Prediction of postoperative complications after elective aortic surgery using stepwise logistic regression analysis

Perioperative risk factors that contribute to postoperative morbidity have been identified for various forms of vascular disease, but it is not clear to what degree each type of disease contributes to morbidity when these diseases are seen in combination. We used stepwise logistic regression analysis to determine the relative importance of 25 risk factors in predicting postoperative complications in 100 consecutive patients undergoing elective intra-abdominal aortic surgery (59% aneurysmal disease, 37% occlusive disease, 26% renal artery lesions). Thirty-one patients developed postoperative complications, including three deaths. The most common complication was deterioration in renal function (17 patients, 24% of those at risk) followed by the need for prolonged endotracheal intubation (>48 hours, 8%). All other events occurred uncommonly (4% or less). Stepwise logistic regression demonstrated four criteria that were significantly associated with complications. In descending order of importance these were: renal artery occlusive disease, intraoperative blood replacement, the preoperative APACHE II score, and a history of heavy smoking. Since both physiologic status and comorbid conditions contribute to morbidity and costs of elective vascular surgery, outcome studies must be adjusted to account for these preoperative characteristics. Additionally, since two-thirds of patients who have renal artery occlusive disease develop complications that prolong hospitalization, our current methods of protecting the kidney during ischemia should be improved to potentially reduce this cause of morbidity.