Automated hospital records tool lifted adverse drug reaction reporting fortyfold

Two Sydney hospitals that once sent almost no adverse drug reaction reports to Australia’s medicines regulator now submit more than 540 per year, after researchers embedded an automated reporting tool directly into the hospitals’ electronic medical record. The retrospective study, published in BMC Digital Health, analysed reporting at Blacktown and Mt Druitt Hospitals between March 2022 and March 2025 and found 1,632 adverse drug reaction reports were submitted following implementation, compared with zero in 2018 and 2019, three in 2020 and nineteen in 2021.

The intervention was deliberately low-friction. Clinicians documented suspected reactions in the allergy and adverse drug reaction tab of the electronic medical record, which is standard clinical documentation practice, and a structured electronic form called a PowerForm captured all the data elements required for reporting to the Therapeutic Goods Administration. Key fields such as patient identifiers, admission type, suspected medicines, reaction descriptions and onset timing were automatically pre-populated from the record. Notably, no training or awareness campaign was provided, and many clinicians were not necessarily aware that their entries would later be extracted for formal regulatory reporting.

Each month, the pharmacy department and the hospital’s medication safety committee downloaded all newly entered reports, reviewed them for completeness and duplicates, and submitted them collectively to the regulator. The authors argue this workflow design, which avoided asking busy clinicians to complete a separate report outside their routine tasks, explains much of the dramatic increase. Under the previous voluntary system, healthcare professionals had to locate a reporting channel, manually re-enter information already documented in the clinical record, and submit it separately, a burden that contributed to persistently low national reporting rates.

The scale of under-reporting in Australia is illustrated by 2017 figures cited in the study: only 19 percent of reports submitted to the Therapeutic Goods Administration that year came from healthcare professionals, comprising hospital pharmacists at 10 percent, community pharmacists at 6 percent and general practitioners at 3 percent. Sponsors accounted for 54 percent of reports and state and territory health departments reporting adverse events following immunisation accounted for 18 percent. Commonly cited barriers include limited knowledge of reporting procedures, time constraints, insufficient resources and high workload.

The reported reactions showed a clear clinical pattern. Antibacterials for systemic use were the most frequently implicated drug class at 43.8 percent of cases, followed by analgesics at 16.5 percent and anti-inflammatory and antirheumatic products at 6.9 percent. Among system organ classes, immune system disorders dominated at 34.5 percent, followed by gastrointestinal disorders at 7.4 percent, a distribution the authors interpret as reflecting antibiotic hypersensitivity as a major contributor to the reported burden.

Patient demographics followed patterns familiar from international pharmacovigilance data. Female patients accounted for 69.1 percent of the 1,632 reports, consistent with global analyses showing women contribute roughly 55 to 60 percent of adverse drug reaction reports and face a 1.5 to 2-fold higher risk of experiencing them, a disparity attributed to sex-related differences in pharmacokinetics and pharmacodynamics as well as higher medication exposure. Adults aged 18 to 64 years made up 60.7 percent of reports, patients aged 65 and over 31.7 percent, and children under 18 7.6 percent. Most reports came from inpatient encounters at 46.0 percent, with emergency presentations at 37.6 percent and outpatient encounters at 16.4 percent, and registered nurses filed 55.7 percent of reports.

Causality assessment using the World Health Organization-Uppsala Monitoring Centre criteria revealed a substantial documentation gap. More than half of cases, 57.3 percent, were classified as unassessable or unclassifiable, with 19.4 percent deemed possible, 13.3 percent probable or likely and 8.2 percent certain. The authors attribute this to incomplete clinical timelines, polypharmacy and limited rechallenge data inherent in spontaneous and automated reports. Preventability, assessed with the modified Schumock and Thornton scale, could be determined for fewer than half of cases: 2.7 percent were definitely preventable and 1.9 percent probably preventable, 37.7 percent were not preventable and 57.7 percent could not be determined, a pattern the researchers say likely reflects documentation limitations rather than true non-preventability.

The economic analysis estimated total adverse drug reaction-related costs at A$800,391 over the study period, calculated from emergency department presentations, hospital admissions and excess bed-days valued using standardised Australian public hospital prices. Admissions accounted for 41.5 percent of costs and excess bed-days 20.3 percent, together nearly two-thirds of the total, while emergency presentations accounted for 38.3 percent. Monthly costs peaked in 2024 at A$29,604 and stood at A$26,188 in the first three months of 2025. The 75 preventable cases, 4.6 percent of the total, corresponded to A$36,783 in potentially avoidable expenditure. A one-way sensitivity analysis varying unit costs by plus or minus 20 percent showed total costs were most sensitive to the admission unit cost, at plus or minus A$66,400, followed by the emergency department unit cost at plus or minus A$61,200 and the bed-day unit cost at plus or minus A$32,500.

The authors caution that the findings carry important limitations. Severity, seriousness and duration of reactions were not systematically captured by the tool and could not be assessed. Causality and preventability classifications were performed primarily by a single assessor without a formal inter-rater reliability exercise, so their reproducibility has not been verified. Pre-implementation figures came directly from the regulator, which noted that hospital-level attribution was not always possible because some reports were submitted under a combined sender name. The cost analysis took a hospital perspective only, excluded downstream and societal costs, and in cases where a reaction occurred as a complication during an admission for an unrelated reason, the full admission cost may overestimate the cost attributable to the reaction, making the estimates a conservative upper bound. The decline in monthly reports in early 2025 also warrants caution, as it reflects only a three-month observation window, though a genuine decline cannot be excluded.

Despite these caveats, the study is presented as the first Australian assessment of an automated electronic medical record-embedded adverse drug reaction reporting tool with direct regulatory submission, and its reporting gains are broadly comparable to international experience, including step-changes seen in the United States Veterans Affairs health system. The authors emphasise that increased reporting reflects improved detection rather than a true rise in reaction incidence, since spontaneous systems are estimated to capture fewer than 10 percent of adverse drug reactions. They suggest that incorporating similar functionality into New South Wales’ forthcoming statewide digital health system could strengthen pharmacovigilance, and they call for prospective evaluations, structured capture of reaction severity, integration with clinical decision support, and broader implementation across primary care and community settings to clarify the full system-level value of automated pharmacovigilance.

Subject of Research: An automated electronic medical record tool for reporting adverse drug reactions at two Australian hospitals

Article Title: Towards safer medication practices: a retrospective analysis of adverse drug reactions

Article References: Goh, J. X., Li, R., Small, F., Kairaitis, L. K., Van, C., & Castelino, R. L. (2026). Towards safer medication practices: a retrospective analysis of adverse drug reactions. BMC Digital Health, 4(1), Article 77. https://doi.org/10.1186/s44247-026-00310-8

Image Credits: AI Generated

DOI: 10.1186/s44247-026-00310-8

Keywords: adverse drug reactions, pharmacovigilance, electronic medical record, Therapeutic Goods Administration, medication safety, antibacterials, hospital admissions, healthcare costs, drug hypersensitivity, retrospective study, Australia, automated reporting