David Geffen School of Medicine at UCLA
Behaviorally-Designed EHR Alerts to Reduce Antipsychotic Prescribing in Dementia: A Confirmatory Pragmatic Trial
Health Care System Partner
UCLA Health
Dr. Mafi is an associate professor of medicine in the Division of General Internal Medicine and Health Services Research at the David Geffen School of Medicine at UCLA and an affiliated adjunct physician policy researcher at the RAND Corporation. He is a practicing general internist and health services researcher whose program of research aims to test innovations and technologies that can improve the value and safety of health care for older adults. He also attends the UCLA Santa Monica Geriatrics Teaching Service. Dr. Mafi’s work focuses on designing and rigorously testing scalable, low-cost interventions, particularly behaviorally informed electronic health record (EHR) nudges and emerging artificial intelligence tools, through embedded pragmatic, randomized trials conducted in real-world health systems. This award will provide the opportunity to conduct a fully powered, confirmatory embedded pragmatic clinical trial of a behaviorally designed EHR alert to reduce inappropriate antipsychotic prescribing to people living with dementia (PLWD), generating the definitive evidence needed to disseminate a low-cost intervention, built in Epic, to the more than half of U.S. health systems that share this platform.
RATIONALE: One in four older Americans with dementia receive psychoactive medications that are often prescribed inappropriately and can have harmful adverse effects (e.g., falls, strokes, hospitalizations, and death). Exploratory post-hoc analysis of our group’s prior embedded pragmatic clinical trial (ePCT) found a behaviorally designed EHR alert intervention embedded in the UCLA Health system, reduced antipsychotic prescribing for PLWD among the sub-group of clinicians who were the highest prescribers. The next step is to conduct a fully-powered ePCT trial intentionally designed to confirm this finding.
OBJECTIVE: To conduct a clinician-randomized ePCT at UCLA Health to evaluate the effectiveness of a behaviorally designed EHR alert to reduce antipsychotic prescribing to community-dwelling PLWD by high-prescribing clinicians.
SETTING: UCLA Health, an integrated academic health system serving Los Angeles County, California, with a single Epic EHR across all its sites of care.
POPULATION Approximately 220 UCLA Health clinicians who prescribe antipsychotics to PLWD, randomized 1:1, with the primary outcome analysis examined in the top tertile of baseline prescribers (n≈73).
INTERVENTION: When a clinician in the intervention arm initiates a new antipsychotic prescription for a PLWD, an Epic-based EHR alert does the following: 1) reminds the clinician that antipsychotics increase mortality risk in dementia; 2) offers an evidence-based handout for care partners to manage behaviors using non-pharmacological approaches; and 3) defaults the order to the lowest dose and shortest supply, which the clinician can change.
OUTCOMES: The primary clinical outcome is cumulative antipsychotic pill-days prescribed per clinician over 18 months among high-prescribing clinicians. Secondary clinical outcomes are pill-days among all randomized clinicians, patient safety events within 90 days of prescribing (emergency department visits, hospitalizations, falls, agitation or psychosis, and mortality), and substitution of other psychoactive medications.
IMPACT: If successful, this ePCT will establish that a low-cost, behaviorally designed EHR alert can reduce antipsychotic prescribing to PLWD by the high-prescribing clinicians. As the alert is integrated into Epic, used by more than half of U.S. health systems, it has the potential for rapid national dissemination, and to reduce harmful effects of antipsychotics among millions of community-dwelling Americans with dementia.