Can short online self-help courses reduce depression?
Two guided online self-help courses, one based on CBT and a shorter one based on problem solving, both reduced depressive symptoms more than waiting, with medium-sized effects and no clear winner between them.
Source
Internet-based treatment for adults with depressive symptoms: randomized controlled trial
Study at a glance
- Design
- RCT — Three-arm RCT with computer-generated block randomisation: 8-week internet CBT, 5-week internet problem-solving therapy (PST), or waiting list; both active arms had weekly email support from trained psychology students; outcomes at 5, 8 and 12 weeks analysed by intention to treat with linear mixed models.
- N
- N=263 · 263 adults randomised across the three arms; attrition from follow-up questionnaires reached 43% by 12 weeks and was handled by mixed-model estimation.
- Population
- Self-referred Dutch-speaking adults recruited via newspaper and web advertisements, scoring 16 or more on the CES-D depression scale; mostly women and highly educated.
- Outcome
- Depressive symptoms on the CES-D (primary); anxiety (HADS subscale) and quality of life (EQ-5D) as secondary outcomes; clinically significant change.
Structured fields used in claim comparison tables when every cited study has a complete layer.
What they did
Adults with elevated depressive symptoms who answered newspaper and web adverts were randomly assigned to an 8-lesson internet CBT course, a 5-week internet problem-solving therapy course, or a waiting list. Both courses came with brief weekly email feedback from trained master's students that encouraged people to keep going but did not give therapy. Depression, anxiety and quality of life were measured online at 5, 8 and 12 weeks.
What they found
Both courses beat the waiting list on depression by 8 and 12 weeks, and problem-solving therapy was already ahead at 5 weeks; the two courses did not differ from each other. At 12 weeks the between-group effect sizes for depression were d = 0.69 for CBT and 0.65 for problem solving, and 34 CBT and 30 problem-solving participants showed clinically significant improvement versus none on the waiting list. Anxiety and quality of life also improved more in the treatment groups, though quality-of-life effects were small.
The limits
What it doesn't show
Dropout from the follow-up questionnaires was high and was greater in the treatment arms than on the waiting list, so results depend on statistical estimation of missing data. Only about 38% of people completed all lessons, participants were self-referred, mostly women and highly educated, and none had a clinical diagnostic interview, so the findings may not apply to clinical or less-educated populations. There was no blinding, the comparison was a waiting list rather than an active control, and follow-up lasted only 12 weeks.
Key terms
- Waiting-list control
- A comparison group that gets no treatment during the trial but is offered it later; it controls for time and natural recovery but not for attention or expectations.
- Problem-solving therapy (PST)
- A structured psychological treatment that teaches people to sort their worries into unimportant, solvable and unsolvable problems and to work through solvable ones step by step.
- Intention-to-treat analysis
- Analysing every randomised participant in the group they were assigned to, whether or not they completed treatment, to keep the benefits of randomisation.
- Cohen's d
- A standardised effect size expressing the difference between two group means in standard-deviation units; about 0.5 is conventionally called medium.
- Clinically significant change
- Improvement that is both statistically reliable for the individual and brings their score below a clinical cut-off, here 16 on the CES-D.
- CES-D
- The Center for Epidemiologic Studies Depression scale, a 20-item self-report questionnaire where scores of 16 or more suggest clinically relevant depressive symptoms.
Flashcards
0 of 11 answers reviewed
Research intelligence for this paper
See its role on concept claims, tensions it is part of, placement history, and related discoveries.
Quiz yourself
What was the primary outcome of this trial?
Common questions
If so many people dropped out, can we trust the results?
Partly. The authors analysed everyone randomised using mixed models that estimate missing scores, and completers and dropouts looked similar at baseline, but the estimates assume data are missing at random, which cannot be checked.
Does this show problem-solving therapy is better than CBT?
No. Problem solving worked faster, showing a benefit at 5 weeks, but the two courses did not differ at any time point, so the study supports them as roughly equivalent.
Did people need to finish the whole course to benefit?
Surprisingly, people who did not complete all lessons still improved more than the waiting list, and their improvement did not differ significantly from completers, although this comparison is not randomised.
More on Depression treatment