Lundström2022 - Effect of Internet-Based vs Face-to-Face CBT for Adults With Obsessive-Compulsive Disorder §

• Type:#article
• Year read:#read2022
• Subject: _thesis defense MOC OCD ICBT
• Bibtex: @lundstrom2022
• Bibliography: Lundström, L., Flygare, O., Andersson, E., Enander, J., Bottai, M., Ivanov, V. Z., Boberg, J., Pascal, D., Mataix-Cols, D., & Rück, C. (2022). Effect of Internet-Based vs Face-to-Face Cognitive Behavioral Therapy for Adults With Obsessive-Compulsive Disorder: A Randomized Clinical Trial. JAMA Network Open, 5(3), e221967. https://doi.org/10.1001/jamanetworkopen.2022.1967

---

Reviewer comments §

46.02 - Response to reviewer comments

• Trial protocol
  • Reduces researcher degrees of freedom, prevents p-hacking.
• Interim analyses
  • We wanted to confirm that the results would be informative and that the assumptions made for the power calculation were correct. They were in line with our power calculation.
• F2F protocol comprehensive compared to clinical practice
  • Yes, but this is the gold-standard protocol so we wanted to compare ICBT to the best possible treatment out there.
• Missingness
  • It’s unfortunate that missingness was relatively high in the F2F-group relative to ICBT. But we still think it’s within acceptable limits. The repeated measures model for Y-BOCS makes the estimates accurate because we have many assessment points even though one may be missing.
• Lower power for interactions
  • Yes, so those should be considered exploratory and not confirmatory.
• 120 participants but 166 guesses made? (issue 4.12)
  • Post-trt and 3-month follow-up (240 possible guesses)
  • Randomization revealed at 12 occassions and rest lost due to administrative error.
• Since unguided was unlocked all at once, could there have been an effect of the pacing beyong lack of therapist support? (issue 4.19)
  • Yeah, the patients may have been overwhelmed by all the content that was immediately available. But on the other hand, for some it may have been positive to go straight to parts that were relevant to them.
  • But in Musiat2022 - Impact of guidance on intervention adherence in computerised interventions for mental health problems, a meta-analysis we see that guidance has a positive effect on adherence.
• Quadratic time term better fit?
  • Pre-registered study protocol, but we reported quadratic in supplement. This did not alter the conclusions from the main analysis.
• Therapist competence only assessed in F2F, not ICBT
  • Yes, ideally we would address competence in ICBT messages too since there are meaningful differences. Since the start of the study there has been some development of scales and this would have added to the study.
  • But, same therapists in both treatment arms so therapist factors are controlled for.

Non-inf issues §

• ITT versus PP analysis
  • ITT are not conservative in non-inferiority since they may show smaller trt differences. Per protocol can be a better approach, we included it in supplement and did not find differences.
• 3-point margin
  • We had power to detect a 3-point difference, we see that the point difference was 2.1 and 90% CI reached to 4.6, so the CI was 2.5 points wide. However, this narrow margin assumed that the point difference would be 0 which was not the case.
• 90% CI
  • We are only concerned about one side of the CI (upper end), so we could increase power by having a 90% CI rather than 95% CI.
  • Sensitivity analyses with 95% CI did not alter the conclusions of the non-inferiority hypothesis.