Over recent weeks, media commentators, government, the health-care professions and unions again have gone into a tailspin over long working hours. At the centre of this lies the vexed issue of extended shifts, often up to 36 or more, in medical staff.

On the surface the debate would seem self-evident. Forty years of research indicates that the effects of extended wakefulness are unequivocal, cognitive performance is impaired, particularly for complex time-pressured decisions. Furthermore, fatigue negatively impacts on emotional tone and can make tired people more irritable, more withdrawn more hierarchical and less sensitive to the needs of patients and other staff.

Moreover, if you talk to doctors and nurses, as we have over many years, they will all enthusiastically recount stories of poor clinical decision making, and adverse patient safety outcomes directly attributed to fatigue-related impairment. Although, not surprisingly, they are more likely to attribute the effects to others rather than themselves. Taken together the qualitative and quantitative research data would clearly predict that long working hours should significantly impair performance and increase the risk of adverse outcomes to patients and doctors.

Not surprisingly, this has lead to increasingly strident calls to reduce the risk of adverse outcomes due to fatigue-related error limit doctors working hours and to improve patient safety primarily by prescribing maximum weekly hours, shift durations and minimum break durations. This has been the case in many other industries for more than a century. In Europe, the working time directive and in the US, the 80-hour week rule, have recently been instituted as a policy initiative that will reduce working hours, reduce fatigue and improve safety outcomes for doctors and patients. Problem solved, right?

Well in fact no. Unfortunately, as H.L. Mencken famously observed, for every complex problem there is a simple solution … and it is usually wrong! Nowhere is this aphorism more appropriate than in the complex area of fatigue risk management and especially in health care. Despite the intuitive appeal of the argument, i.e. that fatigue impairs performance and, therefore, reduces patient safety, or of the corollary, i.e. that we can, therefore, improve safety by reducing working hours in order to reduce fatigue, the truth is not that simple.

If we look past the laboratory-based studies on simple cognitive performance, there is not a convincing body of good research showing that tired doctors compromise patient safety. The reasons for this are complex, but important to consider if we are to implement effective policy. Many older doctors have argued that their training protects them against the effects of fatigue and that they have ‘learned’ to deal with it through experience.

In  fact, there is no experimental evidence to support this. Indeed recent studies published in The New England Journal of Medicine have shown that doctors working long hours are actually far more likely to have a car accident while driving to and from work than controls. It is entirely reasonable to therefore conclude that what we have learnt from 40 years of research may be true. That is, that the effects of fatigue on doctors, when we control for task, are the same as they are for truck drivers and for the population at large. That is, that tired drivers are more likely to fall asleep at the wheel and have an accident.

The key issue, however, is the phrase … when we control for task. What is also unequivocal in the scientific literature is the observation that at the individual level, different tasks show differential susceptibility to fatigue. Some task are significantly impaired by fatigue and others less so.

In general, the tasks that are refractory to fatigue fall into two categories. The first category, for which there is a significant scientific literature, are those that are relatively automatic, well learned and require few attentional resources. The other category is more complex. We can also see performance maintained for tasks that are complex and have significant potential for adaptive plasticity.

In simple terms, we can change the way we do the task in order to compensate for the fact that we are tired. For example, a surgeon may decide to slow down in order to reduce the chance of making an error while operating. A tired airline pilot may prepare for landing by starting preparations 10 minutes early so as to reduce the likelihood of a time-pressured error while tired.

At the moment, the scientific literature is remarkably silent on the issue of adaptive plasticity in fatigued individuals. In the case of tired doctors we think this is a critical area of research since anecdotal evidence suggests the health-care environment is a very good example of how fatigue-related impairment is frequently managed through compensatory protective behaviours. Given the relative proportion of time that doctors work fatigued, the number of errors and adverse outcomes is remarkably low. Much of the current debate reflects the tension between the error rates inferred from relatively simplistic laboratory studies and the everyday experience of senior doctors.

Not unreasonably, many doctors are very aware of the level of performance protection that their experience confers and resent the lack of nuance in inferences drawn from over-simplified laboratory-based studies. If we are to move beyond the relatively sterile debate that characterises current policy deliberations, we must develop a research paradigm that extends beyond the “button pushing” tests of human performance. We need a new conceptualisation of fatigue-related effects on performance that also seeks to understand the role of compensatory protective behaviours. Most importantly, we need an approach to fatigue management that seeks to collect, categorise, evaluate and then disseminate these compensatory protective behaviours systematically as part of a general medical or health care education.

Earlier in this piece I alluded to the fact that there are few studies demonstrating that fatigue negatively impacts on actual clinical performance. It is tempting to conclude that fatigue is an overstated threat because of the argument above. That is, that while fatigue does impact performance, it is by and large compensated for by the use of protective behaviours. This is not the only reason for the lack of clear evidence. There are also some more arcane reasons linked to the cultural prejudices about what constitutes proof in the health care sector.

The current and favoured method of proof in the health-care sector is the randomised control trial (RCT) study. Those who do not believe that fatigue compromises patient safety point to the lack of Cochrane-grade evidence to support the argument linking fatigue with impaired clinical performance. In many cases, this claim is either disingenuous or ill-informed. While there is little doubt that the RCT is the gold standard for clinical intervention trials. The notion of a study using a fatigue “placebo” is nonsensical. Similarly, a first-year psychology student would understand that when you observe a person, their behaviour will change in socially desirable ways.

The traditional research paradigms for testing a new drug or surgical technique are not always the appropriate way to determine scientifically whether there is a link between a complex psychological state and performance. A moment’s reflection will confirm the unsuitability of the RCT for any intervention that influences mental state. By comparison, it is also not possible to develop a placebo control for moderate levels of alcohol intoxication [BAC% 0.05-0.08] since participants will clearly be aware of their changed mental state. Nevertheless, our everyday experience of alcohol [as with fatigue] leads to a reasonable conclusion that there is a clear link between fatigue and performance but that the link to consequence may be less well understood.

Given the fact that most individuals are patently aware of the fact that they are tired and that to participate in a study approved by an institutional ethics committee looking at the link between fatigue and performance will require those involved to understand the purpose of the study, must subvert the experimental paradigm of the RCT. This is not to say that we should not do good science. What it means is that we must look for appropriate designs and converging evidence from a range of perspectives. Those who insist that there is no link if there is no RCT to support the link have failed to understand the diversity of the scientific method. The fact that we cannot run a RCT for evolution or that there is not yet a published RCT for the parachute does not prevent us from accepting the theoretical validity of evolution or the practical [and self-evident] benefits of a parachute.

A coherent and scientifically defensible approach to fatigue risk management in health care is further complicated by factors beyond the individual level. Health care is, in general a complex system and exhibits complex systems behaviour. Nowhere is this more critical than in the area of risk management. Current initiatives to restrict doctors working hours would obviously reduce fatigue and may well, as a consequence, reduce adverse patient safety outcomes. But with any intervention, it is critical to look at the flow on effects into the whole system. That is the non-linearities. From a risk-based perspective every intervention must be considered in terms of its “opportunity cost”. That is, if I restrict doctors working hours what will be the effect more broadly across the system.

One does not need to think too deeply to realise that limiting individual doctors working hours will reduce the aggregate supply of doctors across the system unless additional doctors can be recruited. The same holds for other health-care professionals.

At the moment, most health-care systems in most countries, are struggling to manage chronic shortages of doctors and nurses in the face of increasing service demands. Restricting working hours in this context will produce a reduction in the aggregate labour supply. This will, in turn, lead to a potential set of unintended negative outcomes. For example, a reduction in labour supply can lead to an increase in workload per unit time. Doctors managing a greater concurrent workload can make more errors as a consequence of the increased work load. Alternatively, it may be the case that a reduction in supply could mean the lack of availability of a health-care professional. At its most stark: I am sorry, we cannot operate on this patient because the doctor has ‘run out of hours’. I suspect that faced with such a choice, the average Australian would prefer a tired doctor to no doctor at all.

Similarly, if we reduce doctors working hours then we change the nature of their work and remuneration. In SA in the mid-1990s, reduced hours for junior doctors were introduced to much fanfare. One of the principle motivations for this policy change was to support the “safe hours” initiatives that are so currently popular.

However, in this case there were some important and unintended consequences. For example, in response to the reduction in income, some of the junior doctors decided to seek secondary employment and moonlight in the private hospital system. As a consequence, their actual fatigue levels were significantly higher since they would work one shift at a public hospital, a second shift at a private hospital then return to the same public hospital for a third consecutive shift. A further consequence was a greater “call-in” frequency for senior doctors and VMOs. As a consequence, senior doctors were being interrupted during the night and then would resume their typical day activities with no compensatory breaks to recover.

Unfortunately and despite the somewhat overwrought rhetoric, fatigue risk management remains a complex issue for which there is no simple straightforward solution. Pragmatically, if there was a simple solution we would have thought of it already. God knows, enough really smart people have thought about it for a long time!

Fatigue risk management is the perfect example of what Tony Blair described as a “wicked problem”. That is, one for which the solution is complex, multi-factorial and will require thoughtfulness, flexibility and time. Successful fatigue risk management in health care will require exactly that.

But there are some bright spots in the debate. Queensland Health is a good example of an organisation that has grasped the complexity of the issue and the need for an integrated approach. It has recently started a series of pilot projects that will require individual regions, hospitals and medical units to develop a comprehensive risk-management program suited to the context in which the staff are operating. What will work for a GP in a rural or remote community will not work in a busy teaching hospital in Brisbane.

Reducing working hours without giving due consideration of the impacts on overall patient care and the total risk environment is just plain negligent. QH has realised that good fatigue-risk management will require the professions to rethink many of the cultural assumptions around what constitutes best-practice. While long hours of work have been worn as a badge of honour, they can occasionally lead to bad outcomes as was reported in a recent coronial inquiry in that state.

In those trials, the doctors and nurses who participated were willing to discuss the effects of fatigue on theirs and their colleagues’ performance in a full and frank manner. They were also able to identify, acknowledged and formalise a very rich ecology of adaptive and protective behaviours as a central element of their fatigue risk management systems. Most importantly, some of the researchers were able to engage with the health-care professionals in a way that enabled them to understand the complexity of the real-world health-care environment, the commitment of those individuals to patient well-being and the fact that fatigue was only one of many conflicting pressures they had to deal with.

Yet despite the innovation, commitment and resources associated with this project, it was reported in the tabloid media as QH “suggesting that all doctors had to do to manage fatigue was to drink four cups of coffee a day”. Again we see the need to simplify the debate to a one-dimensional level and to sensationalise through disingenuous misrepresentation.

I have no issue with the need to subject government policy and research to intense public scrutiny. We are all the better for it. But it is also critical that an issue as important as this is afforded the luxury of accurate representation. My concern is that by misrepresenting the issue we will persuade the government and the health-care professionals to retreat from rational, evidence-based policy. By demonising our first attempts to deal with fatigue-related risk honestly and openly in a way that may well improve patient and doctor safety, we may allow populist sentiment and simplistic rhetoric compromise our health-care system.

If we are to solve this particular “wicked” problem, we need to work collectively to understand the risks that fatigue imposes on patients and health-care professionals. This requires the researchers to look carefully at the effects of fatigue, the subtlety of our adaptive behaviours and to understand the overall effects of interventions on aggregate risk at the systems level.

This will require the organisations representing health-care professionals to engage meaningfully in this research and to understand that good research will help us understand the complex effects of fatigue on health-care outcomes and, most importantly, the most cost effective ways in which the risks associated with fatigue can be reduced. In some cases we will need to make changes in work practices to reduce the level of fatigue. In others, we will need to understand the best way to “fatigue proof” the ways our health-care professionals work when funding, geography, continuity of care or commitment require people to work while fatigued rather than allow someone’s health to be compromised.

Most importantly, it will require our politicians and governments to invest in developing solutions that are evidence-based, through better funding of research in the area, by funding better policy development, implementation and evaluation and by providing the leadership that will ensure that this issue is neither ignored nor exploited for the sake of expediency.

Drew Dawson contributed to the development of Queensland Health’s fatigue risk management guidelines.