Effect of Internet-Based vs Face-to-Face Cognitive Behavioral Therapy for Adults With Obsessive-Compulsive Disorder

Key Points Question Are therapist-guided, internet-based cognitive behavioral therapy (ICBT) and unguided ICBT noninferior to traditional face-to-face CBT for the treatment of obsessive-compulsive disorder (OCD)? Findings In this noninferiority randomized clinical trial that included 120 adults with OCD, participants in all 3 treatment groups had significantly improved symptoms, but the noninferiority of ICBT could not be established. The health economic evaluation indicated that both therapist-guided and unguided ICBT were cost-effective compared with face-to-face CBT. Meaning These findings suggest that ICBT can be a cost-effective alternative for the treatment of OCD in health care contexts where access to traditional CBT is in short supply.


26
Introduction Expert guidelines recommend cognitive behaviour therapy (CBT) as a first line 27 treatment for Obsessive-Compulsive Disorder (OCD) but the majority of patients with OCD 28 do not have access to CBT. Internet-delivered CBT (ICBT) has the potential to make this 29 evidence-based treatment more accessible whilst requiring less therapist time than traditional 30 face-to-face (f2f) CBT. Data from six clinical trials suggests that ICBT for OCD is both 31 efficacious and cost effective but whether ICBT is non-inferior to traditional f2f CBT for 32 OCD is yet unknown. 33 34 Methods and analysis A single-blind, randomized controlled non-inferiority trial comparing 35 therapist-guided ICBT, unguided ICBT, and individual (f2f) CBT for adult OCD patients. The 36 primary objective is to investigate whether ICBT is non-inferior to gold standard f2f CBT. 37 Secondary objectives are to investigate if ICBT is equally effective when delivered unguided, 38 to establish the cost-effectiveness of ICBT, and to investigate if the treatment outcome differs 39 between self-referred and clinically-referred patients. Participants will be recruited at two 40 specialist OCD clinics in Stockholm, and also through online self-referral. Participants will be 41 randomized to one of three treatment conditions: F2f CBT, ICBT with therapist support or 42 unguided ICBT. The total number of participants will be 120 and masked assessments will be 43 administered at baseline, bi-weekly during treatment, at post-treatment, and at 3-and 12-44 months follow-ups. The main outcome measure is the clinician-rated Yale-Brown Obsessive 45 Compulsive Scale (Y-BOCS) at 3-month follow-up. The margin of non-inferiority is set to 3 46 points on the Y-BOCS using a 90% confidence interval. 47 48 Ethics and dissemination The study has been approved by the Regional Ethics Board of 49 Stockholm (REPN 2015/1099-31/2) and registered at Clinicaltrials.gov (NCT02541968). The 50 study will be reported in accordance with the CONSORT statement for non-pharmacological 51 trials. The results will be published in peer-reviewed academic journals and disseminated to 52 patient organizations and media. 53 54 55

57
• First study evaluating if two modalities of ICBT are non-inferior to the gold standard 58 face-to-face CBT for OCD. 59 • Full health economic evaluation of therapist-guided ICBT, unguided ICBT and f2f 60 CBT for OCD. 61 • Recruitment of both clinic-referred and self-referred patients, which will help 62 generalise the results to more typical OCD cases. 63 •

154
Study design 155 Single-blind, randomized controlled non-inferiority trial comparing therapist-guided ICBT, 156 unguided ICBT without therapist support, and individual f2f CBT for OCD in adults. The total 157 number of participants will be 120 (40 per group), with stratification according to source of 158 referral (self-vs. clinic referred patients). Block randomization will be performed within each 159 stratum to ensure all participants are equally represented across treatment conditions. 160 Participants will be assessed at baseline, bi-weekly during treatment, at post-treatment, and at 161 3-and 12-months follow-ups. The CONSORT flowchart of the trial is depicted in Figure 1. Sample selection 169 Regular patients referred to two OCD specialist clinics in Stockholm will be assessed for 170 eligibility. The trial will also be advertised online so that interested participants can self-refer 171 by registering on the trial's secure webpage and completing a screening questionnaire. People 172 living in Stockholm, Södermanland or Uppsala County are eligible to participate in the study 173 (these counties are within 1 to 2 hours travel distance to Stockholm). 174 175 After completing an online screening, a clinical psychologist will contact potentially suitable 176 participants by telephone for a brief screening interview. They will then be offered an 177 appointment with a psychiatrist at one of the two OCD specialist clinic for a full psychiatric 178 assessment. The psychiatrist will administer the Mini International Neuropsychiatric 179 Interview (MINI)(27) and The Structured Clinical Interview for DSM-5 (SCID-5)(28) to 180 confirm the diagnosis of OCD, document psychiatric comorbidities, administer baseline 181 instruments, and decide on inclusion/exclusion. The randomization sequence will be generated by Karolinska Trial Alliance (KTA,  187 https://karolinskatrialalliance.se, an independent entity not involved in the study) before 188 inclusion of the first participant, using masked block randomization. Patients will receive their 189 randomization number based on the order of their first psychiatrist appointment. Patients will 190 be stratified based on self-or clinical referral. Sealed envelopes with information on treatment 191 allocation will be stored in a secure locker in case of emergency unblinding. 192 193 Assessors will be blind to group assignment up to the 12-month follow-up. To ensure that the 194 blinding is maintained, patients will be given clear instructions not to disclose which 195 treatment they have been randomized to while being interviewed by the blind assessors. 196 Where blindness is inadvertently broken, raters will be immediately replaced and the 197 participant re-assessed by another rater. Blind raters will be asked to guess each patient's 198 group allocation at each assessment point.(29) This will establish if the blind raters' guesses 199 regarding treatment allocation were better than chance.

201
Interventions 202 Therapist-guided ICBT. Patients will receive 14 weeks of ICBT for OCD using a validated 203 treatment protocol.(13-16) As in regular CBT for OCD, the main treatment component is 204 exposure and response prevention (ERP). The therapists will be licensed psychologists with 205 expertise in treating patients with OCD. Therapists will respond to messages encrypted in the 206 Internet platform at a set time during office hours (8am-5pm) on weekdays, in order to ensure 207 that participants receive a response within 24 hours. Each participant's response rate at the 3-208 month follow-up will be calculated and monitored by the project leaders. The participants 209 who are non-responders (defined as Y-BOCS reduction < 35% and CGI-I >2)(30) at the 3-210 month follow-up will be contacted by telephone and offered face-to-face CBT for 14 weeks. 211 212 Self-guided ICBT. This arm will be identical to the ICBT described above but without any 213 online therapist support. If participants experience any technical problems with the online 214 platform during the treatment, they can contact project leaders for help. In the internet 215 platform, patients will have detailed contact information in case of emergency. Participants in 216 this group who are non-responders at 3-month follow-up will be offered up to 14 weeks of f2f 217 CBT according to the same procedure explained in the previous section. 218 219 Individual f2f CBT. Patients receive 16 sessions of individual f2f CBT for OCD delivered 220 over a time period of 14 weeks, according to a validated protocol.(31) Sessions will be held 221 twice weekly during the first two weeks and once a week for the remaining 12 weeks. The 222 therapists will be licensed psychologists with expertise in treating patients with OCD. The 223 content of the f2f CBT is the same as in the ICBT arms. Sessions will be audiotaped in order 224 to ensure that the therapists adhere to the treatment protocol. Adherence to protocol will be 225 independently rated by a psychologist (not otherwise involved in the study) specialized in 226 CBT treatment for OCD. 227 228 Sample size calculation 229 In order to provide accurate estimates for the power calculation in the current trial, we will 230 used individual-level data from a previous study of therapist-guided ICBT with identical Y-231 BOCS assessments by blinded raters and six repeated observations.(14) To calculate the 232 required sample size, we used a bootstrap simulation with 1000 samples using the following 233 assumptions, based on data from the previous trial: a variance of the random intercept of 10.5, 234 a variance for the random slope of 0.04, and a within-individual residual variance of 20.4. 235 With 3 treatment groups and 8 observations (Y-BOCS) per patient, we estimated that a total 236 of 120 participants would be needed to detect a slope difference between two groups (i.e. 237 group 1 vs. group 2 and group 1 vs. group 3) of 3 points at 3-month follow-up with over 90% 238 power. We will request an interim power analysis by the Karolinska Trial Alliance to test 239 these assumptions, using data from 80 individuals, and adjust sample size if power is lower 240 than anticipated (see supplementary file 1 for a detailed description). 241 242 Measurements 243 Table 2 lists clinician-rated and self-rated assessments at the different time points. 244 The primary outcome measure is the clinician-rated Yale-Brown Obsessive Compulsive Scale 245 (Y-BOCS), the gold standard for assessing the severity of OCD symptoms.(32) Clinicians in 246 this trial will practice together on case examples to establish high inter-rater reliability. The 247 Y-BOCS will be administered by blind raters at baseline, at weeks 2, 4, 6, 8, 10 and 12 during 248 treatment, at post-treatment (week 15), and at 3-and 12-months follow-ups. The primary 249 endpoint is the 3-month follow-up. (PEAS)(39) will be used to quantify compliance with ERP homework and the Working 259 Alliance Inventory -Short Form (WAI-SF)(40) will be used to measure therapeutic alliance 260 in the face-to-face CBT and ICBT with therapist support treatment conditions. The Insomnia 261 Severity Index (ISI)(41) will be used to measure participants sleep patterns and the Treatment 262 Credibility Scale (TCS)(42) will be used to measure how credible participants perceive the 263 treatment to be. Measurements will be administered before and after treatment as well as 264 during 3-and 12 months follow-ups. In order to increase participant retention at follow-up 265 assessments, participants will be notified via text message 48 hours prior to an appointment. 266 Should a participant not attend a follow-up session, a psychiatrist will contact participants via 267 telephone to perform the assessments. Safety and adverse events 281 Data on adverse events and suicidal ideation will be collected by blinded independent raters 282 bi-weekly during treatment, at post-treatment and at 3-and 12-month follow-up. Adverse 283 events will be collected using a standardized checklist, the Safety Monitoring Uniform Report 284 Form (SMURF).(43) If a participant expresses suicidal ideation (i.e. a score on item 9 of the 285 MADRS-S ≥ 4), assessors will initiate a structured suicide risk assessment. If there is an 286 urgent need for psychiatric care, a trial psychiatrist will contact participants to schedule a 287 face-to-face appointment as soon as possible. 288 289 Statistical analysis 290 The main outcome analyses will be conducted according to the "intent-to-treat" principle. 291 Mixed-effects regression analyses for repeated measures with maximum likelihood estimation 292 (MLE) of parameters will be used with the assumption that data are missing at random. The 293 latter assumption will be tested. For each outcome measure, the model will include fixed 294 effects of time (baseline, mid-treatment, post-treatment, and 3-month follow-up [primary 295 endpoint]), treatment group (guided ICBT, unguided ICBT, f2f CBT) and an interaction effect 296 of treatment group x time to allow for the differential change between the three groups from 297 baseline to the 3-month follow-up. The models will include individuals' random intercept and 298 random slope to account for variability between and within participants over time.  and between-group effect sizes will be calculated with Cohen's d.(44) Numbers needed to 300 treat will be calculated based on responder status. 301 302 Alpha for all analyses will be set at 0.05. Non-inferiority is established when the 90% Wald 303 confidence interval for the difference between treatment conditions excludes the pre-specified 304 margin of inferiority, which is set at 3 points on the Y-BOCS (45, 46). This means that if the 305 upper limit of the 90% confidence interval is less than 3 points, we are 95% confident that 306 ICBT will be non-inferior to f2f CBT. The non-inferiority hypothesis will be tested of both 307 therapist-guided and self-guided ICBT against the f2f CBT. Additional analyses of the 12-308 month follow-up data will determine whether the treatment gains are maintained long-term 309 and whether ICBT is non-inferior to f2f CBT at follow-up. 310 311 Cost-effectiveness analysis 312 Health economic data will be collected using the TIC-P(47) and the Swedish National Patient 313 Register, the Swedish Prescribed Drugs Register and the longitudinal integrated data-base for 314 health insurance and work-related research (LISA). Costs will be analysed using a societal 315 perspective i.e. including both sick-leave, hospitalizations, service use, medication, etc. and 316 analysed in relation to outcome (i.e. OCD symptoms and quality-adjusted life years using the 317 Y-BOCS and EQ-5D, respectively). National tariffs will be used to estimate costs from health 318 care visits. Productivity losses will be estimated using gross earnings data from each 319 patient.(48) Treatment costs, i.e. therapist support time per patient logged on the platform and 320 time spent on f2f sessions, will be included in the cost estimation. 321 322 Cost-effectiveness comparisons will be analysed using incremental cost-effectiveness ratios. 323 The "net benefit approach" will also be used. This approach estimates the cost-effectiveness 324 depending on different societal willingness-to-pay values for one unit of improvement.(48) 325 Non-parametric bootstrapping (one thousand replications) will be used to estimate the 326 difference between ICBT (guided or unguided) and gold standard f2f CBT. 327 328 Analysis of predictors and moderators 329 We will analyse predictors and moderators of response and remission status at 3-and 12-month 330 follow-up using repeated k-fold cross validation with 10 folds and 20 repeats to reduce the risk 331 of model instability (49, 50). We then average model performance over the repeats using area 332 under the receiver operating characteristic curve (AUC) of sensitivity and specificity to 333 distinguish between responders/remitters and non-responders/non-remitters (51). 334 335 Limitations 336 There are several potential threats to the validity and generalizability of the current trial 337 results, some of which apply to most clinical trials. First, the trial was designed to maximise 338 the chances of the results being as generalizable as possible. However, despite the best of our 339 efforts to recruit both clinic-and self-referred individuals, it will be difficult to confidently 340 claim that our participants will be representative of the entire population of OCD patients in 341 Sweden

372
The trial will be conducted in compliance with this study protocol, the Declaration of Helsinki 373 and Good Clinical Practice (GCP). Karolinska Trial Alliance (KTA) is an external party that 374 will monitor the study every 6 months and ensure that the study follows GCP, i.e. that all 375 participants give informed written consent and that study related materials are handled 376 correctly. All professionals involved in the study will attend a course in GCP and get certified 377 by the KTA. 378 379 The study has been approved by the Regional Ethics Board of Stockholm (REPN 2015/1099-380 31/2) and registered at Clinicaltrials.gov (NCT02541968), and will be reported in accordance 381 with the CONSORT statement for non-pharmacological trials. OCD is associated with significant suffering, loss of function across multiple life domains, 395 high suicide risk, and large societal costs. ICBT has great potential to increase access to 396 evidence-based care for a large group of sufferers that normally do not receive evidence-397 based psychological treatments. The study outlined in this protocol is the first direct 398 comparison of ICBT and gold standard f2f CBT and is a crucial step before ICBT can be 399 recommended for use within the regular health-care system. The study will provide new 400 insights into the effectiveness of different treatment modalities for OCD and the health 401 economic evaluation will help decision-makers to rationally allocate available resources.