Comparison of SAPS II, MPM II24 and SAPS in intensive care [Confronto fra SAPS II, MPM II24 e SAPS in terapia intensiva]

OBJECTIVE: To compare the performance of the new SAPS II, new MPM2 and SAPS in a cohort of patients admitted to our polyvalent ICU. METHODS: Design: the ability of the SAPS II scoring system to predict the probability of hospital mortality was assessing calibration and discrimination (ROC curve) measures obtained using published coefficients and within relevant subgroups using formal statistic assessment (goodness of fit). Patients: from May 1997 to May 1998, 420 consecutive patients over 18 years old. RESULTS: When the parameters based on the standard model were applied, the SAPS II discrimination (area under ROC curve) was = 0.889 and calibration (chi square test) of SAPS II was = 4.448 with p = 0.879; MPM2 chi 2 = 0.9385, p = 0.402 and SAPS chi 2 = 27.089, p = 0.0001. The performance of SAPS II model was very good. Worst predictive accuracy was achieved in trauma and elective surgery patients. CONCLUSIONS: SAPS II model gave good results in terms of calibration and discrimination. SAPS II has better accuracy then SAPS and MPM2. Concerning the performance of models, large differences were apparent in relevant subgroups: trauma and sepsis patients. Moreover the choice of adequate statistic method to compare intensive care populations appeared to need more research.