Generalised anxiety disorder involves worry that runs most days, across many parts of life, and is hard to switch off. Cognitive behavioural therapy is the first-line psychological treatment for it, but there are more people wanting that treatment than there are therapists available to provide it. Researchers have been testing whether a well-designed app can carry some of that load.
Parsons and colleagues randomly assigned 351 adults with generalised anxiety disorder to either a self-directed CBT program delivered by smartphone or an online psychoeducation program, with everyone assessed remotely and the assessors rating improvement kept unaware of which program a participant had used. The comparison was a fair one.
The CBT app group improved considerably more. At 10 weeks, 71.0% of them had reached remission, compared with 34.6% of the psychoeducation group. By 24 weeks the figures were 77.7% and 52.0%. Self-reported anxiety severity was significantly lower in the app group at both points, and the difference was large.
A program like this is not a substitute for therapy in every case, particularly where anxiety sits alongside other difficulties. It does suggest that structured, self-directed CBT can do real work.
If worry is wearing you down, it is worth speaking with someone about it.
What the study was testing
Generalised anxiety disorder is chronic and disabling, and cognitive behavioural therapy is its first-line psychological treatment. The practical problem is access. There are far more people who would benefit than there are trained therapists to see them, and the gap is widest outside major cities.
Digital CBT is one response to that gap. The question this trial set out to answer was not whether an app is as good as a therapist, which it did not test, but whether a smartphone-delivered CBT program outperforms an active comparison. That comparison matters. Testing an app against a waiting list tends to flatter the app, because simply being enrolled in something changes how people report their symptoms. Here the control group received an online psychoeducation intervention, so both groups got structured material about anxiety.
How it was run
This was a single-blind, parallel-group, decentralised randomised clinical trial with equal allocation to each arm, conducted between March 2023 and February 2024 through Boston University. Decentralised means the whole study ran remotely rather than at a clinic site. Participants were adults aged 22 or older who met diagnostic criteria for current generalised anxiety disorder and scored 15 or higher on the GAD-7, a seven-item self-report questionnaire where higher scores indicate more severe anxiety.
The intervention was a self-directed, tailored digital CBT program designed to be used flexibly for learning CBT techniques by smartphone. There were two primary outcomes: remission, judged by an independent evaluator who did not know the group allocation, using the Clinical Global Impressions-Improvement scale, and self-reported anxiety severity on the GAD-7. Both were assessed at 10 weeks and again at 24 weeks.
What it found
A total of 351 adults were randomised, 175 to digital CBT and 176 to psychoeducation. Their mean age was 40.6 years and 259 of them, or 73.8%, were women.
Anxiety severity was significantly lower in the digital CBT group at both time points. The effect size was 1.09 at week 10 and 0.96 at week 24. An effect size of this kind expresses a difference between groups in standard deviations, and by the usual convention 0.2 is small, 0.5 moderate and 0.8 large. Both figures sit above that threshold.
Remission told a similar story. At 10 weeks, 103 of 145 digital CBT participants reached remission, or 71.0%, against 54 of 156 in the psychoeducation group, or 34.6%. At 24 weeks it was 115 of 148, or 77.7%, compared with 78 of 150, or 52.0%. Both differences were statistically significant, meaning they are unlikely to have arisen by chance.
Reading it sensibly
Two things stand out. The benefit was still present at 24 weeks, well after the program itself, so this is not a short-lived lift. And the comparison group was not idle, which makes the gap harder to explain away.
The trial recruited adults with a confirmed diagnosis and at least moderately severe symptoms who were willing to join a fully remote study, and it tested one specific program rather than digital CBT in general. It does not tell us how such a program compares with seeing a therapist, or how it performs when anxiety sits alongside depression, trauma or substance use. What it does support is that structured self-directed CBT can produce meaningful, lasting change for generalised anxiety.
If worry has become a constant background to your days, a psychologist can help you work out what would suit your situation.