Principal components analysis of drug expenditure and utilisation trends for major therapeutic classes in US Medicaid...

Drug expenditures have been increasing much faster than spending on other medical services and have become burdensome for state Medicaid programmes. The purpose of this study was to analyse these trends across major therapeutic classes and to identify their similarities and differences.

Using national claims data from the Centers for Medicare & Medicaid Services for 1991 quarter 1 through to 2004 quarter 4, expenditures and prescriptions were aggregated across all drugs in 64 different therapeutic classes, providing 128 (64 x 2) different time series, most of length 56 quarters. Principal components analysis (PCA) was then applied to the data.

PCA revealed three principal components that accounted for 90% (92%) of total variation in Medicaid drug expenditure (utilisation) patterns. The first principal component (PC1), explaining 66% (67%) of the variation, is an exponential-like upward trend; PC2, explaining 17% (14%) of the variation, represents an increasing-then-decreasing pattern; and PC3, explaining 7% (11%) of the variation, represents an up-and-down cyclical pattern. Highly correlated with PC1 are antiretrovirals, antiseizure agents and corticoid steroids, among other drug classes.

Most drug therapeutic classes exhibited exponential-like upward expenditure (and utilisation) trends, clearly illustrating the overall rising expenditure burden for Medicaid.