Abstract Introduction Selection and dosing of surgical antimicrobial prophylaxis (SAP) to prevent surgical site infections (SSIs) is variable and often inappropriate in clinical settings resulting in increased risk of SSI. We therefore developed and implemented a novel computer decision-support tool to improve appropriate SAP selection specific to each patient’s procedure and clinical context. Methods The intervention, performed at a multi-site tertiary hospital network, was a computer decision-support tool programmed within the hospital’s electronic health record (EHR) to provide patient-specific SAP recommendations based on four variables: procedure name, patient’s beta-lactam allergy status, MRSA status and weight. Safety of cefazolin prophylaxis among patients with self-reported beta-lactam allergy was established in the pre-operative clinic using a validated simple two-item questionnaire. Before each operation, a best practice SAP recommendation alert was provided to the anaesthesiologist based on the inputs from the four aforementioned variables. Results In the post-intervention period of 2 years, a total of ∼110 recommendation alerts were logged. A random sample audit of 408 surgical encounters were selected from the pre- and post-intervention periods. Overall, appropriate antibiotic administration rose from 77% (161/208) to 92.5% (185/200) (χ2 = 18.0, P 0.001) post-intervention. Significant improvements were made for procedure-specific antibiotic selection (80.5% versus 94.5%; χ2 = 19.3, P 0.001) and weight-based dosing (92.5% versus 98.4%; χ2 = 7.45, P = 0.006). Using National Surgical Quality Improvement Program (NSQIP) auditing of 21 912 surgeries showed a significant decline in SSIs following the implementation of the intervention in June 2022. Conclusion In this first-ever intervention designed to direct SAP prescribing based on patient-specific variables, we significantly improved appropriate SAP selection and reduced SSIs across a comprehensive list of surgical procedures.
McIntyre et al. (Tue,) studied this question.