Low-Dose Trazodone, Benzodiazepines, and Fall-Related Injuries in Nursing Homes: A Matched-Cohort Study

Abstract

Objectives: To evaluate whether risk of fall-related injuries differs between nursing home (NH) residents newly dispensed low-dose trazodone and those newly dispensed benzodiazepines.

Design: Retrospective, matched cohort study in linked, population-based administrative data. Matching was based on propensity score ( ± 0.2 standard deviations of the score as a caliper), age ( ± 1 year), sex, frailty status, and history of dementia. The derived propensity score included demographic characteristics, clinical comorbidities, cognitive and functional status, and risk factors for falls.

Setting: All NHs in Ontario, Canada.

Participants: Propensity score-matched pairs of residents aged 66 and older who received a full clinical assessment between April 1, 2010, and March 31, 2015 (N=7,791).

Measurements: Hospitalization (emergency department visit or acute care admission) for a fall-related injury within 90 days of exposure. Subdistribution hazard functions accounted for competing risk of death. Sensitivity analyses were used to examine falls resulting in hip or wrist fracture only, as well as different lengths of follow-up at 30, 60, and 180 days.

Results: Cumulative incidence of a fall-related injury in the 90 days after index was 5.7% for low-dose trazodone users and 6.0% for benzodiazepine users (between-group change=-0.29, 95% confidence interval (CI)=-1.02-0.44]; hazard ratio=0.94, 95% CI=0.83-1.08). Findings were consistent across sensitivity analyses.

Conclusion: New use of low-dose trazodone was no safer with respect to a risk of a fall-related injury than new use of benzodiazepines. Additional studies to compare the effectiveness and risks of low-dose trazodone with those of a variety of psychotropic drug therapies are required in light of increasing trends in the use of trazodone in NHs.

Keywords: benzodiazepines; fall-related injuries; nursing home; pharmacoepidemiology; trazodone.