Digital and remote delivery¶
TL;DR — Internet-delivered GAD treatment shows large effects against waitlist: Hedges' g=0.79 for anxiety and 0.75 for worry across 20 RCTs (Eilert 2021, PMID 33225589), and a GAD subgroup effect of g=0.62 in a 47-trial anxiety-disorder meta-analysis (Pauley 2023, PMID 34047264). In that broader network, guided and unguided estimates did not differ significantly, and digital versus face-to-face was also nonsignificant (g=0.14, 95% CI −0.01 to 0.30; 9 comparisons, 683 participants). A 2025 smartphone CBT trial used active psychoeducation and found remission of 71.0% versus 34.6% at 10 weeks (OR 4.63, 2.85–7.54) and 77.7% versus 52.0% at 24 weeks (Parsons 2025, PMID 41396602). A separate GAD delivery-format network found individual CBT superior to remote CBT (SMD 0.96, 0.13–1.79) and remote CBT not superior to treatment as usual or waitlist (Liu 2025, PMID 40506439). Differences in comparator, intervention mix and scope prevent treating these estimates as a direct contradiction. Engagement remains a constraint: pooled app uptake 92.4%, adherence 61.8%, post-test attrition 18.6% and follow-up attrition 28.4% across 79 trials (Liu 2026, PMID 41259035).
The GAD-specific digital evidence¶
| Source | Scope | Effect |
|---|---|---|
| Eilert 2021 (PMID 33225589) | 20 RCTs of internet-delivered interventions, GAD-specific, searched to April 2020 | Anxiety g=0.79; worry g=0.75; depression, functional impairment and QoL moderate-to-large; maintenance at follow-up "seems likely"; considerable heterogeneity moderated by intervention variability |
| Pauley 2023 (PMID 34047264) | 47 RCTs, 53 comparisons, 4,958 participants across anxiety disorders vs waitlist/care-as-usual | Overall g=0.80 (0.68–0.93); GAD subgroup g=0.62; panic ± agoraphobia 1.08; social anxiety 0.76; mixed anxiety 0.68; no subgroup differed significantly |
| Pauley 2023, guidance (PMID 34047264) | Same | Guided g=0.84 vs unguided g=0.64 — not significantly different |
| Pauley 2023, head-to-head (PMID 34047264) | 9 comparisons, 683 participants | Digital vs face-to-face g=0.14 (−0.01 to 0.30), favouring digital, not significant |
| Paxling 2011 (PMID 21770848) | 89 randomised, 8-week guided iCBT self-help with applied relaxation, PSWQ primary | Large within- and between-group effects (d>0.8) on all outcomes except quality of life; 13.6% non-completion of post-measures; gains maintained or improved at 1 and 3 years |
| Dahlin 2016 (PMID 26731173) | Internet-delivered acceptance-based behaviour therapy for GAD, RCT | Third-wave content delivered digitally |
| Carl 2020 (PMID 32725848) | 256 randomised, moderate-to-severe GAD symptoms, smartphone digital CBT (Daylight) vs waitlist | Anxiety adjusted difference 3.22 (2.14–4.31), d=1.08; worry, depression, sleep, wellbeing and individualised QoL all improved |
| Parsons 2025 (PMID 41396602) | 351 randomised (175 DCBT / 176 psychoeducation), decentralised, single-blind, GAD-7 ≥15 plus diagnostic criteria, blinded evaluator | GAD-7 Cohen d=1.09 (0.81–1.37) at 10 wk and 0.96 (0.67–1.24) at 24 wk; remission 71.0% vs 34.6% (OR 4.63, 2.85–7.54, p<0.001) at 10 wk, 77.7% vs 52.0% (OR 3.22, 1.95–5.32) at 24 wk |
| Trenoska Basile 2022 (PMID 35403706) | 10 studies, 11 remote conditions, 1,071 participants, GAD-specific | Within-group pre-post g=1.30 (1.03–1.58); between-group at post-treatment g=0.76 (0.47–1.06); low-intensity g=1.36 vs high-intensity g=0.83, not significantly different (Q₁=2.28, p=0.13) |
| Harrer 2025 (PMID 40703853) | Unified meta-analytic series, 168 studies, 22,144 patients with interview-confirmed diagnoses, digital vs inactive controls | Generalized anxiety g=0.80 — among the larger effects, with social anxiety 0.84, insomnia 0.94, panic 1.05, specific phobia 1.18, and depression 0.62, PTSD 0.57, OCD 0.68 lower. Dropout generally ≤20% but higher in intervention arms (RR 1.13–2.66); trials with low risk of bias and care-as-usual comparators were limited across all indications; 16 trials evaluated a prescribable digital therapeutic (g=0.33–1.60) |
| Roy 2021 (PMID 34860673) | 65 randomised, GAD; app-delivered mindfulness training (30 modules over 2 months) added to treatment as usual vs TAU | Median 25.5 of 30 modules completed; 46% completed the program; GAD-7 reduction 67% vs 14% (median change −8.5 vs −1.0, p<0.001). Mediation chain: increased non-reactivity at 1 month → reduced worry at 2 months (p=0.02); reduced worry at 1 month → reduced anxiety at 2 months (p=0.03) |
| Park 2025 (PMID 41086425) | 96 randomised (GAD-7 ≥10), 10-week parallel-group superiority trial of an ACT-based digital therapeutic plus TAU vs TAU, blinded evaluators | GAD-7 adjusted mean difference −2.26 (95% CI −3.78 to −0.75) at week 10; 71% and 65% maintained ≥80% of prescribed usage at weeks 5 and 10 |
| Andersson 2012 (PMID 22728647) | Therapeutic alliance in guided iCBT across depression, GAD and social anxiety | Alliance measurable in guided digital treatment |
| Wang 2026 (PMID 42247657) | Self-guided internet mindfulness-informed stress management for GAD, RCT with longitudinal network analysis | Recent unguided-format trial |
The Parsons 2025 trial matters disproportionately because its control was active (online psychoeducation), not a waitlist. Nearly everything else in the table above is waitlist- or care-as-usual-controlled, which is exactly the comparator problem that inflates psychotherapy effect sizes (cognitive behavioural therapy; Cuijpers 2025, PMID 40238104).
Conversational agents and generative AI¶
A newer layer sits below guided iCBT in intensity and above self-help in interactivity, and its effects are correspondingly smaller.
| Source | Scope | Effect |
|---|---|---|
| He 2023 (PMID 37115595) | 32 RCTs, 6,089 participants, conversational-agent interventions across mental-health problems | Generalized anxiety symptoms g=0.29 (95% CI 0.21–0.36); depressive symptoms 0.29 (0.20–0.38); specific anxiety 0.47 (0.07–0.86); quality of life/wellbeing 0.27; general distress 0.33; stress 0.24 |
| Zhang 2025 (PMID 41401240) | Systematic review of 26 studies of generative-AI chatbots with an RCT-only meta-analysis; search completed March 2025 | The first synthesis restricted to generative rather than rule- or retrieval-based agents; notes that most such interventions were delivered in non-WEIRD countries |
An effect of g≈0.29 for generalized anxiety symptoms (He 2023, PMID 37115595) is roughly a third of what guided internet CBT achieves in diagnosed GAD (g=0.79; Eilert 2021, PMID 33225589) and less than half the digital-intervention effect in interview-confirmed generalized anxiety (g=0.80; Harrer 2025, PMID 40703853). These are also symptom-level rather than disorder-level populations. The distinction between a digital treatment for diagnosed GAD and a chatbot that reduces anxiety symptoms is the same screening-versus-diagnosis distinction that runs through screening and measurement, and it is routinely elided in product claims.
The contradiction¶
| Claim | Evidence | Comparator |
|---|---|---|
| Digital ≈ face-to-face | g=0.14 (−0.01 to 0.30), 9 comparisons, 683 participants (Pauley 2023, PMID 34047264) | Direct digital-vs-face-to-face RCTs, anxiety disorders pooled |
| Digital < individual face-to-face | Individual CBT vs remote CBT SMD 0.96 (0.13–1.79); remote CBT not superior to TAU or waitlist (Liu 2025, PMID 40506439) | Network, 52 trials, 4,361 GAD patients |
A third estimate now sits between them: GAD-specific remote CBT gives a between-group g=0.76 (0.47–1.06) at post-treatment, with low- and high-intensity remote formats statistically indistinguishable (Trenoska Basile 2022, PMID 35403706); and the unified digital series puts generalized anxiety at g=0.80 against inactive controls (Harrer 2025, PMID 40703853). Both are against inactive controls, which is precisely the comparator the format network avoided. Candidate reconciliations, none tested: (i) the format network is GAD-specific while the head-to-head analysis pools anxiety disorders, so this may be a real GAD-specific finding; (ii) "remote CBT" in the format network mixes telephone, video and internet self-help, which are not the same intervention; (iii) the 9-comparison head-to-head analysis is underpowered, with a confidence interval that includes no difference and modest superiority alike; (iv) publication and comparator selection differ between the two evidence bases. Until one of these is tested, this knowledge base treats "guided digital CBT is equivalent to face-to-face CBT in GAD" as unproven (OPEN-QUESTIONS.md, OQ-5).
Guidance, adherence and attrition¶
| Metric | Value | Source |
|---|---|---|
| App uptake (initial activation) | 92.4% (95% PI 48.4–99.4), 56 conditions | Liu 2026, PMID 41259035 |
| App adherence (trial-defined adequate engagement) | 61.8% (95% PI 53.2–69.7), 21 conditions | Liu 2026, PMID 41259035 |
| Post-test attrition | 18.6% (95% PI 2.4–68.3) | Liu 2026, PMID 41259035 |
| Follow-up attrition | 28.4% (95% PI 3.0–83.7) | Liu 2026, PMID 41259035 |
| Attrition vs waitlist | RR 1.49 (95% PI 0.34–6.48) — higher in app conditions | Liu 2026, PMID 41259035 |
| What reduces attrition | Reminders, human contact, and absence of gamification features | Liu 2026, PMID 41259035 |
| Effect of guidance on completion | Amount completed g=0.29 (0.18–0.40); proportion completing log-OR 0.50 (0.34–0.66); full completion 12 percentage points higher with guidance (22 studies) | Musiat 2022, PMID 34802474 |
| Routine-care effectiveness (guided iCBT) | Pooled pre-post g=1.78 (depression) and 0.94 (anxiety) across 19 studies/30 groups; deterioration 2.9% | Etzelmueller 2020, PMID 32865497 |
| Standalone apps vs inactive controls | 72 RCTs, 21,702 participants; efficacy and study quality assessed against waitlist/informational/control-app comparators | Kulke 2025, PMID 41290454 |
| Component-level evidence | Meta-analytic review of intervention components in internet- and mobile-based interventions for anxiety disorders | Domhardt 2019, PMID 30450811 |
The guidance picture is genuinely mixed and the two results are compatible: guidance reliably improves adherence (+12 percentage points full completion; Musiat 2022, PMID 34802474) without a statistically detectable improvement in outcome in the anxiety-disorder subgroup analysis (guided 0.84 vs unguided 0.64, ns; Pauley 2023, PMID 34047264). The Bayesian CBT network reached the opposite conclusion via meta-regression — clinician guidance significantly moderated effect size (Dai 2025, PMID 40367584). Three analyses, three answers.
Where digital delivery is being deployed¶
- Routine care. Guided iCBT delivered in routine services produces clinically relevant pre-post change with acceptable uptake, adherence and satisfaction, and a low deterioration rate (2.9%) — but these are non-randomised pre-post designs (Etzelmueller 2020, PMID 32865497). A Canadian health technology assessment evaluated iCBT for major depression and anxiety disorders at system level (Health Quality Ontario 2019, PMID 30873251).
- University students. iCBT versus treatment as usual for anxiety and depression among Latin American university students (Benjet 2023, PMID 38032621), with a precision-treatment secondary analysis identifying who benefits (Benjet 2023, PMID 37285133). This matters because student populations carry high screened GAD prevalence (epidemiology and burden).
- Transdiagnostic versus disorder-specific. Both formats have been trialled in primary care; comorbidity does not appear to block transdiagnostic internet CBT (Newby 2017, PMID 27396841; Johnston 2013, PMID 23458319).
- Support intensity. Standard weekly versus optional weekly therapist support has been directly randomised (Hadjistavropoulos 2017, PMID 28964994) — the cleanest available test of how much human contact a digital programme needs.
- Economics. Digital interventions in mental health have been assessed with evidence synthesis plus economic modelling (Gega 2022, PMID 35048909).
Open questions¶
- Is guided digital CBT equivalent to face-to-face CBT in GAD specifically? The equivalence estimate pools anxiety disorders (Pauley 2023, PMID 34047264); the GAD-specific network says no (Liu 2025, PMID 40506439).
- Does guidance improve outcome or only adherence? Musiat 2022 (PMID 34802474) and Dai 2025 (PMID 40367584) say guidance matters; Pauley 2023 (PMID 34047264) finds no significant outcome difference.
- What does the Parsons 2025 result (PMID 41396602) imply for the waitlist-controlled literature? A 36-percentage-point remission difference against active psychoeducation is larger than most waitlist-controlled digital effects would predict.
- Are the engagement benchmarks (uptake 92.4%, adherence 61.8%; Liu 2026, PMID 41259035) achievable outside trials? Trial participants are recruited and monitored; service populations are not.
- Does digital delivery close or widen the treatment gap? The gap is largest in low-income countries and among older adults (epidemiology and burden). A PubMed search rerun on 2026-09-02 found no GAD digital-treatment meta-analysis reporting separate effects for low-income-country or older-adult samples; equity effects remain unquantified.
Related pages¶
- Cognitive behavioural therapy — the face-to-face comparator and the format network.
- Other psychological and non-drug therapies — internet-delivered ACT and mindfulness.
- Comorbidity and primary care — where digital programmes are meant to relieve capacity.
- Screening and measurement — GAD-7 as the near-universal digital-trial endpoint.
- Clinical trials landscape — what is registered now.
- Overview — map of the condition.
References¶
- Eilert N, et al. The effectiveness of Internet-delivered treatment for generalized anxiety disorder: An updated systematic review and meta-analysis. Depress Anxiety. 2021;38:196-219. PMID 33225589
- Pauley D, et al. Two decades of digital interventions for anxiety disorders: a systematic review and meta-analysis of treatment effectiveness. Psychol Med. 2023;53:567-579. PMID 34047264
- Paxling B, et al. Guided internet-delivered cognitive behavior therapy for generalized anxiety disorder: a randomized controlled trial. Cogn Behav Ther. 2011;40:159-73. PMID 21770848
- Dahlin M, et al. Internet-delivered acceptance-based behaviour therapy for generalized anxiety disorder: A randomized controlled trial. Behav Res Ther. 2016;77:86-95. PMID 26731173
- Carl JR, et al. Efficacy of digital cognitive behavioral therapy for moderate-to-severe symptoms of generalized anxiety disorder: A randomized controlled trial. Depress Anxiety. 2020;37:1168-1178. PMID 32725848
- Parsons EM, et al. Digital Cognitive Behavioral Treatment for Generalized Anxiety Disorder: A Randomized Clinical Trial. JAMA Netw Open. 2025;8:e2548884. PMID 41396602
- Andersson G, et al. Therapeutic alliance in guided internet-delivered cognitive behavioural treatment of depression, generalized anxiety disorder and social anxiety disorder. Behav Res Ther. 2012;50:544-50. PMID 22728647
- Wang Z, et al. Self-Guided Internet-Based Mindfulness-Informed Stress Management for Generalized Anxiety Disorder: Randomized Controlled Trial With Longitudinal Network Analysis. J Med Internet Res. 2026;28:e91751. PMID 42247657
- Liu S, et al. CBT treatment delivery formats for generalized anxiety disorder: a systematic review and network meta-analysis of randomized controlled trials. Transl Psychiatry. 2025;15:197. PMID 40506439
- Liu C, et al. Uptake, Adherence, and Attrition in Clinical Trials of Depression and Anxiety Apps: A Systematic Review and Meta-Analysis. JAMA Psychiatry. 2026;83:43-50. PMID 41259035
- Musiat P, et al. Impact of guidance on intervention adherence in computerised interventions for mental health problems: a meta-analysis. Psychol Med. 2022;52:229-240. PMID 34802474
- Etzelmueller A, et al. Effects of Internet-Based Cognitive Behavioral Therapy in Routine Care for Adults in Treatment for Depression and Anxiety: Systematic Review and Meta-Analysis. J Med Internet Res. 2020;22:e18100. PMID 32865497
- Kulke JK, et al. Efficacy of standalone smartphone apps for mental health: an updated systematic review and meta-analysis. Lancet Digit Health. 2025;7:100923. PMID 41290454
- Domhardt M, et al. Internet- and mobile-based interventions for anxiety disorders: A meta-analytic review of intervention components. Depress Anxiety. 2019;36:213-224. PMID 30450811
- Health Quality Ontario. Internet-Delivered Cognitive Behavioural Therapy for Major Depression and Anxiety Disorders: A Health Technology Assessment. Ont Health Technol Assess Ser. 2019;19:1-199. PMID 30873251
- Benjet C, et al. Internet-delivered cognitive behavior therapy versus treatment as usual for anxiety and depression among Latin American university students: A randomized clinical trial. J Consult Clin Psychol. 2023;91:694-707. PMID 38032621
- Benjet C, et al. A Precision Treatment Model for Internet-Delivered Cognitive Behavioral Therapy for Anxiety and Depression Among University Students: A Secondary Analysis of a Randomized Clinical Trial. JAMA Psychiatry. 2023;80:768-777. PMID 37285133
- Newby JM, et al. Transdiagnostic versus disorder-specific internet-delivered cognitive behaviour therapy for anxiety and depression in primary care. J Anxiety Disord. 2017;46:25-34. PMID 27396841
- Johnston L, et al. Comorbidity and internet-delivered transdiagnostic cognitive behavioural therapy for anxiety disorders. Cogn Behav Ther. 2013;42:180-92. PMID 23458319
- Hadjistavropoulos HD, et al. Randomized controlled trial of internet-delivered cognitive behaviour therapy comparing standard weekly versus optional weekly therapist support. J Anxiety Disord. 2017;52:15-24. PMID 28964994
- Gega L, et al. Digital interventions in mental health: evidence syntheses and economic modelling. Health Technol Assess. 2022;26:1-182. PMID 35048909
- Dai X, et al. (Third-wave) cognitive behavioral therapy for generalized anxiety disorder in adults: A systematic review and Bayesian network meta-analysis. J Psychiatr Res. 2025;187:134-143. PMID 40367584
- Cuijpers P, et al. Cognitive Behavior Therapy for Mental Disorders in Adults: A Unified Series of Meta-Analyses. JAMA Psychiatry. 2025;82:563-571. PMID 40238104
- Trenoska Basile V, et al. Remote cognitive-behavioral therapy for generalized anxiety disorder: A preliminary meta-analysis. J Clin Psychol. 2022;78:2381-2395. PMID 35403706
- Harrer M, et al. Effectiveness of digital interventions for eight mental disorders: A meta-analytic synthesis. Internet Interv. 2025;41:100860. PMID 40703853
- Roy A, et al. Clinical Efficacy and Psychological Mechanisms of an App-Based Digital Therapeutic for Generalized Anxiety Disorder: Randomized Controlled Trial. J Med Internet Res. 2021;23:e26987. PMID 34860673
- Park C, et al. Efficacy of a Smartphone-Based Digital Therapeutic (Anzeilax) in Generalized Anxiety Disorder: Randomized Controlled Trial. J Med Internet Res. 2025;27:e69981. PMID 41086425
- He Y, et al. Conversational Agent Interventions for Mental Health Problems: Systematic Review and Meta-analysis of Randomized Controlled Trials. J Med Internet Res. 2023;25:e43862. PMID 37115595
- Zhang Q, et al. Generative AI Mental Health Chatbots as Therapeutic Tools: Systematic Review and Meta-Analysis of Their Role in Reducing Mental Health Issues. J Med Internet Res. 2025;27:e78238. PMID 41401240