Department for Psychotherapy and Biopsychosocial Health, Danube University Krems;
Thomas Berger
Department of Clinical Psychology and Psychotherapy, University of Bern
Björn Meyer
GAIA AG, Hamburg, Germany, and Department of Psychology, City, University of London
Christina Späth
Department of Psychiatry and Psychotherapy, Lübeck University
Johanna Schröder
Institute for Sex Research and Forensic Psychiatry, University Medical Center Hamburg-Eppendorf
Fritz Hohagen
Department of Psychiatry and Psychotherapy, Lübeck University
Steffen Moritz
Department of Psychiatry and Psychotherapy, University Medical Center Hamburg-Eppendorf
Jan Philipp Klein
Department of Psychiatry and Psychotherapy, Lübeck University
Acknowledgement: Jan Philipp Klein received payment for presentations, workshops, and books on psychotherapy for chronic depression and psychiatric emergencies. Björn Meyer is employed as research director at GAIA AG, the company that developed, owns, and operates the web intervention investigated in this trial. All other authors and members of the EVIDENT study team report no financial or other relationship relevant to this article. Trial registration: ClinicalTrials.gov (identifier: NCT01636752).
Funding was provided by German Federal Ministry of Health Grant II A 5-2512 FSB 052. The funding body played no role in study design, data collection, analysis, or interpretation of the data. Additional funding was provided by GAIA AG (Hamburg, Germany), which provided all participants in the trial with licenses for the web intervention (Deprexis) as well as technical support. GAIA AG played no role in data collection, data analysis, or the decision to publish the results.
Deprexis is an Internet-based psychological intervention for depression (
For personalized treatment selection (
The EVIDENT study (
The present study evaluated whether specific anxiety-related comorbidities would moderate the intervention effect on depression outcome. Evaluating anxiety-related disorders as moderators of outcomes in interventions for depression is important, not only because these disorders are highly comorbid but also because anxiety-related comorbidities put individuals with depression at risk of a negative outcome (e.g.,
RQ 1: Do specific anxiety-related comorbidities moderate the intervention effect on depression outcome?
The following eight specific anxiety-related comorbidities were evaluated as moderators within one statistical model: generalized anxiety disorder, social phobia, panic disorder, agoraphobia, panic disorder with agoraphobia, specific phobia, posttraumatic stress disorder, and obsessive–compulsive disorder. The statistical model included 12 moderators in total. Besides the eight specific anxiety-related comorbidities, the three previously identified significant moderators in the EVIDENT trial (antidepressant medication, concomitant psychosocial treatment, attitudes toward Internet interventions) were added to control for the known moderators, and initial depression severity was entered to explore whether moderating effects are specific to anxiety or more general to the severity of depression. Continuous scale scores (RQ 1a) as well as clinically important changes (RQ 1b) were used as depression outcomes.
RQ 2: If one or more significant moderator(s) emerged when including all 12 moderators, would this result depend on the presence of the other moderators in the statistical model?
RQ 3: Can moderating effects be replicated in randomly selected split halves of the sample?
The study was approved by the Ethics Committee of the German Psychological Association (SM 04_2012) and registered in ClinicalTrials.gov (NCT01636752). See the
Inclusion criteria were as follows: mild to moderate depressive symptoms, between 5 and 14 points on the Patient Health Questionnaire−9 (PHQ-9;
In total, N = 1,013 participants were randomized to the following two conditions: Deprexis group with access to care-as-usual (Deprexis; n = 509, M = 42.8 years old, 68.8% female) or control group with access to care-as-usual (CG; n = 504, M = 42.9 years old, 68.5% female). Randomization (1:1) to the two groups (Deprexis or CG) was stratified by PHQ-9 (PHQ-9 < 10 vs. PHQ-9 ≥ 10), because participants with PHQ-9 < 10 (mild depressive symptoms) received unguided Deprexis, and participants with PHQ-9 ≥ 10 (moderate depressive symptoms) received guided Deprexis. An independent investigator used a computerized random number generator for the allocation schedule.
Patient Health Questionnaire−9 (PHQ-9;
The PHQ-9 is a self-report including nine items on depressive symptoms. The continuous PHQ-9 score is the primary outcome in the EVIDENT trial. For the present study, the continuous PHQ-9 scores as well as minimal clinically important differences (
Web Screening Questionnaire (WSQ;
The WSQ is a brief online questionnaire consisting of 15 items that screen for common mental disorders. Cutoffs are provided to dichotomize the participants as having either a negative or a positive screening for the disorder (the cutoff scores can be seen in
Attitudes Toward Psychological Online Interventions Questionnaire (APOI;
The APOI assesses attitudes toward Internet interventions on these four dimensions: skepticism and perception of risks, confidence in effectiveness, technologization threat, and anonymity benefits.
Quick Inventory of Depressive Symptomatology—Clinician Rating (QIDS;
The QIDS-C comprises 16 items and assesses the severity of depressive symptoms. The raters were blinded and trained (at least one audiotaped interview and observation of ratings by experienced raters; see
In both conditions (Deprexis and CG), participants had access to care-as-usual and were allowed to use antidepressant medication or psychotherapy or any other treatment, and this was not influenced by the investigators.
Deprexis
In addition to access to care-as-usual, participants with mild depressive symptoms (PHQ-9 scores of 5–9) received unguided Deprexis for 3 months, whereas participants with moderate depressive symptoms (PHQ-9 scores of 10–14) received the same plus regular e-mail support by trained clinicians. The supportive messages aimed primarily at increasing adherence with the intervention. Deprexis includes the following modules: (1) psychoeducation, (2) behavioral activation, (3) cognitive modification, (4) mindfulness and acceptance, (5) interpersonal skills, (6) relaxation exercises, physical activity and lifestyle modification, (7) problem-solving, (8) expressive writing and schema-focused contents, (9) positive psychology and emotion-focused interventions, (10) dream work and emotion-focused interventions. According to
CG
Access to Deprexis was offered after the 12-month follow-up. Until the 12-month follow-up, the CG had access to care-as-usual but not to Deprexis.
For RQ 1a, one linear multilevel model with the continuous PHQ-9 scores as dependent variable, the full maximum likelihood estimation, and a random intercept was performed in SPSS25. A random intercept model was selected because a model including the random intercept and the random slope did not converge. The multilevel model had two levels: assessments as Level 1 nested within study participants as Level 2. The four main assessment points of the EVIDENT trial were included (baseline, 3-month follow-up, 6-month follow-up, 12-month follow-up). As fixed effects, condition was entered as factor (Deprexis = 1 vs. CG = 0) and the following 12 moderators were included within one linear multilevel model. The eight WSQ anxiety-related disorders (screening positive = 1 vs. screening negative = 0), the three already known moderators (antidepressant medication, concomitant psychosocial treatment, attitudes toward Internet interventions), and initial depression severity. Time was included as a fixed effect. The baseline assessment was coded as 0, the exact time (in days) since baseline was used for the three follow-up assessments (3-month, 6-month, 12-month) given that the assessments did not occur at the exact intervals described. Besides main effects, two-way interactions (Condition × Moderators, Time × Moderators, Time × Condition) and three-way interactions (Time × Condition × Moderators) were examined.
To address RQ 1b, we performed one multilevel model for categorical outcomes, specifically, a binary logistic model within Generalized Estimating Equations (GEE) in SPSS25. Before running this model, participants were coded as having reached clinically important PHQ-9 change (coded as 1) or not having reached clinically important PHQ-9 change (coded as 0) at each assessment point (except baseline) using the suggested cutoff of at least 5 points of improvement (
For RQ 2, multilevel models like the ones described for RQ 1a and RQ 1b were performed, but only with the significant moderator(s) entered in these two models.
To address RQ 3, the analyses described earlier were performed in randomly selected split halves. The split halves of the total sample were randomly selected in SPSS using select cases, random sample of cases, approximately 50% of the cases.
For significant moderating effects in models with continuous PHQ-9 scores as dependent variable, R
Anxiety-related comorbidities did not differ between Deprexis and CG (see
Of all 12 potential moderators, only social phobia was a significant moderator when investigating continuous PHQ-9 scores as outcome, t(2186.73) = −2.14, p = .032, R
Of all 12 potential moderators, only social phobia moderated the depression outcome (again in favor of Deprexis) when evaluating clinically important PHQ-9 changes as outcome, Wald (1) = 4.40, p = .036, Exp(b) = 1.650 (see
After removing the 11 moderators not reaching significance in RQ 1, social phobia moderated continuous PHQ-9 scores as outcome, t(2512.52) = −2.41, p = .016, R
Continuous PHQ-9 scores and 12 potential moderators: The moderating effect of social phobia was replicated in the first split half, t(1076.73) = −2.31, p = .021, R
In total, 12 potential moderators were evaluated, and only the participants’ comorbid social phobia moderated the intervention effect on the outcome. Participants with comorbid social phobia had a more favorable depression outcome in response to Deprexis compared to CG. This was the case when continuous PHQ-9 scores as well as clinically important PHQ-9 changes were the outcome. The previously identified moderators (antidepressant medication, concomitant psychosocial treatment, attitudes toward Internet interventions) in the EVIDENT trial were not significant anymore in the statistical model including the anxiety-related comorbidities. Moreover, the moderating effect was specific to social phobia and was not related to general depression severity. Furthermore, social phobia was still a moderator after excluding the other 11 potential moderators. This means that the moderating effect of social phobia did not depend on the other moderators in the statistical model. Yet, the moderating effects did not even reach small effect sizes, accounted for less than 1% of the variance in change in depressive symptoms, and showed limited reproducibility.
To address the concerns about the replicability of findings of post hoc moderation analyses, future studies could prospectively select participants with depressive symptoms and comorbid social phobia as well as those without comorbid social phobia and allocate both samples randomly to either Deprexis or CG. These future studies should assess social phobia (and other anxiety-related comorbidities) with standardized clinical interviews and dimensional scales, because the major shortcoming of the present study is that the WSQ was the only measure to assess anxiety-related comorbidities. The WSQ is a self-report with relatively high false positive rates. Regarding the significant moderator social phobia, positive predictive values for social phobia were .40 (
If the results can be replicated, clinical implications would be to offer Deprexis particularly to adults with mild to moderate depressive symptoms and concurrent social phobia.
Arndt, A., Lutz, W., Rubel, J., Berger, T., Schröder, J., Späth, C., . . .Moritz, S. (2019). Identifying change-dropout patterns during an internet-based intervention for depression by applying the Muthen-Roy model. Cognitive Behaviour Therapy. Advance online publication.
Beevers, C. G., Pearson, R., Hoffman, J. S., Foulser, A. A., Shumake, J., & Meyer, B. (2017). Effectiveness of an Internet intervention (Deprexis) for depression in a united states adult sample: A parallel-group pragmatic randomized controlled trial. Journal of Consulting and Clinical Psychology, 85, 367–380. 10.1037/ccp0000171
Boschloo, L., Cuijpers, P., Karyotakia, E., Berger, T., Moritz, S., Meyer, B., & Klein, J. P. (2019). Symptom-specific effectiveness of an internet-based intervention in the treatment of mild to moderate depressive symptomatology: The potential of network estimation techniques. Behaviour Research and Therapy. Advance online publication. 10.1016/j.brat.2019.103440
Cohen, Z. D., & DeRubeis, R. J. (2018). Treatment selection in depression. Annual Review of Clinical Psychology, 14, 209–236. 10.1146/annurev-clinpsy-050817-084746
Coryell, W., Fiedorowicz, J. G., Solomon, D., Leon, A. C., Rice, J. P., & Keller, M. B. (2012). Effects of anxiety on the long-term course of depressive disorders. British Journal of Psychiatry, 200, 210–215. 10.1192/bjp.bp.110.081992
Dold, M., Bartova, L., Souery, D., Mendlewicz, J., Serretti, A., Porcelli, S., . . .Kasper, S. (2017). Clinical characteristics and treatment outcomes of patients with major depressive disorder and comorbid anxiety disorders—Results from a European multicenter study. Journal of Psychiatric Research, 91, 1–13. 10.1016/j.jpsychires.2017.02.020
Donker, T., van Straten, A., Marks, I., & Cuijpers, P. (2009). A brief web-based screening questionnaire for common mental disorders: Development and validation. Journal of Medical Internet Research, 11, e19. 10.2196/jmir.1134
Fava, M., Rush, A. J., Alpert, J. E., Balasubramani, G. K., Wisniewski, S. R., Carmin, C. N., . . .Trivedi, M. H. (2008). Difference in treatment outcome in outpatients with anxious versus nonanxious depression: A STAR*D report. American Journal of Psychiatry, 165, 342–351. 10.1176/appi.ajp.2007.06111868
Fuhr, K., Schröder, J., Berger, T., Moritz, S., Meyer, B., Lutz, W., . . .Klein, J. P. (2018). The association between adherence and outcome in an internet intervention for depression. Journal of Affective Disorders, 229, 443–449. 10.1016/j.jad.2017.12.028
Gräfe, V., Berger, T., Hautzinger, M., Hohagen, F., Lutz, W., Meyer, B., . . .Greiner, W. (2019). Health economic evaluation of a web-based intervention for depression: The EVIDENT-trial, a randomized controlled study. Health Economics Review, 9, 16. 10.1186/s13561-019-0233-y
Karyotaki, E., Riper, H., Twisk, J., Hoogendoorn, A., Kleiboer, A., Mira, A., . . .Cuijpers, P. (2017). Efficacy of self-guided Internet-based cognitive behavioral therapy in the treatment of depressive symptoms: A meta-analysis of individual participant data. JAMA Psychiatry, 74, 351–359. 10.1001/jamapsychiatry.2017.0044
Klein, J. P., Berger, T., Schröder, J., Späth, C., Meyer, B., Caspar, F., . . .Hohagen, F. (2016). Effects of a psychological internet intervention in the treatment of mild to moderate depressive symptoms: Results of the EVIDENT study, a randomized controlled trial. Psychotherapy and Psychosomatics, 85, 218–228. 10.1159/000445355
Klein, J. P., Berger, T., Schröder, J., Späth, C., Meyer, B., Caspar, F., . . .Moritz, S. (2013). The EVIDENT-trial: Protocol and rationale of a multicenter randomized controlled trial testing the effectiveness of an online-based psychological intervention. BMC Psychiatry, 13, 239. 10.1186/1471-244X-13-239
Klein, J. P., Gamon, C., Späth, C., Berger, T., Meyer, B., Hohagen, F., . . .Schröder, J. (2017). Does recruitment source moderate treatment effectiveness? A subgroup analysis from the EVIDENT study, a randomised controlled trial of an internet intervention for depressive symptoms. British Medical Journal, 7, e015391. 10.1136/bmjopen-2016-015391
Klein, J. P., Späth, C., Schröder, J., Meyer, B., Greiner, W., Hautzinger, M., . . .Berger, T. (2017). Time to remission from mild to moderate depressive symptoms: One year results from the EVIDENT-study, an RCT of an Internet intervention for depression. Behaviour Research and Therapy, 97, 154–162. 10.1016/j.brat.2017.07.013
Kroenke, K., Spitzer, R. L., & Williams, J. B. (2001). The PHQ-9: Validity of a brief depression severity measure. Journal of General Internal Medicine, 16, 606–613. 10.1046/j.1525-1497.2001.016009606.x
Levis, B., Benedetti, A., & Thombs, B. D. (2019). Accuracy of Patient Health Questionnaire-9 (PHQ-9) for screening to detect major depression: Individual participant data meta-analysis. British Medical Journal, 365, l1476. 10.1136/bmj.l1476
Löwe, B., Unützer, J., Callahan, C. M., Perkins, A. J., & Kroenke, K. (2004). Monitoring depression treatment outcomes with the Patient Health Questionnaire-9. Medical Care, 42, 1194–1201. 10.1097/00005650-200412000-00006
Lutz, W., Arndt, A., Rubel, J., Berger, T., Schröder, J., Späth, C., . . .Moritz, S. (2017). Defining and predicting patterns of early response in an internet-based intervention for depression. Journal of Medical Internet Research, 19, e206. 10.2196/jmir.7367
Maas, I. L., Nolte, S., Walter, O. B., Berger, T., Hautzinger, M., Hohagen, F., . . .Rose, M. (2017). The regression discontinuity design showed to be a valid alternative to a randomized controlled trial for estimating treatment effects. Journal of Clinical Epidemiology, 82, 94–102. 10.1016/j.jclinepi.2016.11.008
Meuldijk, D., Giltay, E. J., Carlier, I. V., van Vliet, I. M., van Hemert, A. M., & Zitman, F. G. (2017). A validation study of the Web Screening Questionnaire (WSQ) compared with the Mini-International Neuropsychiatric Interview-Plus (MINI-Plus). JMIR Mental Health, 4, e35. 10.2196/mental.5453
Moritz, S., Hörmann, C. C., Schröder, J., Berger, T., Jacob, G. A., Meyer, B., . . .Klein, J. P. (2013). Beyond words: Sensory properties of depressive thoughts. Cognition and Emotion, 28, 1047–1056. 10.1080/02699931.2013.868342
Rush, A. J., Trivedi, M. H., Ibrahim, H. M., Carmody, T. J., Arnow, B., Klein, D. N., . . .Keller, M. B. (2003). The 16-item Quick Inventory of Depressive Symptomatology (QIDS), clinician rating (QIDS-C), and self-report (QIDS-SR): A psychometric evaluation in patients with chronic major depression. Biological Psychiatry, 54, 573–583. 10.1016/S0006-3223(02)01866-8
Schneider, B. C., Schröder, J., Berger, T., Hohagen, F., Meyer, B., Späth, C., . . .Klein, J. P. (2018). Bridging the “digital divide”: A comparison of use and effectiveness of an online intervention for depression between baby boomers and millennials. Journal of Affective Disorders, 236, 243–251. 10.1016/j.jad.2018.04.101
Schröder, J., Berger, T., Meyer, B., Lutz, W., Hautzinger, M., Späth, C., . . .Moritz, S. (2017). Attitudes towards internet interventions among psychotherapists and individuals with mild to moderate depression symptoms. Cognitive Therapy and Research, 41, 745–756. 10.1007/s10608-017-9850-0
Schröder, J., Berger, T., Meyer, B., Lutz, W., Späth, C., Michel, P., . . .Moritz, S. (2018). Impact and change of attitudes toward Internet interventions within a randomized controlled trial on individuals with depression symptoms. Depression and Anxiety, 35, 421–430. 10.1002/da.22727
Schröder, J., Sautier, L., Kriston, L., Berger, T., Meyer, B., Späth, C., . . .Moritz, S. (2015). Development of a questionnaire measuring attitudes towards psychological online interventions—The APOI. Journal of Affective Disorders, 187, 136–141. 10.1016/j.jad.2015.08.044
Späth, C., Hapke, U., Maske, U., Schröder, J., Moritz, S., Berger, T., . . .Klein, J. P. (2017). Characteristics of participants in a randomized trial of an internet intervention for depression (EVIDENT) in comparison to a national sample (DEGS1). Internet Interventions, 9, 46–50. 10.1016/j.invent.2017.05.003
Twomey, C., O’Reilly, G., & Meyer, B. (2017). Effectiveness of an individually-tailored computerised CBT programme (Deprexis) for depression: A meta-analysis. Psychiatry Research, 256, 371–377. 10.1016/j.psychres.2017.06.081
Web Screening Questionnaire for common mental disorders (WSQ). (n.d.). Web Screening Questionnaire for common mental disorders (WSQ). Retrieved from
The data reported in this article have been previously published and/or were collected as part of a larger data collection (at one or more points in time). Findings from the data collection of the EVIDENT trial have been reported in separate articles.
Submitted: March 18, 2019 Revised: July 26, 2019 Accepted: July 29, 2019