Effect of Internet-Based Cognitive Behavioral Humanistic and Interpersonal Training vs Internet-Based General Health Education on Adolescent Depression in Primary Care

Key Points Questions Does an internet-based depression prevention program (competent adulthood transition with cognitive behavioral humanistic and interpersonal training) lower the hazard for depression in at-risk adolescents relative to health education attention control? Findings In this randomized clinical trial of adolescents with subsyndromal depression or history of depression randomized to receive internet-based behavioral humanistic interpersonal training or an internet-based general health education control, those who received the CATCH-IT intervention did not evidence fewer episodes of depression in the full intention-to-treat sample, but adolescents with subsyndromal depression may have experienced fewer depressive episodes. Meaning Competent adulthood transition with cognitive behavioral humanistic and interpersonal training may be better than health education for preventing depression in adolescents with subsyndromal depression.


Introduction
Approximately 13% to 20% of adolescents experience minor depressive episodes (mDE) or major depressive episodes (MDE) annually. 1 These adolescents have a higher incidence of medical illness 2 than those without mDE and MDE, and are at higher risk for suicide and recurrent depression. [3][4][5] Effective depression prevention programs are essential. 6 Promising findings for depression prevention programs are available. A Cochrane meta-analysis of prevention trials favored the intervention group over the control group with an overall risk difference for depressive disorders of −0.03, and for depression symptoms a standard mean difference of −0.21. 7 A review noted a 22% risk reduction of depressive episodes for adolescents. 7,8 Another meta-analysis involving 19 randomized preventive trials demonstrated significant reduction in depressive symptoms over 2 years among adolescents with higher symptom levels. 9 Another review of traditional therapies augmented with computerized communications demonstrated smallto-moderate effect sizes for depressive symptoms. 10 A systematic review of primary care-based preventative interventions targeting depression identified 14 randomized clinical trials; only 1 included adolescents, and average effect sizes were small. 11 Targeted interventions that show success during trials may not be scalable owing to practical issues such as cost, or prove ineffective in the broader community. 12 The primary care internet-based intervention, competent adulthood transition with cognitive behavioral humanistic and interpersonal training (CATCH-IT) addresses the need for a scalable intervention. [13][14][15][16] Internet-based interventions are accessible, cost-effective, private, and acceptable because they reduce stigma. 12 The CATCH-IT intervention is simple, consumer friendly, and more easily scaled up than more intensive, face-to-face interventions. A randomized clinical trial in China found that CATCH-IT lowered depressive symptoms in adolescents over 12 months. 17 We present a multisite randomized clinical trial testing the efficacy of CATCH-IT (version 3) vs an internet-based health education (HE) attention control in primary care. We aimed to prevent the onset of depressive episodes and lower symptoms in adolescents at intermediate-to-high risk for depression. Our primary hypothesis was that adolescents assigned to CATCH-IT relative to HE would have a lower hazard ratio (HR) for mDE or MDE at 6 months. We chose to evaluate group differences at 6 months to examine the potential of CATCH-IT as an immediate, medium-term response to depressive symptoms, given that follow-up intervals for such interventions in adolescents generally range from less than 6 to 12 months. 12 We also hypothesize that adolescents in CATCH-IT would show improvement in depressed mood and functional status relative to HE. 18  20 Participants were recruited from 2012 to 2016 through a description of the study during doctor visits, recruitment letters, and posted flyers. Adolescents were screened for risk in-person or by phone. After parental consent, adolescents participated in an eligibility assessment by phone. The parent and adolescent attended an enrollment assessment at their primary care office, when written informed consent from parents and assent from adolescents were obtained, and assessments were administered to confirm eligibility. 21 This study followed the Consolidated Standards of Reporting Trials (CONSORT) reporting guideline. The protocol, implementation process, and methods have been described in Supplement 1 and elsewhere. 18 The study was conducted in clinics in Chicago, northern Indiana, and Boston.

Inclusion and Exclusion Criteria
Adolescents aged 13 to 18 years with elevated levels of depressive symptoms on the CES-D 22 (scores 8-17 on the CES-D 10 or scores Ն16 on the CES-D 20 ) at screening or at baseline, and/or a history of depression or dysthymia, [22][23][24] were eligible. Exclusion criteria included the following: current MDE diagnosis or treatment; past cognitive behavioral therapy; CES-D 10 scores of more than 17 22 ; schizophrenia, psychosis, or bipolar disorder; serious medical condition (ie, causing serious disability or dysfunction); significant reading impairment or developmental disability; imminent suicidal risk; and current drug or alcohol abuse. 25,26 Criteria were selected to avoid confounding factors in depression etiology or treatment, consistent with the use of CATCH-IT as a preventive intervention.

Randomization
Participants were assigned randomly to CATCH-IT or HE (1:1 allocation) using a computer generated sequence blocked by site and time of entry (random blocks of size 4 and 6), stratified by risk severity (based on CES-D score, prior MDE, or dysthymia), sex and age (13-14 years or 15-18 years). 26,27 Outcomes Occurrence of first depressive episode was determined by the DSR. A score indicating at least subthreshold major depression (a DSR of Ն3) was considered to be a depressive episode. To test for robustness of findings, we also examined data using a DSR cutoff of 4 or more, indicating probable MDE, and a DSR of 5, indicating the presence of MDE. 28 Symptom outcomes include the CES-D 10 22 and Global Assessment Scale (GAS) scores.

CATCH-IT Intervention
The CATCH-IT intervention includes an internet component (15 adolescent modules, based primarily on the Coping with Depression Adolescent Course, 29 behavioral activation, 30

HE Intervention
The HE intervention is an attention control internet site (14 modules) providing instruction on general health topics. The 14th module discusses mood and mental health treatment, and also addresses mental disorder stigma. 14,34 Up to 3 check-in calls (weeks 1-3) were offered to ensure website access. The caregiver internet program (4 modules) is similar.

Intervention Shared Elements
Both interventions were consistent with guidelines for adolescent depression in primary care including the following: training clinicians in depression identification, diagnosis, and treatment; establishing referral relationships; screening; using a formal tool to determine depression risk; assessing depression; interviewing caregivers and adolescents; educating caregivers and adolescents on treatment; establishing treatment plans; and establishing safety plans. 35 These steps are closely related to the Chronic Care Model. 36 Rates of episodes were extremely low for this high-risk sample.
When episodes were identified, adolescents were referred for treatment, and caregivers and pediatricians were notified.

Instruments
Instruments have been described previously. 18 The 2-question screener was based on the Patient Health Questionnaire for adolescents. 37,38 The K-SADS 39,40 is a semistructured interview assessing current and lifetime psychiatric diagnoses in participants aged younger than 18 years, administered to parents and adolescents. 39,41 The DSRs are obtained from the Kiddie Longitudinal Interval Follow-up Evaluation 41 for each week of the follow-up interval, and GAS ratings were assigned at each assessment. For both the K-SADS and the Kiddie Longitudinal Interval Follow-up Evaluation, precipitating events were reviewed, and episodes secondary to medical concerns were indicated, if they occurred. The CES-D 10 measures the frequency of 10 depressive symptoms over the past week, using a 4-point scale; it was completed at baseline, 2, and 6 months. 22 Demographic information was collected at baseline, including race and ethnicity, using categories defined by the study team.
Fidelity and exposure to the intervention were based on module completion, and completion and rating of the MIs, with 2 trained raters using the MI Treatment Integrity coding manual (version 4.2.1), 42 and number of characters typed into the CATCH-IT website.

Sample Size
We required 200 participants per intervention condition to achieve 80% power based on a conservative application of our pilot study findings. 40 These calculations assumed that in the control group 72% are free from depression after 1-year follow-up, and the second year continues to follow the same exponential rate for controls; for intervention, the hazard is a constant ratio of 0.62, and an attrition rate of 7% for each of the first 4 quarters, and 2% for each of the second 4 quarters (Trial Protocol in Supplement 1).

Statistical Analysis
The trial tested for differences between group medians in website engagement using Wilcoxon rank sum tests. We estimated incidence rates by calculating the number of depressive episodes per We used linear mixed-effect growth models with random intercepts and slopes to examine differences between group change over time in CES-D 10 and GAS. Analyses were adjusted for the covariates listed above. We used propensity scores to account for differences between treatment groups in the per protocol analysis (Ն2 modules completed) that could otherwise confound treatment effect estimates controlling for: site, age, sex, ethnicity, race, mother's education, parents' marital status, number of siblings, firstborn child, times moved, current GAS score, most severe past GAS score, highest past GAS score, and baseline CES-D 10

Missing Data
The percentages of participants missing each K-SADS or CES-D 10 assessment were calculated. We used a logistic regression model to determine whether those missing from follow-up differed from those who were not. We also used multiple imputation to assess the potential for differential follow-up by intervention condition. We constructed 50 data sets for each site and intervention condition with fully saturated specification by condition interacting with the following variables: all CES-D 10 and GAS values (0, 2, and 6 months), screening CES-D 10 , baseline age, sex, ethnicity, race, and maternal education. These were combined into 50 complete imputed data sets and analyzed separately using the growth models described above; results were pooled.

Implementation
We implemented the study in 8 health systems from 31 practices in a defined population of more than 41 000 adolescents. There were 8499 adolescents screened, 2250 phone assessments, 446 enrolled, and 369 randomized. The 2 groups consisted of CATCH-IT (n = 193) and HE (n = 176) ( Figure 1). Among these participants, 28% had both a past episode and subsyndromal depression; 12% had a past episode only, 59% had subsyndromal depression only, and 1% had borderline subsyndromal depression

Fidelity and Intervention Exposure
Intervention use was monitored and recorded. The number of MIs completed was recorded for CATCH-IT participants. Table 2 shows that CATCH-IT adolescents and parents spent more time using the intervention, but CATCH-IT adolescents completed fewer modules than HE adolescents  technical global rating was 3.0 (0.5) on a 1 to 5 scale, and mean (SD) relational global rating was 2.9 (0.6) (eTable 2 in Supplement 2). scores showed a significantly stronger effect of CATCH-IT on time to event relative to those with    lower baseline scores (HR, 0.82; 95% CI, 0.67-0.99; P = .04) (Figure 3; eTable 5 in Supplement 2).

JAMA Network Open | Psychiatry
For example, the hazard ratio for a CES-D 10 score of 15 was 0.20 (95% CI, 0.05-0.77), compared with a hazard ratio of 1.44 (95% CI, 0.41-5.03) for a CES-D 10 score of 5. Sex, ethnicity, race, and age did not predict outcome or interact significantly with the interventions and outcomes. Both CATCH-IT and HE demonstrated reduced depressed mood and improved functional status, with no statistically significant differences at 6 months (eTables 6 and 7 in Supplement 2).

Missing Data
At least 1 follow-up K-SADS was completed for 85% of the sample, at least 1 follow-up CES-D 10 was completed for 80% of the sample, and at least 1 follow-up GAS assessment was completed for 81% of the sample. Dropout between CATCH-IT and HE was different at 2 and 6 months for both K-SADS and CES-D 10 (eTable 8A in Supplement 2) . At 6 months, K-SADS data were missing for 48 participants

Discussion
Overall, we observed a nonsignificant decrease in depressive disorders at 6 months in CATCH-IT as compared with HE. Adolescents and parents devoted substantial time to both interventions, and both conditions experienced decreased depressive symptoms and improved functional status.
However, higher-risk adolescents demonstrated greater benefit from CATCH-IT, achieving as much as 80% risk reduction with a CES-D 10 score of more than 15, but those without symptoms showed no such benefit. While regression to the mean is a possible explanation for the moderating effect of high CES-D on CATCH-IT, other studies have found that preventive effects for depression interventions are stronger for indicated vs universal samples. 44 Moreover, the same effect did not emerge for the HE condition-higher CES-D scores did not moderate the effect of the HE condition-suggesting regression to the mean may not explain the group difference found. For the 66% of adolescent and parent pairs who completed at least 2 modules (63% for CATCH-IT and 70% for HE), the unadjusted  analysis showed CATCH-IT reduced the risk of mDE and MDE by 59%, but this was not significant after adjustment for demographic factors or after analyses incorporating propensity scoring.

JAMA Network Open | Psychiatry
To our knowledge, this is the first clinical trial in adolescents to evaluate whether depressive episodes can be prevented in primary care settings. 11 Our finding that the risk of depressive episodes may be reduced for adolescents with subsyndromal depression is consistent with our earlier phase 2 clinical trial, which only included adolescents with subsyndromal depression. 45 Results were not significant in the intention-to-treat main effect analysis, but this may be the result of heterogeneity of treatment effect whereby CATCH-IT is favored for those with subsyndromal depression, but not for those with prior depressive episode alone. Perhaps CATCH-IT bored or frustrated adolescents without current symptoms, or conversely, elicited increased surveillance of symptoms or stimulated memories of prior episodes. 46,47 Alternatively, adolescents who are not symptomatic may be less motivated to complete CATCH-IT, the more self-directed intervention, and may actually prefer HE, which did not require substantial effort, perhaps even gaining a sense of self-efficacy. 34 This study has a robust prevention design implemented in a population-based model in primary care. 52,53 The implementation of our study at 2 sites and 8 health systems has rarely been accomplished in studies of child psychiatric conditions. 20 This study fits the model by Curran et al 19 of hybrid efficacy and implementation studies and substantially enhances generalizability. The attention control condition, which included guidelines for adolescent depression in primary care and chronic care model elements, no doubt reduced between-group differences. 24,36 However, given the need for ethical care of adolescents at risk for depressive episodes, a "no intervention" or "wait list control" condition is not possible. 53,54

Limitations
This study had limitations, including the relatively low adherence rate of teens and parents. Module completion for CATCH-IT was consistent with pilot findings. A review of internet-based mental health interventions for youths revealed completion ranged from 24% to 85%, and it was not necessary to complete the entire intervention for positive benefits to emerge. 12 In addition, module completion does not correlate with time spent, as the HE modules are significantly shorter than CATCH-IT; overall, CATCH-IT participants spent more time using the intervention. Nevertheless, future research should examine why adolescents did not complete the interventions, and explore strategies for boosting adherence. Also, our incidence rate for depressive disorders was low, thus, increasing the number of participants needed to have adequate power to detect group differences. We do not know for certain whether intervention effects can be attributed to the internet-based modules or to the MIs, although results from our pilot study suggest that adolescents who did not get MIs still evidenced reduced symptoms of depression at follow-up. Other limitations include the findings of differential attrition, which were adjusted analytically, and the fact that researchers enrolled only 92% of the target sample.

Conclusions
Our long-term goal for the CATCH-IT intervention is to provide a first-line program for primary care physicians to offer as part of the guidelines for adolescent depression in primary care, to support adolescents while the need for further intervention can be evaluated. We continue to examine moderators that may explain who responds best to this approach. Future directions include the development of versions for personal devices (eg, tablets and mobile phones), and a version individualized for sexual-and gender-minority teens.
A scalable, population-based approach to preventing depression in adolescents in primary care may be efficacious for adolescents with subsyndromal depression, but not for those with a prior episode alone.