A case study of mediators of treatment effectiveness

Recent research has observed that the use of confrontation in psychosocial treatment for alcohol abuse or dependence has a negative effect on posttreatment alcohol use among patients at average or above average levels of trait anger. It is not known what mediates that negative effect. The current study examines the role of session attendance as a mediator of the effect of confrontation on patients' subsequent alcohol use. In doing so, the study demonstrates the process of testing for mediation as well as planning analyses to meet additional conditions that can lend support to a causal mechanism of change.

Multiple regression analyses were used to test for session attendance as a mediator among 107 individuals with alcohol abuse or dependence who received either cognitive behavioral therapy (CBT) (n = 39), motivational enhancement therapy (MET) (n = 34), or twelve-step facilitation therapy (TSF) (n = 34). Emphasis was placed on achieving the desired temporal sequence of the therapy intervention, the mediator, and the outcome variable.

The data supported the role of session attendance as a partial mediator of the effect of confrontation on future alcohol use among patients who received CBT, but not among patients who received MET or TSF. In CBT, other potential mediators (e.g., therapeutic alliance and in-session resistance) were not supported and did not change the support for session attendance in the model. Beyond mediation, some but not all criteria for a causal mechanism of change were also met.

This study suggests that in CBT the negative impact of confrontation early in treatment is partially explained by a reduction in the number of sessions patients attended. Different processes appear to be occurring in MET and TSF. By carefully constructing analytic models, results can speak both to issues of mediation and to causal mechanisms of change.