See eResult 3 in Supplement 2 for detailed reasons on not receiving face-to-face CBT when being classified as a nonresponder at the 3-month follow-up assessment.
A, Boxes represent the interquartile range of the 25th and 75th percentiles, the middle line represents the median, the circles with connecting lines represent the mean values, the whiskers extend to the highest and lowest values within 1.5 times the interquartile range, and dots represent outliers. Children’s Yale-Brown Obsessive-Compulsive Scale (CY-BOCS) scores range from 0 to 40, with higher scores indicating more severe symptoms. The posttreatment assessment was 16 weeks after the pretreatment assessment, and the follow-up assessments were 3 months and 6 months after the posttreatment assessment. Nonresponders at the 3-month follow-up were offered additional in-person CBT between the 3-month follow-up and the 6-month follow-up. B, Stepped-care treatment was considered noninferior to in-person CBT because the 1-sided 97.5% CI of the mean difference was below the prespecified margin of 4 points (dotted line) on the CY-BOCS.
Children’s Yale-Brown Obsessive-Compulsive Scale (CY-BOCS) scores range from 0 to 40, with higher scores indicating more severe symptoms. The boxplots indicate the interquartile range of the 25th and 75th percentiles, the middle line represents the median, circles represent the mean value, and the whiskers extend to the highest and lowest values within 1.5 times the interquartile range. C, The light blue vertical lines represent the nonresponders at the 3-month follow-up who received additional in-person treatment.
eMethod 1. Interventions
eMethod 2. Assessment of therapist adherence
eMethod 3. Outcome measures (continued)
eMethod 4. Statistical analysis (continued)
eResult 1. Blinding integrity
eResult 2. Reasons not receiving additional treatment as a non-responder
eTable 1. Content of the internet-delivered CBT program for children and adolescents with OCD
eTable 2. Description of therapists
eTable 3. Measures and assessment points
eTable 4. Demographics and clinical characteristics by site
eTable 5. Between-group effect sizes at 6-month follow-up
eTable 6. Observed values and estimated treatment efficacy for ordinal secondary measurements
eTable 7. Treatment response and remission rates across measurement points
eTable 8. Therapist time
eTable 9. Treatment adherence, credibility and satisfaction
eTable 10. Reported adverse events by study group
eFigure 1. Screenshots of the internet-delivered cognitive behavioral therapy program
Data sharing statement
Customize your JAMA Network experience by selecting one or more topics from the list below.
Aspvall K, Andersson E, Melin K, et al. Effect of an Internet-Delivered Stepped-Care Program vs In-Person Cognitive Behavioral Therapy on Obsessive-Compulsive Disorder Symptoms in Children and Adolescents: A Randomized Clinical Trial. JAMA. 2021;325(18):1863–1873. doi:10.1001/jama.2021.3839
Is internet-delivered cognitive behavioral therapy (CBT) implemented in a stepped-care model noninferior to in-person CBT for children and adolescents with obsessive-compulsive disorder?
In this randomized, noninferiority clinical trial, 152 children and adolescents with obsessive-compulsive disorder were treated with an internet-delivered CBT program followed by traditional in-person CBT if necessary vs in-person CBT alone. After 6 months, the mean Children’s Yale-Brown Obsessive-Compulsive Scale score was 11.57 in those treated with internet-delivered CBT vs 10.57 in those treated with in-person CBT, a difference that met the noninferiority criterion of 4 points.
Treating children and adolescents with obsessive-compulsive disorder with an internet intervention followed by traditional face-to-face therapy if necessary was noninferior to in-person therapy alone.
In most countries, young people with obsessive-compulsive disorder have limited access to specialist cognitive behavioral therapy (CBT), a first-line treatment.
To investigate whether internet-delivered CBT implemented in a stepped-care model is noninferior to in-person CBT for pediatric obsessive-compulsive disorder.
Design, Setting and Participants
A randomized clinical noninferiority trial conducted at 2 specialist child and adolescent mental health clinics in Sweden. Participants included 152 individuals aged 8 to 17 years with obsessive-compulsive disorder. Enrollment began in October 2017 and ended in May 2019. Follow-up ended in April 2020.
Participants randomized to the stepped-care group (n = 74) received internet-delivered CBT for 16 weeks. Nonresponders at the 3-month follow-up were then offered a course of traditional face-to-face treatment. Participants randomized to the control group (n = 78) immediately received in-person CBT for 16 weeks. Nonresponders at the 3-month follow-up received additional face-to-face treatment.
Main Outcomes and Measures
The primary outcome was the masked assessor–rated Children’s Yale-Brown Obsessive-Compulsive Scale (CY-BOCS) score at the 6-month follow-up. The scale includes 10 items rated from 0 (no symptoms) to 4 (extreme symptoms), yielding a total score range of 0 to 40, with higher scores indicating greater severity. Assessors were masked to treatment allocation at pretreatment, posttreatment, 3-month follow-up, and 6-month follow-up assessments. The predefined noninferiority margin was 4 points on the CY-BOCS.
Among the 152 randomized participants (mean age, 13.4 years; 94 [62%] females), 151 (99%) completed the trial. At the 3-month follow-up, 34 participants (46%) in the stepped-care group and 23 (30%) in the in-person CBT group were nonresponders. At the 6-month follow-up, the CY-BOCS score was 11.57 points in the stepped-care group vs 10.57 points in the face-to-face treatment group, corresponding to an estimated mean difference of 0.91 points ([1-sided 97.5% CI, −∞ to 3.28]; P for noninferiority = .02). Increased anxiety (30%-36%) and depressive symptoms (20%-28%) were the most frequently reported adverse events in both groups. There were 2 unrelated serious adverse events (1 in each group).
Conclusions and Relevance
Among children and adolescents with obsessive-compulsive disorder, treatment with an internet-delivered CBT program followed by in-person CBT if necessary compared with in-person CBT alone resulted in a noninferior difference in symptoms at the 6-month follow-up. Further research is needed to understand the durability and generalizability of these findings.
ClinicalTrials.gov Identifier: NCT03263546
Obsessive-compulsive disorder (OCD) is a relatively common psychiatric disorder (the lifetime prevalence was estimated to be 1.3% in a meta-analysis including all worldwide studies until 20171) that is associated with marked functional impairment2,3 and increased mortality risk.4,5 Because the disorder typically develops in childhood or adolescence6 and tends to run a chronic course if untreated,7 early intervention should be prioritized.8
Evidence-based treatments for young people with OCD include cognitive behavioral therapy (CBT) and selective serotonin-reuptake inhibitors, but guidelines typically recommend in-person CBT as the first-line treatment for mild to moderate OCD due to its favorable adverse effect profile and greater acceptability.9 CBT for OCD is a highly specialized treatment that may be difficult to access and is time-consuming, typically requiring regular appointments with a qualified therapist over several weeks or months.9,10 One possible solution to the limited availability of specialized CBT is to deliver a low-intensity version of the treatment online with minimal remote support from a clinician, akin to a guided self-help intervention.11 This kind of guided internet-delivered CBT differs from standard telepsychiatry in that the therapist does not actively deliver the treatment content in real time. Instead, the therapist provides minimal support asynchronously via a messaging system built in the online platform. Research has supported the potential efficacy and cost-effectiveness of this approach for managing pediatric OCD.12-15 However, questions remain regarding the feasibility of implementing such low-intensity interventions in regular health care.
It may be possible to effectively use internet-delivered CBT in a stepped-care fashion, in which patients are first offered a low-intensity intervention and higher-intensity treatments (eg, face-to-face CBT) are reserved for patients who do not benefit sufficiently from the low-intensity intervention. Stepped-care approaches are heralded as the ideal model of psychiatric service delivery, but they have rarely been evaluated. The current trial aimed to evaluate whether a stepped-care model could be as efficacious as (ie, not inferior to) traditional in-person CBT for children and adolescents with OCD.
This was a 2-site single-blinded randomized clinical noninferiority trial comparing stepped care with in-person CBT treatment for children and adolescents with OCD. The principal investigator and outcome assessors were blinded to group allocation, but treatment condition was not blinded for participants or therapists. Masked rater assessments were conducted at pretreatment (week 0), posttreatment (week 16), 3-month follow-up, and 6-month follow-up (primary end point) visits. The study was approved by the regional ethical review board in Stockholm, Sweden (DNR 2017/1070-31/1). All participants and their parents/legal guardians received verbal and written information about the study and provided written informed consent before inclusion. The full trial protocol is published16 and available in Supplement 1. A full heath economic evaluation will be published separately.
Participants were recruited from 2 specialist pediatric OCD clinics in Stockholm and Gothenburg, Sweden. Families could also self-refer to the study via a dedicated website. After an initial screening, eligible participants were invited to participate in an in-person clinician appointment for a full psychiatric assessment that included the Children’s Yale-Brown Obsessive-Compulsive Scale (CY-BOCS)17 and the Mini International Neuropsychiatric Interview for Children and Adolescents (MINI-KID).18 Following the initial assessment, and discussion with the multidisciplinary team in cases of uncertainty, families who fulfilled the inclusion criteria and provided informed consent were included in the study.
Eligible participants were children and adolescents with a primary diagnosis of OCD according to the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition,19 a total score of at least 16 on the CY-BOCS,17 age between 7 and 17 years, ability to read and write in Swedish, and daily access to a computer with internet connection. Participants were excluded if they had changed any psychotropic medication in the 6 weeks before the pretreatment assessment; had a comorbid diagnosis of organic brain disorder, global learning disabilities, autism spectrum disorder, psychosis, bipolar disorder, or severe eating disorder; had suicidal ideation; were housebound or in need of intensive or inpatient treatment; completed a course of CBT for OCD in the past 12 months; or were receiving ongoing psychological treatment for OCD or an anxiety disorder.
Participants were randomly assigned in a 1:1 ratio to receive either stepped care or in-person CBT through an online automated randomization system (randomize.net). This was done by several assigned clinicians according to a delegation list. The system was provided and monitored by an independent clinical trials unit, the Karolinska Trial Alliance (https://karolinskatrialalliance.se/en/), and the randomization sequence was inaccessible to the research team. The randomization sequence (block sizes of 6 individuals) was stratified by age (children aged 7-12 years or adolescents aged 13-17 years) and source of participant referral (clinician-referred or self-referred).
The group allocation was concealed from the masked raters who conducted the pretreatment, posttreatment, 3-month follow-up, and 6-month follow-up assessments. To test masking integrity, the raters were asked to note if participants had inadvertently revealed their allocated group and to guess treatment allocation after each assessment. Assessors who became unblinded were replaced and the recorded CY-BOCS interview was re-rated by a new rater. The trial was unblinded after the last participant completed the 6-month follow-up assessment.
Participants in the stepped-care group first underwent an internet-delivered CBT program for 16 weeks.12,14,20 Two age-adapted versions of the program were used: one for children aged 7 to 12 years and another for adolescents aged 13 to 17 years. Both versions had 14 modules that consisted of text, movies, and exercises that were presented in an engaging and varied way. The treatment components were the same as in the face-to-face CBT treatment manuals: education, exposure with response prevention, and relapse prevention. The parents had access to a parallel online intervention that focused specifically on reducing family accommodation and the use of positive reinforcement. Each family was assigned a personal therapist throughout the whole treatment. Communication was in the form of asynchronous, written messages through the encrypted online platform. Additional telephone support was provided on demand. A video illustrating the main features of the program can be found at https://vimeo.com/355965105/b3d5d1c439.
Participants in the control group received manualized in-person CBT21 with up to 14 sessions delivered over 16 weeks by a personally assigned therapist. Adaptations regarding degree of parental involvement, home visits, and longer sessions were made depending on individual needs. The sessions were audiotaped, and a random 20% of the sessions were selected and rated by independent clinical psychologists to ensure therapists’ adherence to the treatment manual. The ratings indicated excellent therapist adherence (eMethods 2 in Supplement 2).
Participants in either group who were classified as nonresponders at the 3-month follow-up were offered up to 12 sessions of in-person CBT between the 3-month follow-up and 6-month follow-up. The reason for waiting until the 3-month follow-up was to capture the full effect of internet-delivered CBT, because previous trials have indicated a continued improvement beyond completion of the treatment.14,22
The therapists involved in the trial were highly experienced and treated participants in both groups (eTable 2 in Supplement 2); had training in the trial protocol; and received supervision by senior clinicians during the weekly team meetings and, occasionally, during the treatment session. A more detailed description of the interventions is in the study protocol16 and in eMethod 1, eTable 1, and the eFigure in Supplement 2.
The primary outcome measure was the masked assessor–rated CY-BOCS score, a semi-structured clinician-administered interview that assesses OCD symptom severity in children and young adults. The CY-BOCS consists of 10 items that are rated on a 5-point Likert scale, ranging from 0 (no symptoms) to 4 (extreme symptoms), that are summed to yield a total score ranging from 0 to 40, with higher scores indicating greater symptom severity.17 All therapists and assessors received both initial and regular (quarterly) training in the use of the measure. The interrater reliability for the CY-BOCS during the trial was high (intraclass correlation coefficient, 0.99 [95% CI, 0.98-1.00]).
Secondary outcome measures were the masked assessor–rated Clinical Global Impressions (CGI) Severity (CGI-S) and Improvement (CGI-I) scales,23 the masked assessor–rated Children’s Global Assessment Scale,24 the Obsessive-Compulsive Inventory-Child version,25 the Children’s Obsessional Compulsive Inventory-Revised-Parent version,26 the Family Accommodation Scale-Self Rated,27 the Work and Social Adjustment Scale-Youth version and Work and Social Adjustment Scale-Parent version,28 the Mood and Feelings Questionnaire Child and Parent versions,29 the Insomnia Severity Index,30 and the Child Health Utility 9D.31 The findings of the Trimbos/iMTA Questionnaire for Costs Associated With Psychiatric Illness will be published separately. Detailed information on these secondary outcome measures, anchors, score ranges, and interpretation can be found in eMethod 3 and eTable 3 in Supplement 2. All self-rated and parent-rated measures were completed online.
Adverse events were recorded online at midtreatment (week 8), posttreatment (week 16), and 6-month follow-up assessments (eMethod 3 in Supplement 2). Safety aspects and data quality were monitored quarterly by the Karolinska Trial Alliance.
Power was estimated based on repeated measures (each participant being assessed 4 times), using a linear random intercept model with 1000 bootstrap samples using data from a previous randomized clinical trial.15 The enrollment of 152 participants provided an estimated power greater than 95% to reject the null hypothesis that stepped-care treatment would be inferior to in-person CBT alone, after allowing for 10% data attrition.
The noninferiority margin was set at 4 points, which was decided a priori based on clinical judgement and following recommendations to use half of the mean controlled effect size from historical trials.32 A 4-point margin corresponds to approximately half of the effect of CBT vs pill and psychological placebo conditions from a meta-analysis.33 From a clinical perspective, the chosen margin is considered to be a small clinical difference in the CY-BOCS score and is more conservative than previous noninferiority trials in OCD.21,34
The noninferiority analysis was based on the CY-BOCS score at the 6-month follow-up (primary end point) and was determined by a 1-sided 97.5% CI of the mean difference between the 2 groups using mixed-effect regression analyses for repeated measures (all assessment points included), which are valid under the assumption that the data are missing at random.35
All randomized participants were analyzed according to their randomization groups and included in the primary prespecified analysis (Figure 1). No data imputation method was used because the percentage of missing data was less than 10%.16,36 The mixed-effect regression model included fixed effects of time, treatment group, and site (Stockholm vs Gothenburg) and an interaction effect of group × time, as well as random intercept and random slope to account for individual differences. An interaction effect of group, time, and source of referral was included in the model to explore whether source of referral was a meaningful moderator of treatment effect. A planned sensitivity analysis was conducted in which participants who violated protocol (ie, made medication changes or received additional psychological treatment for OCD) were excluded.
Secondary outcomes were analyzed with the above-mentioned regression model using 2-sided tests with α = .05. Because of the potential for type I error due to multiple comparisons, findings for analyses of secondary outcomes should be considered exploratory. Ordinal variables (CGI-I and CGI-S) were analyzed with ordinal logistic regression. Other outcome measures (eg, treatment credibility and satisfaction) were analyzed with independent t tests. Between-group effect sizes were estimated with bootstrapped Cohen d (eMethod 4 in Supplement 2). χ2 tests were used to determine whether the masked assessors’ guesses on treatment allocation were better than chance.
The percentage of participants classified as responders and remitters was compared using logistic regression. Responder status was defined as at least 35% reduction on the CY-BOCS and a CGI-I score of 1 or 2, and remitter status was defined as a score of 12 or less on the CY-BOCS and a CGI-S score of 1 or 2.37
All statistical analyses were conducted using Stata, version 16.1. The final data set, statistical analysis plan, and Stata code were locked at the Karolinska Institutet electronic laboratory notebook system prior to unblinding and data analysis.
Of the 889 children and adolescents screened, 152 were enrolled in the trial (74 randomized to receive stepped-care treatment and 78 to receive in-person CBT) between October 6, 2017, and May 24, 2019 (Figure 1). The mean (SD) age at baseline was 13.4 (2.5) years, 94 participants (62%) were females, and 52 (34%) had at least 1 psychiatric comorbidity. A total of 110 participants (72%) were clinician referrals to the 2 study sites and the remainder were self-referrals. Sociodemographic and clinical characteristics of the participants are shown in Table 1. Baseline characteristics were similar across study sites (eTable 4 in Supplement 2).
The data attrition on the primary outcome was very low, with CY-BOCS scores available for 151 of 152 participants (99%) at the 6-month follow-up (primary end point; Figure 1). The mean CY-BOCS score at the pretreatment assessment was 22.96 in the stepped-care group and 22.95 in the in-person CBT group. At the 6-month follow-up, the mean CY-BOCS score was 11.57 points in the stepped-care group and 10.57 points in the in-person CBT group (Table 2; Figure 2A); the estimated mean difference was 0.91 points (1-sided 97.5% CI, −∞ to 3.28), which met the 4-point criterion for noninferiority (P value for noninferiority = .02; Figure 2B). Individual participant outcomes are shown in Figure 3. There was no significant effect of referral source (clinician-referred vs self-referred; B = 0.07; Z = 0.86; P = .39).
In a sensitivity analysis, the main analysis was repeated excluding 10% (n = 16 [8 in each group]) of participants who broke protocol due to clinical needs (made changes in concomitant medication and/or received additional psychological treatment for OCD during the study period). The inference of noninferiority was robust to these protocol deviations (mean [SD] CY-BOCS score of 11.20 [6.18] points in the stepped-care treatment group vs 10.39 [7.39] in the face-to-face treatment group; estimated mean difference, 0.75 [1-sided 97.5% CI, ∞ to 3.20]).
No significant between-group differences were found in any of the secondary outcome measures (Table 2; eTable 5 and eTable 6 in Supplement 2).
At the 6-month follow-up, 50 participants (68%) in the stepped-care group and 52 (68%) in the face-to-face treatment group were classified as responders (odds ratio, 1.00 [95% CI, 0.51-1.98]; P = .99). At the same time point, 36 participants (49%) in the stepped-care group were in remission, compared with 46 (60%) in the face-to-face treatment group (odds ratio, 0.64 [95% CI, 0.34-1.22]; P = .17) (eTable 7 in Supplement 2).
Participants randomized to the stepped-care group completed a mean (SD) of 10.49 (3.69) online modules, and parents of participants in that group completed a mean of 10.97 (3.44) modules. Participants in the face-to-face treatment group received a mean (SD) of 11.65 (2.74) CBT sessions during the first step of treatment. The mean (SD) therapist time for the first treatment step was 336.84 (217.56) minutes per family for internet-delivered CBT and 741.81 (263.54) minutes per family in the in-person CBT group.
At the 3-month follow-up, 40 participants (54%) in the stepped-care group and 53 (71%) in the in-person CBT group were responders (Figure 1). The 29 nonresponders in the stepped-care group who accepted the offer of face-to-face CBT received a mean (SD) of 7.76 (3.15) additional in-person therapy sessions. The 16 nonresponders in the in-person CBT group received a mean (SD) of 7.56 (2.68) additional face-to-face sessions during the second step of treatment. In total, the mean (SD) therapist time for the entire study period was 526.18 (370.99) minutes per family in the stepped-care group and 849.10 (398.55) minutes in the in-person CBT group (eTable 8 in Supplement 2).
The participants deemed the face-to-face treatment more credible than internet-delivered CBT in early stages of treatment (week 2), but reported equal working alliance and satisfaction (eTable 9 in Supplement 2). Adherence to treatment was high in both groups (eTable 9 in Supplement 2). Four participants revealed their group allocation at the 6-month follow-up (eResult 1 in Supplement 2). The masked raters’ guesses of group allocation were correct in 58% of the cases, which is not significantly different from chance (X2(1) = 3.73; P = .053).
The percentage of participants with adverse events was comparable between the groups (47 [64%] in stepped-care group and 52 [67%] in face-to-face treatment group). Most events were mild and subsided between the assessments (eTable 10 in Supplement 2). Two serious adverse events (1 in each group) were reported and judged to be unrelated to the treatment: 1 participant was admitted to inpatient care due to rapid weight loss (anorexia) and 1 was admitted due to increased anxiety, suicidal thoughts, and self-harm following a stressful psychosocial situation.
Among children and adolescents with OCD, treatment with a low-intensity digital intervention followed by a course of in-person CBT if necessary resulted in a noninferior difference in OCD symptoms compared with in-person CBT alone. Improvements in all secondary outcomes were observed, including self- and parent-reported OCD symptoms, depression, insomnia, general function, quality of life, and parental accommodation of OCD symptoms, with no statistically differences between the groups at the primary end point.
Fifty-four percent of the participants were classified as treatment responders 3 months after receiving internet-delivered CBT and did not require further in-person treatment. Thus, by first offering a low-intensity intervention in a stepped-care fashion, it was possible to reduce the number of patients requiring face-to-face CBT. A health economic evaluation of this trial will be published separately.
The results extend the findings of the previous waitlist-controlled trial in adolescents with OCD15 by improving the treatment protocol, expanding the parental support modules, including both children and adolescents, and using a noninferiority design.
Although the majority of participants responded to treatment at the 6-month follow-up, about one-third still did not benefit sufficiently (some having received 2 courses of in-person CBT). The percentage of participants in remission ranged from 49% to 60%. A logical next step would be to attempt a course of a selective serotonin reuptake inhibitor, because there is some support for the efficacy of sequentially offering selective serotonin reuptake inhibitor medication to patients who do not benefit sufficiently from an initial course of CBT.38 These results highlight the importance of continuous research into effective ways to further augment the effects of CBT for pediatric OCD.
Some strengths of this trial were a low rate of data loss, a narrower noninferiority margin compared with previous trials,21,34 and the consideration of the source of participant referrals. Previous internet-delivered CBT trials have primarily recruited self-referred individuals,12,14,20 raising concerns about the generalizability of the results to clinic-referred participants.39 In the current study, only 28% of the participants were self-referred and the source of referral (clinician referral vs self-referral) did not moderate the results. The current trial had a long controlled study phase of 10 months in total, which is unusual in trials of psychological interventions.
This study has several limitations. First, the study was conducted in Sweden, where there is a very limited number of specialist OCD clinics. Although the results may generalize to countries with similar shortages and health care systems, they may translate less well to other health care settings. In countries in which specialist treatment for OCD is more widely available, a stepped-care approach would be less likely to be of value because it would mean delaying symptom remission. In the current study, specialist treatment was delayed 7 months for those who did not respond to the initial low-intensity intervention. Second, the presence of co-occurring conditions, such as autism spectrum disorder, were exclusion criteria. Additional studies are needed to evaluate whether these findings apply to patients with these conditions.
Among children and adolescents with OCD, treatment with an internet-delivered CBT program followed by in-person CBT if necessary compared with in-person CBT alone resulted in a noninferior difference in symptoms at the 6-month follow-up. Further research is needed to understand the durability and generalizability of these findings.
Corresponding Author: Kristina Aspvall, PhD, Child and Adolescent Psychiatry Research Centre, Gävlegatan 22, SE-113 30 Stockholm, Sweden (firstname.lastname@example.org).
Accepted for Publication: February 28, 2021.
Author Contributions: Ms Aspvall and Dr Serlachius had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Drs Mataix-Cols and Serlachius contributed equally to this work and share last authorship.
Concept and design: Aspvall, Andersson, Silverberg-Mörse, Feldman, Lenhard, Mataix-Cols, Serlachius.
Acquisition, analysis, or interpretation of data: Aspvall, Andersson, Melin, Norlin, Eriksson, Vigerland, Jolstedt, Wallin, Sampaio, Bottai, Lenhard, Mataix-Cols, Serlachius.
Drafting of the manuscript: Aspvall, Andersson, Sampaio, Lenhard.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Aspvall, Andersson, Sampaio, Bottai, Lenhard.
Obtained funding: Aspvall, Andersson, Melin, Lenhard, Mataix-Cols, Serlachius.
Administrative, technical, or material support: Aspvall, Andersson, Melin, Norlin, Vigerland, Jolstedt, Silverberg-Mörse, Wallin, Serlachius.
Supervision: Aspvall, Andersson, Feldman, Lenhard, Mataix-Cols, Serlachius.
Conflict of Interest Disclosures: Dr Jolstedt reported receiving grants from The Swedish Research Council for Health, Working Life and Welfare (Forte 2014-4052) and from Stockholm County Council (HNSV 14099) during the conduct of the study. Dr Mataix-Cols reported receiving personal fees from UpToDate, Wolters Kluwer Health, and Elsevier outside the submitted work. No other disclosures were reported.
Funding/Support: This study was funded by the Swedish Research Council for Health, Working Life and Welfare (grant number 2014-4052), ALF Medicin Project Grant by Region Stockholm (grant number 20160247), and Jane and Dan Olssons Foundation (grant number 2016-64). Dr Serlachius was supported by Region Stockholm (clinical research appointment, grant number 20170605). Ms Aspvall was supported by grants from Drottning Silvias Jubileumsfond, Fonden för psykisk hälsa, and Fredrik och Ingrid Thurings stiftelse.
Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Data Sharing Statement: See Supplement 3.
Additional Contributions: We wish to express our gratitude to all the therapists and blind assessors involved in the trial, who did not receive compensation: Johanna Alaeus, MSc; Mathilde Annerstedt, MSc; Moa Holmsved, MSc; Fredrik Johansson, MSc; Martin Johansson Bellander, PhD; Malin Lavner, MSc; Martin Persson, MSc; Helene Ringberg, MSc; Karin Sundström, MSc; Frida Wickberg, MSc; and Simon Åberg MSc (Stockholm Health Care Services, Region Stockholm, Stockholm); Gustaf Brander, PhD (Department of Clinical Neuroscience, Karolinska Institutet, Stockholm); Katarina Evervall, RN; Malin Glänneskog, MSc; Marcus Hedkvist, BSc; Fredrika Jensen, MSc; Rachel Kamienski, MSc; Ingrid Korsoski, MSc; Ann-Christine Lundblad, RN; Anna Lundh, MSc; Kristina Näsström, MSc; and Stina Sundin MSc (Region Västra Götaland, Sahlgrenska University Hospital, Gothenburg). We also wish to thank the following colleagues who received salary for their work in the trial: Josefine Häger, MSc (Region Stockholm), for administrative support and Naomi Lisai, MSc (clinical services in Region Uppsala), and Lina Poltrago, MSc (clinical services in Region Stockholm), for their assessments of therapist adherence. This work used the BASS platform from the eHealth Core Facility at Karolinska Institutet, which is supported by the Strategic Research Area Healthcare Science.