DESEMPEÑO DE ESCALAS PRONOSTICAS DE MORTALIDAD PEDIÁTRICA EN ALTITUD: ANÁLISIS COMPARATIVO E IMPLICANCIAS EN LA PRÁCTICA CLÍNICA DE TERAPIA INTENSIVA
Date
Journal Title
Journal ISSN
Volume Title
Publisher
Rev. Méd. La Paz
Abstract
Objetivo: Comparar el desempeño de las escalas pronósticas PRISM, PIM-2 y PELOD en una Unidad de Cuidados Intensivos Pediátricos (UCIP) situada a 3600 metros sobre el nivel del mar, evaluando su capacidad para predecir la mortalidad en este entorno de gran altitud. Métodos: Estudio observacional, analítico y prospectivo realizado en la UCIP del Hospital de Especialidades Materno Infantil en La Paz, Bolivia. Se incluyeron 201 pacientes pediátricos entre 1 mes y 18 años, admitidos entre noviembre de 2022 y diciembre de 2023. Se calcularon los puntajes PRISM, PIM-2 y PELOD para cada paciente. La capacidad discriminativa se evaluó mediante curvas ROC y área bajo la curva (AUC), mientras que la calibración se analizó con la prueba de Hosmer-Lemeshow. Resultados: La mortalidad observada fue del 4.5% (IC 95%: 2.2-6.8%) y la incidencia de falla multiorgánica del 27% (IC 95%: 20.9-33.1%). PRISM mostró la mejor capacidad discriminativa (AUC 92%), seguido por PIM-2 (AUC 88.1%) y PELOD (AUC 86%). PRISM y PIM-2 presentaron buena calibración (p=0.49 y p=0.29 respectivamente), mientras que PELOD mostró una calibración deficiente (p<0.001). Conclusiones: En este entorno de gran altitud, PRISM demostró ser la escala más efectiva para predecir la mortalidad en la UCIP, seguida de cerca por PIM-2. PELOD, aunque con buena capacidad discriminativa, presentó una calibración inadecuada. Estos hallazgos son cruciales para optimizar la gestión de recursos y mejorar la calidad de atención en UCIPs ubicadas en regiones de gran altitud.
Aim: To compare the performance of the PRISM, PIM-2, and PELOD prognostic scales in a Pediatric Intensive Care Unit (PICU) located 3600 meters above sea level and evaluate their ability to predict mortality in this high-altitude environment. Study Design: Observational, analytical, and prospective study conducted in the PICU of the Hospital de Especialidades Materno Infantil in La Paz, Bolivia. We included 201 pediatric patients between 1 month and 18 years, admitted between November 2022 and December 2023. PRISM, PIM-2, and PELOD scores were calculated for each patient. Discriminative ability was assessed by ROC curves and area under the curve (AUC), while calibration was analyzed with the Hosmer-Lemeshow test. Results: The observed mortality was 4.5% (95% CI: 2.2-6.8%) and the incidence of multiorgan failure was 27% (95% CI: 20.9-33.1%). PRISM showed the best discriminative ability (AUC 92%), followed by PIM-2 (AUC 88.1%) and PELOD (AUC 86%). PRISM and PIM-2 showed good calibration (p=0.49 and p=0.29 respectively), while PELOD showed poor calibration (p<0.001). Conclusions: In this high altitude setting, PRISM proved to be the most effective scale for predicting PICU mortality, closely followed by PIM-2. PELOD, although with good discriminative ability, presented inadequate calibration. These findings are crucial for optimizing resource management and improving the quality of care in PICUs located in high-altitude regions.
Aim: To compare the performance of the PRISM, PIM-2, and PELOD prognostic scales in a Pediatric Intensive Care Unit (PICU) located 3600 meters above sea level and evaluate their ability to predict mortality in this high-altitude environment. Study Design: Observational, analytical, and prospective study conducted in the PICU of the Hospital de Especialidades Materno Infantil in La Paz, Bolivia. We included 201 pediatric patients between 1 month and 18 years, admitted between November 2022 and December 2023. PRISM, PIM-2, and PELOD scores were calculated for each patient. Discriminative ability was assessed by ROC curves and area under the curve (AUC), while calibration was analyzed with the Hosmer-Lemeshow test. Results: The observed mortality was 4.5% (95% CI: 2.2-6.8%) and the incidence of multiorgan failure was 27% (95% CI: 20.9-33.1%). PRISM showed the best discriminative ability (AUC 92%), followed by PIM-2 (AUC 88.1%) and PELOD (AUC 86%). PRISM and PIM-2 showed good calibration (p=0.49 and p=0.29 respectively), while PELOD showed poor calibration (p<0.001). Conclusions: In this high altitude setting, PRISM proved to be the most effective scale for predicting PICU mortality, closely followed by PIM-2. PELOD, although with good discriminative ability, presented inadequate calibration. These findings are crucial for optimizing resource management and improving the quality of care in PICUs located in high-altitude regions.
Description
Vol. 31, No. 2