Geographic Variation in Retail Pharmaceutical Prices: Random or Regressive?

Background: Over 30 years of research has found non-random geographic variations in the cost, quality, and practice of health care across the U.S. Recent research in the state of Florida suggests irrational geographic variation in pharmaceutical prices, with highest mean retail pharmaceutical prices charged in communities with the lowest median household income. It is not known if this trend is isolated or exists nationwide. We sought to determine whether Vermont displays similar pricing irrationality.

Methods: Prices of six commonly prescribed drugs were collected from the Vermont Attorney General's publically accessible database of “usual and customary” Medicaid drug prices charged by individual pharmacies. Drugs and doses mirrored those used in previous research, are among the top-prescribed medicines, and lacked generic alternatives at the time of analysis. A “basket” analysis of all drugs was also performed to overcome small sample size and small differences in prices, and recognizing the frequency of poly-pharmacy. The median household incomes of Vermont counties were gathered from the U.S. Census Bureau and divided into three approximately even terciles: <$5,000; $45,000-$52,000; and >$52,000. Mean retail prices for the drugs were calculated for each county, then compared across income categories with analysis of variance, using Stata IC/10.1 software.

Result: When analyzed separately, only one drug – Seroquel – exhibited a small but statistically significant difference across counties (p<0.001), with the highest drug price found in the middle income counties. Only the difference between middle and highest income categories was statistically significant (p=0.03). This finding held true when controlled for population and price differences between chain and non-chain stores. The basket price mirrored this trend, with middle income counties exhibiting a relatively small yet statistically significant greater price ($1,420) than both the lowest ($1,395) and highest income counties ($1,373). In other words, compared to the lowest income counties, the basket price was 1.8% higher in middle income counties, and 1.6% lower in highest income counties (p<0.001 for each).

Conclusion: There may be non-random, irrational retail pharmaceutical price variation in Vermont, which could adversely impact access to pharmaceuticals. Since sample sizes and price variations were small, and findings varied from previous analysis, further study is warranted to determine whether any trends are generalizable to other contexts, conditions, and communities. The existence of publically available data on the average retail cost of prescriptions provided by some states makes such analyses relatively simple.