Do weekends or evenings matter in a pediatric intensive care unit?

OBJECTIVE: To assess what independent influence, if any, weekend or evening admission to a pediatric intensive care unit (PICU) staffed 24 hrs/day, 7 days/wk by in-house, board-certified pediatric intensivists might have on mortality.

DESIGN AND PATIENTS: A retrospective study of 5,968 consecutive admissions to the PICU from August 1996 to December 2003 for patients aged 0 days to 21 yrs.

SETTING: A single, 14-bed, multidisciplinary PICU at an academic medical center.

MEASUREMENTS: Standardized mortality ratios of observed-to-predicted mortality were derived with their corresponding p values. Multivariate logistic regression was used to test the independent effect of weekend admission, weekend discharge/death, and evening PICU admission on mortality for the entire sample and, separately, for only emergency admissions, controlling for other significant predictor variables or interaction terms.

RESULTS: Overall, crude mortality was significantly higher on the weekend (weekday, 2.2%; weekend, 5.0% [p = .0000]) and in the evening (day, 2.1%; evening, 3.8% [p = .0004]). Assessing the entire sample using multivariate logistic regression, neither weekend admission (p = .146), weekend discharge/death (p = .348), nor evening PICU admission (p = .711) showed a significant relationship with mortality controlling for other significant factors. Limiting the scope to the emergency admissions subset, neither weekend admission (p = .135), weekend discharge/death (p = .278), nor evening PICU admission (p = .867) were significant predictors of mortality. Weekend and evening admissions differed in important ways from weekday and daytime admissions, making simple comparisons of crude mortality rates inappropriate. Weekend and evening admissions were more likely to be emergency, nonoperative patients; have a lower Pediatric Risk of Mortality III score but have a higher overall predicted mortality risk; and differ in the distributions of patients by primary diagnosis.

CONCLUSIONS: Using multivariate logistic regression to control for important clinical differences, neither weekend admission, weekend discharge/death, nor evening admission had a significant independent effect on mortality risk in the entire sample or for the emergency patient subset. Our findings are consistent with previous work demonstrating the benefit of intensive care units staffed 24 hrs/day, 7-days/wk by in-house, board-certified intensivists.