Evaluating Psychological vs. Biomedical Treatments
The elements we have described characterize what is usually termed a randomizedclinical trial (RCT), so called because of the random assignment of participants togroups. In most ways an RCT would be run in the same fashion for evaluating a new medication or for evaluating a specific type of therapy. There are, however, a few differ-ences between RCTs for drugs and RCTs for therapy.
In evaluating biomedical treatments, we need to be alert to the possibility of placebo effects—the influence of a patient’s beliefs and expectations in bringing abouta cure. We generally test for placebo effects by giving the patient a medically neutral sub-stance (the placebo); the placebo often diminishes symptoms, but because it has no active ingredients, its effect has to be understood in terms of the patient’s beliefs: The patient believes in the pill’s powers, and it is his faith in the medication, not the med-ication itself, that brings the benefits.
In the broad practice of medicine, the evidence is mixed regarding how powerful placebos are. In some studies, placebos seem to produce real improvement (Benson & Friedman, 1996); in others, the placebo effect is weak or nonexistent (Hrobjartsson & Gotzsche, 2001). In the study of mental disorders, though, we need to take the placeboeffect seriously because it is surely plausible that a patient’s distress might be alleviated by her belief in the cure rather than by the cure itself.
In studies of a medication, therefore, we give the medication we are interested in— Wellbutrin, in our example—to one group of patients, and a placebo to the other group. This second group will receive the same attention from the treatment staff as the Wellbutrin group, and they will be told exactly the same things about their treatment. They will also be given the same number of pills as the patients in the true-drug group, and on the same schedule. In this fashion, we ensure that the two groups have the same expectations and the same beliefs about the treatment. If we find, therefore, that the groups differ at the end of the study, the difference must be attributed to the one element that distinguished the two groups—the fact that the patients in one group actually took a medication, while those in the other took inert pills.
In principle, the same logic regarding a placebo control applies to the study of psy-chological treatments. In practice, however, it is not clear what the placebo would be in this case. Would we want the placebo group to work with a therapist who listens, but without warmth? A therapist who expresses warmth, but provides no counsel? A thera-pist who offers counsel irrelevant to the patient’s problems? The choice between these (or other) options is not clear, making this one source of ambiguity in interpreting RCT data for psychotherapy (Nathan et al., 2000). For this reason, many RCTs for psychological treatments do not have special placebo control conditions. Instead, the treatment of interest is compared either to other standard treatments or to a wait-listcontrol. The wait-list control is one in which participants are randomly assigned toreceive either immediate or delayed treatment, and the immediate treatment participants’ responses to treatment are evaluated by comparing them to the delayed treatment participants—who have not yet started therapy.
There is one other important way in which RCT procedures differ for psychological and biomedical treatments. When we are testing a medicine, it is easy to ensure that each patient gets the same number of pills and that the chemical constituents of the pills are the same for everyone in the study. But how can we ensure, in a study of therapy, that each patient gets the same treatment and the same “dose”?
Investigators usually address this point by developing a treatment manual that tells the therapists exactly how to proceed during the study—what steps to take in therapy, what instructions to offer, and so on. This manualized therapy denies the therapist the option of improvising or adjusting procedures midstream, which may sometimes undermine the quality of the therapy (Addis & Krasnow, 2000). However, this stan-dardization seems essential if we are to assess whether a particular therapy is effective.
Clearly, there are trade-offs associated with manualized therapy. On the positive side, this procedure seems our best hope for learning whether a particular type of treatment works. On the negative side, it may not accurately represent the way that treatment would be used outside the study, since most clinicians adopt an eclectic approach. Clinicians also often adjust their approach as the therapy proceeds and they learn which steps seem promising for a particular individual and which do not.