To the Editor: While we agree that the general issues raised by Dr. Douglas are important ones, we believe that there is insufficient evidence to accept either of her arguments as fact, nor is there reason to be too skeptical about the findings of randomized controlled trials on these bases. First, it is not clear that randomized controlled trials do not include complex cases. Barber (1) argued that randomized controlled trials will often include patients with pathology that is as significant and comorbid as seen in private practice, because patients who cannot afford private practice fees often seek out research studies. Furthermore, many contemporary randomized controlled trials include a broader range of patients than randomized controlled trials from a few decades ago with the specific intention of being more generalizable and useful to clinicians. It is nevertheless true that randomized controlled trials do focus on patients with a primary diagnosis (depression, generalized anxiety disorder, borderline personality disorder, etc.); however, these patients have comorbidities similar to those seen in the community (2).

Second, we have relatively limited systematic data on how seasoned clinicians really practice or whether adherence to one approach or a blend of approaches is better for patients of all diagnoses under all conditions. The clinicians in contemporary randomized controlled trials are frequently quite experienced themselves, and psychotherapy manuals and adherence measures often allow for appropriate flexibility pairing different strategies to different situations, as long as they fall within the general category to which the treatment belongs. While randomized controlled trials certainly do impose constraints on the treatment (most notably, with random assignment to treatment groups) that may limit generalizability, we believe that they remain the best method we have for minimizing the impact of researcher and therapist bias when evaluating differential treatment outcomes.