Do Prescription Drug Monitoring Program Mandates Reduce Opioid Overdose Deaths? Evidence from Staggered State Adoption

Abstract

We estimate the effect of state-level Prescription Drug Monitoring Program (PDMP) mandatory query requirements on opioid overdose deaths using a staggered difference-in-differences design. Between 2012 and 2020, 36 U.S. jurisdictions in our analysis sample adopted laws requiring prescribers to check PDMP databases before writing controlled substance prescriptions. Using CDC mortality data from 2015–2020 (198 jurisdiction-years across 41 jurisdictions in the TWFE regression sample after dropping singletons), we find no statistically significant effect of PDMP mandates on opioid deaths (TWFE coefficient = 2.0%, SE = 5.8%, p = 0.74). Because mortality data begin in 2015, jurisdictions adopting mandates in 2012–2015 are always-treated and contribute no pre-treatment variation. For the TWFE regression, identification relies on later adopters (2016–2020) compared to never-treated states. For the Sun-Abraham estimator, only 20 later adopters with complete data remain after excluding those with singleton/missing-data issues. Event study analysis reveals some evidence of differential pre-trends, with the $t=-3$ coefficient statistically significant (p = 0.007). The Sun-Abraham heterogeneity-robust estimator—using only 2016–2020 adopters who have observable pre-treatment periods—yields an ATT of $-2.5$% (SE~=~2.8%, p~=~0.38), also statistically insignificant. These findings highlight the challenge of credibly evaluating crisis-response policies with limited pre-treatment data. \medskip JEL Codes: I12, I18, K32 Keywords: Prescription drug monitoring programs, opioid crisis, difference-in-differences, staggered adoption, parallel trends