[Evaluation of prognostic factors in general surgery; methodological aspects of a multicenter prospective study of 1182 patients].

Surgical risk is defined as the probability of occurrence of early or late complications. For each patient, knowledge of factors affecting surgical risk in a basic step when evaluating prognosis after surgery and determining therapeutic decision. A greater and more effective information may be obtained studying variables within the context of the other ones that more or less condition them. Such a "multivariate" approach allows simultaneously investigating all characteristics by taking account of their correlations. Specification of the problem, choice of the relevant variables, data collection and statistical data processing were considered. Pros and cos of multiple logistic regression model in the assessment of surgical risk were reviewed. The index could be used to identify the most important factors in determining surgical risk and quantify their respective importance. In this way surgical patients could be classified according to risk level. This approach also allow for the concurrent investigation of various potential risk factors and their interactions. For this reason it is termed multifactorial.