Objective: To correlate variables recorded perioperatively with morbidity and mortality in an attempt to assess the predictive value of these variables for the outcome of individual patients.
Design: Prospective study.
Setting: Teaching hospital, Germany.
Subjects: 6304 consecutive patients admitted for operation.
Interventions: Recording of American Society of Anesthesiologists (ASA) class, age, sex, operation class (Hoehn), anaemia, hypertension, bronchopulmonary disease, diabetes mellitus, renal failure, major gastrointestinal disease, type of anaesthesia, operation (and whether emergency or elective), and duration of operation. All these factors were subjected to logistic regression analysis.
Main outcome measures: Odds ratio (OR), sensitivity, specifity, and positive and negative predictive values.
Results: We studied 6304 patients of whom 140 died postoperatively and 1432 developed complications from which they survived. The variable that had most influence on the risk of postoperative complications was the ASA class (ASA class 4, OR 4.2, and ASA class 3, OR 2.2) followed by the severity of the operation (OR 1.86). An uncomplicated course was correctly predicted in 96%, but complications were correctly predicted in only 16%. The positive predictive value was 57%, and negative predictive value 80%.
Conclusions: Despite the fact that we considered a large number of variables we were unable to predict the risk of complications for individual patients with any accuracy.