To derive and validate a prediction model for the development of ARDS in burn-injured patients.Summary Background Data:
Burn injury carries the highest incidence of acute respiratory distress syndrome (ARDS) among all predisposing conditions, but few studies exist on risk factors in these patients. Studies employing biomarkers and clinical risk factors for predicting ARDS mortality have recently been examined but none exist for onset of ARDS nor in patients with burn injury.Methods:
This was a prospective multicenter study of 113 patients with isolated burn injury or inhalation injury. Clinical variables and plasma biomarkers representative of endothelial injury, epithelial injury, or inflammation were collected within 24 hours of admission. The most parsimonious model was chosen by considering discrimination, calibration, and model fit.Results:
Among the biomarkers measured in patients with burn injuries, a one-standard deviation increase in log-transformed levels of the A2 domain of von Willebrand factor in the first 24 hours was most strongly associated with the development of ARDS (OR 7.72; 95% CI: 1.64–36.28, P = 0.03). Of candidate models, a 3-variable model with %TBSA, inhalation injury, and von Willebrand factor-A2 had comparable discrimination to more complex models (area under the curve: 0.90; 95% CI 0.85–0.96). The 3-variable model had good model fit by Hosmer–Lemeshow test (P = 0.74) and maintained similar discrimination after accounting for performance optimism (Bootstrapped area under the curve: 0.90; 95% CI: 0.84–0.95).Conclusions:
The 3-variable model with %TBSA, inhalation injury, and von Willebrand factor could be used to better identify at-risk patients for both the study and prevention of ARDS in patients with burn injury.