VALIDATION OF AN ELECTRONIC ALGORITHM TO IDENTIFY CANDIDATES FOR COLON SURGICAL SITE INFECTION REVIEW JA Yegge1, K Gase1, M Hohrein1, H Xu1, R Khoury1, H Babcock2 1BJC HealthCare, St. Louis MO, 2Washington University, St. Louis, MO In response to an increasing surveillance burden, an electronic algorithm was developed in 2009 to identify surgical site infection (SSI) candidates. METHOD RESULTS The algorithm rules look for readmissions after a qualifying procedure in addition to cultures, antibiotic starts and ICD-9 infection codes. The SSI candidates are sent to an Infection Preventionist’s work list for review. Every colon procedure (identified by ICD-9 code) performed between 10/1/2012-12/31/2012 at 10 BJC Healthcare system adult hospitals were included. The electronic medical records were screened by a single clinical abstractor. Based on the abstracted information a single Infection Preventionist reviewed all procedures identified as potentially infected to determine infection status using 2012 National Healthcare Safety Network definitions. Specificity, sensitivity, positive and negative predictive values were calculated. The objective of this study was to examine the algorithm’s accuracy as a surveillance method for colon SSIs. Test Calculation & Result Sensitivity 17/(17+1) = 94.45% Specificity 301/ (301+98) =75.44% Positive Predictive Value (PPV) 17/(17+98) = 14.79% Negative Predictive Value (NPV) 301/(301+1) = 99.67% 28% 115 Triggered Algorithm 417 Procedures Identified 72% 302 Did Not Trigger Algorithm 1 Confirmed Case (Not Identified by Trigger) 17 Confirmed Cases CONCLUSION These results confirm that the algorithm is highly effective in rejecting true negatives for further evaluation. It is also highly effective in capturing true positives within the subset identified for infection investigation. Additional refinement of the algorithm rules is needed to decrease the number of procedures that are flagged for review. This will decrease the time the Infection Preventionist spends on chart review while not missing any infected cases. Nothing to Disclose