In mental health research, it is important to understand the mechanisms underlying the effects of interventions, for example whether cognitive/behavioral factors mediate the effect of a treatment for youth with anxiety disorders, or whether a primary care service intervention reduces depression and suicidality in elderly adults through increasing depression treatment provision, or through improving treatment quality, or both. Such knowledge has the potential to help tailor interventions to optimize mental health outcomes (NIMH strategic objective 3.2). However, commonly used mediation methods are generally not appropriate and can lead to misleading results. Recently developed causal mediation methods have the potential to provide more accurate results but have not been widely adopted by applied researchers. This is likely due to their complexity, lack of methods for certain settings, and lack of clear dissemination of the methods. The goals of the proposed project are to extend causal mediation methods with a focus on common data structures in mental health research, and to disseminate the existing and new causal mediation methods to mental health researchers and provide guidance to facilitate adoption and appropriate use. In particular, the work will 1) Conduct a review of mental health articles published in the past 10 years that include mediation analyses, focusing on the temporal ordering of the variables and the handling of confounding, 2) Extend and tailor causal mediation methods to common data situations in mental health research that are under-covered by existing methods, and 3) Disseminate existing and newly developed causal mediation methods to mental health researchers and provide guidance to facilitate adoption and appropriate use, using applications to mental health studies for illustration. The methods will be developed informed by three randomized trials in mental health: of adolescent depression treatment (TADS), of anxiety (CAMS) treatment, and of a service intervention to improve depression care for elderly individuals (PROSPECT). The work has the potential to help mental health researchers better understand what works, and why, thus improving mental health in the population through better interventions and treatments.
Identifying the mechanisms and pathways that underlie the effects of treatments and exposures can help lead to more effective interventions to improve mental health in the population. New statistical methods, and clear dissemination of those methods, are needed to allow researchers to make full use of studies and examine what works, and why.