Context: Common mental health problems, such as depression, anxiety, and other disorders, affect many people and incur increasing costs to individuals, employers, and government. Yet, access to mental health services is still limited. A previous study by RAND Europe suggested that providing access to online mental health assessment and support, as well as building on computerised cognitive behavioural therapy (cCBT) interventions, could help reach out to the general population and, in particular, those less likely to seek help elsewhere (see van Stolk et al. 2014).1 The current study explores online platforms and mobile applications that offer cognitive behavioural therapy (CBT) for people with mental health needs.
Method: We used a snowballing approach starting from van Stolk et al. (2014) to identify the most recent academic literature available on the subject. We reviewed, coded and analysed 44 studies related to mental health treatments and cognitive behavioural therapy platforms. This study summarises the results of our work and provides an overview of the tools discussed in the literature, characteristics of the participants who tested these tools, and available evidence on the mental health and employment outcomes.
Products: Overall, this review shows that a variety of cCBT products exist which differ both in terms of the conditions that the products aim to address and the ways in which platforms are designed. The tools offer support through a varying number of modules (or lessons) clustered around specific issues that need to be addressed. The number of modules offered by the tools differs slightly by condition: platforms for anxiety disorders and insomnia are on average lengthier, with a median number of modules of 9 and 8 respectively, compared with those for depression, in which the median is 6.5. The majority of tools use a linear structure and offer at least some additional guidance, although the intensity of this extra support is typically low.
Mental health outcomes: The mental health outcomes—such as reductions of social anxiety or depression symptoms—are measured on various scales, such as the Beck Depression Inventory (BDI-II), Beck Anxiety Inventory (BAI) and Insomnia Severity Index (ISI), among others. This review shows that the cCBT tools appear to have a positive effect on mental health outcomes, although this varies depending on the nature of participants and the outcomes being measured. The mean effect sizes varied from large to moderate:
- Improvements measured on BDI-II for the tools for depression show within-group effect size of 1.23.
- Improvements measured on BAI for the tools for anxiety disorders show between-group effect size of 0.79.
- Improvements measured on ISI for the tools for insomnia show within-group effect size of 1.40.
Work-related outcomes: Only a few studies examined work-related outcomes of the cCBT tools. Those that did showed a small mean between-group effect size of 0.35 on the Work and Social Adjustment Scale (WSAS). Three of the examined tools showed positive significant work-related outcomes as well as mental health improvements.
Participants: Computerised CBT products target a wide range of potential users, but some participant characteristics, such as gender and educational background, are likely to affect the uptake of—or the extent of engagement with—cCBT tools. The data did not point to a relationship between the level of employment or education of the treated individuals and the effectiveness of platforms, although there was a relationship between effectiveness and the age and gender of participants: tools appeared to be more effective in older participants and in women. There is therefore a risk that such platforms will fail to reach out to those who need help most; that this type of treatment is simply not well suited for these groups; or that a more effective method of increasing their participation and adherence to treatment needs to be found. However, the limited number of articles does not allow us to make a firm link between these factors or to inform us about the direction of the relationship between participant characteristics and effectiveness. Similarly, we cannot exclude the fact that there was a selection effect, in that the examined studies included participants who were more likely to take part in cCBT research or to benefit from treatment, and that therefore the effects of the tools have been overstated.
Conclusions: The diversity of existing tools reflects the fact that cCBT interventions (supported by technology developments) lend themselves well to—and offer space for—experimentation in terms of their design and delivery mechanisms. There are no evident patterns that would suggest that certain types of tools (or certain elements of their design) determined the level of their effectiveness. While there is no "silver bullet" for future designs of cCBT, one can expect that mobile applications will increase in number and that most cCBT platforms will provide at least a basic level of additional support.