Related Experiment Videos
Semi-Automated Retrospective Surveillance of Surgical Site Infections (sAReS-SSI) - using hospital routine data for
Dominik Sons1,2, Eva Bernauer3, Alexander Schmidt3
1Institute of Hygiene, Cologne Merheim Medical Centre, University Hospital of Witten/Herdecke, Cologne, Germany.
Background:
Surgical site infections (SSIs) place a significant burden on healthcare systems worldwide. Although surveillance is crucial for obtaining accurate data and developing effective prevention strategies, there are still significant gaps due to the frequent reliance on manual and resource-intensive processes. To gain efficient, in-depth insight, we developed an algorithm for semi-Automated Retrospective Surveillance of Surgical Site Infections (sAReS-SSI), which retrospectively identifies SSIs using existing routine hospital data.
Methods:
sAReS-SSI adopts a patient-centric approach to data analysis, using an ICD-10 and OPS catalogue to detect in-house SSIs through temporal linkage to prior in-house surgeries, refining the NWIF algorithm of the German Institute for Quality and Transparency in Healthcare. sAReS-SSI was evaluated against SSIs that were validated through bedside surveillance in three orthopaedic and trauma surgery wards, as published in the HygArzt study. Its performance was also evaluated in comparison with a reconstructed NWIF algorithm, and contextualised using OP-KISS surveillance reports from the German National Reference Center.
Results:
sAReS-SSI correctly identified 61 of 65 in-house SSIs published in the HygArzt study. Re-analysis of initially false-positive cases revealed 10 additional true in-house SSIs not captured in HygArzt, yielding a sensitivity of 94.7%, specificity of 94.3%, positive predictive value of 35.9%, and negative predictive value of 99.8%. sAReS-SSI accurately captured infection dynamics in pre- and post-intervention periods of the HygArzt study, correctly identified problematic surgery types, and demonstrated that relying solely on NWIF or OP-KISS might miss a portion of potential SSI events.
Conclusions:
Automating the analysis of routine information for the detection of SSIs offers opportunities to save resources and enables efficient, large-scale analysis. By narrowing down potential SSI cases, this approach reduces the workload for manual surveillance by hygiene experts. In our cohort sAReS-SSI flagged potential cases and highlighted how routine-data-based case finding may extend OP-KISS surveillance by including additional non-indicator-surgeries.