The Effect of a Prescription Drug Coverage Policy on Risk of Falls and Overdose in Older Adults


Although individual-level associations between injury and benzodiazepine use for older adults are well established, the effects of new coverage for benzodiazepines in 2013 on population-level injury trends in
older adults on Medicare Part D has not been studied. We hypothesize that the new coverage of
benzodiazepines will be associated with an increase in 1) the rate of fall-related injuries and 2) unintentional overdoses. To examine the impact of this policy, we will analyze OptumInsight claims data on 6.1 million Medicare Advantage enrollees between 2006 and 2015. Outcomes are fall-related injuries (e.g., fractures, head injuries) and unintentional overdoses based on diagnostic ICD-9 and E codes from healthcare claims.

The specific aims are: Aim 1. To examine the effect of Medicare coverage for benzodiazepines on the rate of fall-related injuries among individuals age 65+ enrolled in Medicare Advantage plans. Aim 2. To examine the effect of Medicare coverage for benzodiazepines on the rate of unintentional overdose among individuals age 65+ enrolled in Medicare Advantage plans. Sub-Aim 1. To examine change in risk of unintentional overdose at the time of the change in benzodiazepine coverage policy in the subset of patients prescribed opioid analgesics. If an effect of the coverage policy on injury is found, we will explore differential impacts of the policy across population subsets defined by geographic regions and indications for benzodiazepine prescription. If increased coverage for benzodiazepines is found to result in increased injury morbidity across population subsets, it will suggest that such expansions in access to potentially risky medications should consider incorporating risk mitigation strategies. If increased injury burden is largely due to rate increases among particular population subsets, this would suggest more nuanced coverage policies may provide a better balance or risks to benefits over the population. Alternatively, if no change in risk is associated with coverage, it will suggest that the insurer mechanism to reduce non-recommended use of controlled substances via exclusion in coverage may not be effective and interventions at the prescriber and/or patient level may have greater benefit.