Abstract

EULMESEKIAN, Pablo et al. Validación de dos modelos de predicción de mortalidad, PRISM y PIM2, en una Unidad de Cuidados Intensivos Pediátricos. Arch. argent. pediatr. [online]. 2006, vol.104, n.5, pp.387-392. ISSN 0325-0075.

Introduction. Pediatric Index of Mortality 2 (PIM2) and Pediatric Risk of Mortality (PRISM) are models of mortality probability prediction for Pediatric Intensive Care Units (PICU). We aimed to validate both scores in the multidisciplinary PICU of Hospital Italiano de Buenos Aires. Population, material and methods. We prospectively collected all the information to obtain both scores of mortality prediction for each consecutive patient admitted to the PICU from January 1^st to December 31^st 2004. We used original equations for both models. Calibration was assessed with Hosmer-Lemeshow (HL) goodness-of-fit tests and Standardized Mortality Ratio (SMR) for the whole population. Discrimination was assessed with the area under Receiver Operating Characteristic (ROC) curve. Results. 765 patients were included over the 12 months period. 53.6% were male. Median age was 4.1 years (1 month-20 years). Median length of stay (LOS) was 2 days (1-154 days). 22 patients died in PICU (2.9%). PIM2 estimated mortality was 3.3%; SMR= 0.87 (95% CI 0.54-1.21); HL x₂= 15.16 (p= 0.01) and ROC curve of 0.92 (95% CI 0.90-0.94). PRISM estimated mortality was 5.5%; SMR= 0.50 (95% CI 0.28 -0.73); HL x₂= 22 (p= 0.0005) and ROC curve of 0.94 (95% CI 0.92-0.96). Conclusions. PIM2 had a better performance than PRISM in the heterogeneous patient population from our PICU. Although both model discriminated very well, PRISM overestimated mortality remarkably.

Keywords : PIM2; PRISM; Scores; Mortality; Pediatric intensive care.

        · abstract in Spanish     · text in Spanish     · Spanish ( pdf )