[Perioperative risk assessment in surgery - an analysis in 10 395 patients]

Abstract

The aim of this study was to perform a risk analysis on the basis of routinely documented variables (age, sex, ASA-classification, priority of operation, malignant disease, intraperitoneal or intrathoracic operation and duration of operation) to identify surgical patients who benefit from a more complex risk assessment.

Patients and methods: In a prospective observational trial we analysed a consecutive series of 10 395 patients who were operated on in our General Surgical Department from January 1996 until December 2000 in respect to in-hospital mortality. The variables were examined in univariate tests. Factors with significant impact were subsequently included in a multiple logistic regression analysis. This was done for all variables and afterwards for each ASA-class separately. Predictive accuracy of the prediction model was calculated by the area under a receiver operating characteristic curve (AUC (ROC)).

Results: The overall mortality was 3.9 %. For ASA-classes 2 to 4 we were able to establish a prediction model by means of multiple logistic regression that identified ASA-classification (Odds Ratio [OR ] ASA-class 3 = 3.7; OR ASA-class 4 = 22.4), age (OR 1.019 per year), duration of operation (OR for duration > or = 240 min = 2.25), intraperitoneal/intrathoracic operation (OR = 4.6), emergency operation (OR = 3.1), and malignant disease (OR = 1.5) as independent predictive factors. Both risk group 1 and risk group 5 were excluded from the analysis because there was no mortality in risk group 1 and too few patients in risk group 5. We found an AUC (ROC) of 91.6 % for the considered ASA-classes.

Conclusion: The ASA-classification is a good instrument for the assessment of perioperative mortality. Its predictive power can substantially be improved in the classes 2 to 4 by the variables age, duration of operation, intraperitoneal or intrathoracic operation, priority of operation, and malignant disease.