# A targeted analytical framework to optimize posthospitalization delirium pharmacotherapy in patients with Alzheimers disease and related dementias

> **NIH NIH R01** · BRIGHAM AND WOMEN'S HOSPITAL · 2024 · $854,063

## Abstract

Delirium (acute disturbance in mental status) occurs in 46-56% of persons living with dementia (PLWDs)
during hospitalization. Alzheimer’s disease and related dementias (ADRD) are among the strongest risk factors
for developing delirium during hospitalization. Although an off-label use, antipsychotic medications (APMs) are
the most commonly used pharmacotherapy to manage psychological symptoms of delirium. Because PLWDs
often have a prolonged recovery course from delirium due to acute illness, ~30% of the patients who newly
initiate an APM during hospitalization are discharged with them, and >60% of those discharged with an APM
persist for >6 weeks. Since APMs may cause numerous life-threatening adverse reactions, it is critical to
discontinue them after hospitalization in a timely fashion. However, several critical knowledge gaps limit the
necessary evidence generation to guide such a deprescribing process: 1) There is currently no direct data from
randomized control trials (RCT) on discontinuation of APMs used for delirium because it is extremely difficult to
recruit and consent PLWDs or their healthcare proxies when the patient is in an acute delirious state to
participate in an RCT, and any interventional study would severely underrepresent frail PLWDs seen in routine
care. 2) In the non-randomized settings, adjusting for confounding is challenging when comparing different
deprescribing strategies of a medication used for acute delirium, and the detailed clinical information required
for such analyses is not typically available in routine care data. Our objective is to establish an analytical
framework that enables valid causal effect estimation comparing continuation and multiple deprescribing
strategies (e.g., abrupt discontinuation vs. gradual dose reduction) of APMs in PLWDs with delirium after
hospitalization. We will integrate electronic health records (EHR), national claims data, and multiple clinical
assessment data, covering >502,000 PLWDs from 2013 to 2026, and employ high-dimensional machine-
learning aided confounding adjustment and phenotyping algorithms. Our specific aims include 1) To integrate
EHR with Medicare claims data, Minimum Data Set (MDS), Outcomes and Assessment Information Set
(OASIS), and Inpatient Rehabilitation Facility Patient Assessment Instrument (IRF-PAI) and to develop novel
algorithms to determine key clinical phenotypes; 2) To assess APM utilization/discontinuation patterns and risk
factors of prolonged use of APMs for delirium in PLWDs after hospitalization; 3) To assess the health impact of
different discontinuation strategies (considering the amount and rate of dose reduction) of APMs vs. continuing
APMs in PLWDs with delirium after hospitalization. The subgroup effects by key clinical phenotypes, typical vs.
atypical APMs, and type of admission will also be determined. This proposal will generate evidence reflecting
routine care delivery to inform post-discharge APM management in PLWDs with delirium. It...

## Key facts

- **NIH application ID:** 10892936
- **Project number:** 5R01AG081412-02
- **Recipient organization:** BRIGHAM AND WOMEN'S HOSPITAL
- **Principal Investigator:** JOSHUA K LIN
- **Activity code:** R01 (R01, R21, SBIR, etc.)
- **Funding institute:** NIH
- **Fiscal year:** 2024
- **Award amount:** $854,063
- **Award type:** 5
- **Project period:** 2023-08-01 → 2027-04-30

## Primary source

NIH RePORTER: https://reporter.nih.gov/project-details/10892936

## Citation

> US National Institutes of Health, RePORTER application 10892936, A targeted analytical framework to optimize posthospitalization delirium pharmacotherapy in patients with Alzheimers disease and related dementias (5R01AG081412-02). Retrieved via AI Analytics 2026-05-23 from https://api.ai-analytics.org/grant/nih/10892936. Licensed CC0.

---

*[NIH grants dataset](/datasets/nih-grants) · CC0 1.0*