Background: To assess the accuracy of SAPS II on the Italian population and to perform a customization of the model.
Design: observational prospective study.
Patients: 24 participating centers. 6794 patients out of 9185 enrolled in the cohort study.
Measurements: the performance of SAPS II was assessed with calibration and discrimination. In the case of standard model not fitting the data, new logistic regression equation has been calculated using Archidia database.
Results: SAPS II showed a good discrimination, but a bad calibration. The new logit has been calculated on the population examined in 1995 and successively tested on the population collected during 1996. In both samples discrimination and calibration results were good.
Conclusions: Our customized model reaches a high standard in calibration and discrimination. We suggest this model for future application in Italian ICU.