Technology-based CBT for youth anxiety: moderate short-term benefits but uncertainty remains

Blog by Douglas Badenoch of The Mental Elf Service on Dr. Matti Cervin et al Open Access JCPP paper ‘Technology-delivered cognitive-behavioral therapy for pediatric anxiety disorders: a meta-analysis of remission, post treatment anxiety, and functioning’. This paper and topic will be discussed at our FREE informal journal club ‘CAMHS around the Campfire’ register to join the session on Tuesday 24 May 5pm UK time.

CAMH services everywhere are struggling with capacity. We face a post-pandemic wave of increased demand from youth experiencing anxiety-related distress. The idea of harnessing digital technology to deliver therapeutic interventions is therefore very attractive to cash-strapped services. Not only might we be able to provide services at lower cost, we may be able to provide them to more people and improve accessibility.

The potential benefits of technology-based CBT have been explored in several previous studies and systematic reviews. Unfortunately, previous reviews have been limited in scope and methods. Some reviews include non-randomised trials, some include participants without a formal diagnosis of anxiety disorder (AD), some mix participants with anxiety and OCD, and some use sub-optimal outcome measures.

The gold standard is structured diagnostic interviews.  So, this review set out to overcome these methodological limitations and focused on remission rates as assessed by diagnostic interview in young people with a AD diagnosis.


This was a systematic review and meta-analysis of RCTs of technology-delivered Cognitive Behavioural Therapy (tCBT) for young people with anxiety disorders.


Young people less than 18 years of age that meet the criteria for any anxiety disorder on a structured diagnostic interview.


CBT delivered primarily by digital means, via internet, app, cell phone, tablet or computer.  tCBT may be therapist-led and have caregiver involvement.


Treatment as usual, waiting list or placebo.


The primary outcome was remission of AD, established by diagnostic interview (Anxiety Disorders Interview Schedule), pre- and post-treatment in all studies.

Other outcomes included:

  • Overall functioning (seven studies)
  • Spence children’s anxiety scale (6)
  • Qualitative outcomes


Nine studies (711 participants in total) were included in the review.

37.9% of tCBT participants were in remission post-treatment, compared with 10.2% in control groups.

The meta-analysis generated an odds ratio of 4.7 (95% CI 3.1-7.2). That is, tCBT participants were nearly 5 times more likely to achieve remission post-treatment than control participants.

These results were of moderate certainty according to the GRADE method applied by the researchers.

There was no significant difference in the youth-reported outcomes, but there was for caregiver-reported outcomes and clinician-reported functioning.

A post hoc analysis found that higher remission rates were found in studies with smaller sample sizes, lower pre-treatment anxiety levels, higher completion rates, face-to-face contact and caregiver involvement. However, as these results are post hoc, and based on a small number of studies, they should be treated with caution.

Strengths and limitations

Although the review question was formulated in advance by the Swedish national guidelines board, the review protocol was not pre-registered. The reviewer team made some changes to the outcome measures and meta-analysis in the light of what they found.

The literature search was conducted in January 2020, updated in February 2021. Nine trials doesn’t feel like a lot for the 13 years spanned by the search.  This small data set means that we should interpret these results with some caution. It’s also reflected in the wide confidence intervals and heterogeneity around the effect estimates.

There was blind, independent assessment of study inclusion, characteristics, and risk of bias, with disagreements being resolved by consensus. The reviewers presented a nicely transparent analysis of the risk of bias in each study. Broadly, there was low risk of selection bias with some concerns around outcome reporting.  Adherence and attrition were problematic in some trials. We cannot rule out the possibility that bias affected the reported results of these individual studies.

There were a lot of differences (heterogeneity) between the specific interventions used in different studies. Some of the studies included a face-to-face component in the tCBT intervention, and some were not clear about the nature and extent of therapist involvement. This highlights the pressing need for better reporting of digital interventions so that we can be clearer about what goes on.

There were varying degrees of caregiver involvement in the tCBT interventions. Furthermore, there were differences in the way outcomes were measured; three studies used the caregiver version of the ADIS outcome instrument whilst the other six used the caregiver-and child version.

These differences in what was offered and how outcomes were measured could be important, as the studies’ results also varied widely in the reported remission rates – from 14% to 69%.

This also means that we should treat the overall estimate with caution. Although these differences were explored in the post hoc analysis, we should be very cautious about jumping to any conclusions because of the small number of studies available.

Additionally, the duration of follow-up was limited, so the durability of benefits is unknown. Only one study reported adverse events.


Young people with anxiety disorders may benefit from tCBT.  It is not clear how long-lasting these effects may be, how they compare to CBT on its own or as an adjunct, or whether there are any adverse effects.

More research is needed to answer these questions, and to shed further light on what “active ingredients” make tCBT most effective. These relate to both content and delivery. For example, how much therapist support is necessary? And does parental involvement help? Future studies should work on improving uptake, adherence, and follow-up, and making the specific content of the interventions more transparent in research reports.

We also need to improve the methodological quality of trials investigating tCBT, particularly in outcomes reporting.

Join the discussion

This paper and topic will be discussed at our FREE informal journal club ‘CAMHS around the Campfire’ register to join the session on Tuesday 24 May 5pm UK time.

Add a comment

Your email address will not be published. Required fields are marked *