EHR Nudges for Statin Prescribing: Not There Yet
Electronic health record-based clinical decision support (EHR CDS) alerts didn’t increase statin prescribing in a three-arm cluster-randomized trial.
Statin prescribing at guideline-recommended doses increased from 41.2% to 44.5% among cardiologists randomized to 6 months of pop-up alerts requiring action, but that was no better than the slight bump seen among controls receiving no intervention (from 41.8% to 42.6%, P=0.08 between groups).
However, for the subgroup of patients with clinical atherosclerotic cardiovascular disease (ASCVD), the active choice intervention did lead to significantly improved statin prescribing at the optimal dose (from 39.8% to 44.6%, P=0.008), reported Mitesh Patel, MD, MBA, of the University of Pennsylvania in Philadelphia, and colleagues.
Whether there is really a benefit for this subgroup of patients warrants further study, Patel’s group wrote in a manuscript published online in JAMA Cardiology.
The active choice intervention consisted of an unchanging pop-up alert in the EHR that required cardiologists to respond by accepting or declining guideline-directed statin therapy for patients deemed not at goal according to 2013 American Heart Association/American College of Cardiology or 2015 National Lipid Association guidelines.
A passive version of the intervention, in which doctors had to manually navigate within the EHR to access the same CDS alert, did not significantly improve statin prescribing at guideline-directed doses over controls, either (from 39.6% at baseline to 40.6%, P=0.86).
Neither the active nor passive intervention improved statin prescribing overall.
Participating cardiologists complained that the CDS notifications tested did not specify the indications that triggered the alert. The message was also confusing for some who thought the alert was saying the patient was not taking a statin at all, rather than that the dose could be increased, Patel and colleagues reported.
“Future interventions could increase general communications before and after the intervention is implemented, as well as within the alert, so that cardiologists are aware that some alerts are focused on increasing the dose of a statin that the patient is already taking,” they suggested.
Notably, 72.7% of cardiologists assigned to the passive choice intervention reported not even remembering seeing the alert over the 6 months of the study.
“Done well, CDS can greatly enhance personalization, efficiency, and effectiveness of care delivery,” according to Thomas Maddox, MD, MSc, of Healthcare Innovation Lab of BJC HealthCare/Washington University School of Medicine in St. Louis.
“However, CDS has yet to realize its potential. Indeed, clinicians have ignored up to 93% of CDS alerts and are frustrated by excessive rates of false-positive reminders. As a result, only two-thirds of CDS are associated with any measurable change in care, and those that do primarily affect process, rather than outcome, measures,” he wrote in his accompanying editorial.
“CDS interventions require codesign and iterative testing with their intended users. Speaking to clinicians’ needs to understand the clinical rationale behind any recommendations and tailoring them to individual patient characteristics are key. Appreciating the context in which clinicians work and incorporating CDS to accommodate those realities is essential,” he emphasized.
User-centered design principles undoubtedly require large upfront investment in CDS innovation but present the best path forward toward an effective, sustainable CDS intervention, according to the editorialist.
The investigators stressed the importance of improving statin use for prevention in cardiology patients.
“Statins have been demonstrated to reduce the risk of major adverse cardiovascular events. However, about half of patients meeting guideline criteria for a statin have not been prescribed one. Among patients prescribed a statin, about two-thirds are receiving a lower than optimal dose,” the authors cited from the literature.
Patel’s study included 16 cardiology practices in Pennsylvania and New Jersey with 82 participating cardiologists from the same health system. These included general, heart failure, and interventional cardiologists.
Randomization was performed at the individual-cardiologist level.
Investigators compared statin prescribing rates between the pre-intervention period (March 24 to Sept. 23, 2018) and the intervention period (Sept. 24, 2018, to March 23, 2019).
A total of 11,693 patients were included (mean age 63.8 years, 58% men, 66% white). More than two-thirds had an ASCVD clinical diagnosis. Their mean 10-year ASCVD risk score was 15.4%.
The study findings may have limited applicability to other health systems. Furthermore, the authors cautioned that they only had data on statin prescriptions without fill rates or adherence data and that there could have been some contamination of intervention among cardiologists randomly assigned to different intervention arms but working in the same practice.
Last Updated October 07, 2020
The study was supported by institutional funding.
Patel reported personal fees and other from Catalyst Health, HealthMine Services, and Holistic Industries and other support from Life.io.
Maddox disclosed holding a current NIH grant; previously receiving personal fees from NewYork-Presbyterian, Westchester Medical Center, Sentara Heart Hospital, the Henry Ford Health System, and the University of California San Diego; advising Myia Labs; and being a compensated director for the J. F. Maddox Foundation.