Linear logistic regression analysis. The application of a statistical method to the prediction of operative risk in surgical patients

The precise prediction of postoperative mortality is obviously of great interest to the clinician. It is important in defining both high-risk patients, as well as situations in which nonoperative therapy may be desirable. The authors suggest the use of stepwise logistic regression analysis to develop a multifactor prognostic scoring system, permitting an accurate preoperative prediction of postoperative mortality. Although multivariate analysis is capable of producing predictive equations, no predictive model can substitute for qualified surgical judgment, nevertheless, statistical modeling helps us to understand better the relative magnitude of factors that contribute to mortality and morbidity in a variety of surgical conditions.

Duke Scholars

Author Alfonso Torquati Surgery, Minimally Invasive Surgery