| Field | Before | After |
|---|---|---|
| Field Abstract | Before Mental health care access remains limited in low- and middle-income contexts, prompting innovation in task-shifting interventions that leverage existing social networks. This randomized controlled trial (RCT) evaluates the Heal by Hair program in Lomé, Togo (2024), which trains hairdressers, trusted community actors, to deliver mental health promotion and basic support. The intervention comprises two arms: (1) a core training (3-day workshop + follow-up sessions) T1, and (2) an augmented "Heal by Hair Cercles" component, wherein trained hairdressers provide structured peer support to clients exhibiting severe symptoms (T2). Using a cluster-randomized design, we assign 800 hairdressers and 4,000 of their clients to either treatment (T1 or T2) or control groups. Post training the treatment groups will then be further separated between T1 and T2 (1:1 ratio). Primary outcomes include mental health knowledge (hairdressers) and self-reported symptoms (hairdressers/clients), assessed via validated psychometric instruments (e.g., PHQ-9, GAD-7) at 12- and 24-month follow-ups. To address potential reporting bias, a subsample (n = TBD) undergoes clinical validation via 30-minute psychological assessments conducted by trained psychologists. Secondary analyses will explore heterogeneous effects by baseline predictors of our main outcomes using Generalized Random Forest. By testing a scalable, salon-based model, this study advances evidence on (1) the efficacy of mental health interventions led by lay mental health provider and (2) the incremental benefits of structured peer support (Cercles). Findings will contribute to literature on task-shifting in global mental health and inform policies on integrating mental health services into informal community networks in low income countries. | After Mental health care access remains limited in low- and middle-income contexts, prompting innovation in task-shifting interventions that leverage existing social networks. This randomized controlled trial (RCT) evaluates the Heal by Hair program in Lomé, Togo (2024), which trains hairdressers, trusted community actors, to deliver mental health promotion and basic support. The intervention comprises two arms: (1) a core training (3-day workshop + follow-up sessions) T1, and (2) an augmented "Heal by Hair Cercles" component, wherein trained hairdressers provide structured peer support to clients exhibiting severe symptoms (T2). Using a cluster-randomized design, we assign 800 hairdressers and 4,000 of their clients to either treatment (T1 or T2) or control groups. Primary outcomes include mental health knowledge (hairdressers), self-reported symptoms (hairdressers/clients), as well as a range of validated psychometric instruments (e.g., PHQ-9, GAD-7) collected during 12- and 24-month follow-up surveys. To address potential reporting bias, a subsample (n = TBD) will undergo clinical validation via 30-minute psychological assessments conducted by trained psychologists. Secondary analyses will explore heterogeneous effects by baseline predictors of our main outcomes using Generalized Random Forest. By testing a scalable, salon-based model, this study advances evidence on (1) the efficacy of mental health interventions led by lay mental health providers and (2) the incremental benefits of structured peer support ("Cercles"). Findings will contribute to literature on task-shifting in global mental health and inform policies on integrating mental health services into informal community networks in low income countries. |
| Field Trial Start Date | Before September 01, 2025 | After August 17, 2025 |
| Field Last Published | Before September 19, 2025 10:16 AM | After September 24, 2025 02:14 PM |
| Field Randomization Method | Before The randomization will follow a re-randomization procedure: a re-randomization algorithm will run to achieve balance over a number of pre-determined (see PAP) baseline variables, so called rerandomization variables. A rerandomization variable is deemed balanced if having no significant differences between groups at the 10% significance level. The procedure will generate 1,001 random draws. The median one, number 501, will be chosen as our final random draw. If convergence issues arise (i.e. achieving 1000 draws is too slow or impossible), variables will be iteratively removed to reach 1000 randomizations in a reasonable time (typically below 8 hours). | After The randomization will follow a re-randomization procedure: a re-randomization algorithm will run to achieve balance over a number of pre-determined (see PAP) baseline variables, so called rerandomization variables. A rerandomization variable is deemed balanced if having no significant differences between groups at the 10% significance level. The procedure will generate 1,000 random draws. The median one, number 501, will be chosen as our final random draw. If convergence issues arise (i.e. achieving 1000 draws is too slow or impossible), variables will be iteratively removed to reach 1000 randomizations in a reasonable time (typically below 8 hours). |
| Field Public analysis plan | Before No | After Yes |
| Field | Before | After |
|---|---|---|
| Field Document | Before |
After
Pre-analysis Plan_AEA.pdf
MD5:
b5aaedb76e94a6a92b45412d6961b2b2
SHA1:
40019c38023b71ab4e30258e14be3cba7d39fba1
|
| Field Title | Before | After PAP |