Work Hours and Reducing Fatigue-Related Risk: Good Research vs Good Policy

While there is little doubt that physicians-in-training work long hours and experience chronic sleep restriction over many years, the consequences remain unclear. There is not yet a significant body of evidence-based data clearly delineating the relationship between work patterns, sleep duration, and patient safety or practitioner well-being. On the contrary, anecdote and economics have driven much of the public debate over the legitimacy and necessity for extended work hours and any presumed impact on health care consumers and practitioners.

In this issue of JAMA, 2 articles make a significant contribution to a more informed debate. The first article by Fletcher and colleagues¹ reviews the impact of restricted working hours on physicians’ well-being and professional competency. This review focuses on describing the types of interventions by specialty and by subsequent impact.

At best, the cited studies can be characterized as naive examples of the “action research” paradigm.² Few, if any, of the studies had systematic evaluation as a prospective aspect of implementation. The authors rightly assert that the majority of studies in the area can be characterized as poorly controlled and based primarily on subjective, introspective, and poorly defined dependent measures. Indeed, they conclude that only 1 of the cited studies met the basic requirements of an evidence-based trial.³ Thus, the most significant conclusion from this review is the difficulty in drawing any meaningful conclusions from the existing literature that could serve to guide policy.

Nevertheless, setting aside the limitations of and contradictions among individual studies, it is possible to look for converging evidence or broad themes from this body of work. From this perspective, there appears to be a small but significant group of studies for which restricted working hours were associated with self-reported improvements in variables that broadly reflect “quality-of-life” issues. For example, reductions in self-reported stress, increases in the amount and quality of personal and study time, and improved mood and affect were commonly described.

In contrast, the perceptions of senior staff often emphasized negative impacts using poorly defined constructs, such as shift work mentality, experience, continuity of care, and development of professionalism. Several of the cited studies reported fewer procedures or fewer attended patients as a result of restricted working hours and inferred a consequence of reduced patient safety. However, the idea that limiting work hours necessarily results in reduced patient care is empirically unsubstantiated and may, at least in part, be interpreted as a prejudice that rationalizes maintenance of the status quo. As alternative possibilities, better-rested physicians arguably might learn faster and achieve competency after fewer procedures or patients as a result of improved motivation, attention, or memory. Problems with “continuity of care” might be more usefully improved through the quality and standardization of hand-over procedures than by extending shift durations.

Perhaps the most important issue raised in the review by Fletcher et al¹ is why the data pertinent to this topic are comparatively “unscientific.” This is particularly salient in view of the potential importance of this matter to patient safety and the recruitment and retention of medical staff. The obvious response to this question is that the topic is new, the interventions as yet nonstandardized, and the outcome measures not well defined or validated. The limitations of these studies may reflect a lack of collaboration between medical professionals and relevant domain experts. Addressing this would undoubtedly increase experimental rigor.

However, this may not be the only explanation. Anthropological studies of the professions have frequently shown that topics calling into question long-held cultural values of the researchers can create an epistemological “blind spot.”^{4 - 5} This is a more intractable problem and may require researchers to develop a more detached view of the way in which hours of work are linked, for example, to economic factors or cultural markers of commitment and initiation.

Improving the quality of research on the relationship between extended work hours and the quality of working-life is an important professional issue, but the impact on patient safety may be more salient for the public. This topic is addressed by a second article in this issue, in which Arnedt and colleagues⁶ report that, relative to light call (approximately 44 hours per week), heavy call (80-90 hours per week) reduced sleep and by inference increased impairment using standardized measures of neurobehavioral performance. As a relative index of the level of impairment, moderate alcohol intoxication was used as a positive control. It is important to note that the type of work also differed between conditions. Intensive care rotations may have a different workload than elective or selective rotations. Thus, the results need to be interpreted in light of the contribution of the extended work hours as well as the type of work being performed.

Nevertheless, heavier workload was associated with an average reduction in sleep of approximately 10 hours per week and approximately 3 hours for the 24 hours before testing. Reduced sleep was associated with significant declines in neurobehavioral performance. Moreover, the impairment associated with heavy call was broadly equivalent to the impairment reported for blood alcohol concentrations between 0.04% to 0.05%. Also, for heavy call, performance on a simulated driving task was more impaired than for a 0.04% to 0.05% blood alcohol concentration, and residents’ perceptions of fatigue-related impairment were limited and task specific.

Although the authors acknowledge that these laboratory tests of performance have not been validated against medical tasks, the indirect implication is that residents working 80- to 90-hour weeks are at an equivalent or greater risk compared with an intoxicated physician. This is, without doubt, a notable finding and one that should concern those responsible for patient safety and medical training.

As pointed out by the authors, however, response times, vigilance, and driving performance are not measures of clinical efficacy or patient safety. It is difficult and perhaps premature to predict how the seeming equivalence for selected neurobehavioral effects of fatigue and alcohol extend to clinical decision making across diverse working conditions. Nevertheless, studies such as this should motivate clinicians and managers to reflect on the legal and scientific defensibility of current work practices.

It would be reasonable for physicians and other health care professionals to conclude that restricting working hours was a commonsense corollary of the research and would be prudent, if not rational, public policy. Indeed, fatigue management initiatives restricting work hours have flourished in recent years. For example, in the United States, guidelines now restrict work hours for physicians-in-training to 80 hours per week or less.⁷ The United Kingdom issued amendments to the European Working Time Directive pertaining to junior physicians.⁸ Starting in the late 1990s, the Australian Medical Association’s Safe Working Hours Project was developed and was ratified by 2002.^{9 - 10}

Based on the literature to date, there is little doubt that an increased likelihood of sleep restriction, fatigue, and cognitive impairment is associated with many schedules worked by medical staff. Some indirect evidence links fatigue with impaired medical decision making^{11 - 13} and reduced patient safety.¹⁴ The evidence of a similar link between hours of work and patient safety is, however, less clear. Most important, studies demonstrating unequivocally that regional or national regulatory restrictions to working hours improve patient safety remain to be performed. Until then, merely restricting working hours can only be considered a speculative means of managing fatigue-related risk.

Despite the intuitive appeal of restricting working hours, it is important to consider potential negative ramifications. Without a broad, balanced overview, it would be easy to fall victim to the law of inadvertent consequence. In some scenarios, limiting working hours may increase risk to patients and physicians. For example, restricted working hours may lead to restricted access to health care practitioners through a reduction in the labor supply, insufficient clinical preparation for the “real world,” increased sleep restriction in senior physicians, or increases in error rates due to work intensification. Each of these could result in higher overall risk.

It is essential to ensure that the policies to reduce fatigue-related risk do not increase risk due to inadvertent or unforeseen consequences of the policy. To minimize the likelihood of paradoxical outcomes, research and policy to address fatigue should be developed within a broader framework that explicitly also recognizes economic implications, and institutional and professional culture.

An increased emphasis should also be placed on the centrality of occupational safety and health and risk management principles to policy debates over working hours and safety. In the past, senior medical staff and human resources managers dominated the organizational debate over working hours. Fatigue risk management may be embedded more usefully within a broader systems-based approach to safety management.¹⁵ The following principles may prove useful.

First, fatigue-related risk should be seen as a shared responsibility of organizations and employees consistent with their legislated occupational safety and health obligation to provide a safe workplace.

Second, in addition to hours of work, the timing and duration of sleep and wakefulness associated with a specific work schedule should be determined empirically and mandated as the primary indicator of fatigue-related risk.¹⁶

Third, the nature of the task undertaken by a medical practitioner must form part of the risk assessment. Task differences by specialty, seniority, or degree of supervision can contribute significantly toward differences in the overall likelihood or consequence of fatigue-related error and patient safety.

Fourth, to be considered successful, interventions to reduce fatigue-related risk should, as far as is practicable, be required to also demonstrate a net reduction in overall risk.

And fifth, restricting working hours is only one of many ways in which fatigue-related risk can be reduced. Substantial reductions in fatigue-related risk can also be achieved through improved design of standard procedures that are better defended against fatigue-related error.

These principles and guidelines can provide a useful context within which to evaluate the evolving nexus between fatigue research, clinical efficacy, and fatigue management policy. Failure to consider the broader issues carries the considerable hazard that well-intentioned policies to reduce fatigue-related risk may not lead to overall improvements in patient safety.