Drug-drug interaction reports generated by computers – just how valuable are they?

Australian researchers have recently examined the different types of adverse drug events (ADEs) that drug-drug interaction (DDI) alerts are configured to prevent in hospitalized patients. Using a retrospective cross-sectional design, the investigators analysed all DDI alerts presented to prescribers during a 1-month period. The outcome of interest was focused upon the type of ADE the alerts were trying to prevent. 1599 unique DDI alerts were generated for 715 patients, with the most common potential ADEs (not mutually exclusive) that prescribers were alerted to being QTc prolongation or torsades de pointes (n = 1028/1599, 64 %), and extrapyramidal symptoms or neuroleptic malignant syndrome (n = 463/1599, 29 %). Either of these two potential ADEs accounted for 83 % (n = 1329/1599) of unique DDI alerts. The researchers point to a well-known issue associated with alerts generated by computer-assisted order entry in medical systems: the majority of alerts are of low value and are ultimately overridden by prescribers. This, combined with the high number of repetitive alerts can contribute to alert fatigue. See the original research here