Research and evaluation challenges pertaining to medicaid clients with substance abuse: Indeterminacy in data linkage

Medicaid clients (and especially clients with substance abuse problems) are often served by multiple agencies. Research and evaluation can be challenged by the need to link data generated by numerous administrative systems. Medicaid clients involved with the criminal justice system may be especially difficult for data linkage owing to numerous identities. This chapter describes methods used to verify matches pertaining to individuals eligible for drug court. Subjects were identified for the drug court and Medicaid managed care research project described in a previous chapter. Especially helpful were flow charts showing (a) the routes of individuals as they made contact with several agencies and (b) the creation of data elements describing those individuals by the numerous organizations. Building on this background, the chapter discusses matching problems as well as the several uncertainties that can arise. Case examples are provided showing results of assorted decisions made about possible linkages. Of some 2,153 potential cases, automated matching software would have retained only about 1,500 matched cases and many of those retained would have lacked demographic and-or service utilization data. Conversely, using manual methods, 214 potential cases were deleted chiefly due to the research project's inclusion and exclusion criteria. Only about 30 potential cases were deleted because of irresolvable data problems. Ultimately, some 1,939 subjects were deemed valid unique cases along with much of their identifying, demographic, and service utilization data. The manual linkage procedures decreased sampling error substantially which increased precision of the research project's data analyses.