Effects of the 2011 Duty Hour Restrictions on Resident Education and Learning From Patient Admissions
Objective: In July 2011, new duty hour limits for resident physicians were instituted to address concerns about the effects of sleep deprivation on patient care and trainee experience. We sought to evaluate potential educational impacts of these duty hour changes with regard to learning and frequency of attending interactions during patient admissions.
Methods: Forty-nine residents on general pediatric teams participated in a prospective observational cohort study. Intervention residents (n = 23) worked a shift-based schedule compliant with new requirements. Control residents (n = 26) were on call every fourth night and compliant with 2003 work hour limits. Faculty members were present 16 hours daily. Resident surveys assessed learning from admissions (frequency of attending interaction and perceived learning during admissions). Data were analyzed with generalized linear mixed models to account for multiple responses from each resident.
Results: Intervention interns and seniors were less likely to present admissions to faculty during morning rounds, but there were no differences between intervention and control groups in percentage of admissions discussed with faculty at any time. Perceived learning from admissions was not different between the 2 groups.
Conclusions: Faculty-resident interaction decreased during morning rounds; however, overall attending contact did not, suggesting inpatient teaching approaches must adapt to meet learners’ needs throughout the workday.
Resident education has undergone significant changes over the past decade with the goal of improving learning, patient care, and safety of patients and residents. With concerns over continued lack of resident sleep after initial duty hour regulations in 2003,1 the Institute of Medicine (IOM) proposed new duty hour recommendations.2 Subsequently the Accreditation Council for Graduate Medical Education (ACGME) instituted new duty hour requirements for resident physicians in July 2011. The 80-hour workweek was maintained, but shift length was further limited to 28 hours for seniors and 16 hours for interns.3
Although the IOM’s focus was to enhance sleep, supervision, and safety, the impact of the resulting ACGME regulations on resident education is unclear. Despite implementation, there remains a significant lack of evidence on the best practices for resident work schedules.4 A systematic review of trials of schedules eliminating shifts exceeding 16 hours found improvements in resident quality of life, patient safety, and quality of care, but the effects on medical education were mixed.5 Implementation of the latest duty hour restrictions may have adversely affected resident perception of the educational experience.6,7
Overall, the evidence is sparse regarding effects of the regulations on specific educational experiences, including patient admissions. We sought to examine the effect of a schedule compliant with 2011 duty hour limits on resident education in terms of reported learning from admissions and the timing of direct faculty interaction. We chose faculty interactions because both the IOM and ACGME emphasize faculty supervision as fundamental for education and safety.2,8
We conducted a prospective observational cohort study at Cincinnati Children’s Hospital Medical Center, a large academic pediatric tertiary-care facility. Pediatric hospitalist faculty, working in a shift model, are present in-house to supervise general pediatric resident teams and staff admissions between 7:00 am and midnight daily. Attending physicians examine all new patients and discuss admissions with the senior resident during this time. Between midnight and 7:00 am, attending physicians are available by phone for questions that arise on any new or existing patients. The attending staffing model remained constant during the intervention and control periods. Family-centered bedside rounds occur every morning on the general pediatric teams.9 In addition, residents participate in 2 daily didactic conferences during the week, 1 in the morning and 1 at noon.
Residents assigned to 1-month rotations on general inpatient teams in January, February, and March 2011 were eligible to participate. Residents on teams in January were the intervention residents following schedules compliant with 2011 duty hour limits. Residents on teams in February and March served as the control residents and followed standard 2003 duty limit scheduling procedures. The majority of resident rotation assignments were made before the decision to trial an intervention. However, staffing the intervention schedule required more interns and seniors than a typical month under the 2003 duty limits. For sufficient senior resident staffing, 6 residents originally scheduled on elective or night float volunteered to act as senior residents for 2 to 4 weeks during the intervention month. This study was deemed exempt by the institutional review board.
Control Team Structure, Schedule, and Admission Responsibilities
During the 2 traditional schedule months, control residents (n = 26) were on 1 of 4 teams. Each team was composed of 1 senior resident and 2 or 3 interns. All control seniors and interns took traditional 30-hour calls every fourth night. Each team admitted patients every other day until 11:00 pm. A night float team admitted overnight patients (from 11 pm to 7 am) and handed the patients back to the primary team in the morning.
Intervention team structure, schedule, and admission responsibilities
During the intervention schedule month, the intervention group residents (n = 23) were on 1 of 4 teams. Each team was composed of 2 senior residents and 3 or 4 interns. Intern shifts did not exceed 14 hours and had at least 9 hours off between shifts. The interns had an overnight shift every sixth night (see Appendix). Senior residents took call every fourth night with 25-hour shifts. Intervention teams took admissions on a shift schedule without a night float system.
We assessed 2 essential aspects of admission experience: timing of attending interaction and perceived learning. Residents were asked to record data for each admission up to their first 5 admissions per shift. To improve consistency in data collection and facilitate clearer comparisons, residents were instructed to include admissions with diagnoses expected to be high in volume including pneumonia, bronchiolitis, cellulitis, gastroenteritis, and neonates with fever. Additionally, to include additional high-yield educational admissions, residents were instructed to include children with an uncertain diagnosis and children with technology-dependence (eg, gastrostomy tube or need for respiratory support). To document resident-faculty interaction, residents recorded both if and when they discussed admissions directly with faculty. To measure the perception of knowledge gained, residents rated agreement with the statement: “I gained new knowledge, clinical decision making skills, and/or clinical judgment from this admission” on a 5-point Likert scale. The knowledge scale was dichotomized before analysis: 1, 2, or 3 (strongly disagree, disagree, or neutral) were compared with 4 or 5 (agree or strongly agree).
We analyzed differences between the intervention and control group with generalized linear mixed models to account for multiple admissions from each subject. For interns, models included a random team effect to account for learning within team groups.
Because workload affects educational experience,10 we obtained census information from hospital billing data. Because admission data were collected weekly to maintain patient anonymity, we used median daily team census by week to adjust analyses of admission outcomes.
We also conducted a post hoc power calculation based on the outcome response observed in the control group using methodology11–13 to account for the repeated measures from residents. We also completed descriptive analyses of learning from admissions by interns (intervention and control combined) with varied attending interactions. All analyses were performed by using SAS statistical software (version 9.2, Cary, NC).
Data from interns were obtained for 216 admissions during the intervention period and 148 admissions during the control period. Data from senior residents were obtained for 161 admissions during the intervention period and 79 admissions during the control period. The types of admissions were similar for the intervention and control groups (Table 1).
Control and intervention senior residents reported similar rates of discussing admissions with faculty at any time. Intervention interns reported discussing admissions at any time with faculty less frequently than controls (75% vs 91% of admissions), but this difference did not reach statistical significance (adjusted odds ratio [AOR] 0.24; 95% confidence interval [CI]: 0.03–2.04; Table 2). We determined post hoc that we were powered to detect an effect difference of ≥ 23% when comparing intervention to control interns’ discussion of admission with faculty at any time.
Intervention residents, both interns and seniors, presented patients on morning rounds less frequently compared with control residents (interns AOR 0.07; 95% CI: 0.01–0.70; seniors AOR 0.06; 95% CI: 0.003–0.99). Neither intervention interns nor seniors statistically differed in terms of discussing admissions with faculty at the time of admission compared with controls (Table 2).
Interns in the intervention and control group had similar rates of reporting new knowledge gained from admissions (68% vs 58%). Intervention seniors had higher rates of reporting having gained new knowledge from admissions (64% vs 38%). Although this effect for seniors was significant in the unadjusted analysis (unadjusted odds ratio 3.03; 95% CI: 1.42–6.47), it was confounded by census and not significant in the adjusted model (AOR 1.32; 95% CI: 0.25–6.91; Table 2). Interns reported having gained new knowledge from admissions 68% of the time when they had any attending contact; without an attending contact, they reported having gained new knowledge 48% of the time. When interns presented an admission on rounds, they reported having gained new knowledge 69% of the time. When interns did not present on rounds, they reported having gained new knowledge from the admission 57% of the time.
The median daily team census by the week was lower in the intervention period (median = 9; interquartile range: 7.5–10) than during the control period (18; interquartile range: 16–21) (Figure 1).
The intervention schedule decreased the number of new admissions residents discuss with their attending physicians during morning rounds. However, residents reported that overall, they discussed new admissions with an attending at a similar rate in the intervention and control groups in our system where attending physicians are present in the hospital 16 hours per day. No difference was observed in residents’ perceived learning from new admissions in the new work hour system compared with the previous work hour system.
Learning from patient care is a core aspect of resident education. The number of pediatric admissions is correlated with improved pediatric certifying examination scores.14 There is concern that under duty hours restrictions, residents will spend less time in direct care of patients.15 Therefore, our objective findings that interns and seniors had large and significant reductions in the frequency they discussed new admissions on morning bedside rounds are concerning. Morning rounds traditionally serve as an important teaching opportunity with multidisciplinary interaction among families, residents, medical students, bedside nurses, and the attending physician. Interns who discussed admissions on rounds or with the attending at any time more frequently reported gaining new knowledge from the admission; however, this was a post hoc analysis, and we did not perform statistical testing on these differences. The decrease in the number of patients presented during morning rounds was not unexpected given the structural changes to resident schedules.
However, we found no overall difference in the occurrence of at least 1 discussion with faculty about each patient for both interns and seniors. Given the slight (although not significant) increase in the percentage of intervention interns discussing admissions with the attending at the time of admission (36% intervention vs 19% control), it is possible that interns responded to their lack of presence on rounds by seeking out attending input. In our institution, faculty members are present in the hospital 16 hours every day, and their presence was consistent in both the intervention and control periods. A significant decrease in overall faculty contact in the shift model might occur if faculty were present only during daytime hours. Additionally, there may have been real and clinically significant differences that existed in our trial, but we lacked the power to find it. In the era of new duty hours, faculty should be cognizant that in-person discussion of new admissions at times other than rounds is an important learning opportunity given the less frequent opportunities for learning on morning rounds. Furthermore, including the types of teaching more often done during family-centered rounds (ie, communication skills, family education by resident, contingency planning)9 may be needed in the discussion of patients at other times.
Our finding that intervention residents reported similar rates of new knowledge gained from admissions is encouraging. Despite concerns over work compression in a shift model,16,17 it is possible that residents in the intervention group were less fatigued and thus better able to consolidate their learning from admissions.
This study should be considered in the context of several limitations. First, there were unpredictably large differences in census between the intervention and control months, likely from the onset of viral respiratory season. We adjusted our models to account for census differences. Second, this is a trial of 1 intervention schedule over 1 month at 1 institution. Ideally, our findings would be verified in larger interventions with alternative schedules over longer periods of time with less variability in census. However, significant challenges exist in performing future duty limit research4; this study was done in the weeks before mandatory implementation of the new duty limits. Third, there were large differences in the number of admissions recorded in the intervention and control. This difference may be related to timing of admissions and the types of admissions. Given the higher census, the interns in the control group likely had more admissions beyond the first 5 collected. The effect of the lower collection rate from the control group is unclear. Busy control residents may have been more motivated to report on extremely positive or extremely negative educational experiences. Fourth, the subjective educational outcomes we assessed were perceptions of learning and may not capture actual learning. Furthermore, the metrics used to capture learning were novel and not formally validated. However the count and timing of attending and resident interaction is an objective measure of duty hour effects. Fifth, changes to resident schedules had to occur to adequately staff the intervention schedule, and neither faculty nor residents were randomized to intervention or control months, which introduces the possibility of self-selection bias for intervention residents. However, the majority of schedules were made before the decision to trial the intervention, minimizing the risk of selection bias.
Our schedule compliance with the 2011 ACGME duty hour requirements led to significant structural changes in residents’ educational opportunities. Despite the significant decrease in intervention resident presentations on rounds, we did not find differences in the frequency of having at least some contact with faculty or in perceived learning from admissions. The evening presence of faculty (who were present in both the intervention and control periods) may have blunted the impact of the rounds changes. Regardless, the unavoidable shift in resident learning away from morning rounds, a time when a significant amount of inpatient learning has traditionally occurred, should encourage resident educators to capitalize on learning opportunities throughout the workday.
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: Dr Auger received salary support from the Robert Wood Johnson Foundation through the Clinical Scholars program at the University of Michigan.
POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.
- Accreditation Council for Graduate Medical Education
- adjusted odds ratio
- confidence interval
- Institute of Medicine
- Landrigan CP,
- Fahrenkopf AM,
- Lewin D,
- et al
- Ulmer C,
- Wolman DM,
- Johns MME
- 3.↵Accreditation Council for Graduate Medical Education. Common Program Requirements. 2010. Available at: http://www.acgme.org/acWebsite/dutyHours/dh_dutyhoursCommonPR07012007.pdf. Accessed January 23, 2012
- Whalen T,
- Wendel G
- Muething SE,
- Kotagal UR,
- Schoettker PJ,
- Gonzalez del Rey J,
- DeWitt TG
- Brown H,
- Prescott R
- Liu H,
- Wu T
- Diggle P,
- Liang K,
- Zeger S
- Chase LH,
- Highbaugh-Battle AP,
- Buchter S
- Copyright © 2014 by the American Academy of Pediatrics