Background: Multivariable analysis best identifies independent risk factors. Study Design: We conducted a prospective evaluation of 2,605 patients through univariate analysis followed by nonconditional multiple logistic regression analysis on 39 pre-, intra-, and postoperative factors, analyzed according to preoperative factors alone, preoperative and intraoperative factors together, and all 3 combined. The purpose was to identify surgeon-dependent independent risk factors for mortality after elective colorectal surgery, with immediate anastomosis for cancer and nonacute diverticular disease. Results: Overall mortality was 3.5%. Through multivariable analysis, five risk factors were found when preoperative data were analyzed alone. Four remained (age between 60 and 75 years, age greater than 75 years, male gender, and heart failure) and 4 new factors (palliative resection, total colectomy, respiratory failure, and surgeon-dependent fecal soiling [the only surgeon-dependent factor]) appeared when pre- and intraoperative factors were analyzed together. Of the latter, two remained stable when all three categories of risk factors were combined and analyzed (palliative resection and total colectomy), and the two others disappeared. Of the eight pre-, intra-, and postoperative factors combined, two new factors appeared: extrasurgical site (ESS) and surgeon-dependent, organ space surgical site (O/SSS) morbidity. Conclusions: Every effort must be made to collect specific, surgeon-dependent (technical and clinical) data, along with administrative data, for multivariable analysis of risk factors. Classification into three periods (pre-, pre- and intraoperative together, and pre-, intra-, and postoperative combined) enables determination of relevant, surgeon-dependent risk factors (fecal soiling and postoperative morbidity) for which there are direct preventive actions. © 2008 American College of Surgeons.
