A structured-data algorithm for semiautomated surveillance of surgical site infection after colorectal surgery: A diagnostic accuracy study.

Abstract

Background. Manual surveillance of surgical site infections (SSIs) after colorectalsurgery is resource-intensive, limiting scalability. Semiautomated algorithms based onstructured electronic health record (EHR) data may maintain high case-findingsensitivity while reducing workload.Methods. A retrospective diagnostic-accuracy study was conducted in a teachinghospital participating in a nationwide SSI surveillance programme. All electivecolorectal procedures performed between January 2010 and December 2023 wereincluded. SSIs were classified according to CDC-NHSN/ECDC criteria. Eight binaryEHR-derived “alerts” were combined into a composite rule (any alert positive). Manualsurveillance served as the reference standard. Performance was assessed overall, bySSI depth (superficial, deep, organ/space), and by procedure type (colon vs rectal).Discrimination (AUC), sensitivity, specificity, positive predictive value (PPV), andnegative predictive value (NPV) were calculated with 95% confidence intervals (CIs).Results. A total of 1,213 patients (1,085 colon; 128 rectal) were included. The overallSSI incidence was 11.2% (3.1% superficial, 1.2% deep, 6.8% organ/space). Thecomposite alert achieved an AUC of 0.859 (95% CI 0.838–0.878) for any SSI, withsensitivity 0.721, specificity 0.876, PPV 0.424, and NPV 0.961. At this operating point,19% of procedures would be flagged for manual verification, corresponding to anestimated 81% reduction in full chart reviews. Discrimination was highest fororgan/space infections (AUC 0.919; sensitivity 0.831; specificity 0.911). Performancefor deep SSI was intermediate (AUC 0.805), and for superficial SSI, more limited (AUC0.571). Sensitivity was higher for colon surgery (AUC 0.853) and specificity higher forrectal surgery (AUC 0.881).Conclusions. The structured-data algorithm demonstrated strong overall discriminationand excellent performance for organ/space infections, supporting the feasibility ofsemiautomated surveillance without compromising detection quality. External andprospective validation, definition of diagnostic safety thresholds, and workloadreductionanalyses are required to optimise implementation. Exploration of NLP addonsmay be considered where resources permit. ClinicalTrials.gov: NCT07130656.