Estimating the Frequency of Automated Dispensing Cabinet Discrepancy Safety Events Using Markov Models

Background Medication errors are one of the most common types of adverse events in inpatient settings, accounting for nearly 5% of adverse events experienced by hospitalized patients. In this exploratory study, we modeled and estimated the frequency of automated dispensing cabinet (ADC) discrepancy safety events across three different data sources: ADC transaction logs, user ADC discrepancy reports, and user reports to a patient safety event (PSE) reporting system. Methods We collected data from the three sources over a 1-month time period from 5 January to 2 February 2021 and used a Markov model to categorize ADC transactions. Results A total of 1,989,443 ADC transactions were recorded. Of these, 18,943 (0.95%) had a discrepancy difference; of these, 1163 (6.1%) had a user report. In total, 17 (0.09% of 18,943) ADC discrepancy PSEs had discrepancy user reports and appeared in the transaction logs. However, 1146 of 1163 (98.5%) discrepancy user reports did not have an associated PSE report. In addition, 1914 of 3077 of the identified discrepancy events (62.2%) did not have a discrepancy user report or a PSE report. Conclusion The study findings illustrate how PSE reports are only the tip of the iceberg, capturing less than 0.6% of possible ADC discrepancy events. This work can be leveraged to better understand both safety hazards and the effectiveness of interventions.