RESUMO
BACKGROUND: Although polydrug incidents comprise a substantial proportion of overdose deaths, scholarly and popular focus has centered on prescription opiates. This study examines the role of benzodiazepine and opioid prescriptions on overdose-both individually and synergistically-using data from Medicare Part D, a source of prescription drug claims for about 35 million Americans. METHODS: Prescribing data from the Medicare Part D Public Use Files for 2013, 2014, and 2015 (approximately 3.5 billion prescription drug claims) are geolocated using the prescriber's national provider identifier to calculate the proportion of claims for opioids and benzodiazepines in each county. These rates are matched with overdose data and controls to compile an analytic dataset of 9105 county-years. Multinomial logistic regression is used to estimate the probability that a county experiences higher rates of overdose fatalities. RESULTS: A 1% increase in the benzodiazepine proportion of claims is associated with 1.2 odds of high, versus low, overdose (P < .1) and 1.4 odds of very high overdose (P < .05). Moreover, there was a substantial interaction between opioids and benzodiazepines (P < .001). A county with 6% benzodiazepine prescriptions and 12% opioid prescriptions has a .58 predicted probability of very high overdose, significantly higher (P < .001) than the .33 probability for a county with 12% opioid prescriptions but 3% benzodiazepine prescriptions. CONCLUSION: These findings shed light on the polydrug epidemiology of the overdose epidemic. Overdose deaths are highest where elevated opioid and benzodiazepine claims coexist. Overdose levels may reflect polydrug use and misuse, requiring clinical and policy responses beyond reducing opioid prescriptions.
