Impact of a surgeon-trained observer on accuracy of colorectal surgical site infection rates.

ABSTRACT

PURPOSE: The aims of this study were 1) to establish accurate and reproducible baseline surgical site infection rates for our department and 2) to identify risk factors associated with surgical site infection in patients undergoing surgery on a colorectal service. METHODS: Phase I--Surgical site infection grading between the surgeon-trainer and the observer-trainee was validated using a four-point scale for wound evaluation previously used by our institution. Phase II--Patients undergoing colorectal surgery were prospectively monitored. The observed surgical site infection rate was compared with morbidity and mortality reports. Patient and perioperative variables were analyzed for their effect on surgical site infection using the chi-squared test. Risk factors approaching significance on univariate analysis (P < 0.2) were entered into a multivariate stepwise logistic regression model. RESULTS: Concordance on surgical site infection grading between the surgeon-trainer and the observer-trainee improved from an initial 79 percent to 96 percent during the validation period. The surgeon-trained observer reported a surgical site infection rate of 7.2 percent vs. a morbidity and mortality reported rate of 3.3 percent. Among the variables examined, obesity and surgical procedure category were significantly associated with surgical site infection rates. The effect of prophylactic antibiotics and prior chemotherapy, radiation, or steroid therapy on surgical site infection rates approached significance. A logistic regression analysis incorporating these risk factors for surgical site infection accurately predicted infection status 93 percent of the time. CONCLUSION: Use of a surgeon-trained observer doubles the detection rate of postoperative surgical site infection. Accurate, prospective assessment identifies risk factors significantly associated with increased surgical site infection rates in colorectal surgical patients.