Designing a Better EMR to Combat Providers’ Medication Alert Fatigue

By Stephen Spotswood

INDIANAPOLIS, IN — While the medication alerts, automated reminders and warnings that pop up in electronic medical records (EMRs) improve patient care in theory, reality can be quite different: Alerts may be viewed as unhelpful noise by providers and rarely lead to medication changes.

An unidentified nurse at the North Chicago VA Medical Center in Illinois accesses a patient’s electronic medical record. (Photo fromNorth Chicago VA via DoD/VA Good News.)

That insight comes from a new study by VA researchers that offers a better understanding of how prescribers actually interact with the alert system and how the system can live up to its potential of helping prescribers cut down on medication errors. The study was published recently in the International Journal of Medical Informatics. 1

At the Richard Roudebush VA Medical Center in Indianapolis, researchers observed 320 naturally occurring alerts among 30 prescribers (20 primary-care and 10 specialty-clinic providers) and 146 patients.


pencil_white.jpgOpinion poll:
Do you usually respond to medication alerts on your facility’s electronic medical record?

Please click here to participate in this month’s U.S. Medicine readership poll.


Analysis of the data showed there are nine factors influencing how a prescriber interacts with a medication alert. Those are:

  • the logic of the system,
  • how redundant the system was,
  • the content of the alert,
  • how the alert was displayed,
  • cognitive factors, such as awareness and fatigue,
  • pharmaceutical knowledge,
  • medication management,
  • patient workflow, and
  • the reliability of the alert system.

Any or all of these factors could determine how a prescriber responds to an alert.  

For example, information provided or not provided in the alert was a big determinant of how the prescriber responded. Many alerts failed to adequately explain why they were triggered. Data collected from 21 of the 30 prescribers showed that lack of specificity in the alert was a barrier to interpreting and acting on it.

In some instances, the alert did not provide essential patient information, even though that information was available elsewhere in the EMR. Decision-making for some drug-interaction alerts is dependent on knowing patient lab data, which is not included in the alert. Some prescribers relied on memory, while others overrode the alert, went through with the prescription order, then went back afterward to validate their decisions.

Other alert triggers are inappropriate and warn against common treatments, such as duplicating the drug class, antiretrovirals. Patients are commonly on at least three such drugs as part of a cocktail, and duplication of the class is not only safe but standard practice. 

1 2

Share Your Thoughts