Search

UNSTRUCTURED Artificial intelligence (AI) tools have the potential to objectively identify youth in need of mental health care. Speech signals have demonstrated promise as a source for predicting various psychiatric conditions and transdiagnostic symptoms. Thus, we designed a study testing the association between obsessive compulsive disorder (OCD) diagnosis and symptom severity on vocal features in children and adolescents. Here, we present an analysis plan and statistical report for the study. Audio recordings of clinical interviews of 15 children and adolescents with OCD and 15 children and adolescents without a psychiatric diagnosis will be analyzed. Youth were between 8 and 17 years old. We will test the effect of OCD diagnosis on computationally-derived scores of vocal activation and valence using an analysis of variance. To test the effect of OCD severity classifications on the same computationally-derived vocal scores, we will use logistic regression. Finally, we attempt to create an improved indicator of OCD severity by refining the model with more relevant labels. Models will be adjusted for age and gender. Model validation strategies are outlined. A major strength is that we include age and gender in our models to increase classification accuracy. A major challenge is the suboptimal quality of the audio recordings, which are representative of in-the-wild data and a large body of recordings collected during other clinical trials. This pre-registered analysis plan and statistical report will increase validity of the interpretations of the coming results.

In a recent letter to the editor, a group of clinician-researchers posit that the conclusions in our published systematic review1 on cognitive-behavioral therapy (CBT) for pediatric obsessive-compulsive disorder (OCD) are based on inappropriate methodology. In this reply, we address the concerns expressed by Storch et al.2

Objective: To assess benefits and harms of cognitive-behavioral therapy (CBT) versus no intervention or versus other interventions for pediatric obsessive-compulsive disorder (OCD). Method: We searched for randomized clinical trials of CBT for pediatric OCD. Primary outcomes were OCD severity, serious adverse events, and level of functioning. Secondary outcomes were quality of life and adverse events. Remission from OCD was included as an exploratory outcome. We assessed risk of bias and evaluated the certainty of the evidence with the Grading of Recommendations Assessment, Development and Evaluation (GRADE). Results: Nine trials (N = 645) were included comparing CBT with no intervention and 3 trials (N = 146) comparing CBT with selective serotonin reuptake inhibitors (SSRIs). Compared with no intervention, CBT decreased OCD severity (mean difference [MD] = −8.51, 95% CI = −10.84 to −6.18, p < .00001, low certainty), improved level of functioning (patient-rated: standardized MD [SMD] = −0.90, 95% CI = −1.19 to −0.62, p < .00001, very low certainty; parent-rated: SMD = −0.68, 95% CI = −1.12 to −0.23, p = .003, very low certainty), had similar proportions of participants with adverse events (risk ratio = 1.06, 95% CI = 0.93−1.22, p = .39, GRADE: low certainty), and was associated with reduced risk of still having OCD (risk ratio = 0.50, 95% CI = 0.37−0.67, p < .00001, very low certainty). We had insufficient data to assess the effect of CBT versus no intervention on serious adverse events and quality of life. Compared with SSRIs, CBT led to similar decreases in OCD severity (MD = −0.75, 95% CI = −3.79 to 2.29, p = .63, GRADE: very low certainty), and was associated with similar risk of still having OCD (risk ratio = 0.85, 95% CI = 0.66−1.09, p = .20, very low certainty). We had insufficient data to assess the effect of CBT versus SSRIs on serious adverse events, level of functioning, quality of life, and adverse events. Conclusion: CBT may be more effective than no intervention and comparable to SSRIs for pediatric OCD, but we are very uncertain about the effect estimates.