Mediators and Moderators to Treatment Response in Large Clinical Trials

Abstract

Large clinical trials have long been considered the gold standard of behavioral treatment research. Through the use of large sample sizes and meticulous methodology, randomized trials collect enough data to allow researchers to not only determine whether variables of interest are related (e.g., Does treatment with a particular selective serotonin reuptake inhibitor result in a significant reduction in symptoms of anxiety) but to describe the strength of this relationship as well (e.g., Does the combination of psychotherapy and pharmacotherapy result in greater reduction of anxiety symptoms than either treatment alone). By design, large clinical trials attempt to minimize the effects of confounding variables (e.g., age, treatment history) through random assignment to treatment groups. Though clinical trials are performed with strict control of primary efficacy variables to produce reliable data, results frequently identify other environmental or biological factors which may affect a patient’s response to treatment. Factors that affect the relationship between treatment and outcome, known as mediators or moderators, often provide researchers with a better understanding of how, why, or under what conditions treatments are effective. The MacArthur Foundation established definitive criteria for mediators and moderators which state that a moderator must precede treatment, be independent of treatment, and display a significant interaction with treatment while a mediator must occur after treatment, be associated with treatment, and produce either a main effect or an interaction with treatment (Arnold et al., 2010). As we discuss mediators and moderators of large clinical trials, we will be using the definitions outlined by the MacArthur guidelines, which most importantly specify that mediators and moderators affect certain treatment outcomes without affecting all treatment arms. We will discuss a study of risperidone on children with autistic disorder and irritability conducted by the National Institute of Mental Health Research Units on Pediatric Psychopharmacology (RUPP) Autism Network, the Treatment of Early Age Mania (TEAM) study, the Pediatric Obsessive Compulsive Treatment Study (POTS I), and the Treatment of SSRI-Resistant Depression in Adolescents (TORDIA) study (see Table 1); these studies were able to collect enough data to examine for the presence of mediators and moderators of response in analyses so that we may identify the effects that different variables have on response and use the implications in future research and treatment (Arnold et al., 2010; Brent et al., 2008; Garcia et al., 2010; Vitiello et al., 2012). In addition to mediators and moderators, we will explore other predictors of response, such as individual chemistry (i.e., prolactin levels), demographics, adherence, comorbid disorders, family history, treatment duration, and concomitant medications, which can also have significant effects on treatment response (Arnold et al., 2010; Donaldson, Nakamura, & Moinpour, 2009).