During our study of an alternative inpatient resident service schedule, we implemented duty periods of 16 hours or less without increasing resident staffing or altering admission capacity. However, our study demonstrated 3 areas of concern with shortened resident shifts: residents felt less prepared to handle cross-cover issues, education was compromised because of the inability to attend teaching sessions, and the number of times residents had less than 10 hours off between shifts increased significantly.
Previous studies that have examined quality-of-care measures after implementation of the 2003 ACGME duty hour requirements10,11
demonstrated no benefit or harm. Our results, although limited by the short time frame, are consistent with no apparent clinical harm or benefit to patients. The residents in the intervention month indicated that they did not think sign-out on cross-cover patients was adequate. Our sign-out system engages many features that have been recommended elsewhere, including an electronic document automatically populated with clinical data.8
However, the designated sign-out time often coincided with new admissions, which disrupted communication about cross-cover of patients.
Residents thought that the educational environment was negatively affected by the reduced duty periods. The reported effect of duty hours on the educational environment is mixed.12
In the current study, residents thought they missed educational opportunities when they were assigned to the night shift and thus were not able to attend lectures. Overall, when residents evaluated their educational experience, the difference between the 2 groups was not statistically significant. This may reflect the perception that the educational sessions were beneficial when they could be attended, but the intervention schedule limited the opportunity to attend. The timing of required lectures and the need to maintain a curriculum will be important factors as programs devise new schedules related to changes in shift length.
Although the number of hours worked per shift was reduced, the number of “10-hour” violations increased, suggesting that we may have substituted one duty hour requirement violation for a different violation. Overall, resident satisfaction with the intervention month was reduced. Informal discussion with the residents after the completion of the month further elucidated concerns about providing cross-cover care, as well as the length of the admitting day. The busiest time of the admitting day was from 4 pm until 9 pm, which resulted in the daytime resident staying after 9 pm to complete patient admissions.
As with any observational study, there are limits to the interpretation of our data. First, it was conducted during 3 months at a single institution and included a small number of residents (6 in the intervention, 12 in comparator months). Team members may not have had time to become accustomed to the schedule. Second, although intervention residents were less likely to attend the core educational conferences, these individual residents had attended fewer conferences during their intern year compared with the control residents (P
=.008). Third, the residents did not attend continuity clinic during the intervention month, which was allowed through an Educational Innovations Project exemption. Although the ACGME requirements for internal medicine residency programs allow this schedule,13
it may not be feasible for other programs. Fourth, a more formal method of work hour recording, such as a card-swipe system, may have given more accurate work hour results compared to weekly resident surveys and may have shed more light on the exact timing of the 10-hour violations that occurred. Fifth, the 16-hour duty period pilot consisted of PGY-2 residents, whereas the new duty hour guidelines note that these duty hour periods are required only for PGY-1 residents; if this pilot had consisted of PGY-1 residents, some of the results may have been different, but this model could be used with PGY-1 residents if in-house supervision is present. Finally, the residents were surveyed at the end of each week regarding hours spent working, sleeping, researching, and reading, which leads to potential recall bias.