The Accreditation Council for Graduate Medical Education (ACGME) has addressed resident duty hours for the past 20 years. A significant revision to residency standards became effective in July 2003, which included the 80-hour work-week limit and a 24-hour limit to continuous duty, with an additional allotment of 6 hours for continuity/transfer of care and didactic activities (1). In December 2008, the Institute of Medicine (IOM) provided recommendations adjusting the ACGME duty-hour standards to reduce resident fatigue-related errors and improve patient safety and medical training (2). In response, an ACGME Task Force was formed and made the following changes, effective July 1, 2011: duty periods of PGY-1 residents must not exceed 16 hours in duration; duty periods of PGY-2 residents-and-above may be scheduled to a maximum of 24 hours of continuous duty in the hospital, and they are strongly encouraged to use strategic napping after 16 hours of continuous duty; intermediate-level residents must have at least 14 hours free of duty after 24 hours of in-house duty; and residents must not be scheduled for more than 6 consecutive nights of night-float (3).
Recent studies have attempted to determine what effect, if any, the implementation of the 2003 ACGME duty-hour standard had on residents and patient safety (4, 5). The degree to which these changes have affected psychiatry residency training, however, is largely unknown, because the studies of sleep-loss in residents tend to be limited to nonpsychiatric residents (6–8). One study that bridged the ACGME changes sampled residents between 2002 and 2005 and found that psychiatry residents had less burnout than family-medicine residents on both “emotional exhaustion” and “depersonalization” scales (9). One study, from the U.K., found that admissions and assessments made up at least 85% of calls after midnight (10).
Out of concern for residents' general well-being, and with new ACGME changes, it is a worthy goal for administrators to understand why psychiatry residents may not sleep while on call. Furthermore, information gleaned from medical and surgical residencies may not apply to psychiatry programs. We aim in this study to review the workload inventory of on-call psychiatry residents and to evaluate which activities are associated with reductions in on-call sleep.
A prospective cohort study was conducted at the Sheppard and Enoch Pratt Hospital in Baltimore, MD, a 231-bed psychiatric hospital, from July 1, 2008 through June 30, 2009. The hospital comprises multiple inpatient units for adult, child, adolescent, and geriatric patients. It does not have an emergency room or any medical floors.
The 20 participating residents were previously-assigned rotating residents, who volunteered to participate in this study. Participating residents took calls at this hospital, one of the inpatient sites for the University of Maryland/Sheppard Pratt residency in psychiatry. While on call, each resident provided a 24-hour period of coverage for the entire hospital. On weekdays, the on-call resident was the only physician in the hospital after 5 P.M. and began accepting admissions after that time until 8 A.M. On weekends, the resident was responsible for admissions and acute evaluations over the 24-hour period. The on-call frequency was typically every 5th night. Residents rotated through this hospital in 3-month blocks.
New patients referred from local emergency rooms were screened by admissions staff and then admitted to the inpatient units by the on-call resident. On-call residents were limited to four admissions; “moonlighters” managed additional admissions, but were not responsible for subsequent pages about newly-admitted patients. Participating residents were given on-call logs to record their activities. Information recorded included total hours of sleep (i.e., including naps and interrupted sleep), time of page, unit number, and a free text space where they recorded the reason for the page. This information was expected to be completed after each page during the on-call shift. A post-call signout was routinely conducted to discuss overnight issues. On-call logs were de-identified as resident and patient names were redacted and logs were given numeric codes. Institutional Review Board approval for exemption was obtained for this research.
Data were entered into a database using SPSS Version 16. On-call activities were categorized into seclusion/restraint evaluations, general-medical evaluations, and general psychiatric evaluations, based on the request from nursing staff. If, at any point, a patient was put into seclusion/restraints, this was categorized as such, regardless of the reason or inciting event. Seclusion/restraint evaluations included requests for both initial assessments and reassessments. Otherwise, pages were categorized on the basis of the nature of the request from nursing staff. For example, a page to evaluate a patient who fell was categorized as medical, whereas a request to evaluate a patient with anxiety was categorized as psychiatric. These categories were chosen from our previous experience of working with on-call residents. Stepwise linear regression was performed to evaluate how hours of sleep were affected by other on-call activities.
From the 2008–2009 academic year, 298 on-call log sheets were received (81.64% response rate) from 20 participating residents. Of all the collected log-sheets, only two were not included because of incompleteness. The mean amount of sleep per night gotten by each resident was 3.52 hours (standard deviation [SD]: 1.72). The mean number of pages was 17.21 per night (SD: 6.60). The mean number of admissions was 2.96 (range: 0–4; SD: 1.27). On-call activities were categorized into seclusion/restraints, general-medical complaints, or general psychiatric complaints. The majority of pages that residents received were related to seclusion/restraints (N=1,666; 45%) and general-medical evaluations (N=1,637; 44%). General psychiatric evaluations only comprised 11% (N=405) of on-call pages received.
Linear regression was utilized to explore how categorized data explained the variance in residents' sleep. Variables included number of admissions, number of pages, seclusion/restraints, psychiatric evaluations, and medical evaluations. The variance in the hours of sleep was best explained by the number of pages received and the number of admissions (R=0.432; p<0.001), as shown in Figure 1. Interestingly, the reason for the page (i.e., type of evaluation) did not affect residents' sleep.
FIGURE 1.Hours of Sleep by Number of Pages and Number of Admissions
This study was conducted at a large, free-standing psychiatric hospital, with overnight and weekend coverage provided by one psychiatric resident. Participating residents provided detailed information about on-call activities. The response rate was excellent. Some on-call logs were not submitted because of clerical error or resident disorganization. The majority of residents made great efforts to submit legible logs that detailed their shifts.
This study showed that the total number of pages received and the number of admissions best explain the variance in the hours of sleep. This conclusion seems logical for two reasons: First, the more pages the resident received while on call, presumably, the busier he or she was managing staff and patient issues. Second, admitting new patients to the hospital is a time-intensive activity.
During post-call signouts, residents offered their own explanations as to why the previous night of call was difficult. Participating residents attributed a night of less sleep to pages from certain “troubled” units or to the types of page they received (e.g., frequent renewal of restraints, medical emergencies). Our study, however, showed that the residents' subjective experiences did not correlate with the results. We found that the number of pages for seclusion/restraints, psychiatric evaluations, and medical evaluations did not explain the variance in the hours of sleep. Interventions aimed at increasing on-call sleep should be directed at training staff to triage overnight pages and limiting after-hours admissions.
The mean amount of sleep received while on call did not consistently increase for each 3-month rotation during the study period. We were surprised by this because we expected that, as a resident progressed during the rotation, he or she would learn the skills to become more efficient while on call and would subsequently get more sleep. The amount of sleep received during the first month of the academic year, however, was the lowest of the study period; and, remarkably, the amount of sleep received during the last month of the academic year was below the mean for the entire study period. This suggests that residency programs cannot rely on experience alone to translate into more efficient practices.
Training interventions should monitor the resident's ability to cultivate habits so as to receive more sleep, such as performing more focused evaluations, triaging nursing issues, and managing a busy pager. Post-call analysis by supervisors, chief residents, and peers can help a resident become more efficient and track his or her progress during the rotation.
We found that on-call residents managed a broad range of acute and non-acute medical issues. Unexpectedly, the on-call residents received four times as many pages for medical issues as for psychiatric issues. This emphasizes the need for psychiatry residents to receive broad medical training, even when working in settings that are “strictly” managing mental illness.
This study has several limitations. First, this study initially began as a quality-improvement project, in which on-call logs were created and distributed to residents to assist their organization and help identify whether there were specific units or staff members who repeatedly called the resident. The on-call log required free text; the legibility and content of each resident's log varied significantly. Second, the on-call sleep duration was subjectively measured by sample participants. The authors assumed that residents were sleeping during their on-call shifts, if possible. We did not account for leisurely activities that would limit getting sleep. Difficulty falling asleep was also not addressed, which may be common after stressful events while on call. Third, data were self-reported; disgruntled residents might have provided log-sheets misrepresenting the amount of sleep they got. A final limitation was that the study was conducted in a single residency program at one institution and was done over the span of only 1 year. The institution, however, was a large psychiatric hospital with a broad collection of adult, geriatric, and child/adolescent inpatient units.
In response to the 2011 ACGME standards, many residency programs may need to restructure their on-call requirements. In particular, training directors will be required to find methods that increase residents' on-call sleep, decrease consecutive shifts, or include a block of protected on-call sleep. It is not known, however, what effect these duty-hour changes will have on residency education and training. Our study shows that training directors should be cautious about implementing changes without having data that describe the on-call activities of their residents. Although on-call duties may vary among different hospitals and psychiatry residency programs, training directors will share a collective pressure to comply with the new ACGME requirements and answer the question, “What can be done to improve the residents' on-call experience and to increase their amount of sleep received while on call?” It is our hope that this study will join that dialogue in improving residency experiences.
Preliminary results from this paper were presented at a poster session at the Association of Academic Psychiatry Annual Meeting, Washington, DC, October 2, 2009.