The use of prognostic indicators in the development of a statistical model predictive for adrenal insufficiency in trauma patients.

We performed a retrospective chart review of trauma patients admitted to Palmetto Richland Memorial Hospital and identified 63 cases of adrenal insufficiency along with 65 trauma patient controls. Two statistical models, a neural network and a multiple logistic regression, were developed to predict patients with increased risk of developing adrenal insufficiency. Each model had 11 selected independent variables, along with patient demographic data, to make a probabilistic prediction of patient outcome.