Logistic Regression Analysis for Predicting Mortality of Acute Burn Injury in Taiwan

• Main Authors: Shih-Hao Liu, 劉士豪
• Other Authors: Li-Ching Chen
• Format: Others
• Language: zh-TW
• Published: 2012
• Online Access: http://ndltd.ncl.edu.tw/handle/27805880470533123524

碩士 === 淡江大學 === 統計學系碩士班 === 100 === In recent years, due to medical and technological advances, the mortality of burn patients has slowly decreased. Many studies have shown that age, burn size, and inhalation injury are important predictors of mortality following an acute burn injury. Objective estimates of the probability of death from burn injuries would help clinicians to make decisions and also provide patients and others with explicit basis for medical and financial decisions about their care. The goal of the study is to develop a simple and objective model for the prediction of mortality after burn injury in Taiwan population. From 1997 to 2010, a prospective review of 23,147 patients admitted with acute burn injury to 44 contract hospitals of Childhood Burn Foundation of the Republic of China across Taiwan was conducted. Variables examined were age, sex, extent of burn, presence or absence of inhalation injury, flushing time, hospital admission status, referral status, admission to intensive care unit or not, and mortality. The logistic regression analyses were used for evaluation of risk factors. Model performance was evaluated by measures of the validation Hosmer-Lemeshow goodness-of-fit statistic, the receiver operator characteristic (ROC) curve, sensitivity, specificity and accuracy of the models were also discussed.