Today, over 2 billion people are affected by conflict or violence, and caregivers affected by war are particularly at-risk for poor mental health and violence against children. This will be among the first RCTs evaluating a psychosocial, parenting intervention amidst a war crisis. This research is testing if 'Hope Groups' work to help families in war. Hope Groups are peer-led psychosocial and parenting support groups of 4-7 participants for Ukrainian caregivers affected by war, compared to a wait-list control group.

This is a carbon-copy of a preregistration published to Open Science Framework on November 9, 2023: https://osf.io/uvj67/overview.

This study is a two-armed pragmatic cluster randomized controlled trial (CRCT) to test whether 'Hope Groups' (a peer-led psychosocial and parenting support group consisting of 12 sessions) improve caregiver mental health, positive parenting, and prevention of violence against women among the intervention group post-intervention, compared to a wait-list control group. Randomization will be performed at the cluster level (the clustered unit of randomization is each Hope Group consisting of 4-7 individuals) with a 1:1 allocation ratio and pair-matched on the hope group's facilitator. All facilitators recruit 2 or 4 groups, which enables pair-matching on facilitator to account for any strong, unmeasurable confounders within a facilitator's network (see more details in 'recruitment' section). Clusters will be randomized to either receive the Hope Group program now ('intervention group') or be placed on a waitlist to receive the Hope Group after trial endline ('control group').

After a facilitator has recruited all participants, and their participants have completed informed consent and baseline surveys, the trial data analyst will use a randomization package in RStudio to conduct the cluster pair-matched randomization (e.g., if a facilitator recruited 2 clusters, 1 group is randomized to intervention and 1 to control; if a facilitator recruited 4 groups, 2 groups are randomized to intervention and 2 to control). The trial analyst will set the seed to ensure randomization results are reproducible. The trial analyst will be blinded to the identifies of all participants within all clusters at the time of randomization. For each facilitator, their clusters will be randomized at the same time (using a randomization package in RStudio), in order to conceal allocation. After randomization is completed, the trial analyst will share the allocation of clusters with the study coordinator. The study coordinator will inform the facilitator, who will inform the intervention group(s) they will start the Hope Group intervention, and inform the control group(s) they are on a wait-list to start the Hope Group intervention after trial endline.

Endline surveys will be conducted for all clusters 1-week after Hope Group completion.

Inclusion Criteria:

• Participant is Ukrainian (or speaks Ukrainian and has been affected by the war in Ukraine);
• Participant is aged 18 or older
• Participant is a parent or caregiver who spends at least 15 hours per week with at least one child
• Participant has not already participated in a 'Hope Group'.

Exclusion Criteria:

• NA

IPD will be made available, including all primary and secondary outcomes and basic covariates (eg, sex, age, displacement status), but with identifiable information excluded (eg, initials, DOB, geographic location).

De-identified data will be made available to researchers upon reasonable request, subject to appropriate data-use conditions. This is available beginning March 1, 2026. Fully available anonymous data will be posted on Open Science Framework after the main results and moderator results are published, likely in 2027.

De-identified data will be made available to researchers upon reasonable request to Sydney Tucker (sydney.tucker@spi.ox.ac.uk), subject to appropriate data-use conditions. Fully available anonymous data will be posted on Open Science Framework after the main results and moderator results are published (likely 2027), and can be directly accessed on Open Science Framework.

Analysts will be blinded during randomisation and analysis.