Objective: To assess the performance of 3 severity of illness scoring systems (Acute Physiology and Chronic Health Evaluation II (APACHE II), Simplified Acute Physiology Score II (SAPS II) and Mortality probability model II on admission (MPM 0) and after 24h of admission (MPM II)) in Egyptian critically ill patients.
Material and method: A prospective cohort study was conducted in a medical geriatric Intensive Care Unit, in a university hospital, Cairo, Egypt. We included 140 elderly patients (60 years and above) Probabilities of hospital death for patients were estimated by applying the 3 severity of illness scoring system and compared with observed outcomes. The overall goodness-of-fit of the three models was assessed.
Results: The hospital death rate was under estimated by the 3 severity of illness scoring system. The difference in estimated mortality rate among survivors and non-survivors was significantly different for the 3 scoring system. Regarding calibration, assessed by the Lemeshow±Hosmer chi-square statistic, showed that the MPM II had the best calibration (p=0.92) and the APACH II (p=0.64) and MPM 0 (p=0.56) had good calibration, while the SAPS II had the worst calibration (p =0.07), that is they all have accepted calibration but with varying degrees of accuracy While discrimination using the ROC( receiver operating characteristic ) the SAPS II, MPM 0 and the MPM II showed good discriminative power as their ROC was 0.78, 0.79, 0.76 respectively, and the best for MPM 0 (ROC=0.79), while the APACHE II had moderate discriminative power as the ROC was o.67, that the 3 models had good to moderate discriminative power
Conclusion: we can conclude that all the 3 scoring system (APACH II, SAPS II, MPM (0, II)) can be used to predict mortality in critically ill elderly patients, and that they had accepted degree of discrimination and calibration in elderly ICU patients. Hospital mortality was higher than predicted for all 3 models, so it is important to note that a patient with low mortality prediction scores can die in his ICU admission period, that is to say low severity scores cannot guarantee against suspected mortality. In order to improve performance of these models may be alternative mortality prediction approaches might be needed to customize the models according to the geriatric Egyptian ICU patients.
