Understanding results from randomized trials: use of program- and client-level data to study medical and nonmedical...

To address unanticipated results from randomized trials, researchers often focus on client-level data about services received during treatment. Program-level observations can also be helpful, especially in understanding treatment delivered in groups. Using both approaches, this article strives to better understand inconsistent results from a trial comparing medical and nonmedical group-format day treatment.

Dependent treatment seekers were randomized to a hospital-based medical day-treatment program or to one of two community-based nonmedical day treatment programs. Services received during treatment were captured from clients using the Treatment Services Review (N = 230 subjects; 78 women), and group sessions were observed to measure therapeutic style using an Event Form (N = 48 observations).

The trial had found better medical outcomes at the hospital than at either nonmedical program, but most other tests had demonstrated similar outcomes at day hospital and one of the nonmedical sites and worse outcomes (psychiatric, family/social and employment) at the other nonmedical study site. Our analysis of services reported by study participants found a pattern of fewer substance misuse-oriented groups and less serious discussions about medical, psychiatric and family/social problems at the same nonmedical site that had worse outcomes. The way that services were delivered at that site further helped to explain the poorer outcomes there: Groups tended to be more didactic, classroom-like and less discussion-oriented.

Although services received are helpful in explaining treatment outcome, treatment observation adds explanatory value. Without increasing the cost of service provision, programs that rely heavily on didactic approaches might improve their outcomes simply by encouraging more interactive discussions that engage the clients.