“We have to speak the language of medicine and the intellectual language of medicine is the randomized, double-blind, controlled trial,” Powell said. “We must also strengthen behavioral science interventions. Don’t settle for a minimalist behavioral approach.”

Behavioral medicine specialists must adapt if they wish to level the playing field with medical and surgical treatments, agreed Bonnie Spring, Ph.D., of the University of Illinois at Chicago. Design, execution and analysis of clinical trials must be of the highest quality to warrant inclusion in the systematic reviews that ultimately move clinical practice in new directions.

The keys to good trial design, Spring said, are proper randomization and blinding, sufficient power to validate results, intention-to-treat analysis and use of top reporting standards like CONSORT. Intention-to-treat analysis involves using all randomized cases and valid, clearly explained approaches to impute any missing data.

“Intention-to-treat means you have to follow the patient,” she said. “You can’t lose data or outcomes. Phone them or go out and track them down.”

The rule of thumb, Powell said, should be “once randomized, always analyzed.”

Varying dropout rates among participants can undermine the validity of a trial, Powell said. She cited a study of emotional support for patients with ischemic heart disease that showed an apparently positive effect of the intervention. Closer examination of the data found that baseline socioeconomic status differed between the two groups because subjects were randomized before giving consent. Low socioeconomic status subjects had dropped out, confused by the consent form, but so did high socioeconomic status control subjects, who had figured out that they were indeed controls. A better-designed restudy of the same intervention provided more equivocal results.

“At present, we don’t have the consistency to get these interventions incorporated into medical practice,” Powell said. “We need rigorous, randomized behavioral efficacy trials with clinically significant outcomes.”

-- Aaron Levin, Health Behavior News Service