Introduction

Resident work hours have been considered important to the learning and professional socialization of physicians, in an industry where patient needs are not met during the standard workday. Yet these long hours also are of concern for patients and residents, due to the association between sleep loss, medical errors for patients, and risks to residents themselves, including motor vehicle accidents, occupational injuries, and negative effects on well-being. Ideally, decisions regarding work hour limits for resident physicians should be based on the relevant scientific literature regarding approaches that optimize patient safety, resident learning, and resident safety and well-being.

The Accreditation Council for Graduate Medical Education (ACGME) work hour standards, which include the early standards that predate the 2003 common program requirements and the standards that went into effect in 2011, were guided by the available scientific evidence generally gleaned from published studies.^1,2 The aim of this selective, narrative review is to present the seminal articles over the past 4 decades that shaped the community's and the ACGME's thinking about resident hours, including highlighting themes and controversies in this literature. An additional aim is to identify areas where relevant literature is lacking, and suggest directions for future research.

Methods

The review encompasses 30 well-cited, influential articles published in US peer-reviewed journals between 1971 and 2013. Articles were selected from a comprehensive review of the literature on resident work hours and related dimensions. These were compiled by the author for an ACGME Task Force that was charged with a revision of Section VI (Resident Duty Hours in the Learning and Working Environment) of the Common Program Requirements. This compilation consisted of more than 1050 original articles. To be included in this narrative review, a study had to examine resident hours or limits on them, and had to be a frequently cited article that addressed an aspect of resident work hours, such as medical errors, resident safety and well-being, and resident learning, or the positive or negative consequences of work hour limits. Articles also had to meet a basic quality criterion.

Citation volume was assessed independently using Google Scholar. For quantitative studies, quality was examined using the Medical Education Research Study Quality Instrument (MERSQI).³ The MERSQI assesses design, sampling, type of data, validity, data analysis, and outcomes, for a maximum score of 18 and an average score of 9.95 for medical education studies. To be included, studies had to score 11 or above, and the scores for the 28 quantitative articles ranged from 11.4 to 17, with many of the older studies scoring lower on the metric. MERSQI data are not reported individually, and a number of studies have separate scores for different outcome variables, such as the external collection of safety data versus the self-reporting of work hours or educational outcomes. The 2 purely qualitative studies in the analysis scored high on a quality assessment tool for qualitative research.⁴ The organization of the review highlights the historical evolution of the community's knowledge and understanding about resident hours, the focus on several constant themes throughout the 4 decades, and the emerging focus on new outcomes of interest.

Results

As shown in the table, many of the articles included in this review are among the most highly cited articles in this vast literature, with the most cited study garnering nearly 1500 citations.^{4–11,13–30,32–34} In contrast, 2 important qualitative studies included in the analysis^12,31 received fewer citations.

table Study Findings and Contributions to/Impact of the Knowledge About Work Hour Limits

View Fullscreen

The major themes in the early literature on work hours begin the discussion of the impact of long hours on residents' clinical and cognitive performance,^4,8 resident safety and well-being,^{5,17,18,23,24,26} and resident driving safety after call or night shifts.^14,16,21 A second theme encompasses the risks to patients associated with resident work,^{10,20,22,32–24} including higher risks for night admissions⁹ and problems with cross-coverage of patients.¹³ In contrast, data on the impact of duty hour limits on resident learning and professional development are scarce, and studies generally are limited to assessing the impact on board and in-training examination performance and resident self-reporting of educational impact.^8,29,30

In 1989, New York became the first state to regulate resident hours; thus, the third theme is the impact of limits on resident hours on the quality and safety of care. Studies of the impact of New York State's regulation of resident hours predate studies of limits at the national level. These early studies produced some mixed results, including delayed test ordering and increased complications in medical residents' patients,¹⁰ as well as residents reporting conflict between their desire to comply with the regulations and their obligation to patients and colleagues.¹² While these early studies clearly highlight the need for limits on resident work, beginning with that by Friedman and colleagues,⁴ they also present conflicting findings, with no apparent impact of long hours for surgical residents⁷ and an increased number of errors for surgical patients under the New York State limits.²² This marked the beginning of a 3-decades–long debate about differences in the impact of work hour limits in medical versus surgical specialties. Studies of the effect of 2003 ACGME standards generally suggest a benefit of these limits, particularly the 80-hour weekly limits and standards reducing the frequency of overnight call. However, some studies produced mixed findings or suggested no impact on quality and safety of care. For example, 2 large studies using administrative data showed no significant differences in outcomes for medical²⁸ or surgical^26,28 patients.

Several studies showed that residents' actual hours that were worked routinely exceeded scheduled hours, as residents tend to remain to complete tasks. A high-quality study of internal medicine interns limited to 16 hours in an intensive care unit rotation showed that actual hours worked significantly exceeded scheduled hours during the intervention and control weeks¹⁹; 2 studies showed that residents regularly remained beyond scheduled hours to complete clinical tasks^12,31; and an early study showed significant differences in resident work styles and efficiency, with longer and more difficult on-call experiences for some, despite comparable objective workloads.¹¹

Data on resident well-being indicate serious consequences of residency to trainees' well-being, and suggest that the intern year is a particularly stressful time.^6,18,24 Studies also show a benefit of the 2003 ACGME standards on well-being in both medical and surgical residents.^23,26 Research on the 2003 and 2011 limits clearly show that internal medicine residents' weekly hours have declined from more than 85 at the implementation of the 2003 ACGME standards to around 65 hours per week. This reduction in weekly duty hours may be most responsible for the benefits attributed to the 2003 and 2011 common requirements, offering empirical support for the 80-hour weekly limit and limits on the frequency of overnight call. In this context, it is interesting to note that 6 of 22 primary studies included in a systematic review of the benefit of a 16-hour limit³⁵ actually assessed the impact of reductions in call frequency (while retaining 24-hour call), with call patterns and associated hours in the intervention groups approximating the current practice of call every third or fourth night. A concerning finding is that this reduction in weekly hours, and compliance with the work hour limits, appears to have resulted in an increase in clinical demands, increased perceived workload and stress, and reduced participation in educational activities,^32–34 with 1 study finding an increase in self-reported errors by first-year residents.³⁴

The literature clearly shows the negative effect of sleep loss: several highlight that in a 24-hour/7-day industry, under almost any model of work scheduling, including shift-based approaches, some individuals must work at their circadian nadir, with associated consequences for their performance, safety, and well-being.^16,17,20,24 Scheduling may mitigate this decrement, but it is not possible to completely eliminate it. Protected time for napping has been suggested as an intervention; however, 2 studies showed that efforts to “protect” sleep time for on call did not increase sleep, as residents may not use the nap option out of a sense of professionalism and a desire to provide care for their patients.^15,25

Studies that offer insight into the educational impact of work hour limits are scant, with 1 study finding a reduction in performance on the certifying examination for neurological surgery residents,³⁰ and several finding reductions in residents' self-reported participation in didactic activities and satisfaction with their education.^33,34

Discussion

The utility of this influential literature is limited due to small sample sizes, the short-term nature of many interventions, as well as concerns about the ability to generalize from studies performed in a single specialty and, often, a very specific clinical context, such as an intensive care unit, to all specialties and clinical contexts. In fact, 17 of the 30 influential studies used internal medicine residents, and 12 used solely first-year internal medicine trainees, potentially echoing a critique of the psychology literature and its use of sophomore-level college student subjects.³⁶ Three studies each involved residents across multiple specialties, 3 involved surgery residents, 2 involved medicine and surgery residents, 2 involved pediatrics residents, and 1 each involved residents in emergency medicine, family medicine, and neurological surgery.

Many medical, surgical, and hospital-based specialties are represented in a limited way in the large data set from which these studies were selected. In addition, to date few studies have assessed the large number of additional standards added in 2011, including enhanced standards for supervision, the creation of a safe and professional learning environment, educating residents and faculty on fatigue and alertness management, and resident participation in quality and safety improvement.

It also is important to note that positive or negative changes in any outcomes of interest cannot be unequivocally attributed to the limits. Concurrent with the implementation of the 2 sets of common standards, there has been a substantial increase in a focus on quality and safety in teaching settings, as part of the standards themselves,^1,2 through the ACGME's Clinical Learning Environment Review,³⁷ and through a multitude of efforts within the health care sector to enhance the safety and quality of care.³⁸ There also is some indication that the improvement or worsening of quality and safety indicators may be related to greater resident peer or faculty involvement in care. This may enhance patient safety, but it could be problematic for learning. For real learning in a clinical context, residents need to be immersed in systems of patient care, not bobbing gently at the surface in the interest of safety. A letter to the editor on a widely cited trial²⁰ of a 16-hour limit on continuous duty attributed the beneficial outcomes to increased involvement and coverage of first-year residents' work by more senior residents and faculty.³⁹

Efforts to address the negative impact of shortened work hours, particularly the 16-hour limit for first-year residents instituted in 2011, have included night float, improved handoff procedures, and teaching on the night shift, among others. None were universally effective in addressing the negative effects of the work hour limits on patient care or learning.^{13,25,32–24} In fact, some studies have found that night float may produce outcomes inferior to those in other forms of scheduling, and in 1 study night float was terminated early due to concerns about quality and safety of care.³³

To deny the negative impact of long work hours, sleep loss, and fatigue in residents is folly, as these physiological demands are very real. At the same time, to have these physiological limits become the major driver of the system to ensure the learning, professional development, and socialization of the next generation of physicians likely is unrealistic as well. While empirical data is scarce, there is a pervasive sense that the ACGME common work hour standards, particularly those implemented in 2011, have resulted in “work compression,” by wringing out lower intensity work and downtime hours at the hospital that were valuable from an educational, professional socialization, and camaraderie perspective. This is not an indictment of the standards, but more likely of how they were implemented in some resource-constrained, resident-dependent hospitals. There also is some suggestion from citation patterns that certain areas initially thought promising have largely been abandoned as areas for study. For example, a high-quality study of a budget-neutral napping intervention, published in a leading medical education journal, received only 3 citations,⁴⁰ potentially because research priorities have shifted to other interventions with more community buy-in. The work hour reductions instituted in 2003 and 2011 should have brought about changes in the structure of residency training, but this does not appear to have happened in all programs and institutions.

New common work hour standards proposed by the ACGME on November 4, 2016,⁴¹ affirmed continued support for an 80-hour weekly limit, 1 day in 7 free of all program duties, and overnight call no more frequently than every third night. The new standards propose the elimination of the 16-hour limit on continuous duty. The decision was based on evidence that a universal, cross-specialty, cross-clinical setting application of a 16-hour limit has not been beneficial. It is important to note that programs and specialties that have found this limit effective, often as “ramp-up” to allow individuals to transition from the relatively limited clinical experience of many medical students, to the 24-hour, 7-day responsibilities of residency and medical practice, will be able to continue to use it. The new standards seek to address work compression and the shifting of clinical and administrative work to non-counted hours, as well as significantly increase the focus on resident and faculty well-being, and finding meaning in work.⁴¹

Limitations of this review include the potentially subjective selection of articles by the author, with different individuals curating this literature potentially reaching different conclusions as to what constitutes the most influential studies. While citation volume is a fairly undisputed metric, there are other highly cited articles, and there are no accepted metrics of the practical impact of research. In addition, the MERSQI assesses internal validity, yet its dimensions do not address generalizability. Several of the highest-quality studies were conducted with very small samples, in a single specialty, and often in a specific clinical context, reducing generalizability. The review does not include studies on the impact of work hour limits on operative volume, due to the limitations of this literature and the fact that operative volume is a poor proxy for the quality of surgical education. Finally, a number of highly cited studies were omitted from the review because they used identical methods for slightly different populations (eg, patients in Veterans Affairs hospitals versus Medicare patients in nonfederal hospitals) or providing different publications of analyses of a larger data set.

There are a number of underinvestigated areas, with the most prominent being the impact of work hour limits on resident education. This likely is due to many potential confounding variables, and a dearth of accepted, useful metrics for this outcome. A requirement for safe practice in the future is a competent graduate, and a concern is that the safety of patients receiving care in teaching settings today might be achieved at the expense of the future patients that residents will care for after completing training. This will require more research into evidence-based ways to balance these competing demands. There also is a need for research that addresses continuity in residents' exposure to patients, to allow them to develop an understanding of the progression of disease, a patient's response to treatment, or the diagnosis and management of complications. A second area for future research entails addressing the effect of duty hour limits for surgical programs, as some studies suggest a negative impact in the areas of education and quality of patient care,⁴² and recent research suggests benefit in reverting to slightly more liberal standards.⁴³ Finally, studies often use historical comparisons that assume that other attributes of the learning environment have remained stable. However, patient acuity, intensity of service, new technology and, particularly, the widespread use of the electronic health record and resulting documentation needs for residents, have changed how residents spend their time, with only a minority of time spent in direct patient care when compared to time and motion studies a decade or more ago.⁴⁴

Conclusion

Duty hour limits are supported by the scientific literature, particularly limits on weekly hours and reducing the frequency of overnight call. However, many outcomes of interest are underrepresented in the empirical literature. Scientific rigor and established methods have required researchers to restrict their studies to a narrow subset of the large number of variables of interest. These methods are challenged in properly addressing the multiple, often competing, attributes of the learning environment that collectively affect patient safety, resident learning, and resident well-being. Empirically derived limits must be examined for implementation and fit with context and stakeholder buy-in; all limits will benefit from a broader examination of their implementation that allows for a concurrent assessment of several outcomes of interest. Residents may play an important role in this. In addition to being affected, they are an energetic and insightful group, who may see the benefits and drawbacks of approaches in ways their program and institutional leaders and faculty may be blinded to.