CRD summary

This review concluded that computer-based psychological treatments were effective for depression, in various settings and populations. Due to high attrition rates and shortcomings and variation between trials in some of the analyses, the authors' conclusions may not be reliable. Their call for more research to investigate confounding factors was appropriate.

Authors' objectives

To evaluate the overall effectiveness of computer-based psychological treatments for depression and to examine the impact of additional support.

Searching

PubMed, EMBASE and PsycINFO were searched for relevant studies published in English between March 2001 and March 2011; search terms were reported. Reference lists of relevant studies were handsearched for more studies.

Study selection

Studies were eligible for the review if they were of adults (aged 18 years or older) with depression (self-reported or diagnosed)that was confirmed using reliable and valid measures, with or without comorbidity, who were recruited from the general population or a clinical group. It appears that studies of any design were eligible for the review, but only randomised controlled trials (RCTs) were eligible for the meta-analysis. Studies were required to be delivered by computer, directly to the patient, via a therapist, or a combination of both. Outcomes of interest were self-reported measures of depression, the proportion of participants achieving a clinically significant reduction in depression and the proportion of participants who recovered from depression. Preliminary research into newly developed computer-based treatments was excluded.

In the included studies, participants were recruited from primary and secondary care or the community. Some studies recruited from specific populations, such as those over 50 years old or those with HIV or diabetes. Most participants were assessed for depression using a valid screening instrument or an official diagnosis. Eighteen different interventions were described, mostly based on cognitive-behavioural therapy. Most of them were delivered online; four studies delivered interventions on a compact disc (CD). Most interventions were delivered to individuals; one was delivered to groups of participants. Most studies involved support of some type, either by therapists or administrators. Support delivery varied including at the same time or another time, by telephone, e-mail, or face-to-face. Follow-up ranged from one month to over a year. Studies were undertaken in Sweden, Australia, the Netherlands, Germany, Canada, the USA and the UK.

One reviewer selected studies for the review; any uncertainties were resolved by consultation with a second reviewer.

Assessment of study quality

The RCTs were assessed for quality using the Cochrane risk of bias tool. The criteria included sequence generation; allocation concealment; blinding of participants, personnel and assessors; incomplete outcome data; and selective reporting.

The authors did not state how many reviewers performed the quality assessment.

Data extraction

Data were extracted on the outcomes. Mean differences (Cohen's d) were calculated for continuous outcomes and odds ratios were calculated for dichotomous outcomes, with 95% confidence intervals. An effect size of d=0.8 was considered large, d=0.5 was moderate and d=0.2 was small.

The authors did not state how many reviewers extracted data for the review.

Methods of synthesis

The outcomes of those RCTs that included sufficient data for intervention and control groups were pooled in meta-analyses, using a random-effects model, to produce summary effect odds ratios and standardised mean differences, with 95% confidence intervals. Intention-to-treat analyses were undertaken. Heterogeneity was assessed using Χ² and Ι² and publication bias was assessed by inspection of the funnel plot for overall improvement in depression. Subgroup analyses were undertaken on study characteristics (support type, number of sessions, clinical setting, communication mode, population type and control group type). Those studies that were not RCTs were combined and discussed in a narrative.

Results of the review

Forty studies, 24 RCTs (7,059 patients) and 16 other studies (3,440 patients), were included in the review. Nineteen of the RCTs (3,885 patients) were included in meta-analyses and were considered to be high quality, but about half these trials were at high or unclear risk of bias for incomplete short-term data.

Overall findings: In therapist-supported studies, the mean post-treatment effect size across all depression measures was d=1.35 (18 effects) and at follow-up d=1.29 (13 effects). In administrative-supported studies, the mean post-treatment effect size was d=0.95 (23 effects) and at follow-up d=1.20 (10 effects). In studies with no support, the mean post-treatment effect size was d=0.78 (17 effects) and at follow-up d=1.13 (11 effects). The overall dropout rate was 57% (40 studies). Satisfaction rates varied, but were generally high, ranging from 71% to 93%.

Meta-analysis: Compared with controls (mainly waiting list or treatment as usual), there was a statistically significant effect on self-reported depression after treatment (d=-0.56, 95% CI -0.71 to -0.41; Ι²=81%; 19 trials) and at follow-up (d=-0.20, 95% CI -0.31 to -0.09; 14 trials). Computer-based interventions were associated with a significant increase in the odds of clinical improvement in depression (OR 3.68, 95% CI 2.12 to 6.40; seven trials) and recovery from depression (OR 4.14, 95% CI 2.01 to 8.53; eight trials).

Subgroup analyses: All subgroup analyses found statistically significant effects of treatment on self-reported measures of depression, except for those using a synchronous mode of communication. Significant effects were found for all the following subgroups: support type (therapist, administrative or no support), number of sessions (less than eight versus eight or more), clinical setting (community versus primary or secondary care), asynchronous communication, population (specific or general) and control group (treatment as usual or waiting list).

Inspection of the funnel plot suggested that publication bias was possible.

Authors' conclusions

The review and meta-analysis supported the efficacy and effectiveness of computer-based psychological treatments for depression, in various settings and populations.

CRD commentary

The review addressed clear research questions, supported by appropriate broad inclusion criteria. Relevant sources were searched to identify studies published in English, but no specific attempts were made to find unpublished studies, so it is possible that some relevant studies were missed. Inspection of a funnel plot suggested that publication bias was possible. Appropriate methods were used to select studies, but the authors did not state how many reviewers assessed quality and extracted data, so reviewer error and bias cannot be ruled out.

A valid tool was used to assess the quality of the RCTs and they were generally considered to be high quality, but about half had either unclear or high risk of bias for incomplete data. The quality of the other studies was not reported, which means the reliability of the overall results is not clear. The authors noted that many of the other studies were small, with no control for confounding factors, and the outcomes were often self reported, which limits the reliability of the conclusions. The synthesis of the high-quality RCTs in meta-analyses and of the remaining studies in a narrative was appropriate. Considerable heterogeneity was identified and this was explored in pre-specified subgroup analyses, but many of these included only a few trials and the authors noted that other potential confounders were not taken into account. They noted that supported interventions had greater effects than those that were unsupported.

Due to high attrition rates and shortcomings and variation between trials in some of the analyses, the authors' conclusions may not be reliable. Their recommendations for further research seem appropriate.

Implications of the review for practice and research

Practice: The authors did not state any implications for practice.

Research: The authors recommended further research on the maintenance of the benefits of computer-based treatments and role of therapist factors. More research was needed on the nature of support, the type, content and deployment of interventions, specific populations, delivery of support and reasons for dropouts.

Funding

Not stated.

Bibliographic details

Richards D, Richardson T. Computer-based psychological treatments for depression: a systematic review and meta-analysis. Clinical Psychology Review 2012; 32(4): 329-342. [PubMed: 22466510]

Original Paper URL

http://www.sciencedirect.com/science/article/pii/S027273581200027X

Indexing Status

Subject indexing assigned by NLM

MeSH

Depressive Disorder /therapy; Humans; Patient Dropouts /statistics & numerical data; Psychotherapy /methods; Telemedicine /methods /statistics & numerical data; Treatment Outcome

AccessionNumber

12012023711

Database entry date

29/10/2012

Record Status

This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.