Dr. Lesser supports our conclusion of the possible lack of a relationship between the use of benzodiazepines and hip fractures in the elderly with additional data from the published literature. Since the landmark studies of 1987 (1) and 1989 (2), results of research on the benzodiazepine–hip fracture relationship have become increasingly contradictory. We concur with Dr. Lesser that differences in results are probably due to study design issues, such as benzodiazepine exposure misclassification in prospective cohort studies (3) and imperfect control for potential confounders in large, claims data–based, case–control studies (4). On the basis of these results and our study, we believe that broad-based policies, such as the payment restrictions encompassed in Medicare Part D, which are in part based on these controversial results, are misdirected.

The ideal randomized, controlled study of the benzodiazepine–hip fracture relationship is unlikely to be conducted, for at least 2 reasons. First, benzodiazepines are inexpensive drugs that have been on the market for a long time, making funding of a costly randomized, controlled trial unlikely. Second, benzodiazepines have proven efficacy; therefore, the equipoise principle required for randomly assigning patients would not be met.

Without a randomized trial, quasi-experimental studies like ours are the best research design option to assess the relationship. Potential confounders of a longitudinal quasi-experimental study would need to be related to the outcome of interest (hip fractures) and to happen at the same time as the policy change that gave rise to the quasi-experiment. We believe that no such confounders could have explained our finding of stable rates of hip fractures at a time when benzodiazepine use suddenly declined by about 60% after a statewide policy restricting access to the drugs.

However, we also believe that clinicians should evaluate risks and benefits of all medications, including benzodiazepines, on the basis of each patient's unique clinical circumstances and caution against overinterpreting the findings of our study in applying them to individual patient clinical decision making.