Does Online CBT Actually Work for Anxiety? What the Trials Show

The Question Worth Asking Before You Start

Structured CBT for anxiety has decades of evidence behind it in its original, in-person form. Moving it online -- video sessions, messaging with a therapist, structured modules -- is a real change in delivery, and it's reasonable to ask whether something gets lost in the switch. Two lines of controlled evidence address this directly, rather than just assuming the online version is "just as good."

The Meta-Analysis: Eight Trials, Head to Head

A 2021 systematic review and meta-analysis in the International Journal of Preventive Medicine pooled 8 randomized controlled trials that directly compared internet-delivered CBT against face-to-face CBT for anxiety disorders, covering 348 participants in the internet-delivered arms and 316 in face-to-face arms. The pooled post-treatment effect size was essentially zero -- Hedges' g = 0.01 (95% CI: -0.16 to 0.18) -- meaning the two delivery formats produced statistically equivalent outcomes. The confidence interval is narrow enough to rule out a large difference in either direction, though the authors themselves noted more research across a wider range of settings would strengthen the conclusion.

A Direct Noninferiority Trial in Practice

A more rigorously designed single trial, published in JAMA Psychiatry in 2020, randomized 204 patients with health anxiety to either internet-delivered or face-to-face CBT and was specifically built as a noninferiority trial -- meaning it set a pre-defined margin (2.25 points on the Health Anxiety Inventory) and tested whether internet-delivered therapy fell within it. The result: an effect size difference of Cohen d = 0.00, with confidence interval upper limits (1.98 in intention-to-treat, 2.17 per-protocol) comfortably inside the noninferiority margin in both analyses. This is one of the more convincing head-to-head comparisons in the anxiety literature specifically because it was designed from the outset to test equivalence rather than just report it after the fact.

The Honest Caveat: Guided, Not Just Self-Directed

Both bodies of evidence here involved therapist-supported programs -- a real clinician reviewing progress, giving feedback, and staying involved in the case -- not a purely self-guided app with no human contact. That distinction matters: research on unsupported self-help anxiety apps generally shows smaller effects than either face-to-face or guided internet-delivered therapy. The equivalence shown above is specific to programs where a licensed therapist remains part of the process, just delivered remotely rather than in person.

What This Means If You're Considering Online Therapy

• For anxiety disorders specifically, the best current evidence shows no meaningful difference in outcomes between internet-delivered and face-to-face CBT, when a real therapist is involved throughout
• This is not evidence that any anxiety app or unsupervised program works equally well -- the trials above used structured, therapist-supported programs
• Convenience and access (no commute, more flexible scheduling, lower cost in many programs) can be a genuine advantage without being a compromise on outcome
• If a program offers ongoing licensed-therapist contact rather than just static self-help content, it more closely matches what these trials actually tested