#KeyLIMEPodcast 281: ROSTERS, RCTs, & the Complexity of Meded Outcomes

The KeyLIME team are back with new episodes! They return with a look at physician fatigue  and how scheduling affects patient care — the current research leaves things unclear. The researchers decided to delve further into the topic with a multicentre look at the effect of 16-hour shifts vs “extended duration” duty hours for trainees in pediatric ICUs on “serious medical errors”.

Listen to the co-hosts discuss here.


KeyLIME Session 281

Listen to the podcast


New KeyLIME Podcast Episode Image

Landrigan et. al., Effect on Patient Safety of a Resident Physician Schedule without 24-Hour Shifts  N Engl J Med 2020 Jun 25;382(26):2514-2523.


Jason R. Frank (@drjfrank)


Does clinician fatigue harm patients? It’s a no-brainer, right? We should all be modifying schedules for all involved in healthcare to ensure patient safety…Wait, not so fast. The literature doesn’t really support that assertion. Modified duty schedules in various forms have been shown to have positive effects on physician wellness, mixed effects on learning, and (as regular KeyLIMErs know from multiple past episodes!) the evidence of an effect of scheduling on patient safety is super-muddy. The NEJM has in recent years published big studies, such as the FIRST and iCOMPARE trials that showed no benefit to patients. In fact, there is some signal that increasing the number of handoffs and the number of patients per clinician is a sure-fire way to add risk. So far, the lit says make humane schedules to keep your clinicians human, but not necessarily to help their patients…So we need another study, right? (NB. Past episodes on duty hours, fatigue & patient safety can be found in KeyLIME # 206,165, 115, 113, 104, 97, 62, & 22)



Enter Landrigan et al. They are back! They are bigger! They are in the NEJM! This Boston group has produced multiple previous papers relating to modified resident duty hours vs various outcomes. This time they share a multicentre study to look at the effect of 16-hour shifts vs “extended duration” duty hours for trainees in pediatric ICUs on “serious medical errors”, the ROSTERS study.

Key Points on the Methods

This was a cluster-randomized, multicentre, crossover trial comparing 16-hour shiftwork vs usual overnight call schedules for pediatric ICU residents. The study ran from 2013-2017 in 6 US centres, each running for 2 years. Each site was randomized to start on one of the 2 schedules, then crossed over after a year to the other schedule. Participants were 333 PGY2 & above trainees during 410 rotations (172 controls, 188 intervention). 38,821 patient days were included in the trial, with about 3500 patients in each arm.

Data gathering was in real time and extensive. The primary outcome was the number of “serious medical errors” attributed to the trainees, as detected in chart reviewers or by independent MD observers, or voluntarily reported. There were 72,102 hours of observation. Serious medical errors were defined as “a medical error that causes harm”, or could have caused harm, or near misses. Secondary outcomes were serious medical errors by unit, the number of patients/resident (as a measure of workload), sleep, sleepiness, work hours, and neurobehavioural performance on vigilance tasks. There were no corrections for multiple comparisons in the stats.

Key Outcomes

The authors were surprised by the major results…Intervention group residents worked fewer hours per week (61.9 vs 68.4), and reported more sleep (52.9 vs 49.1 hrs/week). The prevalence of having <4hrs of sleep was 25% in the call group and 9% in the intervention. However, there were ~500 more “serious errors” in the intervention group. The rate was significantly greater for the shiftwork intervention (97.1 vs 79.0 / 1000 patient-days; relative risk 1.53). This was also true when looked at unit by unit: 181.3 vs 131.5 tilted to the shiftwork (RR 1.56).  When workload was controlled for, there was no difference in error rate between shiftwork vs call. Finally, while the patients seen in all the ICUs were similar, it was noted how much variability there was between sites.

Key Conclusions

The authors conclude that their shiftwork design paradoxically led to pediatric trainees making more serious errors than those doing call-type schedules. The intervention was associated with fewer work hours, more sleep, and better neurobehavioural performance. However, the number of patients per trainee, and increased handoffs seemed to be factors. The authors speculate that excessive workloads and poor handoffs may be a greater risk to patient safety than fatigue.

Spare Keys – other take home points for clinician educators

  1. The authors deserve kudos for the elaborate time and motion-type methodology they used to obtain data in real time.
  2. However, RCTs in meded are difficult to perform, as behavioural interventions can have many more confounders than other “medical” interventions
  3. In social science interventions, the scenario is inherently complex, so complex methods are needed
  4. Fidelity in meded interventions is a critical factor in interpreting outcomes. Despite the authors’ assurances, there is evidence that the intended effects of this trial may not have been universally applied (eg sleep).
  5. Instead of regulating duty hours, fatigue risk management is the future, and comes late to meded from other industries. However, performance factors are complex, and relate to much more than fatigue.

Access KeyLIME podcast archives here


The views and opinions expressed in this post and podcast episode are those of the host(s) and do not necessarily reflect the official policy or position of The Royal College of Physicians and Surgeons of Canada. For more details on our site disclaimers, please see our ‘About’ page