• Inmaculada Hernandez, Meiqi He, and Yuting Zhang.
• Department of Pharmacy and Therapeutics, School of Pharmacy, University of Pittsburgh, 3501 Terrace St, Pittsburgh, PA 15261, USA. Electronic address: inh3@pitt.edu.
• Drug Alcohol Depend. 2018 Oct 1; 191: 141-144.

BackgroundConcurrent opioid and benzodiazepine use is associated with a three-fold increase in the risk of opioid-related overdose. No study has evaluated geographic variation in the concurrent use of opioids and benzodiazepines in US Medicare. We compared state, hospital-referral region (HRR), and county-level variation in concurrent opioid and benzodiazepine use among US Medicare opioid users and examined the heterogeneity in concurrent use within states.MethodsUsing 2013-2014 US Medicare Part D claims, we identified non-cancer beneficiaries who used opioids in 2014 (n = 268,678). The outcome was concurrent opioid and benzodiazepine use. We constructed logistic regression models to isolate state, HRR, and county-level variation not explained by patient characteristics, and evaluated how county and HRR quintiles are distributed within state quintiles.ResultsThe adjusted probability of concurrent use ranged from 16.7%-29.6% across states, 12.1%-37.0% across HRRs, and 0%-65.2% across counties. State-level variation masks substantial county-level variation: only 18% of counties located in the lowest state quintile were in the lowest county quintile, and only 23% of counties located in the highest state quintile were in the highest county quintile. We also observed variation in concurrent use across HRRs within states, but it was not as dispersed. For example, 52% of the HRRs located in the highest state quintile were in the highest HRR quintile.ConclusionsLarge variation in concurrent use of opioids and benzodiazepines exists across the US. State variation masks substantial local variation, which beckons for polices to monitor concurrent opioid and benzodiazepine use at the county level.Copyright © 2018 Elsevier B.V. All rights reserved.

28 keywords...

hide…