Initiation of benzodiazepine or antipsychotic therapy during hospice enrollment was associated with increased 180-day mortality among patients with Alzheimer disease and related dementias, according to a recent study.
Specifically, benzodiazepine initiation was linked to a 41% higher hazard of death, and antipsychotic initiation was associated with a 16% higher hazard. These associations persisted in propensity score–weighted models, which demonstrated hazard ratios of 1.92 for benzodiazepines and 1.86 for antipsychotics. A dose-response pattern was also observed: each additional filled prescription further increased mortality risk by hazard ratios of 1.15 with benzodiazepines and 1.06 with antipsychotics.
These findings were derived from a large national case-control study using Medicare claims linked to Minimum Data Set assessments. Led by Lauren B. Gerlach, DO, MS, of the Department of Psychiatry at the University of Michigan Medical School and the Institute for Healthcare Policy and Innovation at the University of Michigan in Ann Arbor, the researchers identified 139,103 long-stay nursing home residents who were enrolled in hospice between 2014 and 2018 and had no benzodiazepine or antipsychotic exposure in the prior 6 months. A new-user design was applied, and incident users were matched 1:1 with nonusers on hospice enrollment timing, age, sex, comorbidity burden, cognitive function, and baseline central nervous system–active medication use. All-cause mortality within 180 days of hospice enrollment served as the primary outcome, and was estimated using Cox proportional hazards regression models stratified by matched sets and adjusted for demographic, clinical, and hospice-level characteristics.
Of 100,058 residents who were at risk for benzodiazepine exposure, 47,791 initiated therapy, while 15,314 of the 114,933 residents at risk for antipsychotic exposure initiated treatment. “While this study was not designed to estimate national prevalence, it is worth noting that nearly half of the cohort initiated use of a benzodiazepine (47.8%), and 13.4% initiated use of an antipsychotic during hospice,” noted Dr. Gerlach and colleagues. After matching, the benzodiazepine cohort included 26,872 pairs and the antipsychotic cohort included 10,240 pairs. The mean age across matched cohorts was 89 years, and most residents exhibited severe cognitive impairment and substantial comorbidity. Medication initiation occurred early in the hospice course, with a median of 3 days from enrollment for both drug classes.
The authors added that prescribing practices for patients with Alzheimer disease and related dementias in hospice settings showed substantial variation. Benzodiazepine use ranged from 12% to 80% and antipsychotic use ranged from 6% to 62% across agencies, even after adjustment for patient characteristics. These differences appeared to reflect agency-level norms rather than clinical presentation, and clinicians reported divergent views regarding the appropriateness of these medications. Because nearly 20% of patients with Alzheimer disease and related dementias live beyond the 6-month hospice eligibility prognosis and sedating agents may compromise goals related to cognition and function, standardized prescribing guidance and broader access to nonpharmacologic interventions may support more consistent, safer care, the authors suggested. Further, the expiration of Medicare’s 2014 to 2018 hospice medication reporting requirement has left prescribing unmonitored and highlights the need for renewed efforts to track medication use, as well as guide evidence-based practice, inform policy, and support patient safety.
The study’s findings should be considered in the context of several limitations, including potential unmeasured confounding, reliance on prescription claims that may not represent actual medication use, restriction to Medicare fee-for-service nursing home residents, and the use of data from 2014 through 2018. The analysis also did not assess symptom burden or quality of life, both of which are important outcomes in hospice care.
Donovan T. Maust, MD, MS, reported equity in Predictably Human and compensation from the American Society for Addiction Medicine unrelated to this study, and no other conflicts were noted.
Source: JAMA Network Open
