External validation and updating of a prediction model for nosocomial pneumonia after coronary artery bypass graft surgery.

Abstract

The generalizability of a prediction model from North America for incident nosocomial pneumonia following coronary artery bypass graft surgery was assessed for 23247 patients on the Australian and New Zealand Society of Cardiac and Thoracic Surgeons (ANZSCTS) registry. The performance of the North American model was evaluated using measures of calibration and discrimination. The model had reasonable discrimination (area under the receiver-operating characteristic curve, AUC=0·69), but unsatisfactory calibration (Hosmer-Lemeshow test, P<0·001) in the ANZSCTS patients. An update of the model coefficients yielded a model with AUC=0·71 and good calibration (P=0·46).