Quantifying true error rate in assisted reproduction: a multi-site retrospective cohort analysis

Abstract Background We conducted a multisite retrospective cohort analysis of 414,500 assisted reproductive technology (ART) procedures from ten clinics using a barcode-based electronic witnessing system. Identity management is fundamental to ART safety and is emphasised in international laboratory guidance on patient identification and traceability issued by professional societies and regulators. Estimates of error frequency vary across detection methods (manual witnessing, barcode, and radio-frequency identification [RFID]) and settings using these systems. Results Among 414,297 analysed procedures drawn from 414,500 total barcode-based electronic witnessing system (EWS) procedures, we observed 2,399 true identity-related errors (overall rate 0.58%, 95% confidence interval [CI] 0.56%–0.60%) after excluding 203 non-clinical system entries (175 Error events and 28 Incomplete cycles). Subtype as a share of true errors counts and were reported as: no match (unclassified) 1,452 (60.5%); wrong patient 718 (29.9%); wrong item 146 (6.1%); too many barcodes 83 (3.5%). Conclusions Within barcode-EWS-using clinics, identity-related mismatches were detected at approximately 0.58% of procedures. Because 60.5% of true errors were unclassified in historic exports, clinical prioritisation should be cautious. Findings apply to ART laboratories using similar barcode-EWS and should not be generalised to non-EWS laboratories. Transparent limitations regarding vendor-specific data, selection bias, and lack of chart-level adjudication are detailed. Importantly, in this dataset, no identity mismatch progressed to a completed procedure with the wrong patient; the EWS blocked progression until the mismatch was resolved.