Trajectories of Medication Use Prior to Death in US Nursing Home Residents With Dementia

Polypharmacy and questionably beneficial medication use are common among nursing home (NH) residents with dementia. Deprescribing patterns in this population are not well described. This study describes trajectories of medication use in NH residents with dementia in the year prior to death.

Repeated cross-sectional analysis.

US NH decedents with dementia.

Trajectories of medication use at 90-day intervals in the year before death using the 2014 to 2018 Minimum Data Set 3.0 (MDS 3.0). Outcomes are based on a prespecified MDS 3.0 checklist of medications, including select chronic disease medications (anticoagulants and diuretics), psychoactive medications (antipsychotics, anxiolytics, antidepressants, and sedative hypnotics), and antibiotics. Chi-square test of trend examined medication use trajectories. Multivariable logistic models identified factors associated with (1) overall deprescribing, (2) deprescribing, and (3) initiation of each medication class in the 3 months prior to death.

A total of 1.12 million decedents with dementia were included in the study. Comparing medication use 12 months prior to death to medication use in the MDS 3.0 assessment closest to death, there were significant trends in medication use for all drug classes, including increases in anticoagulants, anxiolytics, and antibiotics, and decreases in antidepressants, diuretics, antipsychotics, and hypnotics. In multivariable analysis, factors associated with deprescribing of at least 1 drug prior to death included greater functional impairment (severe, adjusted odds ratio [aOR], 1.48; 95% CI, 1.39-1.58; moderate, aOR, 1.11; 95% CI, 1.04-1.18), worse cognitive impairment (severe, aOR, 1.25; 95% CI, 1.23-1.28; moderate, aOR, 1.08; 95% CI, 1.05-1.11), and hospice enrollment for dementia (aOR, 1.22; 95% CI, 1.19-1.25) or for nondementia diagnosis (aOR, 1.29; 95% CI, 1.26-1.32) vs no hospice.

Increases in chronic disease medications such as anticoagulants in the year prior to death among NH decedents with dementia suggest that further work is needed to guide end-of-life prescribing.