Project Summary. During the last six years, our team conducted four randomized clinical trials of
psychosocial treatments for adolescents with ADHD with a cumulative sample size of 876 youth (ages of 11-
17; Sibley et al., 2016, under review, in preparation, in progress) with funding from the NIMH (R34 MH092466,
R01 MH106587), Institute of Education Sciences (R305A150433), and the Klingenstein Third Generation
Foundation. In the current proposal, we outline a plan to conduct an Integrative Data Analysis (IDA; Curran &
Hussong, 2009) that will pursue a linked series of person-level treatment outcome analyses. We will combine
multimethod (i.e., observational, parent-rated, self-rated, teacher-rated, official records) data from the four
recent RCTs (cumulative N=876) that evaluate brief psychosocial treatments for adolescents with ADHD
across a range of settings, interventions, and referral characteristics. Common across studies is a Baseline
(BL)-Post Treatment (POST)-Follow up (FU) design and a standard battery was delivered across studies that
includes core primary and secondary outcome measures (see Table 3). In the first months of Y01, Drs. Sibley
& Coxe will work with Dr. Curran to create a harmonized data set. Following completion of the final database,
the team will conduct a series of Latent Class Analyses, Mixture Models, and Structural Equation Models as
outlined in the study aims. These analyses will identify treatment-related phenotypes and family-risk profiles
using latent class analysis that organize the heterogeneous population into clinically meaningful subgroups.
We will then examine questions of treatment moderation, such as whether latent classes predict treatment
engagement (% of prescribed treatment course attended), and whether latent classes and adjunctive supports
(i.e., medication, parent involvement style, school accommodations) predict treatment response trajectories on
two primary outcomes (parent-rated ADHD symptoms and GPA). We will also examine whether there are
treatment-level variables (time of year, setting of treatment, content of treatment) that moderate the
relationship between phenotype and response. Finally, we will use structural equation modeling to identify key
treatment mediators (i.e., teen organization skills, parent contingency management, parent-teen conflict,
parental well-being) and to examine whether significance of mediators varies by phenotype, treatment-level
variables, and parenting profile. We believe that pursuing personalized medicine questions for adolescents
with ADHD could provide useful information that promotes improved treatment engagement and response—
leading to meaningful changes in long-term outcome for individuals with ADHD, whose adult trajectory is highly
related to experiences in adolescence (Barkley et al., 2008; Molina et al., 2012; Sibley et al., 2014). Treatment
research for adolescents with ADHD lags behind research on treatments for children with ADHD. As a result,
complex person-level and mediator x moderator analyses have not been conducted with this population—
namely due to sample size limitations in extant studies.