OBJECTIVES: Community/Retail pharmacies accounted to 168,624 million U.S. dollars of pharmaceutical expenditures in 2013. The purpose of this study was to summarize and determine the extent to which pharmaceutical expenditures are affected by various market segments. Pharmaceutical expenditures and out of pocket costs should be conditioned on participation, due to the large zero outcomes. Thus, predictors of pharmaceutical expenditures can be best evaluated as segments of a market (usage, business type, or consumer), with health economic spending behavior, conditioned on the marketing mix (predicting pharmaceutical purchase event).

METHODS: A retrospective, 2-year panel study design was used, and conducted using data from the 2013-14 Medical Expenditures Panel Survey (MEPS) to analyze pharmaceutical events. A fixed effects, two-part model with clustered, robust standard errors was used to analyze pharmaceutical expenditure by controlling for the complex, stratified, weighted, clustered sample design. The outcome measure was total prescription drug expenditures and patient level out of pocket costs; the key predictor used was pharmacy provider (online, mail order, retail, hospital/HMO/clinic, and store).

RESULTS: Individuals who used online and mail order pharmacies were slightly more likely (5-20%) to have out of pocket costs on their medications; while those who used hospital/clinic based pharmacies or retail, or store based pharmacies were 10-20% less likely to have out of pocket costs. Insurance-based incentives are likely to have a major role in pharmaceutical expenditures and utilization of various pharmacy provider types, due to the various tiers of pharmaceutical coverage. Online pharmacies resulted in increased pharmaceutical expenditures, likely due to the increased use of these pharmacies for high cost medications.

CONCLUSIONS: This paper illustrated how pharmaceutical expenditures are affected by numerous market factors, including patient characteristics, pharmacy type, and medication characteristics. The use of a two-part model had high predictive value for predicting the presence of pharmaceutical expenditures.