Abstract: Background: In China, it is resigned and difficult for PICU doctors to accept discharge against medical advice. Objective: To explore the clinical characteristics of dead and surviving children who left PICU against medical advice, and analyze the factors influencing the post-discharge death. Design: Multicenter prospective cohort study. Methods: Consecutive patients discharged against medical advice were recruited from PICU of 8 children's specialty hospitals in the East China from August 1, 2016 to July 31, 2017 as the cohort population. The outcome of survival or death according to the telephone follow-up within 28 days after discharge was the cohort's end points. Demographic data, clinical symptoms, reasons for discharge, and evaluation parameters for PCIS and PRISMⅢ were collected. The logistic risk model was used to analyze the influencing factors of post-discharge death. Main outcome measures: Mortality rate within 28 days of children discharged from the PICU against medical advice. Results: A total of 4,952 cases from PICU of 8 hospitals were included into analysis with in-hospital mortality rate of 5.6%(279/4,059). Among the 893 cases(18.1%) discharged from the hospital against medical advice, there were 518 males(58.0%) and 375 females, with a median age of 1.4 years. Three cases were lost to follow-up within 28 days after discharge. In total, 550 cases(61.6%) died and 340 cases survived. The proportion of discharged cases in rural areas was higher than that in cities(62.2% vs 378%), and the proportion of death cases was higher than that of survival cases(65.0% vs 57.8%) in rural areas during the 28-day follow-up. The differences were statistically significant. The main cause of death in discharged cases was infection accounting for 49.2%, followed by unknown causes, tumor, congenital malformation and genetic metabolism accounting for about 10%, respectively. There was a statistically significant difference in the PRISMⅢ scores between death and survival of discharged cases ［8(3,15) vs 3(0,7)］ . A single factor analysis was performed on the clinical characteristics of discharged deaths and hospital deaths. Variables with statistically significant differences were entered into logistic regression analysis. The risk of discharged rural cases was 55% higher than that of urban cases(OR=1.554, 95%CI: 1.112-2.173). The risk of discharged cases without medical insurance increased by 169% compared with those with medical insurance(OR=2.686, 95%CI: 1.910-3.778). The risk of death from hospital discharge was reduced by 53% for children with a history of cardiopulmonary resuscitation before hospitalization(OR=0.467, 95%CI: 0.271-0.802). For every 1 point reduction in the severity of disease score PRISMⅢ, the risk of death from hospital discharge was reduced by 4%(OR=0.962, 95%CI: 0.946-0.978). Conclusion: The fatality rate excluding cases discharged against medical advice in PICU of 8 hospitals in East China was 5.6%(279/4,059), while the general one was 16.8%(829/4,959). Living in rural areas and no medical insurance increased the risk of post-discharge death. A history of cardiopulmonary resuscitation before hospitalization could reduce the risk of death.

Key words: Children, Intensive care unit, Die, Survive, Leave against medical advice