Electronic health record alerts: well-designed?
Three years ago on this blog, I commented on the importance of design in alerting systems, citing a study that indicated that the most important factor in alert acceptance was the quality of the display of the alert. This factor had an odds ration of 4.75, far outweighing the level of the alert (high, moderate or low risk), which had an odds ratio of 1.74.
Recently my colleague Shobha Phansalkar and her co-authors Zachariah, Seidling, Mendes, Volk, and Bates published a study evaluating how well medication alerts are designed in a cross section of electronic health record (EHR) systems. They used an instrument they had previously developed called I-MeDeSA to evaluate these different systems. This instrument consists of 26 questions covering 9 areas of alert design. Each question is scored as 0 (absent) or 1 (present), so the maximum possible score is 26. For example, one question is whether the prioritization of alerts is indicated appropriately by color. The study included 14 EHRs, consisting of 8 developed in-house at various healthcare institutions, and 6 commercial EHRs. The commercial EHRs included Cerner (two versions), Epic (two versions), GE, and NextGen. The article provided the complete list of the 14 EHRs, but otherwise the study results masked the identity of each EHR.
How well did the EHRs do? The average score was 13.6/26 (52.3%), with a range from 8/26 (30.8%) to 18.4/26 (70.8%). Four systems received a score of less than 13/26. Systems generally did well in the areas of alert visibility, alert placement, and the ability to access relevant information directly from the alert. Systems generally did poorly in the areas of transparent alert philosophy (i.e. describing how priority levels are assigned), indication of alert priority (i.e. using color, signal words or shapes) and learnability/confusability (e.g. visual characteristics to distinguish alert severities).
With an average score barely exceeding 50%, this study clearly indicates that more work is needed on the design of these systems. Alert fatigue continues to be a major issue in medication decision support systems, so it’s imperative for EHRs to review the designs of their alerting systems with respect to these human factors principles, and to make improvements accordingly. I would note however that some of the EHR versions studied were a few years old, so it may well be that if the study were repeated using the most recent version of each system, the average score would be higher. Nevertheless, it’s almost certain that additional improvements could and should still be made.
You may also like:
The latest pharmaceutical market data paint an interesting picture of the pharmacy landscape for the remainder of 2020, especially new data showing the impact of COVID-19. Industry watcher
Targeted cancer therapies, also called precision medicine or personalized medicine, involve tailoring therapy to the specific genetic variants present in a tumor. The
Filter by Type
Sign up to receive the latest news and information!