Lidan Zhang1, Huimin Huang, Yucai Cheng, Lingling Xu, Xueqiong Huang, Yuxin Pei, Wen Tang, Zhaoyuan Qin. 1. Pediatric ICU, the First Affiliated Hospital, Sun Yat-Sen University, Guangzhou 510080, Guangdong, China (Zhang LD, Huang HM, Cheng YC, Xu LL, Huang XQ, Pei YX, Tang W); Department of Pediatrics, the First Affiliated Hospital, Sun Yat-Sen University, Guangzhou 510080, Guangdong, China (Qin ZY). Corresponding author: Qin Zhaoyuan, Email: zqqinzhaoyuan@163.com.
Abstract
OBJECTIVE: To assess the performance of pediatric clinical illness score (PCIS), pediatric risk of mortality score III (PRISM III), pediatric logistic organ dysfunction score 2 (PELOD-2), and pediatric multiple organ dysfunction score (P-MODS) in predicting mortality in critically ill pediatric patients. METHODS: The data of critically ill pediatric patients admitted to Pediatric Intensive Care Unit (PICU) of First Affiliated Hospital of Sun Yat-Sen University from August 2012 to May 2017 were retrospectively analyzed. The gender, age, basic diseases, the length of PICU stay were collected. The children were divided into survival group and non-survival group according to the clinical outcome during hospitalization. The variables of PCIS, PRISM III, PELOD-2, and P-MODS were collected and scored. Receiver operating characteristic (ROC) curve was plotted, the efficiency of PCIS, PRISM III, PELOD-2, and P-MODS for predicting death were evaluated by the area under ROC curve (AUC). Hosmer-Lemeshow goodness of fit test was used to evaluate the fitting degree of each scoring system to predict the mortality and the actual mortality. RESULTS: Of 461 critically ill children, 35 children were excluded because of serious data loss, hospital stay not exceeding 24 hours, and death within 8 hours after admission. Finally, a total of 426 pediatric patients were enrolled in this study. 355 pediatric patients were survived, while 71 were not survived during hospitalization, with the mortality of 16.7%. There was no significant difference in gender, age, underlying diseases or length of PICU stay between the two groups. PCIS score in non-survival group was significantly lower than that of survival group [80 (76, 88) vs. 86 (80, 92)], and PRISM III, PELOD-2 and P-MODS scores were significantly increased [PRISM III: 16 (13, 22) vs. 12 (10, 15), PELOD-2: 6 (5, 9) vs. 4 (2, 5), P-MODS: 6 (4, 9) vs. 3 (2, 6), all P < 0.01]. ROC curve analysis showed that the AUCs of PCIS, PRISM III, PELOD-2, and P-MODS for predicting death of critical ill children were 0.649, 0.731, 0.773, and 0.747, respectively. Hosmer-Lemeshow test showed that PCIS predicted the mortality and the actual mortality in the best fitting effect (χ 2 = 7.573, P = 0.476), followed by PELOD-2 and P-MODS (χ12 = 9.551, P1 = 0.145; χ22 = 10.343, P2 = 0.111), while PRISM III had poor fitting effect (χ2 = 43.549, P < 0.001). CONCLUSIONS: PRISM III, PELOD-2 and P-MODS can discriminate between survivors and moribund patients well, and assessing the condition of critically ill pediatric patients with relatively accuracy. PCIS was the best fitting effect in predicting mortality and actual mortality, followed by PELOD-2 and P-MODS, while PRISM III had poor fitting effect.
OBJECTIVE: To assess the performance of pediatric clinical illness score (PCIS), pediatric risk of mortality score III (PRISM III), pediatric logistic organ dysfunction score 2 (PELOD-2), and pediatric multiple organ dysfunction score (P-MODS) in predicting mortality in critically ill pediatric patients. METHODS: The data of critically ill pediatric patients admitted to Pediatric Intensive Care Unit (PICU) of First Affiliated Hospital of Sun Yat-Sen University from August 2012 to May 2017 were retrospectively analyzed. The gender, age, basic diseases, the length of PICU stay were collected. The children were divided into survival group and non-survival group according to the clinical outcome during hospitalization. The variables of PCIS, PRISM III, PELOD-2, and P-MODS were collected and scored. Receiver operating characteristic (ROC) curve was plotted, the efficiency of PCIS, PRISM III, PELOD-2, and P-MODS for predicting death were evaluated by the area under ROC curve (AUC). Hosmer-Lemeshow goodness of fit test was used to evaluate the fitting degree of each scoring system to predict the mortality and the actual mortality. RESULTS: Of 461 critically ill children, 35 children were excluded because of serious data loss, hospital stay not exceeding 24 hours, and death within 8 hours after admission. Finally, a total of 426 pediatric patients were enrolled in this study. 355 pediatric patients were survived, while 71 were not survived during hospitalization, with the mortality of 16.7%. There was no significant difference in gender, age, underlying diseases or length of PICU stay between the two groups. PCIS score in non-survival group was significantly lower than that of survival group [80 (76, 88) vs. 86 (80, 92)], and PRISM III, PELOD-2 and P-MODS scores were significantly increased [PRISM III: 16 (13, 22) vs. 12 (10, 15), PELOD-2: 6 (5, 9) vs. 4 (2, 5), P-MODS: 6 (4, 9) vs. 3 (2, 6), all P < 0.01]. ROC curve analysis showed that the AUCs of PCIS, PRISM III, PELOD-2, and P-MODS for predicting death of critical ill children were 0.649, 0.731, 0.773, and 0.747, respectively. Hosmer-Lemeshow test showed that PCIS predicted the mortality and the actual mortality in the best fitting effect (χ 2 = 7.573, P = 0.476), followed by PELOD-2 and P-MODS (χ12 = 9.551, P1 = 0.145; χ22 = 10.343, P2 = 0.111), while PRISM III had poor fitting effect (χ2 = 43.549, P < 0.001). CONCLUSIONS: PRISM III, PELOD-2 and P-MODS can discriminate between survivors and moribund patients well, and assessing the condition of critically ill pediatric patients with relatively accuracy. PCIS was the best fitting effect in predicting mortality and actual mortality, followed by PELOD-2 and P-MODS, while PRISM III had poor fitting effect.