Abstract
Objective
Accurate naloxone distribution data are critical for planning and prevention purposes, yet sources of naloxone dispensing data vary by location, and completeness of local datasets is unknown. We sought to compare available datasets in Massachusetts, Rhode Island, and New York City (NYC) to a commercially available pharmacy national claims dataset (Symphony Health Solutions).
Data Sources and Study Setting
We utilized retail pharmacy naloxone dispensing data from NYC (2018–2019), Rhode Island (2013–2019), and Massachusetts (2014–2018), and pharmaceutical claims data from Symphony Health Solutions (2013–2019).
Study Design
We conducted a descriptive, retrospective, and secondary analysis comparing naloxone dispensing events (NDEs) captured via Symphony to NDEs captured by local datasets from the three jurisdictions between 2013 and 2019, when data were available from both sources, using descriptive statistics, regressions, and heat maps.
Data Collection/Extraction Methods
We defined an NDE as a dispensing event documented by the pharmacy and assumed that each dispensing event represented one naloxone kit (i.e., two doses). We extracted NDEs from local datasets and the Symphony claims dataset. The unit of analysis was the ZIP Code annual quarter.
Principal Findings
NDEs captured by Symphony exceeded those in local datasets for each time period and location, except in RI following legislation requiring NDEs to be reported to the PDMP. In regression analysis, absolute differences in NDEs between datasets increased substantially over time, except in RI before the PDMP. Heat maps of NDEs by ZIP code quarter showed important variations reflecting where pharmacies may not be reporting NDEs to Symphony or local datasets.
Conclusions
Policymakers must be able to monitor the quantity and location of NDEs in order to combat the opioid crisis. In regions where NDEs are not required to be reported to PDMPs, proprietary pharmaceutical claims datasets may be useful alternatives, with a need for local expertise to assess dataset-specific variability.