Adherence to Antimicrobial Prophylaxis Guidelines for Elective Surgeries Across 825 US Hospitals, 2019–2020

Abstract

BACKGROUND: There are limited US data assessing adherence to surgical antimicrobial prophylaxis guidelines, particularly across a large, nationwide sample. Moreover, commonly prescribed inappropriate antimicrobial prophylaxis regimens remain unknown, hindering improvement initiatives. METHODS: We conducted a retrospective cohort study of adults who underwent elective craniotomy, hip replacement, knee replacement, spinal procedure, or hernia repair in 2019-2020 at hospitals in the PINC AI (Premier) Healthcare Database. We evaluated adherence of prophylaxis regimens, with respect to antimicrobial agents endorsed in the American Society of Health-System Pharmacist guidelines, accounting for patient antibiotic allergy and methicillin-resistant Staphylococcus aureus colonization status. We used multivariable logistic regression with random effects by hospital to evaluate associations between patient, procedural, and hospital characteristics and guideline adherence. RESULTS: Across 825 hospitals and 521 091 inpatient elective surgeries, 308 760 (59%) were adherent to prophylaxis guidelines. In adjusted analysis, adherence varied significantly by US Census division (adjusted OR [aOR] range: .61-1.61) and was significantly lower in 2020 compared with 2019 (aOR: .92; 95% CI: .91-.94; P < .001). The most common reason for nonadherence was unnecessary vancomycin use. In a post hoc analysis, controlling for patient age, comorbidities, other nephrotoxic agent use, and patient and procedure characteristics, patients receiving cefazolin plus vancomycin had 19% higher odds of acute kidney injury (AKI) compared with patients receiving cefazolin alone (aOR: 1.19; 95% CI: 1.11-1.27; P < .001). CONCLUSIONS: Adherence to antimicrobial prophylaxis guidelines remains suboptimal, largely driven by unnecessary vancomycin use, which may increase the risk of AKI. Adherence decreased in the first year of the COVID-19 pandemic.