Wallace et al.
A Novel Approach for Developing and Interpreting Treatment Moderator Profiles in Randomized Clinical Trials
Wallace ML, Frank E, and Kraemer HC
Mental health practitioners need better methods for personalizing treatment selection for individual patients. In the treatment of depression, for example, initial treatments may succeed as infrequently as one-third of the time, with additional treatments increasing the success rate to only two-thirds. An alternative to this costly trial-and-error approach is to identify treatment moderators – i.e., baseline variables that suggest which treatment will be optimally suited for each patient. Unfortunately, searches for moderators of depression treatment have been mostly unsuccessful. Only a handful of weak moderators have been identified, and the decision as to the preferred treatment would still be unclear if these moderators indicated different treatments. Thus, Dr. Meredith Wallace and colleagues demonstrated how multiple weak moderators can be combined to produce a clear and consistent algorithm that mental health practitioners can use to make treatment decisions for individual patients.
In their demonstration, Dr. Wallace and her colleagues studied 291 adults from a randomized clinical trial that compared an empirically supported psychotherapy (interpersonal psychotherapy) and selective serotonin reuptake inhibitor pharmacotherapy as treatments for depression. They selected eight relatively independent individual moderators out of 32 possible variables. Then, using novel statistical methods, they combined these eight individual moderators to produce a single moderator, M*, with a much stronger moderator effect size than any of the individual moderators examined.
After validating a combined moderator such as M* in a future randomized clinical trial, it can be used for clinical decision making. For example, mental health practitioners could input each patient’s relevant baseline variables into a hand-held computer, which would then output each patient’s M* value and suggested treatment. Furthermore, although only sociodemographic and clinical variables were considered in this demonstration, the methods are entirely applicable to and should take on greater importance when used when genetic markers, brain imaging parameters, and biologic markers of other kinds. Incorporating these types of measures may lead to even stronger and more meaningful combined moderators.
Meredith L. Wallacea, PhD; Ellen Franka, PhD; Helena C. Kraemera,b, PhD (aDepartment of Psychiatry, University of Pittsburgh; bDepartment of Psychiatry, Stanford University)