Application of the APACHE III prognostic system in Brazilian intensive care units: a prospective multicenter study

To compare patients and their outcomes at ten Brazilian intensive care units (ICUs) with those reported from the United States.

Prospective multicenter inception cohort study.

Ten Brazilian adult medical-surgical ICUs.

1734 consecutive adult ICU admissions.

We used demographic, clinical and physiologic information and the APACHE III prognostic system to predict risk of hospital death for 1734 ICU admissions. We then divided the observed by the predicted hospital death rate to calculate standardized mortality ratios (SMRs) for patient groups and each ICU. Hospital mortality for Brazilian patients (34%) was double that found in the United States (17%, p < 0.01). Discrimination of survivors from non-survivors using APACHE III was good (area under a receiver operating characteristic curve = 0.82), but the predicted risk of death was significantly (p < 0.0001) lower than observed outcome (SMR = 1.67). Three of the ten Brazilian ICUs, however, had SMRs of 1.01 to 1.1 and no significant difference between observed and predicted outcomes; the remaining seven ICUs had significantly higher SMRs, ranging from 1.50 to 2.30.

The APACHE III prognostic system was a good discriminator of hospital mortality for ICU admissions at 10 Brazilian ICUs. There was substantial and significant variation, however, in SMRs among the Brazilian ICUs, which suggests that further evaluations of international differences in intensive care using a common risk assessment system should be performed and factors associated with variations in risk-adjusted mortality scrutinized.