Matching patients with treatments: conceptual and methodological issues.

The current enthusiasm for matching patients with optimal treatments rests on limited conceptual analyses. In addition, much of the existing research on patient-treatment matching has been based on methodological assumptions that are not commensurate with the complexity of the matching problem. Six key conceptual and methodological issues that underlie attempts to match patients with optimal forms of treatment and to conduct research on patient-treatment matching are outlined. The conceptual issues are: selecting effective matching variables, specifying the end result that matching is to enhance and determining the stage(s) in the treatment process at which matching decisions are to be made. The three methodological issues deal with the type of patient-treatment match or interaction effect and include: nonlinear interaction effects, higher-order interaction effects and multilevel interaction effects. Examples clarifying these issues are drawn from the literature on treating alcohol-dependent persons, but the issues are discussed at a broad level that permits generalization to treatment for many disorders. Implications for research on and the practice of "prescriptive treatment" are considered.