1074 — Association of Timely Administration of Prophylactic Antibiotics for Major Surgical Procedures and Surgical Site Infection

Hawn MT (Birmingham VAMC Research Enhancement Award Program), Itani KM (VA Boston Healthcare System), Gray SH (Birmingham VAMC Research Enhancement Award Program), Vick CC (Birmingham VAMC Research Enhancement Award Program), Henderson W (Colorado Health Outcomes Program), Houston TK (Birmingham VAMC Research Enhancement Award Program)

Objectives:
Prophylactic antibiotic (PA) administration 1-2 hours prior to surgical incision (SCIP-1) is a publicly reported process measure proposed for performance pay. The purpose of this study is to determine if SCIP-1 was associated with surgical site infection (SSI) rates in Veterans Affairs (VA) hospitals.

Methods:
Patients with External Peer Review Program SCIP-1 data with matched National Surgical Quality Improvement Program data were included in the study. Patient- and facility-level analyses comparing SCIP-1 and SSI rates were performed. We adjusted for clustering effects within hospitals, validated SSI risk score and case mix (% colon, vascular, orthopedic) using generalized estimating equations (GEE) and linear modeling (GLM).

Results:
The study population included 9,195 elective procedures (5,981 orthopedic, 1,966 colon, and 1,248 vascular) performed in 95 VA hospitals. Timely PA occurred in 86.4% of cases (88% orthopedic, 83.7% vascular, and 80.9% colon (p < 0.001). Overall, SSI occurred in 4.7% of cases (1.6% orthopedic, 8.1% vascular, and 12% colon (P < 0.001) Timely PA was associated with 4.6% SSI compared to 5.8% in the untimely group (OR = 0.78, 95% CI 0.60, 1.01) in unadjusted analysis. Odds of SSI varied by procedure type with orthopedic OR = 1.08, 95%C.I. 0.56, 2.06; vascular OR = 0.72, 95% C.I. 0.44, 1.19; and colon OR = 1.14, 95% C.I. 0.79, 1.62. Patient-level risk-adjusted multivariable GEE modeling found the SSI risk score was predictive of SSI (p < 0.001), whereas SCIP-1 was not associated with SSI. Hospital-level multivariable GLM found that the proportion of a facility’s cases that were colon (p < 0.0001) but not SCIP-1 rate or facility volume was associated with facility SSI rate. The study had 80% power to detect a 1.75% difference for patient-level SSI rates.

Implications:
Timely PA did not significantly contribute to overall patient or facility SSI rates. There were significant confounders that were associated with whether the patient received a timely antibiotic and whether the patient developed a SSI, with type of procedure being the most notable.

Impacts:
Timely PA did not discriminate whether patients got a SSI nor facility SSI rates. These data are important for the ongoing discourse on how to measure and pay for quality of surgical care.