Objective: Comparative assessment of performance of pediatric intensive care.
Design: Open, prospective multicenter study.
Setting: All pediatric intensive care units (n = 10; six tertiary and four nontertiary) in the Netherlands.
Patients: 1063 consecutive unselected admissions, < or = 18 yrs old, during a 4-month period.
Interventions: None.
Measurements and main results: Effectiveness was defined as the ratio of observed to expected (Pediatric Risk of Mortality-score-derived) mortality. Efficiency was determined by two objective criteria: mortality risk of > 1%, or administration of at least one intensive care unit (ICU)-dependent therapy. In the total population, observed and expected mortality rates were similar. Between hospitals, crude mortality showed wide variations (mean 7.1%, range 1% to 10%). However, in each center, observed and expected mortality rates were similar (mean ratio of observed/expected mortality 0.99, range 0.8 to 1.5). In tertiary care centers, severity of illness-corrected mortality rates in high-risk patients were less than in a United States reference population. Paradoxically, in low-risk tertiary care patients, the observed mortality rate was higher than expected. The relatively high mortality rate in this group is probably the result of the large number of low-risk tertiary care patients suffering from severe, incurable chronic disease. The average number of efficient ICU days was 72%, although large fluctuations between units were found (range 22% to 95%), suggesting that in several centers efficiency rates might be improved by a better selection of high-risk patients requiring ICU-dependent therapies.
Conclusions: Differences in mortality rates among pediatric ICUs were largely explained by differences in severity of illness. High efficiency rates combined with adequate effectiveness were found in several centers, indicating that admission and discharge decisions might be improved in less efficient centers.