Seasonality of surgical site infection rates across a broad surgical sample and diverse health system.

Abstract

Background: Prior data suggest that surgical site infections (SSIs) occur more frequently in warmer months. Although several studies have shown this, most do not adequately account for confounding factors or the non-parametric nature of seasonal trends. This study examined SSI rates across multiple hospitals within a single healthcare system, using previously published statistical models applied to electronic health record (EHR) data.

Methods: We retrospectively analyzed all surgeries from 2014 to 2019. Preoperative risk and postoperative probabilities of SSIs were estimated using our Automated Surveillance of Postoperative Infections (ASPIN) models. Observed to expected (O/E) ratios were calculated and plotted by week, with cubic smoothing splines visualizing trends. Seasonality was modeled using generalized linear mixed models with sine and cosine transformations of the week of the year or quadratic transformations as predictors.

Results: 348,289 surgeries were analyzed. Most SSIs occurred between July and August. The risk-adjusted O/E ratio and postoperative SSI probabilities significantly increased (P < 0.05) during the summer in Region 1, an academic hospital, but not in the other two regions.

Conclusions: Our study supports prior work suggesting a summer peak in SSIs but not a distinct July spike. We incorporated novel statistical models to predict preoperative risk and postoperative probability of SSI.