Risk of Fractures Requiring Hospitalization After an Initial Prescription for Zolpidem, Alprazolam, Lorazepam, or Diazepam in Older Adults

OBJECTIVES: To determine whether zolpidem is a safer alternative to benzodiazepines. DESIGN: Retrospective cohort study. SETTING: Community based. PARTICIPANTS: Health maintenance organization members with an initial prescription for zolpidem (n = 43,343), alprazolam (n = 103,790), lorazepam (n = 150,858), or diazepam (n = 93,618). MEASUREMENTS: Zolpidem and benzodiazepine prescriptions were identified from pharmacy databases. Rates of nonvertebral fractures and hip fractures requiring hospitalization were compared before and after an initial prescription for each treatment, adjusting for confounders using doubly robust estimation. RESULTS: In patients aged 65 and older, the rates of non-vertebral fractures and dislocations were similar in the pretreatment intervals. The rate ratios (RRs) for the 90-day posttreatment interval relative to the pretreatment interval were 2.55 (95% confidence interval (CI) = 1.78-3.65; P < .001) for zolpidem, 1.14 (95% CI = 0.80-1.64; P = .42) for alprazolam, 1.53 (95% CI = 1.23-1.91; P < .001) for lorazepam, and 1.97 (95% CI = 1.22-3.18; P = .01) for diazepam. The ratio of RRs (RRR)-the RR in the posttreatment period adjusted for the corresponding RR in the pretreatment period-were 2.23 (95% CI = 1.36-3.66; P = .006) for zolpidem relative to alprazolam, 1.68 (95% CI = 1.12-2.53; P = .02) for zolpidem relative to lorazepam, and 1.29 (95% CI = 0.72-2.30; P = .32) for zolpidem relative to diazepam. The RRs decreased with time from the initial prescription (trend P < .001), as would be expected if the association is causal. CONCLUSION: In older adults, the risk of injury with zolpidem exceeded that with alprazolam and lorazepam and was similar to that with diazepam. If the associations are causal, then the high incidence of these fractures implies that these treatment induce a substantial number of fractures and consequential costs. Further study of the association is imperative. J Am Geriatr Soc 59:1883-1890, 2011.