In brief
Twenty studies, more than 1,700 children and adolescents, thirteen countries, ten randomised trials. The first systematic review of this literature called the findings "decidedly mixed." Read strand by strand rather than pooled, the picture is clearer than that.
Strongest. Anxiety. Every randomised trial that measured it found an effect, across subclinical, clinical, health-related and HIV-related anxiety, in four countries and in group, individual, school and telehealth formats. In the two trials with follow-up, the effect grew rather than faded.
Promising. Real-world behaviour, not just self-report: school attendance improved (g = -0.50 at follow-up) and physical activity was still higher a year later than in an active control. Obsessive compulsive disorder, where DNA-V within an intensive package produced the largest symptom reductions in the literature. Living with a chronic illness, where trials in Uganda, Australia and New Zealand show gains in stigma, health anxiety, quality of life and progress toward valued living. Social and emotional competence. And acceptability, which is the most consistent finding of all: young people in very different countries and circumstances turn up and engage.
Open. Depression, which almost no study was designed to test. Process measurement, where the field's standard instrument confounds inflexibility with distress. Social view and the social discoverer, untested anywhere. And independent replication, which is thin.
What DNA-V is
DNA-V describes four classes of behaviour and two contexts in which learning happens. Getting the description right matters, because the model is often evaluated as though it were only the first three.
DNA-V is transdiagnostic and process-based. It does not target a diagnosis. It targets the skills a young person uses to build a life, which is why it turns up in trials of anxiety, obsessive compulsive disorder, HIV medication adherence, diabetes distress, sleep, anger, caring responsibilities and school stress. More on the model.
Where to find the model itself
This page is about the evidence. For the model, the training, the books and the free practitioner resources, go to dnav.international, the home of DNA-V, run by Louise Hayes, Joseph Ciarrochi and Ann Bailey.
Where the evidence stands in 2026
The first systematic review of DNA-V was published by Byrne and Sherlock (2026) in the Journal of Contextual Behavioral Science. It found 14 studies covering 820 children and young people across eight countries, and rated five of them in the highest quality band on a standard appraisal tool. Their summary called the findings "decidedly mixed."
That review is a useful first map, and the scrutiny is welcome. But the phrase understates what the data show once three things are taken into account: what was counted as evidence in the first place, how psychological flexibility was measured, and what the depression studies were actually designed to test.
What the review counted
The review is titled as an evaluation of DNA-V "for children and young people with mental health difficulties." Its included studies are broader than that, and the breadth shapes the verdict.
- Twelve of the 14 studies used prevention samples rather than clinical ones. Only Schneider and Petersen (2024) recruited young people with a formal diagnosis. One study had no presenting difficulty at all.
- Several primary outcomes were not mental health outcomes: sleep hygiene and physical activity, adherence to HIV medication, learning motivation and study skills, social problem solving.
- The review's stated objectives include evaluating the model "in other associated areas like academics and social pursuits." That is a reasonable thing to do. The difficulty is that the headline verdict then averages across mental health, academic and physical health outcomes, and reports the result as a judgement about mental health.
Separate those strands and the picture changes. On anxiety, the outcome DNA-V was most often used to treat and the one the model's authors would predict it should move, the findings line up. On academic outcomes, which two studies examined and which the review itself reports two different ways, they do not. On depression, almost nothing was tested. "Decidedly mixed" is an accurate description of a set of studies pooled across those three questions. It is not an accurate description of any one of them.
A related point about breadth. DNA-V is transdiagnostic and process-based, so its application across anxiety, OCD, HIV adherence, diabetes distress, weight management, anger, sleep, caring responsibilities and school stress, in thirteen countries and in face-to-face, group, telehealth and online formats, is what the model should produce. An intervention that worked for one diagnosis in one country would be a poorer transdiagnostic model, not a better one. Variation across studies limits pooling. It does not indicate weakness.
Anxiety: a coherent signal, not a mixed one
Every controlled study in the review that examined anxiety found a treatment effect on its anxiety measure. The highest-rated trial in the literature, an RCT with an active control, found an effect that grew at follow-up rather than fading.
| Study | Country | Design | Sample | Anxiety type | Effect | Follow-up |
|---|---|---|---|---|---|---|
| Shao et al. (2024) | China | RCT, active control; quality rating 92% (high) | 139 adolescents | General, subclinical | d = 0.24 post, d = 0.56 at follow-up | 2 months, effect grew |
| Petersen et al. (2023) | USA | RCT vs waitlist; quality rating 69% (good) | 26 adolescents | School-based anxiety | Between-group g = -0.38 post, -0.63 at follow-up | 1 month, effect grew |
| Schneider & Petersen (2024) | USA | Pre-post, no control; quality rating 64% (good) | 25 adolescents with OCD | Anxiety in an OCD sample | Within-group g = 1.17 child report, 1.31 parent report | None; assessed at 12 weeks |
| Petersen, Donahue et al. (2024) | USA | RCT vs waitlist; quality rating 79% (high) | 30 adolescents | Health-related anxiety | Between-group g = -0.22 to -0.27 child report, -0.44 to -0.64 parent report | 1 month |
| Musanje, Kasujja et al. (2024) | Uganda | RCT vs control; quality rating 82% (high) | 122 adolescents with HIV | Health anxiety | Significant reductions; effect sizes not reported | None |
| Woo et al. (2025) | Malaysia | Pre-post, no control; quality rating 53% (good) | 19 adolescents | Anxiety and stress | No significant change | None |
That is four randomised trials across three countries plus two uncontrolled pilots, spanning subclinical, clinical, health-related and HIV-related anxiety, delivered face to face, in groups, individually and by telehealth. Every randomised trial found an effect on its primary anxiety measure. The one null came from a pre-post pilot of 19 young people with no control group, whose authors noted developmental problems with the response format.
Two qualifications belong here. The telehealth trial found its effect on health anxiety, the outcome it targeted, but not on a general measure of child anxiety. And Liu et al. (2023) found no between-group difference on stress at treatment end, though the face-to-face group was significantly lower than both the online and control groups by follow-up. Neither of these overturns the pattern, but a reader is entitled to see them. This reads as a coherent body of early evidence rather than a mixed one.
Beyond anxiety: what the other trials found
Physical health and long-term conditions
Two trials have used DNA-V with young people managing a chronic illness, and a third took it into physical health promotion in schools. Musanje, Kamya et al. (2024) randomised 122 Ugandan adolescents living with HIV (median age 17) to four weekly 90-minute group sessions or control, delivered in a public health facility. The DNA-V arm showed a significant reduction in psychological inflexibility at three-month follow-up, although self-reported antiretroviral adherence did not differ between arms. The companion paper, Musanje, Kasujja et al. (2024), reported significant reductions in depression, health anxiety and stigma in the same trial. It is the only one of the four studies in the review that measured depression to report a significant reduction in it. Panton et al. (2025) ran a pilot RCT with 22 Australian children aged 11 to 14 with type 1 diabetes, six 90-minute sessions in a hospital clinic, and found favourable trends in quality of life and diabetes-related stress. The trial was not powered for statistical testing. The quality-of-life trend held at three-month follow-up; the diabetes-stress gain had returned to baseline by then. Faulkner et al. (2018) embedded DNA-V in a high-school health and physical education curriculum with 115 US students; at one-year follow-up the DNA-V group was spending significantly more time physically active than the active control, with no effect on sleep.
Obsessive compulsive disorder
Schneider and Petersen (2024) delivered a telehealth intensive outpatient program to 25 adolescents with OCD, in which DNA-V sat alongside exposure and response prevention and DBT skills, at roughly ten hours per week for eight weeks. OCD symptoms fell substantially by 12 weeks (Hedges' g from 0.18 to 1.28 across child-report measures), with large reductions in anxiety (child g = 1.17, parent g = 1.31), a medium reduction in depression (g = 0.44), and a large reduction in family accommodation (g = 1.31). Two cautions. These are uncontrolled within-group changes in a 25-person pilot with no comparison arm, so they are not comparable to the between-group estimates elsewhere on this page. And the design cannot isolate what DNA-V contributed. Read as a signal about an intensive package that included DNA-V, not as an effect size for DNA-V itself.
School, learning and academic stress
This is the weakest and most contradictory part of the evidence base, and the review is inconsistent with itself about it. Liu et al. (2023) randomised 136 Chinese students (mean age 13.1) to internet-delivered DNA-V, face-to-face DNA-V, or control across six group sessions. The review's summary table reports significant improvement in learning motivation and stress, though the effect sizes given are negligible (0.04 and 0.05), while the review's narrative states there was no significant change on learning motivation, strategies or confidence across the three groups. What is not in dispute is that the face-to-face group had significantly lower stress than either the online or control group at two-month follow-up. Beni et al. (2023) randomised 132 Iranian adolescents (mean age 14) to ten 90-minute DNA-V group sessions or control in educational centres, and reported improvement in emotion regulation, social skills and academic skills. That result is stratified by family income: the social-skills and academic-skills gains held for adolescents from higher-income families and not for those from low-income families. Petersen et al. (2023) found that school-based DNA-V reduced class absences as well as anxiety, with a negligible effect at treatment end (g = -0.11) growing to a medium one at follow-up (g = -0.50). Absences are a behavioural outcome rather than a self-report one, which makes this finding worth more than its size suggests. Nisar et al. (2025) took a DNA-V-based curriculum to scale: a cluster-randomised trial of Connect PSHE across 20 North Wales primary schools and 745 children aged 7 to 11. Every school delivered Connect; what was randomised was whether staff received additional implementation support. That support made no difference to the primary wellbeing measure, improved two SDQ subscales, and wellbeing rose in both arms. The trial tells us a great deal about implementing a DNA-V curriculum at scale and nothing about DNA-V against a control, because there was no control.
Social and emotional competence, and anger
Saffarinia et al. (2023) ran a pilot RCT with 40 Iranian boys over eight weekly sessions and found significant improvement in social-emotional competence and social problem solving relative to control. Marino et al. (2019) worked with Italian students aged 12 to 14 at high psychosocial risk, all functioning in the borderline intellectual range, and found statistically significant improvement in psychological inflexibility, with no change on the mindfulness total score. This is one of only two studies in the review to move the AFQ-Y, and one of the few to build in a self-view focus. O'Driscoll et al. (2020) reported a single case, an 11-year-old boy in Northern Ireland seen for five individual sessions, whose anger score fell from the moderately elevated to the average range alongside improvement in psychological flexibility. No statistical analysis was undertaken.
Acceptability, attendance and fit
This is the most consistent finding in the literature and the one most easily overlooked. Attendance was high across the reviewed studies, with three reporting full attendance and an 89.5% completion rate in the highest-rated trial. Only one study reported attendance below 60%. Faulkner et al. (2018) found 70% of participants reported gaining something of lasting value. Cox et al. (2025) engaged adolescents in a tier-3 weight management clinic, a group that is hard to retain. Petersen et al. (2026) found adolescents in a Bronx community clinic rated the treatment above the acceptability cutoff and described enjoying the experiential exercises and the character of the Noticer, though attendance in that pilot averaged 5.4 of 8 sessions and its authors flagged attendance as something to design for. Young people across very different countries and presenting problems will do this work.
The psychological flexibility findings are a measurement problem
Seven of the reviewed studies measured process with the AFQ-Y or AFQ-Y8, and only two found significant change. It is tempting to read this as evidence that DNA-V does not move process. The measure is the more likely culprit. The AFQ-Y is unidimensional, it measures inflexibility rather than flexibility, and it conflates psychological inflexibility with general distress, a problem Cherry and colleagues (2021) set out in detail.
When studies use measures that are not entangled with distress in this way, change does appear. Shao et al. (2024) found d = 0.51 on the Cognitive Fusion Questionnaire at follow-up. Burley and McAloon's (2024) meta-analysis of group ACT for adolescent anxiety did find reductions in psychological inflexibility. There is a further possibility, worth testing rather than asserting: DNA-V works on a young person's awareness of how their own thinking functions, and a more aware adolescent may report troubling thoughts more readily, which would push an inflexibility score in the wrong direction while the underlying skill improves. No study has yet examined this, and it should be examined rather than assumed.
The accurate statement is that the field does not yet have adequate process measures for adolescents, and building them is a priority. Treating null AFQ-Y findings as evidence against DNA-V is the wrong inference.
Depression: absent evidence, not negative evidence
Only one of four studies with a depression measure found significant reductions (Musanje et al., 2024, in adolescents living with HIV). This is a real gap and worth stating plainly. The design context matters though. Depression was not the primary outcome in any of the other three studies and none of the samples were selected for depression: Petersen et al. (2023) recruited for anxiety, Petersen, Donahue et al. (2024) for health anxiety, Schneider and Petersen (2024) for OCD. Young people who do not start elevated have little room to improve, and floor effects make significant reductions hard to detect.
One wrinkle in the review is worth flagging, since it cuts the other way. Its summary table credits Schneider and Petersen (2024) with a medium reduction in depression (g = 0.44) at 12 weeks, while its narrative lists that same study among those finding "none or negligible" improvement. Both cannot be right. The figure is an uncontrolled within-group change, so it does not settle anything, but the count of one study in four is softer than it appears.
So the question of whether DNA-V helps adolescent depression remains open, because no trial has yet tested it directly. That is a different conclusion from "DNA-V does not work for depression," and it should shape the next wave of trials.
Studies published since the review's search
Byrne and Sherlock searched to April 2025 and used databases that do not index every relevant journal. Several substantial studies fall outside their 14. How closely each is tied to DNA-V varies, so that is stated for each one rather than left to the reader.
Children and adolescents
Emerging adults
Two recent studies apply DNA-V above the adolescent range. They belong in the evidence base, but they are not evidence about children, and they are listed separately for that reason.
Every study at a glance
The 14 studies in the Byrne and Sherlock review, followed by six more that fall outside it. Quality ratings are the QATSDD scores reported in the review (high, good or weak). Effect sizes are as reported in the original papers. Every study name links to the paper.
| Study | Design and quality | Sample and setting | Focus and dose | Findings |
|---|---|---|---|---|
| Shao et al. (2024)China | RCT, three arms: DNA-V face to face, DNA-V online, active control. High (92%) | 139 students, urban middle school | Anxiety. 6 group sessions per arm | Face-to-face arm: anxiety d = 0.24 post, 0.56 at 2-month follow-up; cognitive fusion d = 0.37 and 0.51. No significant change in the online or control arms. |
| Musanje, Kamya et al. (2024)Uganda | RCT, 61 v 61. High (87%) | 122 adolescents with HIV, median age 17, public health facility | Antiretroviral adherence. 4 weekly 90-min group sessions | Significant reduction in psychological inflexibility at 3-month follow-up. No significant difference in self-reported adherence. |
| Liu et al. (2023)China | RCT, internet DNA-V v face-to-face DNA-V v control. High (84%) | 136 students, mean age 13.1, school | Stress and learning behaviour. 6 group sessions | Face-to-face group had significantly lower stress than online or control at 2-month follow-up. On learning motivation the review's table reports a significant but negligible effect (0.04) while its narrative reports no significant change between groups. |
| Musanje, Kasujja et al. (2024)Uganda | RCT, same sample as above. High (82%) | 122 adolescents with HIV | Mental health. 4 weekly 90-min group sessions | Significant reductions in depression, health anxiety and stigma. The only one of the four studies in the review measuring depression to find a significant reduction in it. |
| Petersen, Donahue et al. (2024)USA | RCT, 15 DNA-V v 15 waitlist. High (79%) | 30 adolescents aged 12 to 17, at home by telehealth, parents included | Health-related anxiety. 10 weekly individual sessions | Significant time-by-condition effect on health anxiety, between-group g = -0.22 to -0.27; parent-reported child anxiety g = -0.44 to -0.64. No significant effect on general child anxiety, psychological inflexibility or anxiety sensitivity. (The review reports this trial's effect sizes differently in its table and its text; these are the text figures.) |
| Petersen et al. (2023)USA | RCT, 13 DNA-V v 13 waitlist. Good (69%) | 26 adolescents, mean age 15.7, school | Anxiety. 8 weekly face-to-face group sessions | Anxiety g = -0.38 at treatment end and -0.63 at follow-up. Class absences g = -0.11 and -0.50. No between-group differences on depression, psychological flexibility or student wellbeing. |
| Schneider & Petersen (2024)USA | Pre-post pilot. Good (64%) | 25 adolescents with OCD, mean age 15.7, outpatient clinic by telehealth | OCD. 8 weeks, roughly 10 hours per week, DNA-V alongside ERP and DBT | All within-group, no control arm. OCD symptoms g = 0.18 to 1.28 at 12 weeks. Child-reported anxiety g = 1.17, depression g = 0.44. Parent-reported anxiety g = 1.31 and family accommodation g = 1.31. The review's narrative elsewhere lists this study among those finding negligible depression change. |
| Marino et al. (2019)Italy | Pre-post. Good (56%) | 13 of 18 students, aged 12 to 14, at high psychosocial risk and functioning in the borderline intellectual range, school | Universal; values, goals, mindfulness and self-view focus. 12 face-to-face group sessions | Statistically significant improvement in psychological inflexibility, a mean 19% decrease. No change on the mindfulness total score, though one item improved significantly. |
| Panton et al. (2025)Australia | Pilot RCT. Good (56%) | 22 children aged 11 to 14 with type 1 diabetes, hospital clinic | Diabetes-related stress. 6 face-to-face 90-min sessions | Favourable trends in quality of life at post-intervention and at follow-up. Diabetes-stress gains returned to baseline by 3-month follow-up. Not powered for statistical testing; no analysis completed. |
| Woo et al. (2025)Malaysia | Pre-post pilot, no control. Good (53%) | 19 adolescents, mean age 15, school | Anxiety and stress. 6 weekly 1-hour online group sessions | No significant change in anxiety, stress or psychological flexibility. Authors noted developmental problems with the response format. |
| Beni et al. (2023)Iran | RCT, 66 v 66. Weak (48%) | 132 adolescents, mean age 14, educational centres | Emotion regulation and study skills. 10 face-to-face 90-min group sessions | Improvement in emotion regulation, social skills and academic skills in the treatment group, no change in control. Effect sizes not reported. Results are stratified by family income: the social-skills and academic-skills gains held for higher-income families and not for low-income ones. |
| Faulkner et al. (2018)USANo indexed record located | Controlled trial, 71 DNA-V v 44 active control. Weak (31%) | 115 students, grades 9 to 12, school | Sleep hygiene and physical activity. 6 sessions in a health and PE curriculum | At 1-year follow-up the DNA-V group spent significantly more time in physical activity. No effects on sleep or psychological inflexibility. 70% reported gaining something of lasting value. |
| Saffarinia et al. (2023)Iran | Pilot RCT, 20 v 20. Weak (31%) | 40 boys, clinic | Social-emotional competence and social problem solving. 8 weekly sessions | Significant improvement on both outcomes relative to control. |
| O'Driscoll et al. (2020)Northern Ireland | Single case study. Weak (30%) | One boy aged 11, school | Emotional and behavioural difficulties. 5 individual face-to-face sessions | Anger moved from the moderately elevated to the average range, with improvement in psychological flexibility. No statistical analysis. |
| Nisar et al. (2025)North Wales, UK | Cluster RCT, 20 primary schools. Not in the review | 745 children aged 7 to 11 | Universal wellbeing. Connect PSHE, a curriculum the paper describes as directly informed by DNA-V. Randomised contrast was standard v additional implementation support, not DNA-V v control | Wellbeing improved in both arms. Additional support did not change the primary outcome (β = 0.22, 95% CI -0.59 to 1.03, p = 0.59); small gains on the conduct problems and prosocial behaviour subscales of the SDQ. The largest evaluation of a DNA-V-based curriculum to date. |
| Hlebec et al. (2024)6 European countries | Cross-national RCT. Not in the review | 217 adolescent young carers aged 15 to 17 randomised, 213 analysed | Mental health and wellbeing. Psychoeducational program using DNA-V-adapted tools | Positive but largely non-significant changes in wellbeing and skills. Qualitative data supported the fit of the approach for young carers. |
| Petersen, Petersen & Pimentel (2026)USA | Pre-post pilot, weekly process measurement. Not in the review | 7 racially diverse adolescents, mean age 16.3, Bronx community mental health clinic | Anxiety. 8 weekly 90-min group sessions | Anxiety fell 5% on aggregate, with 4 of 7 reporting individual decreases of 3 to 37%. Psychological inflexibility fell for 3 of 7. Flexibility peaked at week 6, after all DNA-V components had been taught. Rated acceptable and positive; the Noticer was singled out. |
| Cox et al. (2025)UK | Co-development and qualitative study, not an efficacy trial. Not in the review | 14 recruited, 9 completed, median age 15, tier-3 paediatric weight management clinic; 63 interviews | Intrinsic motivation to change weight. Template intervention explicitly built on DNA-V | Acceptable to a group that is hard to retain. Participant-reported gains in self-awareness, coping strategies and social engagement. These are qualitative themes, not outcome data. |
| Above the adolescent range, listed separately | ||||
| Styles et al. (2026)New Zealand | Single-arm pilot. Not in the review | 16 enrolled, 14 received the intervention; emerging adults aged 18 to 25 with type 1 diabetes and above-target HbA1c | Glucose self-management. DNA-V informed a coping-strategies handout within a values card-sort and ACT matrix intervention | Significant improvement in progress toward valued living (standardised mean change 0.72, 95% CI 0.27 to 1.16). Small gains in diabetes-specific acceptance and psychological flexibility. Time in range rose 2.2 percentage points, not statistically significant (95% CI -1.2 to 5.7). |
| Helmus et al. (2026)Netherlands | Uncontrolled pre-post pilot. Not in the review | 61 enrolled, 33 completed both assessments; young adults with severe mental illness, mean age 27.4, range 19 to 36 | Transdiagnostic ACT training structured on DNA-V | Significant improvements in personal recovery and functional impairment. No change in psychological flexibility or self-stigmatisation. |
What the evidence does not yet show
Being clear about the limits is part of taking the model seriously.
- Most trials are small, and the strongest ones cluster in a few research groups. Independent replication is thin.
- No trial has recruited for depression and tested DNA-V against it directly.
- The school and academic findings are inconsistent, and the review itself reports the Chinese trial's learning outcomes two different ways.
- Social view, which is central to how young people learn in relationship, was not included in any study in the review. Neither was the social discoverer. Self-view has appeared in only one small pilot.
- Follow-up is the exception rather than the rule. Nine of the 14 reviewed studies had none at all, so claims about durability rest on a handful of trials.
- No study has established which components are doing the work, and dosage varied from four sessions to a full school year.
- The field lacks a process measure for adolescents that separates psychological flexibility from distress.
None of this makes the model unproven. It makes the next studies obvious: larger trials with control groups, independent replication outside the two or three groups doing this work, longer follow-up, a process measure built for adolescents, and a trial that recruits for depression and tests it directly.
Key sources
Related work
- The DNA-V model explained
- All publications, searchable, with free PDFs
- Process-based therapy and idionomic analysis
- DNA-V International
Prepared by Joseph Ciarrochi and Louise Hayes (ORCID 0000-0003-0471-8100). Last updated September 2026. Effect sizes are reported as given in the original papers, and where the source review reports a figure two different ways that is stated rather than resolved silently. Please cite the published versions. For the model itself, see dnav.international.