Overrides of medication-related clinical decision support alerts in outpatients
This is a review of the 2014 article by Nanji et al, Overrides of medication-related clinical decision support alerts in outpatients.
A large incidence of alert overrides in a computerized provider order entry medication clinical decision support (CDS) prompted the authors to investigate the reason and appropriateness of the overrides. Both the electronic health record (EHR) and CDS were internally developed, and had ordering alerts for:
- Patient allergies
- Drug-drug interactions
- Age-based recommendations
- Renal recommendations
- Formulary substitutions
- Duplicate drug
- Drug-class interaction
- Class-class interaction
The incidence of CDS overrides was obtained from the EHR’s audit logs. The appropriateness of the overrides was determined based on criteria created by four clinicians: a physician, nurse, and two pharmacists. A sample of 600 alert overrides from a three-year time period was evaluated.
Over three years, out of 2,004,069 medication orders, 157483 CDS alerts were received and 82,889 (52.6%) of these alerts were overridden. 53% of the overrides were appropriate, and they were in the categories of patient allergy, drug-class, duplicate drug, and drug formulary. The inappropriate overrides were in the categories of renal recommendations, drug-drug interactions, and age-based recommendations. In a different study to check the incidence of delirium in cognitively impaired older adults, alerts were overridden by ICU staff to the point that "Alert Fatigue" occurred. The link to that study is "Clinical decision support system and incidence of delirium in cognitively impaired older adults transferred to intensive care".
The authors concluded that the CDS needed to be evaluated to make it relevant and to decrease alert fatigue. They also surmised that education needed to be given to clinicians on the categories where the alert overrides were inappropriate. Also, a study done by Khan BA et al. contended that the interruptive alerts were best handled if presented to ICU nurses or allied health workers and not to the physicians. They also opined that added human intelligence along with the CDSS alerts would make a difference,rather than the CDSS alerts alone. Please check my article review: 
I found this article to be very informative as it quantified and classified the CDS alert overrides, which would result in a targeted modification of clinical decision support, and at the same time produce basis for intervention in the form of education for clinicians. It highlighted one of the benefits of an EHR: having audit logs that can be mined for research geared towards the improvement of the safety of healthcare delivery.
- Nanji, K. C., Slight, S. P., Seger, D. L., Cho, I., Fiskio, J. M., Redden, L. M., . . . Bates, D. W. Overrides of medication-related clinical decision support alerts in outpatients. doi: 10.1136/amiajnl-2013-001813.t</http://jamia.oxfordjournals.org/content/21/3/487.short