All Times EDT
Keywords: prescription drug dispensing, clustering, geographical analysis
The crisis of opioid nonmedical use and overdose in the United States has involved unprecedented levels of opioid analgesic (OA) prescriptions and opioid-related mortality. While the overall national OA prescribing rate declined from 2012 to 2019, the trend varies by geographical location. The prescribing rates continue to remain very high in certain areas across the country. Such geographical ‘hotspots’ may be associated with localized clusters of adverse events, which may inform early signals of drug safety issues. Acknowledging the large influence of community-level factors on substance use and related health outcomes, increasing the geographical granularity of analyses can help identify the community-based factors influencing the opioid crisis and reveal geographic sub-populations in which to target surveillance. Several proprietary drug utilization databases are utilized by the U.S. Food and Drug Administration to better understand the scope of prescription OA use and evaluate patterns. Such drug dispensing data include the number of prescriptions dispensed in aggregate over a time frame by geographical locations. One useful approach to enhance pharmacovigilance using these data may be through identification of areas with notable change over time in prescription OA dispensing. However, beyond evaluations at the state level, characterizing temporal patterns of prescription drug dispensing at granular geographic levels can be a challenge and necessitate specialized statistical methodology. In this poster, we describe a two-step approach to: 1) identify geographic areas that had significant change over time in prescription drug dispensing, and 2) characterize clusters of geographic areas that share similar temporal change patterns. Results from a simulation study and analyses of naloxone and buprenorphine prescription dispensing data are discussed to evaluate the utility of the proposed method.