12-month follow-up of a four-armed adolescent RCT (n=100, 32% retained): VRET gains for public speaking anxiety were maintained, while a non-VR online CBT arm improved further
How this was rated
Long-term follow-up of a pre-registered, two-phase, four-armed RCT (NCT04396392; n=100 randomised, ages 13-16) analysed with intention-to-treat linear mixed models (N=99) under a missing-at-random assumption. Strengths: randomised design, pre-registration, a clinically important and under-studied population, and the first follow-up of adolescent public-speaking-anxiety VRET to extend beyond 3 months. Certainty is held LOW by serious limitations: (a) only 32% completed the 12-month assessment, with uneven, treatment-related attrition - the VRET-only arm retained the fewest participants, raising selective-attrition bias - and very small follow-up cells (VRET n=4); (b) phase 2 was underpowered after COVID-19 cut the planned sample from 160 to 100; (c) the waitlist received delayed treatment, so there is no clean control across the whole follow-up; (d) outcomes were self-report only; (e) the sample was 84% female and excluded non-Norwegian speakers and people with dyslexia, limiting external validity; (f) the COVID-19 period limited the real-life exposure practice that normally reinforces gains.
Ratings use a simplified four-tier scheme (High, Moderate, Low, Very Low) informed by the GRADE working group. Learn more about how studies are rated.
A 12-month follow-up of a pre-registered, four-armed RCT in 100 adolescents (ages 13-16) with public speaking anxiety, comparing self-guided VR exposure therapy (VRET) with online psychoeducation and online exposure programs. Across the whole sample, public speaking anxiety kept falling after treatment (6.5-point drop to 12 months, Cohen's d = 0.94). The VRET groups maintained their gains over the year, while the online psychoeducation-plus-exposure group was the only arm to improve significantly further (within-group d = 1.02) - so this is not a VRET-superiority result. Only 32 of 100 participants completed the 12-month assessment, so the authors present it as an early indication, not proof.
A long-term (12-month) follow-up of a four-armed adolescent RCT (n=100) suggesting that public speaking anxiety reductions from a short course of self-guided VRET were maintained over a year, while a non-VR online psychoeducation-plus-exposure arm continued to improve (the only group with a significant further reduction, within-group d=1.02). It is an encouraging but low-certainty signal for the durability of adolescent VRET, not evidence that VR outperforms other digital options. Only 32% completed the 12-month assessment (some VRET cells as small as n=4), attrition was uneven and lowest in the VRET-only arm, outcomes were self-reported, and the COVID-19 period limited the real-life exposure practice that usually consolidates gains. Most useful as rare long-term context alongside the same group's 3-month follow-up (Kahlon 2019) and adult long-term VRET data.
Key findings
- Long-term (12-month) follow-up of a four-armed RCT (n=100, ages 13-16) of self-guided VRET versus online CBT programs for adolescent PUBLIC SPEAKING ANXIETY (PSA) - the first to extend beyond a 3-month follow-up in this population
- RETENTION WAS LOW: only 32 of 100 participants (32%) completed the 12-month assessment, and attrition was UNEVEN - the VRET-only group had the lowest response rate - raising a real risk of selective-attrition bias, with very small 12-month cells (VRET n=4)
- ACROSS THE WHOLE SAMPLE: public speaking anxiety (PSAS) continued to fall from post-treatment to 12 months by 6.5 points (95% CI -10.57 to -2.43, p=0.003; Cohen's d = 0.94), most of it between post-treatment and 3 months
- The VRET GROUPS MAINTAINED their gains: both VR arms held their post-treatment PSA reduction across the 12 months but did not improve further and did not differ significantly from the waitlist-plus-psychoeducation reference on the follow-up slopes
- The ONLINE PSYCHOEDUCATION + ONLINE EXPOSURE group was the only arm with significant CONTINUED improvement: a total PSA reduction of 8.32 points from post-treatment to 12 months (95% CI -14.12 to -2.52, p=0.006; within-group d = 1.02) - this is NOT a VRET-superiority result
- SOCIAL PHOBIA (SPS-6) symptoms showed no significant change over the follow-up (the original trial's gains were maintained). SOCIAL INTERACTION ANXIETY (SIAS-6) showed a small but statistically significant whole-sample reduction across the full period (-2.02 points, p=0.028) that was not significant within either follow-up interval and gave no group-level benefit; notably, one VR arm (VRET followed by online exposure) showed a transient significant INCREASE in SIAS-6 versus waitlist from post-treatment to 3 months (0.31 points/week, p=0.023) that reversed to a non-significant change over the full follow-up - the VR scenarios involved public speaking only, with no interaction with the virtual audience
- The authors frame the result as an EARLY INDICATION that both VRET and online psychoeducation-plus-exposure can sustain PSA reductions in adolescents over a year, not as proof, and call for larger, controlled, less attrition-prone trials
- Pre-registered (NCT04396392) and funded by the Norwegian Research Council; the VR scenarios were built by Attensi AS on an Oculus Quest 1 (a Norwegian classroom with 14 virtual classmates); authors declared no commercial conflict of interest (co-author Philip Lindner was a Frontiers editorial board member)
Background
Public speaking anxiety (PSA) is a common, performance-focused form of social anxiety that often begins in adolescence and, left unaddressed, can limit classroom participation, harm educational outcomes, and in some cases develop into generalised social anxiety disorder. Cognitive behavioural therapy with exposure is the first-line treatment, and virtual reality exposure therapy (VRET) is an established, effective option for PSA in adults. What has been missing is evidence on whether VRET’s benefits last for young people: before this study, only one trial had followed adolescents beyond the end of treatment, and only to a 3-month follow-up (Kahlon et al. 2019).
This paper reports the 12-month follow-up of a pre-registered, two-phase, four-armed randomised controlled trial run by the same Bergen group, asking whether the post-treatment reductions in public speaking anxiety, social phobia, and social interaction anxiety were maintained, continued to fall, or returned over a full year.
What the researchers did
One hundred adolescents aged 13-16 (mean age 14.2; 84% female) with public speaking anxiety were recruited in Bergen, Norway, and randomised to one of four conditions, each delivered as two consecutive three-week phases (six weeks total):
- VRET only
- VRET followed by an online exposure program
- Online psychoeducation followed by an online exposure program
- Waitlist followed by online psychoeducation (the reference group)
The VRET was self-guided at home on an Oculus Quest 1, using scenarios built by Attensi AS that placed the adolescent in a prototypical Norwegian classroom with 14 virtual classmates. Participants worked through 15 public-speaking exercises whose difficulty the program varied by audience reaction, audience size (1 to 14), task length (30 seconds to 2 minutes), and task type (reading aloud or giving a presentation). The two online programs were text-based CBT, three modules each over three weeks.
Public speaking anxiety was measured with the Public Speaking Anxiety Scale (PSAS, the primary outcome); social interaction anxiety and social phobia were measured with the short SIAS-6 and SPS-6 (secondary outcomes). The 12-month data were analysed with intention-to-treat linear mixed models (N=99), with the waitlist-plus-psychoeducation group as the reference. The trial was pre-registered (NCT04396392).
What they found
- Retention was low. Only 32 of 100 participants (32%) completed the 12-month assessment, and the drop-out was uneven - the VRET-only group retained the fewest participants - so the long-term cells are very small (VRET n=4) and selective-attrition bias is a real concern.
- Across the whole sample, public speaking anxiety kept falling from post-treatment to 12 months, by 6.5 points on the PSAS (95% CI -10.57 to -2.43, p=0.003; Cohen’s d = 0.94), with most of the change happening between post-treatment and 3 months.
- The VRET groups maintained their gains. Both VR arms held their post-treatment reduction across the year but did not improve further, and on the follow-up slopes they did not differ significantly from the waitlist-plus-psychoeducation reference.
- The online psychoeducation-plus-exposure group was the only arm that kept improving, with a total PSA reduction of 8.32 points from post-treatment to 12 months (95% CI -14.12 to -2.52, p=0.006; within-group d = 1.02). This is not a VRET-superiority finding.
- Social phobia (SPS-6) did not change significantly over the follow-up for any group (the original trial’s gains were maintained). Social interaction anxiety (SIAS-6) showed a small but statistically significant whole-sample reduction over the full period (-2.02 points, p=0.028), though this was not significant within either follow-up interval and there was no group-level benefit. One VR arm (VRET followed by online exposure) even showed a transient significant increase versus waitlist from post-treatment to 3 months (0.31 points/week, p=0.023), which reversed to a non-significant change overall. The VR scenarios were public-speaking only, with no interaction with the virtual audience, which the authors suggest may be why social-interaction anxiety barely moved.
Why this matters
This is only the second known long-term follow-up of VRET for adolescent public speaking anxiety, and the first to reach a full year, so it begins to fill a genuine gap. The honest reading is encouraging but narrow: a short course of self-guided VRET appears to leave adolescents no worse off a year later and may help hold their gains, but the study does not show VR to be more effective than a non-VR online program - the online psychoeducation-plus-exposure arm was the one that continued to improve. For services, the useful signal is that several low-intensity, scalable digital formats may sustain reductions in adolescent public speaking anxiety, and choosing between them can rest on access, engagement, and preference rather than on an assumption that VR is inherently better.
Limitations
- Severe attrition. Only 32% completed the 12-month assessment, and retention was lowest in the VRET-only arm; the small, possibly non-representative completer sample is the central threat to every conclusion.
- Underpowered. COVID-19 cut the planned sample from 160 to 100, and the second phase was underpowered, weakening the follow-up analyses.
- No clean control. The waitlist received delayed treatment, so there was no fully naturalistic control across the whole follow-up; the time-by-group slopes are not corrected for that delay and should be read cautiously.
- Self-report only. All outcomes were self-reported, which can overestimate treatment effects relative to objective measures.
- Limited generalisability. The sample was 84% female and excluded adolescents without sufficient Norwegian language proficiency or with dyslexia; cultural and ethnic background was not recorded.
- Pandemic context. Data were collected during COVID-19, which restricted the real-life public-speaking practice that usually consolidates exposure gains and may partly explain why the VR groups maintained rather than extended their improvement.
Implications for practice
For clinicians and services weighing self-guided digital options for adolescent public speaking anxiety, this adds rare long-term (12-month) follow-up data and suggests that gains from a short course of self-guided VRET can hold for a year. It does not show VRET to be superior to a non-VR online psychoeducation-plus-exposure program - if anything, that online arm was the only one to keep improving - so the practical read is that several low-intensity digital formats may sustain PSA reductions, and the choice can be guided by access, engagement, and preference rather than an assumption that VR is more effective. The small, attrition-affected sample means none of this should be presented to young people or families as settled. For speech and language professionals supporting students whose public-speaking fear limits classroom participation, it is supportive context for offering controllable speaking practice, framed around access and participation rather than a cure.
Implications for research
Larger, adequately powered trials with a maintained control group and active retention strategies are needed before adolescent VRET durability can be considered established; the 32% 12-month completion rate is the central threat to inference here. A direct, dose-matched comparison of VRET against an online psychoeducation-plus-exposure program would clarify whether VR adds anything beyond a generic digital-CBT format for PSA. Adding interactive (for example AI-driven) virtual characters might extend effects to social-interaction anxiety, which neither the VR nor the online arms moved. Objective outcomes (not only self-report) and more representative samples (gender balance, language and cultural diversity) would strengthen external validity.
Cite this study
If you reference this study in your work, the canonical citation formats are:
@article{hauge2026,
author = {Hauge, P. R. and Gjestad, R. and Lindner, P. and Nordgreen, T. and Kahlon, S.},
title = {Long-term effects of virtual reality exposure therapy for adolescents with public speaking anxiety: a one-year follow-up of a randomised controlled trial},
journal = {Frontiers in Virtual Reality},
year = {2026},
doi = {10.3389/frvir.2026.1792043},
url = {https://withvr.app/evidence/studies/hauge-2026}
}TY - JOUR
AU - Hauge, P. R.
AU - Gjestad, R.
AU - Lindner, P.
AU - Nordgreen, T.
AU - Kahlon, S.
TI - Long-term effects of virtual reality exposure therapy for adolescents with public speaking anxiety: a one-year follow-up of a randomised controlled trial
JO - Frontiers in Virtual Reality
PY - 2026
DO - 10.3389/frvir.2026.1792043
UR - https://withvr.app/evidence/studies/hauge-2026
ER - Know of research that should be in this hub? If a relevant peer-reviewed study is not listed here, send the reference to hello@withvr.app. The hub is kept up to date as the literature grows.
Funding & independence
Funded by the Norwegian Research Council (NFR grant 259293). The virtual reality scenarios were developed by Attensi AS (a commercial developer) and delivered on an Oculus Quest 1; Attensi is not listed as a funder. The authors declared no commercial or financial conflict of interest; co-author Philip Lindner declared that he was a Frontiers editorial board member at the time of submission, with no impact on the peer-review process, and generative AI was not used in preparing the manuscript. No withVR BV involvement in funding, study design, or authorship, and this study used a different VR platform (Attensi), not Therapy withVR. Summary prepared independently by withVR using the published open-access peer-reviewed paper.