific stations for 5–8 min. Each station has a structured 'score card' that students must address to get the points. This system of examination in no way gives any indication on a student's ability and competency to comprehensively take a history, perform a physical examination, synthesise these findings into a meaningful problem list and finally and actually least importantly come up with a diagnosis [76]. It has got to the point now in the undergraduate curriculum that the clinical process of whole patient assessment is variably taught and certainly not examined, in a sufficiently stringent manner to motivate students to spend long hours doing patient histories and examinations. Having competent health-care professionals spend time with and understanding our patients is the single biggest step to

Second, priority needs to be given to the core business of hospital care, the interaction at the bedside and clinic between the patient and the various health-care professionals [4, 5, 61]. Clinical futile cycles give a practical platform to understand this culture. We need to accept that an abnormal or an inappropriate workplace culture is at the heart of every major inquiry into poor hospital care [77–82]. Every report into these enquiries recommends change. Yet, 30 years on from Bristol [81], we have mid-Staffordshire [80]. So, what have we really learned from the reports and thousands of pages of recommendations? Nothing. We need a different strategy: one that puts the patient and their well-being first. This should be followed by the implicit understanding that our core business is that of interaction with the patient from the most basic and junior levels. The bedside health-care team needs to be trained, credentialed and supported to deliver better health care, not as individual players, but as members of a team.

Despite the hundreds of millions of dollars spent on patient safety, we have very little to show for it except the fact that we know that the problem is real, common and universal to all health-care settings. In this chapter, we propose that the reason why we have not been able to improve patient safety is because we really do not understand what is going on at the point

The organisational response is based on mandated requirements, which look at system and operational issues. Rarely do we focus on the quality of the judgements made by the individual clinicians involved in adverse events and usually never do we question the clinical

they will be working in a team environment.

making health care safe.

**6. Conclusion**

of clinical intervention.

culture in which these events occur.

In support of the CFC model is the study that has looked at the causation of adverse events in hospitals [13, 37, 49, 50]. All these studies can assign almost all causation to three human factor issues at the patient interface: competency, cognition (or failure thereof) and culture. Perhaps, the most disturbing example of this was described in the MERIT study, a randomised cluster control study of Medical Emergency Teams (MET) [60] in 23 Australian hospitals (including private and rural hospitals) in 2002. In the nearly 500 cardiac arrests that occurred during the study, in more than a third of instances staff took abnormal (that broached MET activation criteria) patient observations in the 15 min prior to the cardiac arrest, but did not activate an emergency response. The first thing of note with this phenomenon was that the incidence of not calling for help in an abnormal patient situation was high at 30% in the intervention hospitals and 40% in the control hospitals. Put in another way, in the average Australian hospital in 2002, if a patient had documented abnormal signs, in the 15 min before a cardiac arrest, in up to 40% of the time the staff did nothing about this. Another thing of note with these findings is that in intervention hospitals that had an intense education process on the new MET activation policy and procedure, the incidence of calling for help was only 10% greater than the control hospitals [60]. It is here at the bedside with the pre-cardiac arrest patient that the staff are trapped in a CFC, unable to get out of it due to either clinical incompetency (not able to recognise and act for the pre-arrest patient) and/or culture, whereby calling for help maybe considered not the norm in that ward, on that shift at that time [4, 5, 61–64].

The 'Swiss Cheese' response when RRS fails at the sharp end, for whatever reason, the response is to assume policy and procedure failure, despite the fact that there is no direct evidence for the benefit of Rapid Response Systems (RRS) [62–64]. However, it is well documented that there may be problem with the face validity of RRS due to the very low specificity of the activation criteria [65–67]. Furthermore, there may be problems around staff competency or cultural issues around staff losing face by calling for help. As a result, rather than trying to understand or deal with this very real issue of face validity, possible competency issues and probable cultural issues, the administrative response, all too often, is just to alter the policy and procedure and make the activation criteria mandatory for the bedside staff [68].
