Introduction: Most discharges from Canadian ICUs occur in the daytime, with night-time discharges occurring primarily because of demand for an ICU bed or lack of nursing staff, which can result in premature transfer to the ward. The literature on the effect of discharge time and mortality is conflicting.

Purpose: To determine the impact of night-time ICU discharge on patient outcome.

Methods: Multi-centre, retrospective observational cohort study using a prospectively collected dataset of 79,090 consecutive admissions from 31 Canadian community and teaching hospitals. Discharges were categorized into daytime (07:00 − 20:59) and night-time (21:00 − 06:59). Admissions were excluded if the patients were (1) ≤ 16 years of age (392), (2) admitted following cardiac surgery (6,641) or due to a shortage of available beds elsewhere (457), (3) readmitted to the ICU within the same hospital stay (3,632), or (4) transferred to another acute care facility (7,724).

Results: 62,056 patients were discharged to the ward following the initial ICU admission. Of the 47,062 discharges eligible for analyses, 10.1% were discharged at night. The crude hospital mortality rates were 9.0% and 11.8% for those discharged during the day and night, respectively. The unadjusted odds of death for patients discharged from ICU at night was 1.35 (95% CI 1.23, 1.49). After adjusting for illness severity, source, case-mix, age, gender, and hospital size, the mortality risk was increased by 1.22-fold (95% CI 1.10, 1.36) for night-time discharges. Multivariate regression analyses also revealed that (1) night-time discharges have a significantly shorter ICU length of stay (LOS) than daytime discharges (p<0.001), (2) hospital LOS was similar for daytime and night-time discharges that survived hospital stay (p=0.335), and (3) patients discharged at night that did not survive hospital stay had a significantly shorter hospital LOS (p=0.002).

Conclusions: Patients discharged from the ICU at night have an increased risk of mortality. To understand why night-time discharge compromises patient outcomes researchers have to prospectively examine process, system and staffing factors that may differ during the daytime and night-time subsequently contributing to the disparity we have seen. These data could also be used to develop models that identify predictors of poor outcome for patients discharged at night and provide guidance to ICU physicians to discharge ²low-risk² patients during times of intense pressure for ICU beds.