Benzodiazepine, Antipsychotic Use Linked to Increased Mortality in Hospice Patients With Dementia

Lauren Gerlach, DO

Courtesy of the University of Michigan

A national case-control study has found that initiating benzodiazepines or antipsychotics in nursing home residents with Alzheimer disease and related dementias (ADRD) enrolled in hospice is associated with significantly increased 180-day mortality. The findings, published in JAMA Network Open, raise important questions about prescribing practices for this vulnerable population.

Researchers analyzed Medicare claims and Minimum Data Set assessments from July 1, 2014, through September 30, 2018, examining 139 103 long-term nursing home residents with ADRD newly enrolled in hospice. The cohort was 75.8% women with a mean age of 87.6 years. Participants had no benzodiazepine or antipsychotic use in the 6 months prior to hospice enrollment.

Using 1:1 matching, investigators created 2 cohorts: 26 872 matched pairs for benzodiazepine analysis and 10 240 matched pairs for antipsychotic analysis. Matching criteria included enrollment timing, age, sex, comorbidity burden, cognitive function, and baseline central nervous system medication use.

Key Findings

The study revealed striking associations between medication initiation and mortality risk:

Benzodiazepines: Initiation was associated with a 41% increased hazard of death at 180 days compared with nonuse (hazard ratio [HR], 1.41; 95% CI, 1.38-1.44). Among benzodiazepine initiators, 73.58% died within 180 days compared with 58.3% of matched nonusers.

Antipsychotics: Initiation was associated with a 16% increased hazard of death at 180 days (HR, 1.16; 95% CI, 1.12-1.20). Mortality occurred in 70.7% of antipsychotic initiators versus 63.3% of nonusers.

The median time to medication initiation following hospice enrollment was 3 days for both drug classes (interquartile range, 1-23 days for benzodiazepines and 1-32 days for antipsychotics).

The associations persisted across multiple sensitivity analyses. Propensity score–weighted models showed even stronger associations (benzodiazepine HR, 1.92; 95% CI, 1.88-1.96; antipsychotic HR, 1.86; 95% CI, 1.79-1.94).

When restricted to patients with ADRD as the primary hospice-qualifying diagnosis, findings remained consistent (benzodiazepine HR, 1.31; 95% CI, 1.27-1.36; antipsychotic HR, 1.20; 95% CI, 1.13-1.27).

Cumulative exposure analysis demonstrated that each additional prescription fill increased 180-day mortality for both benzodiazepines (HR, 1.15; 95% CI, 1.12-1.18) and antipsychotics (HR, 1.06; 95% CI, 1.01-1.10).

Nearly half (47.8%) of the at-risk cohort initiated benzodiazepine use during hospice, while 13.4% initiated antipsychotic use. The lower antipsychotic prescribing likely reflects nursing home policies aimed at reducing antipsychotic use in this setting.

The mean hospice length of stay exceeded 130 days, underscoring the challenge of prognostication in ADRD. Unlike cancer, ADRD follows a prolonged, unpredictable trajectory, with approximately 20% of hospice enrollees with ADRD outliving the 6-month prognosis required for eligibility.

Most participants had severe cognitive impairment, high comorbidity burden, and functional dependency. The mean Gagne comorbidity score was 4.8, and 39.7% to 43.7% of participants had the most severe functional impairment (Minimum Data Set Activities of Daily Living Scale score 13-16).

Implications for Practice

The findings highlight several critical issues for hospice prescribing:

Risk-benefit considerations: While these medications may provide symptom relief for agitation, anxiety, and terminal delirium, their use carries substantial risks including sedation, falls, confusion, and—as this study demonstrates—increased mortality risk.

Timing matters: For patients earlier in the hospice trajectory who may have months to live, sedating effects may compromise quality of life, impair communication, and interfere with functional independence when patients and families may prioritize alertness and interaction.

Need for guidelines: Benzodiazepine prescribing rates in hospice range from 12% to 80%, and antipsychotic prescribing rates range from 6% to 62% across agencies, even after adjusting for patient characteristics. This variation suggests that agency practice norms, rather than clinical presentation, may drive prescribing decisions.

Study Limitations

The investigators acknowledged several limitations. Residual confounding from unmeasured factors such as symptom severity remains possible despite extensive matching. Prescription claims may not fully capture actual medication use, and hospice claims files do not contain days supplied. The findings are limited to Medicare fee-for-service beneficiaries in nursing homes and may not generalize to other settings. The data reflect prescribing practices from 2014 through 2018, predating more recent trends. Finally, the study examined all-cause mortality but did not measure symptom burden or quality of life.

The authors emphasized the need for dementia-specific hospice prescribing guidelines, investment in nonpharmacologic alternatives, and Medicare policy reform to track and inform safe medication use in hospice. From 2014 through 2018, Medicare briefly required hospice agencies to report medications billed to the hospice benefit, but that program has expired, leaving hospice medication use unmonitored.

"While these medications may provide symptom relief in appropriate clinical scenarios, their use is associated with substantial risks," the authors concluded. "These findings highlight the need for careful prescribing decisions and the development of dementia-specific hospice prescribing guidelines."

References:

1. Gerlach LB, Zhang L, Kim HM, Teno J, Maust DT. Benzodiazepine or Antipsychotic Use and Mortality Risk Among Patients With Dementia in Hospice Care. JAMA Netw Open. 2025 Oct 1;8(10):e2537551. doi: 10.1001/jamanetworkopen.2025.37551