Naloxone dispensing & opioid-overdose mortality
Asking how the relationship changes from one place to another.
A question from a screen
Watching Dopesick drew me toward the opioid crisis and the policy trade-offs behind it. Hospital fieldwork and a medical-equipment internship had already made me interested in what happens when systems meet people; this project brought that question into data.
Data & method
With guidance from Professor JiaNing Wang at Harvard Medical School, I built a U.S. panel covering 2012–2023 and analyzed 610 state-year observations in R.
I used mixed-effects models, generalized estimating equations (GEE), and state and year fixed-effects models to examine the relationship between naloxone dispensing and opioid-overdose mortality. The work considers policy context and geographic heterogeneity, rather than assuming one national association tells the whole story.
What the models show
Mixed-effects models showed positive associations in the Northeast and South, while regional contrasts varied across specifications.
Those are associations, not evidence that naloxone causes mortality to rise. Changing model specifications also changes parts of the regional picture. That sensitivity is part of the finding and a reason to be careful about interpretation.
Revision is part of research
I have authored a manuscript, now under revision following first-round review at the National High School Journal of Science (NHSJS).
Working through different specifications has made me more attentive to the gap between a striking coefficient and a defensible conclusion. Public health is not only a question of whether an intervention exists, but how policy, local conditions, and implementation interact.