PROJECT SUMMARY
Antipsychotics are frequently prescribed to older adults with Alzheimer’s disease and related dementias (ADRD)
to treat behavioral symptoms, but clinical guidelines suggest their use should be limited in this population due to
the possibility of adverse side effects. Recently, interventions have harnessed tools of behavioral economics
such as “nudges” through letters and e-mails to raise prescribing quality. Organizations can deploy these
interventions quickly, cheaply, and at scale. However, little evidence exists on using this work to address
antipsychotic prescribing to patients with ADRD in spite of the enormous potential value it could have in this
context. Moreover, even where evidence exists that interventions successfully cause “deprescribing”, given the
possibilities for patient harms from indiscriminate cutbacks, these interventions merit careful evaluation.
Our proposed study seeks to fill this evidence gap. We will analyze a novel trial conducted by the Centers for
Medicare and Medicaid Services that randomized over 5,000 high-volume primary care physician prescribers of
quetiapine, the most commonly used antipsychotic in the U.S., to overprescribing letters informed by behavioral
economics. The letters included a peer comparison “nudge” and an overuse warning, and they reduced new
initiations of quetiapine by 24% over two years. Our work focuses on the effects of these letters on clinical quality,
quality of life, and health care use of dementia patients of these physicians. The physicians had over 40,000
patients with ADRD, a large cohort that will facilitate detection of both benefits and harms of deprescribing. We
will exploit rich Medicare data, including nursing home assessments and health care claims, and will study effects
for up to 5 years.
We will proceed in three steps. In Aim 1, we will analyze effects of the intervention on ADRD patients of study
prescribers who live in nursing homes, a particularly vulnerable population for whom extensive data on patient-
centered outcomes is available. Through this data, we will closely track clinical and quality of life outcomes,
including cognitive function, behavioral symptoms, and depression. In Aim 2, we will broaden our analyses to
include dementia patients who live in the community, using claims data to track effects on health outcomes such
as hyperlipidemia and stroke and health care use such as inpatient hospitalizations and primary care visits.
Finally, in Aim 3 we will test whether effects of the letters differ for patients based on care setting,
sociodemographic group, or nursing home quality; these results will show whether letters address or expand
gaps in care quality and will determine which groups, if any, are appropriate for similar interventions in the future.
Taken together, our results will show whether simple, behaviorally informed deprescribing efforts can encourage
guideline-concordant care and improve outcomes for patients with dementia, or whether such interventions lead
to unintended harms. Our findings will guide future policy on deploying low-cost and scalable interventions to
improve the quality of care for older adults with ADRD.