OBJECTIVES: To identify factors that increase the risk of sternal surgical wound infection after cardiovascular surgery and to develop a bedside clinical risk index using these factors. DESIGN: A risk index was developed using clinical data collected from a cohort of 11,508 cardiac surgery patients and validated using three independent subsets of the data. With two of these subsets, we derived a logistic regression equation and then modified the scoring algorithm to simplify the calculation of patient risk scores by clinicians. The final subset was used to validate the index. The area under the receiver operating characteristic (aROC) curve was the primary measure of goodness of fit. SETTING: Toronto General Hospital, a teaching hospital and the largest center for cardiac surgery in Ontario, Canada. PATIENTS: Cardiac surgery patients receiving cardiopulmonary bypass between April 1, 1990, and December 31, 1995, who survived at least 6 days after surgery. RESULTS: Variables that were used to construct the risk index included reoperation due to complication (odds ratio, 4.3; range, 1.9 to 8.5), diabetes (odds ratio, 2.4; range, 1.5 to 3.7), more than 3 days in the intensive care unit (odds ratio, 5.4; range, 3.2 to 8.7), and use of the internal mammary artery for revascularization (odds ratio, 3.2; range, 1.7 to 5.8). Validation showed that the index had an aROC curve of 0.64. CONCLUSIONS: The risk index described in this article allows clinicians to quickly stratify patients into four risk groups associated with an increasing risk of sternal surgical wound infection. It may be used perioperatively or as part of a wound infection surveillance system.
OBJECTIVES: To identify factors that increase the risk of sternal surgical wound infection after cardiovascular surgery and to develop a bedside clinical risk index using these factors. DESIGN: A risk index was developed using clinical data collected from a cohort of 11,508 cardiac surgery patients and validated using three independent subsets of the data. With two of these subsets, we derived a logistic regression equation and then modified the scoring algorithm to simplify the calculation of patient risk scores by clinicians. The final subset was used to validate the index. The area under the receiver operating characteristic (aROC) curve was the primary measure of goodness of fit. SETTING: Toronto General Hospital, a teaching hospital and the largest center for cardiac surgery in Ontario, Canada. PATIENTS: Cardiac surgery patients receiving cardiopulmonary bypass between April 1, 1990, and December 31, 1995, who survived at least 6 days after surgery. RESULTS: Variables that were used to construct the risk index included reoperation due to complication (odds ratio, 4.3; range, 1.9 to 8.5), diabetes (odds ratio, 2.4; range, 1.5 to 3.7), more than 3 days in the intensive care unit (odds ratio, 5.4; range, 3.2 to 8.7), and use of the internal mammary artery for revascularization (odds ratio, 3.2; range, 1.7 to 5.8). Validation showed that the index had an aROC curve of 0.64. CONCLUSIONS: The risk index described in this article allows clinicians to quickly stratify patients into four risk groups associated with an increasing risk of sternal surgical wound infection. It may be used perioperatively or as part of a wound infection surveillance system.