Panel Paper:
Pharmacy Deserts and Medication Adherence
*Names in bold indicate Presenter
This paper examines the effects of a potential driver of non-adherence: geographic access to pharmacies. I argue that distance costs contribute significantly to prescription drug refill decisions. Access is not universal; previous studies document “pharmacy deserts” (neighborhoods with low access to pharmacies) concentrated in low income, minority communities.
Using Oregon all-payer pharmacy claims data, I examine the effect of local pharmacy openings on medication adherence. I find that pharmacy openings cause a 2-4% increase in local neighborhood pills dispensed, pharmacy claims filed, and total unique patients. For comparison, I find patent expirations cause a 5-7% increase in the same outcomes, an effect that takes several months to appear, while the store opening effect is largely immediate. I find especially large store opening effects among symptomatic drugs (such as those for migraine headaches) and drugs that improve mobility (such as cholesterol reducers and antidepressants). Store opening effect size is uncorrelated with patent expiration effect size, suggesting that policies designed to improve adherence must cater to drug type.
I argue that given the correct set of fixed effects, the effect of pharmacy entry is plausibly exogenous, a claim I justify by showing zero pre-opening trends in outcome variables. As a further check, I examine the effect of grocery store openings (such as Target), which include internal pharmacies, but likely open for reasons other than medication adherence. I find similar effects when limiting to these openings.
I find no effect of store opening when I limit to prescriptions received through mail-order, and a larger effect when I remove the mail order prescriptions from my baseline results. This result is consistent with the theoretical cost of distance to adherence, given that mail order patients are likely less sensitive to changes in distance costs. I find heterogenous effects by the number of existing local pharmacies, with the largest effect in neighborhoods with zero pharmacies, suggesting that pharmacy deserts are a serious policy concern. I further document heterogeneous effects by pharmacy type, zip code wealth, and patient insurance types.
This paper has three main contributions. First, it provides a benchmark measure of the benefit of subsidizing pharmacy entry for policy makers seeking to improve adherence. Second, it contributes to a rational model of medication adherence, highlighting the differences between individual versus social costs/benefits. Finally, it shows that pharmacy deserts are a legitimate policy concern that warrant attention from policy makers.