Secondary Logo

Journal Logo

Innovation Reports

Achieving Balance on the Inpatient Internal Medicine Wards

A Performance Improvement Project to Restructure Resident Work Hours at a Tertiary Care Center

Cohee, Brian M., MD; Hartzell, Joshua D., MD; Shimeall, William T., MD, MPH

Author Information
doi: 10.1097/ACM.0000000000000213
  • Free



Graduate medical education has become increasingly regulated. The Accreditation Council for Graduate Medical Education (ACGME) implemented reforms in 2003 and 2011 that focused primarily on regulating resident work hours. Compared with these reforms, the Next Accreditation System, implemented in 2013, has a broader focus, using a model of continuous program evaluation to effect changes in measured outcomes.1 This approach reflects a growing under standing that “one size does not fit all” in resident education.

Evidence demonstrates that sleep depri vation negatively affects performance, but evidence also suggests that additional transitions of care increase errors.2,3 These effects often compete, and the balance that optimizes patient safety, resident education, and resident well-being has not been adequately studied and may be dependent on the quality of handoffs, degree and timing of work intensity, opportunity for sleep, information resources, and quality of supervision. Because many of these factors are specific to the training environment, the rotation structures that optimize this balance may vary by location.

Differences in these factors might also explain why changes in inpatient structure have occasionally yielded unanticipated results. In a report on the implementation of a night float system at Children’s Hospital in Boston, Chua and colleagues4 found that total sleep time decreased under the night float system, and ultimately their program reverted back to a traditional call system. Similarly, Desai and colleagues5 reported testing two systems—both compliant with the 2011 ACGME Common Program Requirements—on the inpatient internal medicine wards at Johns Hopkins. Compared with the 2003 model, they found that both systems resulted in increased sleep while on call, but decreased educational opportunities, continuity, and perceived quality of care, resulting in cessation of both systems. These studies highlight the need for shared experience on how to best design ward rotations in order to optimally balance the many competing interests of graduate medical education.


In this Innovation Report, we describe the methods used in a performance improvement project to redesign the call schedule of the inpatient internal medicine wards at Walter Reed National Military Medical Center (WRNMMC), a tertiary care center. Recognizing that local factors strongly influence the optimal call structure for a given setting, we conducted a brief period of data collection to understand the alignment of resident resources with institutional demands. The data collection and intervention process, described below, demonstrates how a program can find a balance amongst the many competing factors that influence quality of care, and resident education and well-being. Publication of this project was exempted from review by the institutional review board for WRNMMC (project 386614-1).

Preintervention structure

In the night float system at WRNMMC prior to the intervention, one of six teams admitted daily from 6:15 AM until 5:00 PM, with sign-out to the night team at 5:00 PM, and the night team admitted until 6:15 AM the following morning. The night team consisted of two interns who managed acute issues (i.e., cross-cover), a second-year resident who completed admissions, and a third-year resident who supervised the interns, answered consultations, and assisted the second-year resident during high admission periods. Patients admitted by the night team were transferred in the morning to a short-call team on the third day of their call cycle.

Anecdotally, residents reported that on-call teams routinely violated the ACGME Common Program Requirement that time out of the hospital between shifts exceed 8 hours, with a goal of more than 10 hours. Because the majority of admissions occurred during the night shift, the night float admitted more patients than the day teams, and these patients had rushed initial evaluations and poor continuity of care. The emergency room experienced delays in resident evaluations, leading to prolonged disposition times.

Baseline data

To clarify these concerns, we used an eight-day period from November 14 to 21, 2011, to establish baseline data. We gave on-call and night float residents a spreadsheet for each call to document the time of admission requests, the location requesting admission, the time of the admission order, and individual work hour compliance of all team members.

Data validated the anecdotal reports of house staff. Ninety-one patients were admitted during the eight-day period. Full data were not available for all patients; Figure 1 shows the timing of initial page for admission for the 83 patients for which the data were available, as well as the breakdown of admissions by night and day teams. We found that an increase in admissions coincided with the time of scheduled handoff to the night float team at 5:00 PM. The time from initial page until admission order was less than 90 minutes for only 7 of 18 (39%) of emergency department patients for which disposition time was recorded, which we thought was due to the clustering of admissions and the temporal association of admissions with sign-out. In 24 of the 42 (57%) emergency department admissions, the resident did not report the time of admission order. In 22 (92%) of those cases, it was the night float resident who did not record the time; these residents verbally reported that this was due to being overwhelmed by the work intensity. Night float teams admitted 48 of the 91 (53%) patients. Violation of the 10-hour goal between duty periods occurred on 55% of all admitting shifts (16 of 29). There were no violations of the 8-hour standard.

Figure 1
Figure 1:
Admissions by time and admitting service before and after intervention. In the preintervention graph, the dotted lines indicate the times of morning and evening sign-out. Admissions occurring before 5:00 PM that were admitted by the night float were typically transfers from outside facilities that arrived after the evening sign-out period. Before the intervention, the night float team admitted a slight majority of the patients. The bar graph demonstrates the late afternoon increase in workload that came around the time of sign out. Post intervention, the night float team admitted a much smaller portion of the patients. An afternoon increase in workload still occurred, though it was shifted earlier than what was seen in the previous data collection.

Intervention description

After reviewing the temporal distribution of admissions (Figure 1, preintervention), we sought to design a new call structure where most patients would be admitted by the teams that would oversee the majority of their care, and where the available manpower matched the anticipated work intensity, while still incorporating factors to mitigate fatigue. We created a long-call system for the resident team leader with a graded transition to night float from 5:00 PM to midnight. The goal was to increase the number of patients admitted by the primary resident, but allow for overlapping concurrent admissions by the night float during the busiest period. This system was implemented on January 16, 2012.

The new system continued to have six teams who admitted during the day, with no change in composition. A night team consisted of one resident and two interns, eliminating one resident from the team, which meant one less resident on inpatient rotations per block. The number of residents present at night stayed the same, however, because an on-call resident remained in the facility. Teams took call every sixth day. The on-call resident arrived at 6:00 AM and admitted until midnight when the night float resident took all further admissions until 6:00 AM. Residents followed the 2011 ACGME regulations, which limit admissions to 10 new patients and 4 additional transfers. If the resident had accepted at least 10 admissions or transfers by the arrival of the night float at 5:00 PM, they would alternate admissions until reaching the ACGME limit or until midnight. The admitting team interns arrived at 6:00 AM and signed out patients to two night interns before departing the hospital by 8:00 PM. Patients admitted by the night float were transferred to a short call team the next morning.

From midnight until 6:00 AM, the on-call resident did not admit new patients or accept transfers, and the night interns were paged initially for patients on their service. The on-call resident carried his or her pager and was available for questions from interns and nurses; however, the time was relatively protected to allow the resident to complete notes, read, develop management plans, and sleep.

After two weeks of this intervention, a second data collection occurred for 12 days and 12 nights, from January 29 through February 9, 2012. Because there were few admissions over a long weekend, we extended the duration of data collection compared with the collection in November.


Postintervention data collection

A total of 91 admissions occurred during the postintervention assessment. Figure 1 shows the timing of admissions, as well as the breakdown by night and day teams. On-call residents and their teams admitted a higher percentage of patients (82%, 75 of 91) than the night float teams (18%, 16 of 91) (P < .01 compared with preintervention). Self-reported adherence to duty hours improved. The percentage of violations of duty hours regulations decreased from 55% to 6.8% of admitting shifts (16 of 29 and 3 of 44 shifts, respectively, P < .01) (Figure 2). For emergency room admissions, the time from initial page until admission order improved: 70% (31 of 44) of admissions occurred in 90 minutes or less postintervention, compared with only 39% (7 of 18 for which data was available) preintervention (P = .02) (Figure 3).

Figure 2
Figure 2:
Pre- and postintervention duty hours compliance. The “10-hour rule” (10hr) refers to the 2011 Accreditation Council for Graduate Medical Education (ACGME) Common Program Requirements’ recommendation that house staff should spend at least 10 hours out of the hospital between shifts. No residents on either model violated the common duty-hours-mandated 8-hour limit between shifts. The “28-hour rule” (28hr) refers to the ACGME common duty hours requirement that residents who are postgraduate year 2 or above can work a total of 24 hours, followed by 4 hours that are dedicated to transitioning care and academics.


After the intervention, we conducted an anonymous survey. We invited all interns, residents, and attendings who had served on ward or night float teams during an eight-week period, including the rotation block in which the second data set was collected as well as the subsequent block. Requests to complete the survey were sent electronically. Of the 65 recipients, 28 responded (43%).

Both attendings and residents felt the new system improved their knowledge of their patients and ability to make management decisions: 58% (11/19) agreed or strongly agreed, 37% (7/19) were neutral, and 5% (1/19) disagreed. Interns (n = 9) were split on whether the changes improved their knowledge of the patients: 3 (33%) agreed or strongly agreed, 3 (33%) were neutral, and 2 (22%) disagreed or strongly disagreed (one did not answer). Ten of 16 (63%) house staff and 4 of 12 (33%) attendings reported improved perceptions of quality of care; all others perceived no change. Respondents at all levels of training reported improved or unchanged ability to comply with duty regulations (10 of 28 [36%]) agreed or strongly agreed, 16 of 28 [57%] were neutral, and 2 did not answer). When asked about the quality of handoffs in care, 11 of the 28 (39%) felt they were improved or much improved, and 1 (3.6%) thought that they had worsened. Twenty-two of the 28 respondents (79%) reported difficulty attending morning report on postcall days. No other interferences with academics were noted.

Next Steps

We report here a method for tailoring the call schedule of an inpatient internal medicine ward to optimize the match between number of admitting physicians and the typical work intensity. This resulted in changing our call structure from a night float system to a resident long-call system with a graded transition to a night team. Specific aspects of our institution shaped the structure we implemented. Our model, therefore, does not represent a universal ideal. More research is needed to understand how structural change affects education and patient safety. However, we feel there is benefit from sharing the strategies taken to quickly and accurately assess and address the demands of an institution.

We ultimately chose a model that reinstated extended shifts but introduced interventions to share the intensity of the busiest period and created a relatively protected period for completing work and sleeping. This improved the continuity of care and trainee perception of being involved in the initial decisions while simultaneously improving compliance with duty hours regulations, without requiring additional ancillary staff.

We postulate that our approach—identifying variation in the intensity of workload and designing a system that responds by increasing the number of admitting physicians—is an effective strategy for optimizing the balance of fatigue and continuity, which influence patient safety and the quality of education.

There are limitations to our approach and outcomes. No data were collected on error rates or patient outcomes. Available evidence suggests that differences in patient outcomes would be small, if present.6 Our sample size would not be adequately powered to assess these end points.

Additionally, our data sets compared different rotation blocks, and thus different attendings, trainees, and patients. These cohorts may have differed in their abilities and preferences. The second data collection also occurred two months later, and house staff may have developed enough in that time to affect the data. Aside from not attending morning conference the day after call, there were no changes in the timing or structure of rounds or academic conferences.

Finally, the overall response to the survey was poor, and more physicians with prior experience with an extended call structure answered the survey. House staff with more experience with an extended call system might view it more favorably, as was demonstrated on a survey before the merger that formed WRNMMC.7

Our performance improvement initiative better aligned resident resources with work demands, using a brief data collection to characterize work intensity. The improved alignment of resources with demand resulted in increased resident continuity of care, faster disposition, increased adherence to work hours, and improved reports of quality of patient care and handoffs. Our results suggest that the impact of schedule design may be influenced as much by institution-specific variables as the inherent interplay of fatigue and handoffs. A brief assessment of work intensity may represent an expeditious method to create a location-specific schedule. Publication of more experiences will increase the ability of programs to select the right systems for their institutions.

Figure 3
Figure 3:
Emergency room disposition times pre- and postintervention: changes in the timing between initial page from the emergency department and the time of the admitting order. Numbers within the blocks represent percentages of the number of patients for which data were available. The numbers within the unavailable data are the percentage of all emergency department dispositions.

Acknowledgments: The authors would like to acknowledge the residents of the internal medicine residency at Walter Reed National Military Medical Center for their support and participation in this initiative.


1. Nasca TJ, Philibert I, Brigham T, Flynn TC. The next GME accreditation system—rationale and benefits. N Engl J Med. 2012;366:1051–1056
2. Lockley SW, Cronin JW, Evans EE, et al.Harvard Work Hours, Health and Safety Group. Effect of reducing interns’ weekly work hours on sleep and attentional failures. N Engl J Med. 2004;351:1829–1837
3. Horwitz LI, Moin T, Krumholz HM, Wang L, Bradley EH. Consequences of inadequate sign-out for patient care. Arch Intern Med. 2008;168:1755–1760
4. Chua KP, Gordon MB, Sectish T, Landrigan CP. Effects of a night-team system on resident sleep and work hours. Pediatrics. 2011;128:1142–1147
5. Desai SV, Feldman L, Brown L, et al. Effect of the 2011 vs 2003 duty hour regulation-compliant models on sleep duration, trainee education, and continuity of patient care among internal medicine house staff: A randomized trial. JAMA Intern Med. 2013;173:649–655
6. Volpp KG, Rosen AK, Rosenbaum PR, et al. Did duty hour reform lead to better outcomes among the highest risk patients? J Gen Intern Med. 2009;24:1149–1155
7. Kelly WF, Cuneo BM, Shimeall WT. Merging cultures and curriculum: An anonymous survey of Army and Navy physicians before the new Walter Reed Bethesda internal medicine residency program. Mil Med. 2012;177:624–626
© 2014 by the Association of American Medical Colleges