Standardization and decomposition analysis: A useful analytical method for disparity studies

A key of health disparity study is to understand population-specific differences in health outcomes. When comparing outcome measures among different populations, the compositions of confounding factors should be taken into account. Standardization and decomposition analysis (SDA) is a useful analytical method for such comparisons. SDA has some explicit advantages. First, its results can be presented in a manner that is easier to understand than statistical parameter estimates. Outcome difference can be decomposed into component effects that are attributed to the “real” outcome difference and effects of confounding factors; and the relative contributions of all component effects sum up to 100%. Second, SDA has no constraints on the specification of relationship (e.g., linearity), the nature of the variables (e.g., random), or the form of variable distributions (e.g., normality). Third, SDA conducts multiple pairwise comparisons simultaneously free of internal inconsistency. And finally, SDA can be applied to study differences in a wide range of outcome measures such as rate, percentage, proportion, ratio, arithmetic mean, and categorical measures among multiple populations/groups. Using data collected from a multi-site natural history study of rural drug users in the U.S., this study will present the observed and adjusted differences in methamphetamine use practice among rural drug users in different U.S. states; and show how much of the differences in the observed outcome measures between specific populations are attributed to what specific confounding factors.