Evaluating Resident Duty Hour Reforms: Title and subTitle BreakMore Work to Do

The long work hours of medical residents have received increasing attention in recent years due to concerns about patient safety and the health and education of residents themselves. Although patient safety has been heavily emphasized in the media coverage of duty hour reforms, the evidence that long resident duty hours adversely affect patient outcomes is relatively poor.¹ Indeed, when the Accreditation Council for Graduate Medical Education (ACGME) decided to implement duty hour reforms, some expressed concern that patient care could suffer, emphasizing increased discontinuities of care, or the costs of added staffing needed to provide coverage following duty hour restrictions.^{2 - 3} Such concerns make empirical analyses of the consequences of duty hour restrictions especially important.

The 2 articles by Volpp and colleagues^{4 - 5} in this issue of JAMA are among the first studies to examine changes in patient outcomes following ACGME duty hour reforms. One study⁴ finds no evidence of effects on mortality among Medicare patients. The other study⁵ finds some evidence for reduced mortality among patients in US Veterans Affairs (VA) hospitals, in particular those patients with acute myocardial infarction (AMI), in the second year after duty hour reforms. These results, together with another recent large study⁶ that found some evidence of mortality reductions in medical patients in teaching hospitals following duty hour reforms using data from a large fraction of US hospitals, may be reassuring to those who feared that duty hour reforms would adversely affect patient outcomes. These studies may also provide some encouraging news for others who had hoped that duty hour reforms would improve outcomes.

Despite the suggestions of improvement in mortality in some subgroups, the overall impression remains that duty hour restrictions have had little effect on patient mortality. This impression is strengthened in considering that in the VA study the mortality effect in the AMI subgroup in the second year of duty hour reform implementation was one of several subgroup analyses performed. Moreover, the subgroup is not necessarily one for which it seems likely that reduced duty hours would affect mortality, because most of the interventions documented to reduce AMI mortality (ie, early recognition and initiation of therapy such as thrombolysis or percutaneous coronary intervention) are often initiated prehospital or in the emergency department, where duty hours do not likely affect practice patterns.⁷ Indeed, in the regionalized VA system, there is evidence that these cardiac technologies tend to be limited to teaching hospitals, and that patients with AMI initially admitted to these hospitals have improved outcomes.⁸ This is especially interesting given the finding of improved outcomes for VA teaching hospitals in 2004 relative to other VA hospitals, because updated American College of Cardiology/American Heart Association guidelines for the management of patients with AMI were released in July 2004,⁹ and might well have been more rapidly adopted in these teaching hospitals. Similarly, it is possible that other secular trends that differentially affected larger or predominantly teaching hospitals in the VA system, such as the more aggressive coding of comorbidity in medical patients during this time period reported by Volpp et al,⁵ could also account for differences observed in medical patients. These possibilities reinforce the impression that there is still not clear evidence for an effect of duty hour reforms on mortality.

This is just part of the evidence that is still needed to develop a comprehensive understanding of the effects of duty hour reform. With respect to patient care, outcomes such as morbidity and the cost of care also may be affected, and process measures, such as rates of medical errors, may be more easily identified if effects on costs and outcomes of care are small. At least 2 single institution studies^{10 - 11} have found evidence of beneficial effects on these other measures. Duty hour reforms likely also offer direct benefits for medical residents through reductions in fatigue that could improve resident learning and reduce fatigue-related injuries, such as motor vehicle crashes and needlesticks.^{12 - 13}

On the other hand, the potential for the development of a “shift-work” mentality under duty hour reforms could undermine personal responsibility of individual physicians, a core component of medical professionalism.¹⁴ Even such theoretical concerns are complex to assess because it could be countered that professionalism should increasingly be defined in terms of a physician's function within teams, and that duty hour reforms provide an appropriate opportunity to teach residents effective handoffs and other aspects of professional behavior that are critical when care is provided by teams.¹⁵ All of these potential effects of duty hours deserve consideration in a comprehensive analysis of duty hour reform, which should also consider all alternative strategies. For example, providing nighttime coverage to allow call-night naps can increase on-call sleep and reduce post-call fatigue, and could potentially improve continuity of care if longer post-call shifts were possible.¹⁶ In contrast, in 2 national studies^{17 - 18} residents reported less sleep on-call and more handoffs after duty hour restrictions.

One effect of duty hour reforms about which there can be little question, however, is that these reforms have resulted in significant changes in the staffing in teaching hospitals. Some of the work previously performed by housestaff is now performed by attending physicians and fellows, as well as newly hired staff who can provide additional coverage, including hospitalists and nonphysician clinicians.^{19 - 20} These changes all carry costs, whether explicit (such as the cost of hiring new staff) or implicit (such as that of taking the time of attendings and fellows from other activities). Such costs also need to be studied to gain a comprehensive understanding of the effects of duty hour reforms. One study suggested that it would require a substantial decline in adverse events (between 5.1% and 8.5%) to make duty hour reforms cost-neutral from a societal perspective, and an even larger reduction (between 18.5% and 30.9%) to make them cost-neutral for teaching hospitals.²¹

Indeed, the reallocation of work to teaching attendings, fellows, staff hospitalists, or other clinicians may be an important factor to consider in understanding any changes in patient outcomes following duty hour reforms, and especially improvements in outcomes, because those clinicians may often be more experienced than residents. If such adjustments in staffing are needed to maintain outcomes following duty hours reforms, it is possible that less affluent teaching institutions may have been especially likely to experience a decline in outcomes relative to more affluent institutions following duty hour reforms. In addition, the long-term effects of this reallocation of work on residents' clinical training and experience and their subsequent outcomes have yet to be determined. Despite the extensive changes made by teaching hospitals, adherence with duty hour restrictions is not complete.¹⁷ This raises questions of how to interpret studies of duty hour limitations, such as those by Volpp et al,^{4 - 5} that do not assess adherence. Studies that assess both adherence and outcomes in patients cared for by residents before and after duty hours are needed.

With duty hours already in place, even if evidence suggesting adverse effects of duty hours reform were identified, it seems unlikely that these changes are likely to be reversed because of their advantages for house officers. Nevertheless, understanding what institutional responses to duty hour reforms are most effective, and cost-effective, is likely to be valuable. The importance of this point is amplified by suggestions by some sleep experts that currently allowable working hours still jeopardize patient safety and resident education, so that shortening working hours even further should be considered.²² Although such concerns certainly have some appeal from the perspective of sleep science, variable adherence and the relative lack of data concerning the consequences of the current duty hour restrictions suggest that better data should be sought on the effects of the initial limitations on duty hours before additional limits are implemented.

How such data are best obtained is a complex issue. On the one hand, broad policy changes such as the ACGME reforms can be exploited to study large numbers of patients affected by duty hour reforms, but do not offer ideal control groups. On the other hand, narrower studies with randomization to varying limits on duty hours could be conducted by individual institutions if granted necessary waivers. Although these studies would offer better control groups, they would require substantial effort to execute with sufficient power for relevant outcomes. Most likely, studies of both types will be valuable. Studies of the effects of duty hours such as those by Volpp et al^{4 - 5} will need to be augmented so that decisions about duty hour limitations can be made that are well informed about their consequences.