Effects of Work Hour Reduction on Residents’ Lives:  A Systematic Review FREE

Concern about negative effects of sleep deprivation on residents’ well-being is one of several factors behind the mandatory work hour restrictions instituted by the Accreditation Council for Graduate Medical Education (ACGME) in July 2003.¹ Two groups promoting such restrictions represented residents² and medical students.³ While also concerned about patient safety, these groups specifically called attention to the effect of long work hours on residents’ lives by lobbying Congress and petitioning the Occupational Safety and Health Administration to modify resident work hours.⁴ However, some residency training faculty remain apprehensive about what effect the new work hour restrictions will have on education.⁵

Sleep deprivation affects myriad aspects of residents’ lives. In observational studies, safety issues such as risk for automobile crashes have been related to prolonged work shifts.⁶ Issues of well-being such as mood,^7- 8 stress,^9- 10 and relationship-related stress¹⁰ have been linked to sleep deprivation, as have educational outcomes such as worse performance on simulated tasks^11- 12 and standardized tests.^13- 15 It is possible that residents’ well-being, task performance, and test taking will improve in the short term with fewer work hours, but the long-term impact of these changes is unknown.

We have described a conceptual model that balances the potential benefits and drawbacks of continuity of care with those of interventions to decrease resident work hours.¹⁶ In a systematic review of the impact of decreasing resident work hours on patient safety,¹⁷ we found insufficient evidence to fully inform this issue because the interventions were variable, the study quality suboptimal, and the results conflicting, with some indicators remaining unchanged, some improved, and some worsened after change.

We now consider 2 broader areas of resident education and quality of life. Resident education includes resident performance (eg, operative cases and test scores) and resident satisfaction. Quality of life includes resident health and cognitive function. We performed a systematic review of the literature to describe the impact of interventions (mostly made before the ACGME restrictions) intended to decrease resident work hours on residents’ educational experience and quality of life.

We initially searched the English-language literature for studies about resident work hours for the period 1966 to mid 2002 using MEDLINE, PREMEDLINE, EMBASE, and Current Contents. In April 2005, we updated the MEDLINE and EMBASE searches using the same strategy. We used combinations of terms related to work (workload, work schedule, workschedule tolerance, fatigue, mental fatigue, work hours, personnel staffing and scheduling), sleep (sleep; sleep deprivation; sleep disorders; sleep disorder, circadian rhythm; chronobiology), and residency (education, medical, graduate; internship and residency; nightfloat).

We hand-searched journals that are not indexed for certain years between 1966 and 2002 (Medical Teacher, Medical Education, the British Journal of Medical Education, and Teaching and Learning in Medicine). We examined the reference lists of all articles included in the review as well as those of review articles. We reviewed the abstracts and papers from the Association of American Medical Colleges Research in Medical Education for the years 1989-2004 and the Society of General Internal Medicine national meeting for the years 1989-2005. Whenever possible, we corresponded with the authors of the abstracts to clarify questions. Finally, 3 studies that were too recent to be indexed at the time of the search were identified at the time of their publication.

Inclusion criteria were assessment of a system change designed to counteract the effects of work hours, fatigue, or sleep deprivation; an outcome directly related to residents; and study completion in the United States.

Starting with more than 1200 citations for 1966-2002 and 790 from 2002-2005, we reviewed the abstracts of all relevant articles (Figure). Of those, 528 were abstracts or titles for articles appropriate for detailed review to determine if they met inclusion criteria. Articles eliminated without full review were either not research reports, on the wrong topic, or were editorials. Two authors independently reviewed each article from the first search to ascertain if they met inclusion criteria. Disagreement was resolved by consensus, with input from a third author for 1 article. The articles from the 2002-2005 search were reviewed by 1 author using the same inclusion criteria.

Of the 528 articles reviewed, 86 described relevant interventions. We excluded 21 studies that were not performed in the United States and 11 studies that assessed only patient outcomes. We report on 54 studies that evaluated interventions to decrease resident work hours in the United States and assessed an outcome related to residents’ lives.

The data from each included article were abstracted by 2 of 3 authors. We used a standardized abstraction form that included number of subjects, presence or absence of a control group, study design, outcomes, and methodological concerns. All disagreements were resolved by consensus. We were able to contact authors from 4 of the 6 studies for which we needed clarification.

For study designs other than surveys, we used a modification of the US Preventive Services Task Force (USPSTF)¹⁸ criteria to judge the study quality (eTable 3). The USPSTF uses a hierarchy of overall study design as well as a rating of the internal validity. In that hierarchy, well-done randomized controlled trials constitute the highest tier of evidence, while nonrandomized controlled trials, case-control studies, and cohort studies provide second-tier evidence. All but 1 of the studies we identified provided second-tier evidence by these standards.

For survey studies, we developed criteria (eTable 4) based on several resources.^19- 21 A few studies included both a survey component and a nonsurvey component. In those instances, we rated each component using the appropriate set of criteria. At least 1 author rated the quality of each included study, and all disagreements were resolved by consensus.

Results of the published studies are summarized in the Tables. Comprehensive details about each published study and the abstracts, as well as detailed methodological evaluations, are available in the online eTables.

Twelve studies addressing internal medicine residents were included, with study quality ranging from poor to good (Table 1).

Education. Resident education was assessed entirely by survey and interview studies. Issues studied included the impact of new systems on education, professionalism, and satisfaction.

Residents rated the impact of fatigue on learning as better with a nontraditional float system compared with a traditional every-fourth-night call system (on a 1-7 scale with 1 = strongly agree that learning is impaired by fatigue: 3.4 vs 2.0, respectively).²⁸ In this system, an intern floated with a team for 5 days before moving to a different team.

Professionalism was rated in 2 studies. In one,²³ faculty thought that residents had developed a “shift-work”mentality after a night float was added, although residents disagreed. In the other,³² residents reported maintaining their sense of responsibility for their patients after night float was added. (Night float is a system that has 1 or more residents work shifts at night to cover some part of patient care responsibilities. These could include taking calls about patients already in the hospital and/or admitting new patients. If new patients are admitted by night float residents, another team assumes responsibility in the morning.)

In a survey study that included factor analysis, satisfaction with a factor representing faculty, learning, and environment decreased from 3.81 to 2.85 (on a 1-5 scale with 5 as the most positive score) after changes in the intensive care unit and elective rotations significantly increased call nights for second- and third-year residents.³⁰ This factor included items on feedback, career counseling, faculty teaching, personal support from faculty, and clarity of expectations. In another study,³¹ 81% of program directors agreed that morale was improved when night float systems were used.

Quality of Life. Issues addressed included the impact of new systems on resident health (amount of sleep and mood), number of hours worked, and cognitive function.

In one program,²⁴ the addition of a night float system resulted in residents sleeping 2 hours more on-call when the night float resident was on duty.²⁴ In addition, weekly work hours decreased significantly from 91 to 84. In a different program with a night float, residents did not sleep more, but they did have better sleep efficiency.²⁷ However, no significant difference was detected on tests of attention between the residents who had night float coverage and those who did not.

With a new system that decreased the length of the longest shift from 30 hours to 16 hours and added another intern to intensive care unit rotations, residents slept about 6 hours per week more than with the traditional schedule.²⁶ Weekly work hours decreased significantly from 85 to 65, and nighttime attentional failures as measured by electroencephalogram decreased significantly for those working the intervention schedule compared with the traditional schedule (0.33 vs 0.69 per hour, respectively).

Having night float coverage did not significantly change anxiety, hostility, or depression scores in one program.²⁵ In addition, no difference was demonstrated between residents on night float and residents on a normal schedule when their fine motor skills, attention, concentration, memory, or concept formation were tested. In another study, residents felt less stressed and depressed after a night float and other changes were made (on a 1-5 scale with 5 = strongly agree that the resident felt stressed or depressed: 2.77 before vs 1.87 after changes).²⁹ However, further changes that increased call nights for the second- and third-year residents resulted in increased stress levels in that same program.³⁰

We found 6 studies that evaluated obstetrics/gynecology residents (Table 2). Study quality ranged from poor to moderate.

Education. The issues studied included the impact of interventions on perception of education in general, operative experience, test scores, perception of ambulatory experience, and satisfaction.

One study compared fourth-year residents before and after restructuring teams (including the formation of a night float system) and found that on average each resident performed significantly fewer laparoscopies (41.6 vs 26.4 per year, respectively), primary cesarean deliveries (48 vs 25 per year, respectively), vaginal births after cesarean deliveries (10.1 per year vs 1.8 per year, respectively), and several other procedures.³⁵ The total number of cesarean deliveries and sterilization procedures per resident per year significantly increased (from 48 to 78.2 and from 15.6 to 37.8, respectively).³⁵ When resident procedure volume was divided by institutional volume, residents performed a significantly smaller percentage of the total abdominal hysterectomies, hysteroscopies, primary cesarean deliveries, and vaginal births after cesarean deliveries after the intervention; sterilizations remained increased.

In a program that added a physician assistant, midwives, and a night float, the investigators demonstrated no change in the number of major procedures performed by the resident service before and after the changes.³⁸ Perceived education was rated as improved in that program. However, a nonstatistically significant decline in in-service test scores was demonstrated.

Another program instituted a night team and sent residents home post-call to decrease work hours, and a slight upward trend in in-service scores was seen after those changes.³⁶ However, significance testing was not performed.

Two cross-sectional survey studies assessed perceived operative experience and satisfaction after interventions to decrease hours. In the first study,³⁷ 72% of residents reported unchanged surgical experience after changes that included increased number of faculty, physician assistant and midwife coverage, and the addition of a night float system. Eight-six percent of residents rated their satisfaction as improved, and 47% and 53% rated their obstetrical experience as improved or unchanged, respectively. However, 35% and 58% rated their ambulatory experience as unchanged or worse, respectively, after the interventions.

In the second study,⁴⁰ a survey of program directors who had implemented night float systems found that the directors of most New York programs rated surgical and ambulatory experience between impaired and unchanged, while non–New York program directors thought that both improved. Although all program directors rated resident satisfaction as improved, the directors from non–New York programs rated resident satisfaction as significantly greater than did those from New York programs. The authors attributed the more negative perceptions held by the New York programs to the rigid rules of Code 405³⁴ to which the New York programs had to comply.

Quality of Life. No change in resident stress was noted in a program after the addition of a night rotation,³⁹ but neither was there a change in perceived resident support after the change. After the intervention, however, residents perceived less peer resentment of their own pregnancies.³⁹

We found a total of 7 studies that evaluated pediatrics training, with quality ranging from poor to moderate (Table 3)

Education. Impact on education was addressed in these studies. One pediatrics program instituted an evening continuity clinic once monthly for residents so that post-call afternoon clinics could be avoided.⁴⁴ Eighty-six percent of residents felt that they could learn better and 82% thought that they could display more empathy during the evening clinics as compared with post-call clinics. They also felt that the evening clinics contributed more to their primary care education than would other educational activities that could be scheduled instead of cancelled post-call clinics, such as visits to homeless shelters.

A statewide survey that was conducted before implementation of Code 405 in New York, and repeated for several years, showed that by the third year after the regulations, residents’ rating of time available for reading was not statistically significantly different than before implementation (15% agreed that they had enough time for reading in the year before Code 405 was enacted compared with 26%, 31%, and 22% in the next 3 years).⁴⁶ Residents also thought that lack of sleep interfered with their clinical judgment less in the first 3 years following enactment of Code 405 compared with previous years; however, by the last year of the survey, this item was rated no differently than in the year prior to Code 405 enactment. The authors suggested that the lack of difference in later years could have been due to reduced enforcement of the rules over time.

Quality of Life. Issues addressed by these studies included resident health (amount of sleep and mood) and cognitive function. Two studies were conducted in the same program to assess the impact of a night float rotation.^41- 42 When residents who worked as the night float tracked their sleep using diaries, they reported less sleep when acting as the night float than when on a typical ward rotation.⁴² In contrast, a study of the same residents using a survey rather than diaries demonstrated no difference in the amount of sleep between the 2 conditions.⁴¹ More residents reported that falling asleep was difficult and that there were more sleep interruptions while on night float than during regular rotations.⁴¹ Residents expressed more depressed feelings while on night float⁴¹ and scored lower on vigor and higher on fatigue in the Profile of Mood States.⁴² Cognitive function tests of alertness and attention were no different when performed in night float residents vs residents on regular rotations.⁴²

In another program with night float coverage 5 nights per week, sleep for senior residents increased when night float coverage was available compared with nights without coverage.⁴⁵ The amount of sleep for interns did not change regardless of night float coverage.

Twenty-five studies assessed the impact of interventions to reduce work hours on surgical residents. Study quality ranged from poor to moderate (Table 4).

Education. Issues addressed in these studies included perceived impact on education in general and impact on operative experience. In a statewide survey of surgical residents to evaluate the impact of Code 405 in New York, 22% of residents thought that training had improved and 35% thought it had been harmed.⁷¹ Residents reported that time to read increased in a program that incorporated 12-hour night shifts and home call to decrease hours,⁶⁸ as did residents in a program that changed to day teams and night teams.⁵⁷

Three programs demonstrated an increase in operative experience after interventions.^48,60,65 In the highest-quality study of this group,⁴⁸ a program that added a night float and weekend cross-coverage reported significantly more cases for the chief residents as a group (an average of 1015 cases per year for the 3 years before and 1116 for the 3 years after the changes).⁴⁸ Another program added health technicians to the surgical service, resulting in each resident spending an average of 9.8 hours per week in the operating room after the intervention compared with 3.3 hours per week before.⁶⁵

On the other hand, 2 studies suggested that operative experience may not change with interventions to reduce resident hours. In the first, residents left the hospital on post-call mornings, and the number of operative cases did not change before vs after the changes.⁶⁶ The findings of a national study of surgical programs for the 2 years prior to the ACGME work hour rules⁶³ suggested that the number of operative cases for chief residents did not vary depending on whether the program had in place an 80-hour per week model, or whether it had experimented with changes to comply with the impending ACGME work hours rules.

Finally, 3 studies suggested that operative cases may actually decrease with certain interventions.^51,57,70 In the highest-quality study of this group,⁵⁷ after the formation of day and night teams, fourth-year residents performed significantly fewer cases (270 per year vs 207 per year). The numbers for second- and third-year residents also showed a downward trend, while first- and fifth-year resident experience remained the same.

Quality of Life. Issues assessed included quality of life in general, resident health (sleep, well-being, and mood), and number of hours worked. In the 5 studies that measured resident work hours after interventions, they were decreased.^{51- 52,57- 58,65} In another study, residents reported a perceived decrease in hours in the hospital and workload after physician assistants were added to the surgical team.⁶⁹

Eighty-five percent of residents agreed that quality of life improved after night float was added and cross-coverage was increased in 1 program.⁴⁸ In a statewide survey,⁷¹ 66% of New York residents thought that quality of life improved after Code 405. In a program that surveyed residents before and after addition of night shifts and home call,⁶⁸ residents reported improved quality of life (on a 1-5 scale with 5 = strongly agree: 3.1 vs 3.6, respectively), adequate rest more frequently (2.3 vs 3.0, respectively), and more time for family (2.1 vs 2.9, respectively) and for socializing (2.4 vs 3.1, respectively).

Residents’ spouses reported that residents were more available for family events after separate night and day teams were instituted.⁵⁷ Residents in that program also reported decreased fatigue after the interventions. In a study designed to increase protected time for sleep or reading at night, a “communication notebook” for the nurses to record nonurgent issues for the interns was instituted.⁵⁴ This resulted in more perceived sleep on call, a fewer number of pages at night (86 pages/week before vs 53 after), and more nights without any pages between 1 AM and 5 AM (25% before vs 100% after).

Family medicine, psychiatry, and radiology had only 4 studies between them, and all were survey studies of fair to moderate quality (Table 5). In a cross-sectional survey of family medicine program directors,⁷³ 88% reported increased resident satisfaction with night float, 90% thought that resident alertness improved, and 76% reported that residents’ personal lives improved.

In a survey study of psychiatry residents on the impact of night float, there was a mean rating of improvement in their well-being (on a 1-5 scale with 1 = strongly agree: 2.10), education (1.71), and clinical duties (1.86).⁷⁴

Two studies assessed radiology residents. In a multi-institutional survey of chief residents, 85% reported improved call experience and 90% reported better educational experience after the ACGME work hour rules were put into effect.⁷⁵ The number of in-hospital call sessions increased from 104 in 2000 to 114 in 2004, while the number of home calls increased from 57 to 63.

In the second study, in a single program that instituted a night float system, 45% of residents reported that their clinical judgment was better during the night float than on a 24-hour long call shift; the others did not think that their judgment was affected by night float.⁷⁶ However, residents also reported that it took them about 2 days to acclimate to beginning night float and about 2 days to return to a normal schedule afterward.

Summaries of the relevant abstracts from the national Society of General Internal Medicine meetings and the Research in Medical Education of the Association of American Medical Colleges are shown in eTable 5.^77- 91 Their findings corroborate those of the published studies.

Limiting resident work hours may impact faculty members, nurses, medical students, and patients, as well as the residents themselves.¹⁶ Just as the safety of patients may not always be improved by interventions intended to decrease resident work hours,¹⁷ our review indicates that there may not be uniform benefits for residents from these changes.

Policy changes such as these have both intended and unintended consequences. The ACGME Statement of Justification for the work hour limits states that the intended consequences for residents include improvements in their education and well-being.⁹² However, potential unintended consequences include inadequate development of professionalism⁹³ and worse patient-physician communication skills,⁹⁴ and these have not been explored.

Home call was part of 4 programs’ interventions to decrease work hours. No specific assessment of the impact on the residents taking home call was performed, and these are hours that are not counted toward the 80-hour per week limit. The impact of home call on sleep and personal life may be substantial. Some of the studies that assessed the experience of residents assigned to night float found some disadvantages. Residents reported more difficulty sleeping⁴¹ and mixed feelings about the educational value of the night float rotation.³³ The effect of home call and night float should be accurately assessed in programs that use them.

Another potential unintended consequence of reducing resident hours is a decrease in experience, and some educators argue that reduced hours for residents may necessitate more years of training.⁹⁵ Actual experience was studied in surgery and obstetrics/gynecology programs, but not in the other specialties. Some of these studies demonstrated reduced experience,^35,57 while some demonstrated increased operative experience.⁴⁸ The findings of these studies raise the possibility that some interventions intended to decrease work hours may have varying impact at different levels of training. A benefit of reduced fatigue may be worthwhile if operative experience evens out over the course of training. However, it may be problematic if residents have decreased experience in their final year before leaving training, as was found in one study.³⁵

Resident experience could be thought of as a process-of-care measurement because its real importance is in how it affects patient care.⁹⁶ Research in this area should focus on the link between resident experience and patient outcomes. For example, tracking outcomes of recent graduates in their first 5 years after training, such as surgical morbidity and mortality, using specific operative experience during residency as a predictor, would identify the amount of experience needed during residency to achieve clinical competence.

The debate on resident work hours focuses on the balance between protecting residents and caring for patients. In several studies, we found that residents thought that fatigue had less impact on care after their work hours were decreased. As suggested by the link between resident burnout and self-perceived quality of care,⁹⁷ improved quality of life for residents may ultimately result in better patient care, but this will need to be empirically established.

This systematic review should be interpreted within the context of several limitations. First, the experience of many individual institutions is likely unpublished. Publication bias could influence the results of our review, especially if the unpublished studies were more likely to have shown no significant effect of an intervention.⁹⁸ However, unpublished abstracts showed results similar to the published studies. Second, the methodology of the included studies differed and was often suboptimal. Only 1 study used a randomized controlled design,²⁶ and most of the other studies did not measure or adjust for possible confounders. Many of the studies relied exclusively on surveys, and some were done at a single time point, asking the residents to reflect on their experiences before and after interventions. Asking participants to remember events in the past can lead to problems with validity, especially when events occurred remotely or when respondents must recall many events.⁹⁹ Third, the interventions evaluated were divergent, making it challenging to draw firm conclusions about the relative merits and drawbacks of individual systems for reducing resident work hours.

These limitations suggest steps to improve the quality of research in this area. First, programs instituting substantial changes should rigorously study these changes and report them, whether or not they demonstrate an impact. Second, interventions should be studied at multiple institutions to improve the generalizability of the findings. Third, whenever possible, intervention studies should be conducted as randomized controlled trials to reduce the effect of confounding. Survey studies can provide valuable insight into the impact of an intervention, but they must also be performed rigorously. Investigators should use standardized methods for developing and validating the survey, maximizing the response rate, and properly reporting response rate and results.^21,99

Objective outcomes rather than perceived outcomes should be measured. Several studies in this review surveyed residents about perceived amount of sleep, even though more reliable methods are available and were used in other studies; these approaches may produce different results.^41- 42 While actual clinical experience was measured in some studies of surgery and obstetrics/gynecology residents, we found no studies that measured actual experience in internal medicine, pediatrics, family medicine, or psychiatry. An important outcome to consider is not only the number of patients with a given condition (eg, acute myocardial infarction) that are cared for by a resident during his or her training, but also the number of such patients followed up from admission to discharge. This continuity experience is particularly vulnerable to use of a night float system.

A valid, reliable survey tool for measuring resident quality of life needs to be developed if there is to be accurate measurement of the impact of programmatic and policy-level changes on residents. It is also necessary to study the link between resident quality of life, burnout, and patient care.¹⁰⁰

Despite these limitations, this study provides a comprehensive review of the literature and our current understanding of the effects that attempts at work hour limitations have had on the quality of life and educational experiences of residents. It suggests that residents’ quality of life may improve with the ACGME limitations on resident work hours. The short-term effects of interventions to decrease resident work hours on education are mixed, however, and longer-term studies will be needed to understand the impact of the changes on resident experience, knowledge acquisition, and the link between these and patient outcomes. We have outlined several areas of research that could make important contributions to our understanding of how best to nurture residents while also providing excellent patient care both during training and when these residents become practicing physicians.