(A-154) Predicting Response to Mono and Dual Antidementia Pharmacotherapy Among Nursing Home Residents

Background: Half of nursing home (NH) residents have a diagnosis of Alzheimer’s disease (AD) or related dementia (RD). Cholinesterase inhibitors (ChEIs) and memantine have been widely used for the symptomatic treatment of ADRD for decades. Evidence is lacking on heterogeneity of treatment effects.

Objectives: To identify characteristics of NH residents associated with favorable outcomes under different treatment scenarios.

Methods: Using 2011-2018 Minimum Data Set (MDS) and Medicare Parts A and D, we selected a new-user cohort of US NH residents with AD or selected RD diagnoses and Medicare fee-for-service enrollment. New-users titrated to a therapeutic dosage as of day 90 (index) were assigned to ChEI, memantine, or dual treatment groups. Day 90 after a random MDS assessment was nonusers’ index date. Effectiveness outcomes (higher scores are worse) included 1-unit reductions (minimum clinically important difference) in the activities of daily living (ADL) score (range 0-28) and the cognitive function scale (CFS; range 0-3). Residents with scores of zero were excluded. Resident sociodemographic and clinical characteristics at baseline (120 days pre-index) were selected using logistic regression with backwards selection (p < 0.05) starting with all two-way interactions with treatment. With these selected variables, modified Poisson models with robust standard errors were fitted to report adjusted relative risk (aRR).

Results: We studied 538,936 non-users compared to 31,348 ChEI, 11,133 memantine, and 5,927 dual users. The median age ranged from 84-87 years across groups; 67-73% were women and 25% had an AD diagnosis.

Across treatment groups, less time from NH admission and more baseline ADL limitations were associated with a higher probability of ADL improvement with treatment, while more severe baseline cognitive impairment and more extensive ADL limitations were each associated with a higher probability of CFS improvement with treatment. Several other interactions between diagnoses (e.g., AD, diabetes), medications (e.g., count, antipsychotics) and clinical characteristics (e.g., weight loss, ulcers) with treatment were significantly associated with outcomes in one or more models, which may inform shared decision-making. For example, although ChEIs were associated with ADL improvement overall (aRR: 1.19; 95% CI: 1.17-1.21), the ChEI aRR compared to non-users was 0.99 (95% CI: 0.80-1.21) for a resident with a pressure ulcer who resided in the NH for 500 days.

Conclusions: The benefits of anti-dementia medication regimens varied based upon resident characteristics. We are developing algorithms to make individualized predictions based upon such characteristics to facilitate shared decision-making by the care team for NH residents with dementia.