Navigating medication safety with electronic medical records: insights from a dual-phase implementation in paediatric, neonatal and maternity care

OBJECTIVE: Electronic medical record (EMR) implementations can disrupt clinical workflows and impact medication safety. This study evaluated the effect of a two-phase EMR roll-out on medication safety events within the Women's and Children's Division of a large tertiary public hospital. METHODS: We conducted a retrospective real-world difference-in-differences analysis using negative binomial regression to accommodate overdispersion in incident counts. Two activation phases (30 June 2021 and 1 January 2024) were compared against non-activated areas. Data spanning January 2020 to February 2024 were extracted from the Safety Learning System. Natural language processing (spaCy) identified both direct mentions and semantically similar references to 'electronic medical record,' 'EMR' and 'electronic prescribing' for supplementary sensitivity analyses. RESULTS: Following each activation, minor (Incident Severity Rating 3) and near miss (Incident Severity Rating 4) incidents showed a short-lived rise - peaking within 1 month of go-live - then returned to pre-implementation levels. No statistically significant long-term changes in overall medication safety event trends were observed (all P > 0.75). Natural language processing-detected EMR mentions increased immediately post-activation, but did not persist. CONCLUSIONS: Phased EMR implementation was not associated with sustained alterations in medication safety event rates. Transient near miss increases likely reflect heightened reporting awareness rather than true error surges. Future evaluations should broaden text-mining filters, incorporate manual content review and use mixed-methods designs to ensure comprehensive detection of EMR-related safety signals and to capture clinician workflow impacts.