Sleep and Alertness Among Interns in Intensive Care Compared to General Medicine Rotations: A Secondary Analysis of the iCOMPARE Trial

High-quality sleep of sufficient duration is essential for optimal daytime functioning.¹  Recommended sleep duration is 7 to 9 hours per night; however, shift workers, including medical residents, regularly fail to achieve sufficient sleep, which may impair neurobehavioral function and attention.^2–12  Despite long hours with often inconsistent schedules, residents are expected to maintain a high level of alertness in order to make critical clinical decisions and maintain patient safety.^2,9,13,14 

While medical residents' sleep is recognized as habitually insufficient, few studies have investigated the impact of different hospital rotations on sleep.¹⁵  Inpatient rotations have differing workloads, call schedules, and patient acuity that may influence residents' sleep and subsequent alertness. Given that first-year residents (interns) have the unique challenge of adjusting to life as residents, are often scheduled for more hospital-based rotations than upper-level residents, and are nearly always called first, interns may be at particular risk for insufficient sleep.^3,15–19  Understanding the effect different rotations have on sleep would allow for implementation of targeted sleep-promoting strategies on rotations where interns are at highest risk for sleep deprivation, which may in turn protect neurobehavioral function.

We aimed to examine differences in objectively measured sleep duration and alertness among internal medicine resident interns during inpatient intensive care unit (ICU) compared to inpatient general medicine (GM) rotations.

This was a secondary analysis of the iCOMPARE trial that, in academic year 2015–2016, randomized 63 US internal medicine residency programs to either standard work hour (16-hour shift limits) or flexible (no shift length limit) policies. Within that trial, 12 medium-to-large programs were selected to assess sleep and alertness (6 programs from each arm).^20–23  Interns were observed for 2 weeks during GM, cardiology, or ICU rotations.

Sleep/wake was continually assessed using wrist actigraphy, which has been used to study sleep in medical residents.²  Each morning (generally between 6 and 9 am) interns completed a brief (5-part) self-report survey on a trial-issued smartphone, which included the type of shift they were working that day (day, night, continuing extended overnight, finishing extended overnight, day off, other), a sleep log (self-reported sleep periods during past 24 hours, used to aid in determination of sleep timing from actigraphy), and a rating of sleep quality (on a 5-point scale from 1= bad to 5 = good). They also reported any excessive sleepiness during a past period of 24 hours (options were presented from midnight to midnight in 6-hour intervals, and interns were asked to select all applicable intervals where they experienced excessive sleepiness, or to select no intervals), and current level of sleepiness using the Karolinska Sleepiness Scale²⁴  (KSS; a 9-point scale from 1 = extremely alert to 9 = extremely sleepy–fighting sleep). A 3-minute Psychomotor Vigilance Test (PVT-B),^4,6  a measure of the stability of vigilant attention that is sensitive to the neurobehavioral effects of acute and chronic sleep loss with some validity evidence for use in medical residents,²  was then completed on the issued smartphone. Full study methods, including copies of survey tools, smartphone interface, and participant selection details can be found elsewhere.^20,23 

For this secondary analysis, only interns in GM (wards, GM) or ICU (medical, cardiovascular, or combined medical/cardiac ICU) rotations were included. The primary outcome was mean sleep duration per 24 hours from actigraphy recordings. Secondary outcomes were mean sleepiness (KSS) and alertness (number of PVT-B lapses where reaction time was ≥355 msec, with higher number of lapses indicating lower levels of alertness).

For each outcome, data were averaged across the study period (thirteen 24-hour periods). Separate linear mixed-effect models with random program intercepts were used to determine associations between each outcome (sleep duration, sleepiness, and alertness) by rotation, adjusting for age, sex, and work hour policy followed (standard 16-hour or no shift length limit). To evaluate differences in each outcome during workdays compared to days off, outcomes were also stratified by shift type (work shift or day off). Statistical analysis was performed using Stata version 14 (StataCorp LP, College Station, TX). All analyses were 2-sided, with a significance level of α < .05. This study was approved by the institutional review board at the University of Pennsylvania.

Of a potential 398 interns from the iCOMPARE trial, 12 were excluded for being on cardiology rotations, and 386 were included (n = 261 [67.6%] GM, and n = 125 [32.4%] ICU). Mean age ±SD was 27.9±2.1 years, with 194 (50.3%) men (Table 1). Mean sleep duration ± standard error (SE) was 7.0±0.1 hours, and 6.8±0.1 hours for GM and ICU, respectively (P = .09; 95% CI -0.02, 0.33 hours; Table 2). ICU interns had a higher percentage of days with sleep duration <6h (27.6% vs 23.4%, respectively). Mean frequency of days with self-report of excessive sleepiness was higher for ICU from 12am–6am (ICU: 2.6; GM: 1.7; P < .001; 95% CI -1.41, -0.43) and 6am–12pm (ICU: 2.6; GM: 1.9; P = .013; 95% CI -1.30, -0.16). GM interns reported more days with no excessive sleepiness (GM: 5.3; ICU: 3.7; P < .001; 95%CI 0.82, 2.46). Mean (±SE) KSS was 4.8±0.1 for both rotations. Mean (±SE) number of PVT-B lapses were 5.5±0.5 and 5.7±0.7 for GM and ICU, respectively (P = .83; 95% CI -1.48, 1.18 lapses). When stratified by shift type (day shift, night shift, extended overnight shift [>16-hour shift], day off, or other), ICU interns slept less going into an extended overnight and reported a higher proportion of extended overnight shifts compared to GM (see online supplementary data).

This is one of few studies that have compared objective sleep and alertness between different intern rotations. In this secondary analysis from 12 internal medicine residency programs, sleep duration measured by actigraphy over a 2-week period did not differ between GM compared to ICU rotations. Despite ICU interns having a slight deficit of 12 minutes per day on average, and having more extended overnight shifts, there were no differences in sleepiness or alertness between rotations.

ICU interns reported being more excessively sleepy in morning hours (12am to 12pm) and had a higher percentage of days with insufficient sleep (< 6 hours). This is likely because ICU interns had a higher proportion of extended overnight shifts compared to GM interns, resulting in less opportunity for sleep and more opportunity to report excessive sleepiness early in the morning. Regardless, ICU intern alertness remained similar to GM interns, suggesting that they were able to overcome the slight reduction in sleep duration.

Regardless of rotation, interns showed similar patterns of sleep duration, sleepiness, and alertness (number of PVT-B lapses), except ICU interns beginning an overnight shift (likely because ICU interns had a higher proportion of overnight shifts overall). Interns finishing an overnight shift had the least amount of sleep (∼5 hours) and lowest alertness (highest number of PVT-B lapses) compared to all other shift types. Sleep duration among residents has been shown to vary by shift type and length, with extended overnight shifts having the most prominent effect on neurobehavioral function.^2,10,12,19  Further, extended-duration shifts are known to impair performance and safety.⁹  As residents must maintain a high level of alertness in order to make critical clinical decisions, residency programs should consider promoting strategies that offer opportunities for more sleep, such as protected sleep periods, during extended overnight shifts.^13,25 

This study included internal medicine interns from 12 residency programs. Given the heterogeneity of call schedules and workloads across internal medicine residency programs, findings from this study may not be generalizable across programs. We also collapsed all categories of GM and ICU rotations together. There may be unique demands for different types of rotations that differentially impact sleep and thus were not captured. Finally, data were collected for a single 2-week period from 2015 to 2016. Given the rapidly changing clinical landscape, findings may not reflect the current training environment.

In future research on the effects of resident work hours on sleep, it may be valuable to consider how sleep patterns change over time in longer longitudinal studies, include a wider range of specialties, and include outpatient experiences with variable call patterns.

Although ICU interns experienced more excessive sleepiness in morning hours and a higher percentage of days with insufficient sleep (< 6 hours per day), objectively measured sleep duration and alertness did not differ between ICU and GM rotations.