Multifactorial index of preoperative risk factors in colon resections

Our aim was to analyze the predictive value of a variety of preoperative risk factors on operative outcomes. We reviewed all colorectal resections performed in a single hospital between January 1985 and May 1990. Nine hundred seventy-two resections were performed on 825 patients. We studied 17 preoperative risk factors generated from various medical risk categories. Using the multivariate discriminant function analysis, we calculated that 11 of the 17 risks were of significance in predicting outcomes (all with P less than or equal to 0.031). These factors included emergent operation, age greater than or equal to 75 years, congestive heart failure (CHF), prior abdominal or pelvic radiation therapy, corticosteroid use, albumin less than 2.7 g/dl, chronic obstructive pulmonary disease (COPD), previous myocardial infarction (MI), diabetes, cirrhosis, and renal insufficiency. The classification function generated by the discriminant analysis was used to categorize patients into one of four risk groups depending on their "risk score." The index used to develop each patient's "risk score" ranged from six points for an emergency operation to one point for diabetes. The mortality rates for the various risk groups were as follows: Group 1, zero to four points, 1 percent; Group 2, five to eight points, 10 percent; Group 3, 9 to 13 points, 19 percent; Group 4, greater than 13 points, 33 percent. In contrast to previous reports, we showed that age greater than or equal to 75 years alone is not a major preoperative risk factor but, rather, acts as a modifier for the other predictors of postoperative complications. We then assessed clinical questions concerning specific preoperative risks, such as steroid use, obesity, inflammatory bowel disease, COPD, and prior laparotomy, and their associated specific postoperative complications and have developed prevention strategies based on these findings. Through the use of the risk index, we also were able to assess an individual patient's operative risk more accurately.