Modified abbreviated burn severity index as a predictor of in-hospital mortality in patients with inhalation injury: development and validation using independent cohorts

Purpose: The ability to accurately evaluate the severity of inhalation injury can help to optimize patient care. However, there is no accepted severity grading system, especially for inhalation injury. Methods: We screened a multicenter burn registry and included adult patients who required oxygen treatment or mechanical ventilation. After the patient data were divided into development and validation cohorts, missing values were replaced with multiple imputation. Twelve potential predictors were analyzed using multivariate logistic regression to identify prognostic variables for in-hospital mortality and scores were assigned to each predictor based on odds ratios to develop the Modified Abbreviated Burn Severity Index, mABSI. The mABSI was validated using c-statistics and calibration curves. Results: We randomly assigned 1377 and 919 patients to the development and validation cohorts, respectively. Age, self-inflicted injury, cutaneous burn area, and mechanical ventilation requirement were identified as independent predictors, and the mABSI (1–17 scale) was, thus, developed. The mABSI has a high discriminatory power (c-statistic = 0.94; 95% CI 0.92–0.97), and both estimated and observed in-hospital mortalities increased from 1% at score ≤ 5 to almost 100% at score ≥ 14 with linear calibration plots. Conclusions: We developed and validated the mABSI which accurately predicts in-hospital mortality.