Our Recommendation
Friendship Bench Zimbabwe is a high‑performing, evidence‑rich mental‑health organization that squarely fits our framework: it has an explicit Theory of Change, a tightly aligned intervention, and multiple RCTs showing substantial improvements in depression and functioning, plus wellbeing‑based cost‑effectiveness estimates.
The idea is famous because it is so simple: a wooden bench outside a clinic, and a trained grandmother sitting on it. The reason it belongs in this directory is that the idea was put to a cluster randomized clinical trial across 24 primary care clinics in Harare, with 573 patients and an enhanced usual care control arm, and published in the Journal of the American Medical Association. At six months, intervention participants scored 3.81 on the Shona Symptoms Questionnaire against 8.90 in the control arm, an adjusted mean difference of −4.86. A secondary analysis found persistent depression in 17.4 percent of the intervention arm against 58.6 percent of controls. The organization reports these as a 78 percent reduction in depression and suicidal ideation and a 60 percent improvement in quality of life. Depression, suicidal ideation, and quality of life are ultimate outcomes under our framework, not proxies for them, and they were measured against a randomized comparison.
Two further trials followed: one in rural Bikita measuring antiretroviral therapy adherence and viral suppression, and one with Zvandiri across 842 adolescents where common mental disorder symptoms fell from 68 percent to 2 percent. Two caveats belong in a donor's file. The flagship trial ran in 2014 and 2015 with 573 people; the program now reports 1,093,104 clients seen nationally, and the 78 and 60 percent figures sit on the homepage under a heading reading "Impact to Date in Zimbabwe," beside that million-plus number, which invites a reader to attribute trial results to the whole program. And the outcome the organization names first in its own tagline, belonging, is the one it has never measured, despite operating 995 peer support groups built to create it.
Friendship Bench Zimbabwe's Fierce Certification score is 100/100 points based on our criteria:
✔ Has Ultimate Outcome Goals (20 pts)
✔ Measures Intermediate Outcomes (15 pts)
✔ Measures Ultimate Outcomes (20 pts)
✔ Measures Intermediate and/or Ultimate Counterfactual (20 pts)
✔ Shows Continual Learning & Adaptation (25 pts)
The Social Problem
Zimbabwe has a handful of psychiatrists for a population of roughly sixteen million. Common mental disorders, meaning depression and anxiety, are a leading cause of disability globally, and in low-income countries the overwhelming majority of people who have them receive no treatment at all.
The founder's account of why the organization exists is part of the model's origin: a patient who died by suicide because she could not afford the bus fare to the clinic. That is the shape of the problem. The treatment is not expensive in itself. What is missing is anyone trained to deliver it, anywhere a person can reach, at a cost they can pay, in a form that does not require them to identify as mentally ill in front of their community.
The trial population shows who carries this. Eighty-six percent were women. The median age was 33. Forty-two percent were living with human immunodeficiency virus. These are people managing poverty, caregiving, and often a chronic stigmatized illness simultaneously, screened positive for common mental disorder while attending a primary care clinic for something else.
The consequences run inward and outward. Inward: depression, anxiety, suicidal ideation, impaired daily functioning, and the loss of any expectation that things will change. Outward: antiretroviral medication not taken, work not done, and the isolation of having nobody to tell.
The Solution
A wooden bench in the grounds of a clinic, and a trained grandmother sitting on it.
The counselors. Older women from the community, known as ambuya utano or health grandmothers, trained and supervised to deliver a structured, locally adapted problem-solving therapy. They are lay health workers, not clinicians, and their presence in the community is the reason the model works where a clinic referral would not.
The course. Six individual sessions delivered over four to six weeks, using a structured approach to identify problems, including practical ones, and build a positive attitude toward resolving them. Screening and outcome measurement use the Shona Symptoms Questionaire, a 14-item instrument developed and validated in the Shona language rather than imported.
Circle Kubatana Tose, meaning holding hands together. After the individual sessions, clients join peer-led group support where members are trained in income-generating skills such as making bags from recycled plastic. 995 groups are active. This is the step-down layer and it addresses the practical problems that caused the distress in the first place.
Extensions: online talk therapy, a Youth Friendship Bench, and the ongoing Friendship Bench Plus trial training general nurses to prescribe antidepressants alongside the bench.
Mapped against a Theory of Change, the chain is short: train a trusted community member, put her somewhere people already go, deliver structured therapy free, follow with peer support and a livelihood, and depression lifts and quality of life rises. Every link has been tested against a randomized control, and the endpoint of the chain is an ultimate outcome rather than a service statistic.
Key Outputs
As of April 2026:
- 2,773 community health workers trained
- 1,093,104 clients seen on the Friendship Bench
- 995 active Circle Kubatana Tose peer support groups
- Free talk therapy delivered in primary care clinic grounds nationally, plus an online service
- The model is nationally embedded and government-owned in Zimbabwe, reached through a documented progression from proof of concept, to real-world validation, to structured scale-up, to system integration via phased technical assistance and government transition
- Three completed randomized trials and one in progress: the 2016 Journal of the American Medical Association trial, the Bikita antiretroviral therapy trial in JAMA Network Open, the 842-adolescent Zvandiri trial in PLOS Medicine, and the Friendship Bench Plus trial (296 adults, 12 Harare clinics, testing fluoxetine prescribed by general nurses alongside the bench)
- Research partners including the University of Zimbabwe, the London School of Hygiene and Tropical Medicine, and King's College London
- Happier Lives Institute Top Charity 2025
Key Intermediate Outcomes
One intermediary outcome carries randomized counterfactual evidence. The Bikita district cluster randomized trial, published in JAMA Network Open, set mean antiretroviral therapy adherence during months 2 to 6 as its primary outcome, comparing Friendship Bench plus enhanced standard of care against standard of care alone, with 244 participants in the intervention arm, in a population screened positive for common mental disorder and on first-line therapy for at least six months. That trial tests whether treating depression improves adherence to a different treatment, which is a question the global health field has asked for two decades and rarely answered experimentally.
Reach and workforce. 2,773 community health workers trained and 1,093,104 clients seen. For a task-shifting model, the trained workforce count is the more meaningful of the two, because it is the constraint.
Treatment completion. Six structured sessions, with 90.9 percent of the 573 randomized participants completing six-month follow-up. Retention is the usual failure point of lay-delivered psychotherapy, and this figure is unusually high.
The peer support and livelihood layer. 995 active Circle Kubatana Tose groups, teaching income-generating skills. This component was part of the randomized package and has never been isolated, so nobody knows how much of the effect it carries.
System integration. Government ownership is the strongest structural result here. A model that a ministry of health runs and funds outlasts the organization that built it, and it is the reason the reach figure can be in the millions.
Key Ultimate Outcomes
Depression and common mental disorder symptoms, measured against a randomized control group.
The design: a cluster randomized clinical trial with six months of follow-up, conducted September 2014 to May 2015, in which 24 primary care clinics in Harare were randomized 1:1 to the intervention or to enhanced usual care, meaning usual care plus information, education, and support on common mental disorders. That is a hard comparison, because the control arm received something real. 573 patients enrolled, all scoring 9 or higher on the Shona Symptoms Questionnaire.
The result: intervention participants scored 3.81 (95 percent confidence interval 3.28 to 4.34) against 8.90 (8.33 to 9.47) in the control arm; adjusted mean difference −4.86 (−5.63 to −4.10), p<0.001. A secondary analysis of the same trial found persistent depression at six months in 17.4 percent of the intervention arm against 58.6 percent of the control arm, meaning 82.6 percent of intervention participants had remitted below the depression cut-point against 41.4 percent of controls.
Suicidal ideation sits inside the Shona Symptoms Questionnaire and is covered by the 78 percent headline. No separate effect size is published, which is a gap worth closing given how central that outcome is to the founding story.
Quality of life improved 60 percent at six months in the same trial. The instrument and the effect size are not named on the organization's site, which is the one substantive documentation gap in an otherwise well-evidenced claim.
Anxiety was measured with the seven-item Generalized Anxiety Disorder scale in a secondary analysis, and separately in the Zvandiri collaboration, where common mental disorder symptoms among 842 adolescents living with human immunodeficiency virus fell from 68 percent to 2 percent under Friendship Bench problem-solving therapy delivered by trained peer counsellors, against 72 percent to 10 percent under standard counselling.
Viral suppression, a physical health outcome, was measured at 6 and 12 months against the randomized control arm in the Bikita trial.
Continual Learning & Adaptation
The progression from pilot to government ownership was deliberate and is documented. The organization's own account describes a trajectory from small-scale research initiative to nationally embedded, government-owned model, through proof of concept, real-world validation, structured scale-up, and system integration via phased technical assistance and government transition. Very few organizations lay out their scaling logic as a sequence of evidentiary stages and then complete it.
It published the protocol before the trial. The cluster randomized trial protocol appeared in the International Journal of Mental Health Systems in 2015, describing the design and the pilot evidence (Shona Symptoms Questionnaire falling from 11.3 to 6.5 in 320 patients) that justified it. Pre-committing to the design makes the 2016 result harder to reframe.
It kept testing after the famous result. Most organizations with a Journal of the American Medical Association publication would stop. Friendship Bench went on to test the model in a different population (people on antiretroviral therapy in rural Bikita), in a different age group through the Zvandiri collaboration, and on a different question entirely through the Friendship Bench Plus trial, which asks whether adding an antidepressant prescribed by a general nurse improves on the bench alone. That last trial is an organization testing its own ceiling.
It built the instrument locally rather than importing one. The Shona Symptoms Questionnaire was developed and validated in Shona, covering symptoms as they present in that population, including a locally meaningful concept of kufungisisa, or thinking too much. Measurement designed for the setting is why the effect sizes are credible.
The model grew a second layer in response to what the first one could not fix. Circle Kubatana Tose exists because individual therapy did not address the material problems that produced the distress, and the income-generating component is a direct response to that.
It extended delivery channels. Online talk therapy, a Youth Friendship Bench, and international replication through a separate global entity.
Its measurement philosophy is stated by the founder in one line on the homepage: if we can't measure what we're doing then we really don't know if it's working. The publication record supports the claim rather than decorating it.
Further Reading
Organization Sources
- Friendship Bench Zimbabwe homepage, source of the 2,773 community health workers, 1,093,104 clients, 995 groups, 78 percent, and 60 percent figures as of April 2026, and the mission, vision, and values. friendshipbenchzimbabwe.org
- Friendship Bench Zimbabwe, About Us and Problem Solving Therapy. About Us and Problem Solving Therapy
- Friendship Bench Zimbabwe, Circle Kubatana Tose peer-led support groups. friendshipbenchzimbabwe.org/ckt
- Friendship Bench Zimbabwe, Impact Reports. friendshipbenchzimbabwe.org/impactreports
- Friendship Bench Zimbabwe, "The Evolution of Friendship Bench: Two Decades of Community-Led Mental Health Care in Zimbabwe," May 2026. Source of the proof-of-concept to government-transition progression. Blog post
- Friendship Bench Zimbabwe, Publications, The Grandmothers, Youth Friendship Bench, and Online Talk Therapy. Publications, Grandmothers, Youth Bench, Online
- Friendship Bench Global. friendshipbench.org
Randomized Evidence
- Chibanda, D., Weiss, H. A., Verhey, R., Simms, V., Munjoma, R., Rusakaniko, S., Chingono, A., Munetsi, E., Bere, T., Manda, E., Abas, M., and Araya, R. (2016). "Effect of a Primary Care-Based Psychological Intervention on Symptoms of Common Mental Disorders in Zimbabwe: A Randomized Clinical Trial." Journal of the American Medical Association. Source of the 24-clinic design, the 573-patient sample, the Shona Symptoms Questionnaire scores of 3.81 versus 8.90, the adjusted mean difference of −4.86, and the 90.9 percent retention. jamanetwork.com
- Chibanda, D., Bowers, T., Verhey, R., Rusakaniko, S., Abas, M., Weiss, H. A., and Araya, R. (2015). "The Friendship Bench programme: a cluster randomised controlled trial of a brief psychological intervention for common mental disorders delivered by lay health workers in Zimbabwe." International Journal of Mental Health Systems 9:21. The pre-published protocol and pilot results. ncbi.nlm.nih.gov
- "The effect of co-morbid anxiety on remission from depression for people participating in a randomised controlled trial of the Friendship Bench intervention in Zimbabwe." EClinicalMedicine. Source of the persistent depression figures of 17.4 percent versus 58.6 percent and the Generalized Anxiety Disorder scale measurement. sciencedirect.com
- "Effect of the Friendship Bench Intervention on Antiretroviral Therapy Outcomes and Mental Health Symptoms in Rural Zimbabwe: A Cluster Randomized Trial." JAMA Network Open. Source of the Bikita design, the adherence primary outcome, and the viral suppression secondary outcome. jamanetwork.com and preprint
- Simms, V., et al. (2022). "Peer-led counselling with problem discussion therapy for adolescents living with HIV in Zimbabwe: A cluster-randomised trial." PLOS Medicine. The Zvandiri collaboration, 842 adolescents, source of the 68-to-2 percent finding. journals.plos.org
- Friendship Bench Plus trial registration, ClinicalTrials.gov NCT06384209. 296 adults, 12 Harare primary health care centres, testing fluoxetine prescribed by general nurses alongside the bench. clinicaltrials.gov
Outside Assessment and Background
- Happier Lives Institute, Friendship Bench Zimbabwe Top Charity assessment, 2025. happierlivesinstitute.org
- Mental Health and Psychosocial Support Network, "The Friendship Bench Model," Repository of Scalable Psychological Interventions for Humanitarian Settings Toolkit. Source of the 80 percent and 60 percent characterization of the trial results. mhpss.net
- Grand Challenges Canada coverage of the trial results. neurosciencenews.com
Written by
