We've Moved!
Visit SDSU’s new digital collections website at https://digitalcollections.sdsu.edu
Description
Rationale: Pediatric anxiety and depression are prevalent, impairing, and highly comorbid. Available evidence-based treatments have an average response rate of 60%. One path to increasing response may be to identify likely non-responders midway through treatment to adjust course prior to completing an episode of care. The aims of this study, thus, were to identify predictors of post-intervention response assessing: (a) mid-treatment symptom severity (Aim 1), (b) mid-treatment and session-by-session treatment process factors (Aim 2), and (c) a model optimizing the combination of these (Aim 3). Design and Method: Data were drawn from the treatment arm of a randomized trial of a 12-session transdiagnostic intervention for pediatric anxiety and depression (N = 95, ages 8-16). Mid-treatment measures of youth- and parent-reported anxiety and depression were collected, and mid-treatment and session-by-session therapist-ratings were collected of homework completion, youth and parent engagement in session, and youth therapeutic alliance. Measures were entered into logistic regression models to predict response on the Clinical Global Impression Improvement Scale (CGI-I < 2 indicating response) rated by independent evaluators blind to treatment condition. Results: Mid-treatment symptoms were significant predictors of response, including mean scores on both continuous (all ps < .001) and dichotomized (cut-point on continuous measure derived via ROC analysis; all ps < .05) measures of youth- and parent-rated anxiety and depression. Therapist-ratings of youth and parent engagement, therapeutic alliance, and homework completion were significantly associated with response, when tested as a mean rating across Sessions 1-8 (all ps < .004;) and at critical session points (all ps < 0.05). The optimized model included youth-reported anxiety above cut-point, parent-reported depression, child engagement at Session 2, and parent engagement at Session 8. This model was able to correctly classify 76.5% of youth as non-responders and 91.3% as responders (Nagelkerke R2 = .59 (c2 (4, 80) = 46.54, p < .001). Conclusions and Implications: Adaptive treatment strategies may provide guidelines for clinicians to personalize interventions to optimize outcome but are lacking foundational evidence on which to base decisions. This study provides initial evidence that symptom severity and therapy process factors may accurately predict treatment non-response by mid-treatment.
