Late Breaking News
Designing a Better EMR to Combat Providers' Medication Alert Fatigue Cont
Designing A Better System
The researchers include a list of recommended actions, such as:
- giving prescribers options for what point in the process they receive the alert;
- indicating how well each alert is supported by the literature;
- developing a standardized rating for risk;
- displaying essential EMR data in the alert; and
- eliminating the use of technical, programmer-specific and pharmacist-specific language.
The last recommendation stems from the fact that the medication alert system evolved out of a pharmacist-specific scenario and was never adapted for the nurse practitioners and physicians who commonly use EMRs.
“The general progression, not specific to VA, seems to be that alert systems were developed for pharmacists and used in pharmacies as medications were dispensed for patients,” explained Alissa Russ, PhD, lead researcher on the study. “These alert systems were created using medication databases and terminology geared toward pharmacists. As computerized provider order entry systems were developed, the data systems used in pharmacies were transferred over to EMRs. As a result, the EMR alert systems do not necessarily account for the varying training and expertise of physicians and nurse practitioners. A good system would have alerts that support this range of expertise.”
Such a system also would have fewer, more easily understandable alerts, although how that could be implemented is uncertain.
“The ideal strategy to reduce the number of alerts is still unknown,” Russ said. “Participants in the study offered some ideas, including a smarter alert system that can remember the patient and previous alerts, coordinating with pharmacists to turn off individual alerts that are appropriately overridden and suppressing alerts under specific circumstances. But these ideas have not been tested.”
While this study provided valuable insights into the interaction between alerts and providers, what is needed now are evidence-based design recommendations for how to make the system better. Russ and her team are wrapping up a simulation study in which they have tested experimental designs for medication alerts based on their previous findings, as well as input from an expert VA advisory panel.
“Twenty prescribers were recruited for this simulation study, and redesigned alerts were compared against the VA alert system in the VA Health Services Research and Development Human-Computer Interaction Laboratory,” Russ said. “The simulation study has focused on the display format of alerts, how to increase the salience or noticeability of alerts and how to better integrate alerts into clinical workflow.”
Results from the simulation will be shared with the advisory panel, as well as other VA leaders, Russ said.
- Russ AL, Zillich AJ, McManus MS, Doebbeling BN, Saleem JJ. Prescribers' interactions with medication alerts at the point of prescribing: A multi-method, in situ investigation of the human-computer interaction. Int J Med Inform. 2012Apr;81(4):232-43. Epub 2012 Jan 31. PubMed PMID: 22296761.