(PS5-A5) Alliance-outcome Relations in an Exposure-based CBT for Pediatric Irritability Delivered in Person versus via Telehealth

Evidence shows that the therapeutic alliance is a significant predictor of outcomes in traditional face-to-face treatments. However, less is known about the relations between the therapeutic alliance and outcome in treatments delivered via telehealth. An ongoing exposure-based cognitive-behavioral therapy (CBT) trial for severe irritability in youth was adapted to telehealth to comply with COVID-19 pandemic restrictions. Here, we compare changes in therapeutic alliance and its relation to outcome in in-person vs. video telehealth-delivered CBT. The treatment involves exposures with child to anger-provoking situations plus parent management training. Telehealth adaptations included naturalistic in-home exposures, parent assisting with exposures, and activity schedules related to social distancing orders. Forty youth (M age = 11.2 y.o., SD = 1.85; 62.5% male; 77.5% white) with severe, impairing irritability and their parents completed 12 weekly sessions either in-person (N = 19) or via telehealth (N = 21). The Therapeutic Alliance Scale for Children-revised (TASC-r; Shirk & Saiz, 1992; has 2 subscales: Bond, bond between therapist and child; and Task, agreement on tasks) and the Working Alliance Inventory (WAI; Horvath & Greenberg, 1989; has 3 subscales: Goal, agreement on goals; Task, agreement on tasks; and Bond, bond between therapist and parent) were used. Irritability was measured using the child-, parent-, and clinician-rated Affective Reactivity Index (Stringaris et al., 2012; Haller et al., 2020). Three types of analyses were conducted: 1) HLMs tested change in alliance over time at level-1 with treatment format at level-2; 2) Lagged analyses of child and parent alliance at current session (T) predicting next session irritability (T+1); 3) HLMs with pre-, mid-, post-treatment alliance predicting pre-, mid-, post-treatment clinician ARI.  Analysis 1: Findings showed that the WAI total and its 3 subscales (Bond, Task, and Goal) significantly increased during treatment (WAI Total: β = 0.376, t(39) = 3.48, p < .001; WAI Bond: β = 0.094, t(39) = 2.70, p = .01; WAI Task: β = 0.104, t(39) = 3.05, p = .004; WAI Goal: β = 0.181, t(39) = 4.085, p < .001). There was a greater change in the WAI Task subscale in telehealth vs. in-person (β = 0.143, t(38) = 2.28, p =.029).  Analysis 2: Increases in WAI total score predicted significant decreases in parent-rated irritability at the next session (WAI Total: β = −0.078, t(39)= −2.51, p = .016). Analysis 3: Both child and parent alliance measures were found to significantly predict clinician-rated irritability (TASC-r: β = −1.379, t(39)= −3.15, p = .003 and WAI: β = −1.156, t(39)= −3.34, p = .002). Treatment format moderated these relations (TASC-r: β = 2.204, t(38)= 2.59, p = .014 and WAI: β = 1.225, t(38)= 2.33, p = .025), such that alliance was more predictive of clinician-rated irritability in the telehealth group. No other significant differences emerged with treatment format. Overall, the findings suggest that telehealth adaptations increased alliance and alliance-outcome relations for clinician-rated irritability; the naturalistic in-home setting and parents assisting with exposures may have helped to increase overall alliance and agreement on tasks to make exposures more potent.