Whether cognitive behavioural therapy helps generalised anxiety disorder is settled. How it should be delivered is not, and that is the question people actually face: the waiting list for individual therapy is long, the group is starting next month, and the app is available tonight. A network meta-analysis published in Translational psychiatry in 2025 attempted to rank them. [pubmed-anxiety-jul13-2026-source]
What the researchers did
They gathered every randomised trial of CBT for generalised anxiety disorder they could find across four major research databases, ending up with 52 trials covering 4,361 people, average age 43, about seven in ten of them women.
Then they used a network meta-analysis, which is worth understanding because it is doing something clever. Very few trials compare individual therapy directly against group therapy. But many compare each of them against a waiting list, and that shared comparison lets the analysis estimate how they stack up against each other indirectly. It squeezes far more out of the literature than a straight pooled comparison, at the cost of leaning on more assumptions.
What they found
Individual CBT came out ahead. It beat remote CBT, it beat treatment as usual, and it beat waiting lists by a wide margin.
Group CBT was next, clearly better than a waiting list and close enough to individual therapy that the authors argue it should be treated as a genuine alternative rather than a fallback.
Remote CBT is the awkward one. It did not outperform treatment as usual or, in this analysis, waiting-list control.
Acceptability was flat across the board. People dropped out of one format about as often as another, so the differences here are about effect, not about tolerating the format.
How much weight to give it
Moderate. The authors rate three quarters of the included trials as low or unclear risk of bias, which is a polite way of saying the underlying evidence is patchy. Indirect comparisons inherit every weakness of the trials they are built from.
“Remote CBT” is also doing a great deal of work as a category. It spans a self-guided website with no human involvement and a course of video sessions with a therapist, and pooling those into one estimate almost certainly hides real differences between them.
What this means in practice
Two useful things follow.
If you are offered a group, the reflex to hold out for individual therapy is not well supported by this evidence, and holding out usually means waiting. Group formats also provide something individual therapy structurally cannot, which is several other people in the room who recognise exactly what you are describing.
If you are using a self-guided app or website, treat it as a starting point rather than the treatment. It is the format with the weakest showing here, and the right follow-up question is what the next step is if it has not helped in a couple of months.
Which format actually fits your situation?
Tick what is true. This is about what will suit you, not about what is best in the abstract.
0 of 6 ticked
Long waits, difficulty sustaining self-guided work, and never having asked what else exists is the combination where group CBT tends to be the overlooked option. On this evidence it performs close to individual therapy and usually starts sooner.
The useful question at your next appointment is not "can I have therapy" but "what formats do you offer and what is the wait for each". The answers are often more varied than the first offer suggests.
The format that starts soonest usually beats the format that ranks highest, because a treatment you are actually receiving outperforms one you are waiting for.
The source
These findings are drawn from “CBT treatment delivery formats for generalized anxiety disorder: a systematic review and network meta-analysis of randomized controlled trials” (Liu S, Xiao H, Duan Y, et al., 2025), published in Translational psychiatry. Read the full study on PubMed.