The use of multiple logistic function to predict surgical risk.

A method for preoperative assessment of surgical risk, in function of several risk factors, was developed using the multiple logistic function, as a model of multivariate statistical analysis. This model has the advantage to express the two considered outcomes (perioperative complications and death) in numerical terms of probability. The data were obtained from 8630 consecutive patients admitted to 62 hospitals located in various regions of Italy. The patients were divided at random into two groups: group A (6040) was used to develop the method, and group B (2590) was used as control. The method is based on the role played by seven risk factors, that were shown to be significantly related to the two outcomes. These factors are age, systolic blood pressure, pathological changes of renal function, presence of concomitant diseases, estimated duration of surgical procedure, type (major or minor) and timing of surgery, (elective, delayed or emergency). The relative and independent role of each of these factors (beta) and the value of constant alpha were estimated by maximum likelihood method. The surgical risk is assessed by solving the multiple logistic equation, and multiplying the beta values by the corresponding individual coded values of the risk factors represented by the various x's. When the method was tested by applying the computed values of alpha and beta's to assess the outcomes in the control group, the predictions resulted highly accurate (r = 0.99). To simplify the calculation of surgical risk, in appendix a computer program in Basic, already predisposed to receive in code the individual values of x's, is reported