Several Ohio University Psychology professors presented a symposium on “Moderators, Mediators, and Predictors of Psychosocial Treatment Response Among Children and Youth With ADHD” at the Association for Behavioral and Cognitive Therapies annual convention Nov. 20-23 in Philadelphia.
Steve Marshall, Anne Dawson and Joanna Sadler are graduate students in Psychology at Ohio University. Dr. Brian Wymbs is Assistant Professor of Psychology. Dr. Steven Evans is Professor and Assistant Chair for Graduate Studies in Psychology.
Psychosocial interventions which are effective in treating children and adolescents with attention-deficit/hyperactivity disorder (ADHD) include behavioral parent training, behavioral school interventions, combined family and school treatments, and organizational skills training interventions (Evans, Owens, & Bunford, 2014). These treatments are effective across developmental levels (preschool through adolescence) with children having a wide range of impairments (academic, behavioral, family, and peer relationships). Although the Multimodal Treatment Study of ADHD (MTA study) contributed to our knowledge of moderators, mediators, and predictors of treatment response, research on factors influencing treatment outcomes is highly underdeveloped.
This symposium included presentations about the findings of several research teams investigating moderation, mediation, and prediction of response to psychosocial interventions for ADHD. Investigators reflect a diverse group of clinical, school, and applied developmental psychologists conducting research based primarily on principles of social learning theory, cognitive-behavioral psychology, attachment theory, and developmental-ecological psychology. Studies were conducted with students in elementary and middle school from diverse cultural and socioeconomic backgrounds spanning urban, suburban, and rural populations. Two of the studies were school-based and the other two were family-focused and school-linked. Outcomes of treatment were assessed across multiple domains of impairment, including academic, behavioral, peer, and family functioning.
The symposium was unique in its consideration of multiple factors that potentially influence treatment outcomes. The first study was presented by Anne Dawson, and she described the moderating and predictive effects of family factors (i.e., family socioeconomic status, single-parent status, size of family) and child diagnostic status (i.e., ADHD subtype, comorbid anxiety disorder, comorbid disruptive behavior disorder, and presence of learning disability) on treatment response to a family-school intervention for elementary-age students with ADHD. The second presentation examined the impact of parental ADHD on outcomes at post treatment and follow up in response to the same family-school intervention for elementary-aged students with ADHD. The third presentation investigated the mediating and predictive effects of treatment dosage on response to a multi-component, school-based, after-school program for middle school students with ADHD that was conducted at Ohio University. The final presentation examined the mediating and predictive effects of specific treatment components on the effectiveness of a classroom-based, teacher-mediated intervention designed to improve the peer relationships of young students with ADHD.
These results have important implications for research and practice in clinical, educational, and community settings and for 5 through 14 year old children with ADHD from diverse backgrounds. Symposium findings enhance our understanding of which children are likely to benefit the most from currently available evidence-based treatments. Findings were discussed with regard to implications for intervention development to maximize the benefits of treatment for individual children with their unique patterns of impairment.
Moderators and Predictors of Combined Family-School Intervention for Children With ADHD: Child and Family Characteristics (presented by Anne Dawson)
Multiple randomized controlled trials (RCTs) have found that behavioral parent training and combined family and school interventions are effective treatments for children with attention-deficit/hyperactivity disorder (ADHD). However, with the notable exception of the MTA study, few studies have examined child characteristics (e.g., subtype, comorbidity) and ecological variables (e.g., family characteristics) that moderate and predict the effects of treatment. This presentation described follow-up analyses of the RCT of the Family School Success (FSS) intervention, investigating moderators and predictors of treatment response. Targeting 2nd through 6th grade students, FSS builds on the expertise of child clinical and school psychologists by combining behavioral parent training, family-school consultation, and homework interventions for children with ADHD. Relative to an active control group receiving support and psychoeducation, FSS was found to significantly improve the quality of the family-school relationship and children’s homework performance, and to reduce the amount of negative/ineffective discipline used by parents (Power et al., 2012).
Child characteristics investigated in this study included ADHD subtype, comorbid anxiety disorder, comorbid disruptive behavior disorder, and comorbid learning disorders. Family factors included number of children in the home, single-parent family status, and socioeconomic status. Preliminary analyses indicated that none of these child and family variables moderated the effects of treatment. However, the presence of a comorbid anxiety disorder was associated with a more favorable response to both treatments with regard to parent-reported positive parenting practices (t = 3.98, p = .000). Also, the presence of a comorbid disruptive behavior disorder was associated with a poor response to treatment with regard to parent-reported changes in negative/ineffective discipline (t = -2.93, p = .004). With regard to family variables, single-parent status predicted a poorer response to treatment with regard to parent-reported homework problems (t=3.104, p = .002). The results affirm the importance of child diagnostic status and family variables in predicting response to treatment for children with ADHD.
Examining Dose Effects in a Randomized Controlled Trial of Psychosocial Interventions for Adolescents With ADHD
Dose-response relationships are rarely examined in the psychotherapy literature, even though dose gradients are considered evidence for causation in other scientific fields (Hill, 1965). This presentation reported dose-response mediation analyses in an RCT. Using data from a large (N = 326), multi-site RCT of the Challenging Horizons Program (CHP)—a school-based treatment program for adolescents with ADHD—we examined the impact of dosage on outcomes across two treatments: a twice weekly after-school program and school-based behavior consultation at Ohio University and the University of Cincinnati.
Preliminary intent-to-treat (ITT) analyses suggested that, as predicted, the after-school and behavior consultation versions of the CHP resulted in significant reductions in academic impairment when compared to a treatment-as-usual condition (Schultz, Evans, & Langberg, 2014). However, ITT analyses provide conservative estimates of treatment effects that ignore dosage (Lochman et al., 2006). Within the after-school program, for example, a relatively higher rate of voluntary withdrawal (22%; Mantel-Cox X2[2] = 39.29, p < .001) and other factors resulted in attendance ranging from 0 to 60 total sessions (M = 31.9; SD = 18.8), but the degree to which treatment gains were dependent on dose is unclear.
Using main outcomes from the CHP, we examined competing analytical strategies to address dose-response questions in psychotherapy trials. Using dose as a covariate in the ITT analysis, we have noted significant treatment effects on measures of homework completion (Homework Problems Checklist; Anesko et al., 1987) in the after-school program when compared to the treatment-as-usual group, but only for those students who completed more than 80% of the sessions offered (t = -2.85, p = .014). Such linear models may under- or over-estimate the impact of dose, depending on which level is examined. For this reason, we explored supplemental analyses, including probit/logit regression, propensity scores, and the complier average causal effect (CACE) approach, which can address questions relating to curvilinear dose gradients and optimum dosage estimation (Kopta et al, 1994; Lochman et al, 2006). Difficulties relating to the measurement of psychotherapy dose and the potentially confounding influence of treatment acceptability, accessibility, and client motivation were discussed.
Comments