Background/Purpose: Surgical site infection surveillance to determineincidence is a key infection control activity. Case detection is labour-intensive, therefore most infection control programs use manual or simple electronic mechanisms to “trigger” chart review. However, such “trigger” mechanisms are also labour-intensive, and often of poor specificity. Our objective is to develop a complex trigger mechanism using data from an electronic data warehouse, to improve specificity of surveillance of surgical site infection compared to current trigger mechanisms.
Methods: We will derive an electronic trigger tool for cardiac surgical site infection surveillance using a nested case-control design, among a cohort of all patients undergoing coronary artery bypass grafting, cardiac valve repairor replacement, or heart transplant at the University of Ottawa Heart Institute, from July 1 2004 to June 30 2007.
We will perform a systematic literature review to identify potential trigger factors to include in the model, then construct the trigger tool by backwards stepwise logistic regression. The best-fit model will be used to calculate the probability of surgical site infection. We will select the threshold probability to use in surveillance by visual inspection of receiver-operator-characteristic curves. The accuracy of this electronic trigger mechanism will be compared to pre-existing manual and simple electronic mechanisms using relative true positive ratios and relative false positive ratios.
Results/Conclusions: We have selected 200 cases of surgical site infection and 541 controls from among 3744 procedures performed during the study period. As of the date ofthis abstract we are still undertaking the systematic review.
