Risk-adjusted outcome evaluation in a multidisciplinary intensive care unit

OBJECTIVE: To evaluate the outcome of a multidisciplinary Intensive Care Unit (ICU) by applying the Acute Physiology and Chronic Health Evaluation (APACHE II) and Paediatric Index of Mortality (PIM)--version-2 scoring systems. SUBJECTS AND METHODS: Two-hundred and seventeen patients admitted consecutively to the ICU during a period of one year were included for prospective data collection. Data recorded were demographics, diagnoses at admission, APACHE II score for adults and PIM -2 score for children, the duration of ICU stay and hospital outcome. Predicted mortality and standardized mortality ratios were calculated. Calibration and discriminant function of the systems were done by Hosmer-Lemeshow analysis and Receiver Operating Characteristic (ROC) curves. RESULTS: In adults, the mean APACHE II score was 14.3 +/- 8.3; in survivors, it was 8.7 +/- 5.9 (SD) when compared to 21.2 +/- 5.9 (SD) in non-survivors (p < 0.0001). The predicted mortality in adults by APACHE II was 16.5%, the observed mortality being 19.8%. The predicted mortality by the PIM-2 in children was 34.8% with the observed mortality rate being 30%. The overall mean duration of stay was 5.2 +/- 7.5 days. The goodness-of-fit for APACHE II and PIM-2 systems were fair (HL chi-square, p = 0.71, 0.69, respectively). The area under the ROC curve was 0.88 for APACHE II and 0.62 for PIM-2. CONCLUSION: Evaluation of risk-adjusted outcome in multidisciplinary ICUs is challenging because of the need to apply more than one prognostic scoring system.