BACKGROUND
Medicare Part D excludes benzodiazepines from coverage, and numerous state government policies limit use of benzodiazepines. No data indicate that such policies have decreased the incidence of hip fracture.
OBJECTIVE
To assess whether a statewide policy that decreased the use of benzodiazepines among elderly persons by more than 50% has decreased the incidence of hip fracture.
DESIGN
A quasi-experiment comparing changes in outcomes before and after a policy change in a study U.S. state (New York) and a control state (New Jersey).
SETTING
Two U.S. state Medicaid programs, 1988-1990.
PATIENTS
Medicaid enrollees in New York (n = 51 529) and New Jersey (n = 42 029) who received or did not receive a benzodiazepine.
MEASUREMENTS
Benzodiazepine prescribing and hazard ratios for hip fracture, adjusted for age and eligibility category.
INTERVENTION
A statewide policy, implemented in New York in 1989, that required triplicate forms for benzodiazepine prescribing to allow surveillance by health authorities.
RESULTS
The triplicate prescription policy immediately resulted in a 60.3% (95% CI, -66.3% to -54.2%) reduction in benzodiazepine use among women and 58.5% (-64.3% to -52.8%) among men. Benzodiazepine use in New Jersey remained stable. Hazard ratios for hip fracture that were adjusted for age and eligibility category did not change in New York or New Jersey when the periods before and after use of the triplicate prescription policy were compared (change from 1.2 to 1.1 among female benzodiazepine recipients [P = 0.70], 1.3 to 1.1 [P = 0.08] among female nonrecipients, 0.8 to 1.1 [P = 0.56] among male recipients, and 1.1 to 1.3 [P = 0.46] among male nonrecipients).
LIMITATIONS
Information was lacking on race, benzodiazepine dose, and other potential determinants of continued benzodiazepine prescribing.
CONCLUSIONS
Policies that lead to substantial reductions in the use of benzodiazepines among elderly persons do not necessarily lead to decreased incidence of hip fracture. Limitations on coverage of benzodiazepines under Medicare Part D may not achieve this widely assumed clinical benefit.