2D-04 - Outcomes Associated with Deprescribing Versus Continuing Bisphosphonates for Older Nursing Home Residents with Dementia

Background: Observational studies show that many older nursing home (NH) residents continue to receive bisphosphonates, despite low likelihood of continued benefit in the setting of reduced life expectancy, dementia severity, or reduced mobility.

Objectives: We conducted a longitudinal analysis to quantify the risks and benefits of discontinuing bisphosphonates in this population.

Methods: We conducted a longitudinal analysis of Medicare claims, Part D drug dispensing, and Minimum Data Set (MDS) assessments from 2016-2019. We identified long-stay (≥90 days) NH episodes (n=5,941) for adults ≥67 years old enrolled in Medicare Parts A, B, and D with dementia and a bisphosphonate dispensing overlapping the first 30 days of the episode . Discontinuation was defined as a 90-day gap in supply. Primary outcomes were hospitalizations for hip or osteoporotic fractures. Secondary outcomes were hospitalizations for severe esophagitis or gastritis (i.e., adverse effects). Residents were followed until the outcome of interest, discharge, death, or end of data. Covariates included demographics, fracture risk, mobility, bisphosphonate duration and adherence, prognosis (e.g., comorbidities, dementia severity), co-prescribed medications, provider characteristics, and NH and community characteristics (e.g., rurality). We constructed a longitudinal dataset to simulate a trial of discontinuing versus continuing bisphosphonates (i.e., target trial emulation). We conducted a series of sequential trials emulated at the start of each week for up to 2 years of follow-up. We employed a combination of inverse probability of treatment weighting (IPTW) and inverse probability of censoring weighting (IPCW) to estimate the effect of bisphosphonate discontinuation on outcomes while accounting for confounding and informative censoring, respectively. We report hazards ratios across all trials analogous to an intent-to-treat approach.

Results: Among 5,941 eligible NH episodes, the prevalence of discontinuing bisphosphonates was 36.3%. In target trial emulation, discontinuing bisphosphonates versus continuing was associated with a reduced likelihood of hip fracture (HR=0.56 [0.40-0.75]) or osteoporotic fractures (HR=0.50 [0.31-0.68]) over a median of 364 days of follow-up in weighted analyses. Discontinuation was also associated with a reduced likelihood of hospitalizations for severe esophagitis (HR=0.59 [0.27-0.95]) and gastritis (HR=0.62 [0.31-0.68]).

Conclusions: Our findings suggest that bisphosphonates are a reasonable target for deprescribing to reduce medication burden and risk for medication-related adverse effects, particularly among individuals living in the NH with dementia.