Objective: To assess the performance of pediatric intensive care by an international standard.
Design: Prospective, cohort study.
Setting: Nine-bed multidisciplinary pediatric intensive care unit (ICU) within a 174-bed, tertiary care children's hospital.
Patients: Consecutive, unselected patients (n = 612) during a 16-month period.
Interventions: None.
Measurements and main results: Collected data included age, major diagnostic category, primary clinical specialty, severity of illness (daily Pediatric Risk of Mortality [PRISM] score), administration of ICU-dependent therapy (daily), and vital status at discharge. The PRISM score allowed the calculation of overall and daily mortality risk. After testing the applicability of the predictive model, effectiveness was determined by comparing severity of illness-based predicted mortality rate with vital status at discharge. Efficiency was defined by two criteria: a) the administration of at least one ICU-dependent therapy; or b) a mortality risk of > 1%. Five-hundred ninety-three patients were included (19 survivors were excluded because of incomplete data). Mean age was 55.0 months (median 24). Mean length of stay was 4.4 days (median 2). Overall ICU mortality rate was 8.4%. The overall performance of the PRISM score-based predictive model was found to be good (goodness-of-fit test chi 2[5] = 5.49; p = .35; area under receiver operating characteristic curve 0.92). Subgroup analysis showed that the best model performance was in nonoperative patients. Decreased performance was found in operative patients. In cardiovascular patients, the mortality rate was higher. In other surgical patients, the mortality rate was lower than expected. Of 593 patients, 489 (82.5%) admissions were efficient, as were 2,393 (76.5%) of 3,130 patient days in the ICU. The ranking of daily efficiency according to clinical specialty was as follows: postoperative cardiovascular surgical patients (86.3%); nonsurgical patients (76.6%); and other postoperative patients (49.6%).
Conclusions: In our setting, effectiveness and efficiency of pediatric intensive care appeared to be validly determined using explicit criteria (mortality risk, administration of ICU-dependent therapy). Overall effectiveness met the standard set forth in an American study; validity in stratified analysis of diagnostic subgroups remains to be further established. Efficiency showed marked, specialty-related differences. The low efficiency in other (noncardiovascular) surgical patients was probably caused by the recovery function of the ICU. A more general application of these criteria might be considered in modifying admission and discharge policy, as well as in quality control.