Statistical Significance Testing and Clinical Effectiveness Studies

Effectiveness studies, by definition, must reflect patient and treatment variables that exist in the real world of practice. This includes identifying the covariates that contribute to both positive and negative outcomes. The use of clinically significant change and reliable change indices is reviewed to demonstrate that patient changes can be made on the basis of normative comparisons, using outcome relevant variables that reflect the diversity of problems that are of concern to our clients. Therefore, neither agreement on a single outcome index nor plotting the entire outcome frequency distributions of the comparison group, as called for by Krause, is required. Various methods currently exist and are used to address the limitations Krause cites with respect to randomized trials, outcome relevant covariates, file-drawer effects, and so forth. The identification and prediction of an individual's treatment response can be done in treatment as usual studies, where outcome relevant covariates have not been controlled, by studying change at the individual level. By studying such individuals and their combined continuums of treatment response, we can then identify clinically relevant outcome variables and alter the course of treatment accordingly.