Background Rigorous evidence of the effectiveness of male engagement interventions, particularly on how these interventions impact relationship power dynamics and women’s decision-making, remains limited. This study assessed the impact of the Bandebereho gender-transformative couples’ intervention on impact on multiple behavioral and health-related outcomes influenced by gender norms and power relations. Methods We conducted a multi-site randomised controlled trial in four Rwandan districts with expectant/current fathers and their partners, who were randomised to the intervention (n = 575 couples) or control group (n = 624 couples). Primary outcomes include women’s experience of physical and sexual IPV, women’s attendance and men’s accompaniment at ANC, modern contraceptive use, and partner support during pregnancy. At 21-months post-baseline, 1123 men and 1162 partners were included in intention to treat analysis. Generalized estimating equations with robust standard errors were used to fit the models. Findings The Bandebereho intervention led to substantial improvements in multiple reported outcomes. Compared to the control group, women in the intervention group reported: less past-year physical (OR 0.37, p<0.001) and sexual IPV (OR 0.34, p<0.001); and greater attendance (IRR 1.09, p<0.001) and male accompaniment at antenatal care (IRR 1.50, p<0.001); and women and men in the intervention group reported: less child physical punishment (women: OR 0.56, p = 0.001; men: OR 0.66, p = 0.005); greater modern contraceptive use (women: OR 1.53, p = 0.004; men: OR 1.65, p = 0.001); higher levels of men’s participation in childcare and household tasks (women: beta 0.39, p<0.001; men: beta 0.33, p<0.001); and less dominance of men in decision-making. Conclusions Our study strengthens the existing evidence on male engagement approaches; together with earlier studies our findings suggest that culturally adapted gender-transformative interventions with men and couples can be effective at changing deeply entrenched gender inequalities and a range of health-related behavioral outcomes.
We conducted a two-arm multi-site randomized controlled trial to assess the impact of the Bandebereho couples’ intervention on our outcomes of interest. Couples were recruited from local communities in Karongi, Musanze, Nyaruguru and Rwamagana districts in Rwanda from February 19 to March 17, 2015, and followed over a period of 21 months for this study. Men were interviewed at three time points: baseline, 9 months post-baseline, and 21 months post-baseline; due to funding constraints, women were interviewed at only two time-points, at 9 and 21 months post-baseline. In order to highlight the longer-term effects of the intervention, this paper presents the findings from 21 months post-baseline. The Rwanda Men’s Resource Center, a local non-governmental organization implementing the intervention, selected the sites in collaboration with district authorities. For the study, a total of 1199 men were recruited from 48 pre-selected sites within 16 sectors (sub-district administrative units) in the districts selected for the intervention. Couples’ inclusion in the study was determined by men’s eligibility for the intervention. Eligible men were aged 21–35 years, married or cohabitating, expectant and/or fathers of children under-five years (based on self-reports), living within accessible distance of the meeting site, and were not previous Bandebereho intervention participants. The legal age of marriage (21 years) in Rwanda served as the minimum age for participation. Community volunteers facilitating the intervention worked with local community health workers to identify 25 eligible men in each of the 48 sites. We conducted a power analysis prior to study recruitment, in June 2014, to assess ability to detect intervention effects on selected outcomes. We calculated power for outcomes similar to those we planned to measure, including perpetration of IPV, communication about family planning, and gender attitudes, using estimates from the 2010 Rwanda Demographic and Health Survey [18] and the 2010 International Men and Gender Equality Survey [19], assuming an intervention sample size of 576 couples (48 groups with 12 couples each). We conservatively calculated power for a 5–10% difference in these outcomes, using a two-sided test of equality of two proportions with adjustment made for design effects due to clustering, assuming an intra-class correlation coefficient of less than 0.1 and an alpha of 0.05. We found that the indicators would provide enough power (between 65% and 99%, depending on the indicator). Randomization to either the intervention or control group was done after baseline interviews using the individual as the unit of randomization. In each of the 48 sites, 12 men were randomly assigned to the intervention arm (n = 575), and the remaining men were assigned to the control arm (n = 624). Laterite, an independent firm collecting the data, randomized the participants using a random number generator in Stata 12. Bandebereho community facilitators notified men of their assignment. All recruited men remained eligible for randomization to the intervention regardless of participation in baseline data collection: in total, 1199 men out of a possible 1200 were invited to participate in the study, and 1195 men were surveyed at baseline. After randomization, we discovered two facilitators from neighbouring sites had mistakenly recruited the same participant, who was randomized twice into the intervention, resulting in 575 men randomized to the intervention, out of a possible 576. Due to the nature of the intervention, it was not possible to mask group assignments for participants. Group assignment was also not masked for the data collectors, who were not involved in the intervention. Specific measures to track spillover effects were included in the study design because the intervention and control groups reside in the same communities, and the intervention promotes community outreach. However, we posited that the effects of participation in the intensive intervention would outweigh any spillover effects and that such effects would result in underestimation, rather than over-estimation, of the intervention’s impact. Structured questionnaires were administered to male participants at baseline from 19 February to 17 March 2015. As noted above, men’s partners were not surveyed at baseline due to funding constraints. After the baseline and randomization, the Bandebereho intervention was implemented with the intervention group from March to July 2015. Follow-up surveys were conducted with men and their current partners at 9 months, from 9 November to 17 December 2015 (4 months post-intervention), and again at 21 months, from 7 November to 23 December 2016 (16 months post-intervention). At 21 months, 99.6% of the women surveyed were the same partner identified at baseline. At each follow-up, the participation of both partners was not required: either partner could be interviewed even if the other was unavailable. Study participants received a 2000 Rwandan franc transport stipend (about US$2.50) for each interview. Sex-matched interviewers from Laterite, who had no involvement in the intervention, conducted the interviews in Kinyarwanda in centrally located settings such as schools. Data were collected on password-protected tablets. All efforts were made to ensure study participant safety, privacy and comfort. Informed consent was obtained from all participants. The interviewer reviewed the consent form with each participant and answered any questions; participants signed a written consent if they were literate, or provided a thumbprint if they were not. The study was conducted in accordance with international ethical guidelines on researching violence against women, including not interviewing members of the same household about IPV [20]. At follow-up, we asked women about their experiences of IPV, but did not ask men about violence perpetration, and men were not informed of the inclusion of questions about violence in the women’s questionnaire. To minimize risk of harm, we obtained men’s consent to disclose their participation in the study before contacting their partners, and interviews with men and women were conducted on different days. Participants were offered a list of locally available support services after the interviews. Male and female interviewers received ethics and safety training and a female Rwandan counselor met with the female interviewers before, during and after data collection. The study protocol received approval from the Rwanda National Health Research Committee (25 August 2014, NHRC/2014/PROT/0193), the Rwanda National Ethics Committee (24 October 2014, 346/RNEC/2014), and the Rwanda National Institute of Statistics (9 February 2015, 0082/2015/NISR) prior to study recruitment and data collection. As per Rwandan government requirements, study approval was renewed annually with the Rwanda National Ethics Committee (19 October 2015, 338/RNEC/2015; 21 October 2016, 883/RNEC/2016) and the Rwanda National Institute of Statistics (2 November 2015, 0794/2015/10/NISR; 27 October 2016, 0806/2016/10/NISR). The trial was retrospectively registered at clinicaltrials.gov on February 29, 2016 ({"type":"clinical-trial","attrs":{"text":"NCT02694627","term_id":"NCT02694627"}}NCT02694627) after study enrolment began in February 2015, but before collection of the 21-month follow-up data (reported here) or study completion. The delay in trial registration was due to the authors’ lack of awareness of this requirement for journal publication. We registered the study as soon as we were aware of this requirement. No major changes to the study protocol or study outcomes were made. The authors confirm that all related trials to this intervention were registered; there are no ongoing trials related to this study. At 21-month follow-up, 1123 men (94% of the sample) and 1162 women (97%) were surveyed. Respondent attrition was slightly higher for men in the intervention group compared to the control group (7.3 vs. 5.4%), and was essentially identical for women (3.1 vs. 3.0%) (Fig 1). Reasons for loss to follow-up were predominantly inability to find participants due to relocation and respondent unavailability. Men who dropped out were more likely to be out of work and looking for work at baseline compared to men who remained in the study. All available data were included in analyses. The Bandebereho couples’ intervention engaged men and their partners in participatory, small group sessions of critical reflection and dialogue. The Rwanda Men’s Resource Center (RWAMREC), a local Rwanda non-governmental organization, implemented the intervention as part of MenCare+, a four-country initiative to engage men in sexual, reproductive, and maternal health. The MenCare+ program was coordinated by Rutgers and Promundo, and financed by the Dutch Ministry of Foreign Affairs. In Rwanda, the MenCare+ program was known as Bandebereho, or “role model”, as it aimed to transform norms around masculinity by demonstrating positive models of fatherhood. The intervention used a structured 15-session curriculum adapted from Program P, an open source manual for engaging men in maternal and child health, created by Promundo, CulturaSalud, and REDMAS (2013) which includes a curriculum for fathers/couples, resources for designing health provider training and community campaigns [21]. Men participating in the Bandebereho intervention were invited to 15 sessions (maximum 45 hours) and their partners to 8 (maximum 24 hours). Sessions addressed: gender and power; fatherhood; couple communication and decision-making; IPV; caregiving; child development; and male engagement in reproductive and maternal health (See Table 1 for details on curriculum content by session.) Promundo and RWAMREC adapted the curriculum between May 2013 and January 2014, informed by formative research, and input from the Rwanda Ministry of Health, which approved the curriculum for implementation, and from community pilot implementations. The intervention draws on sociological theories of gender and masculinities that highlight how gender inequalities are reproduced–or transformed–through “everyday interactions in [the] home” [22,23]. The intervention creates a structured space for men and women to: 1) question and critically reflect on gender norms and how these shape their lives; 2) rehearse equitable and non-violent attitudes and behaviors in a comfortable space with supportive peers; and 3) internalize these new gender attitudes and behaviors, and apply them in their own lives and relationships. We hypothesize that becoming aware of inequalities, reflecting on the costs of rigid norms, and learning and practicing new skills (e.g. couple communication and joint decision-making) in a safe, non-judgmental peer environment, can lead to changes across a range of health and relationship behaviors. Community volunteers (local fathers) met with the same group of 12 men/couples on a weekly basis. The volunteers received a two-week training, material support, and refresher trainings from RWAMREC. Local nurses and police officers co-facilitated the sessions on pregnancy, family planning, and local laws, respectively. Sessions were conducted in local schools and administrative offices. A transportation stipend of 2000 Rwandan francs (about US$2.50) was provided to men/couples for each session attended. RWAMREC staff monitored implementation of the group sessions and mentored the facilitators. Three intervention cycles, each with 570–576 couples, were implemented between March 2014 and July 2015. This study assessed the third cycle, in which men attended on average 14.1 out of 15 sessions, and women 6.8 out of 8 sessions. The control group received no group intervention, though it did have access to community activities and campaigns related to the broader MenCare+ project. We assessed five sets of outcomes specifically targeted by the intervention, each captured through multiple variables: (1) reproductive and maternal health behaviors, including men’s participation in ANC visits; (2) women’s experiences of IPV; (3) use of physical punishment against children; (4) gendered division of childcare and household tasks; and (5) men’s dominance in household decision-making. Table 2 summarizes the key outcome measures. a MICS surveys can be accessed at http://mics.unicef.org/surveys We compared men’s characteristics at baseline using frequencies and descriptive statistics. To estimate the effects of the intervention on outcomes measured at 21-month follow-up, we conducted intention-to-treat analysis using regression models with normal, Bernoulli, and Poisson response distributions and identity, logistic, and log link functions. We used generalized estimating equations to fit the models, and used robust standard errors with clustering by facilitator for hypothesis testing and confidence interval construction. For each outcome we fit both unadjusted and adjusted models; the latter included controls for age, education, and baseline socio-economic status (defined as having basic needs met). All analyses were conducted using Stata/SE 14. In our presentation of results, we use standard abbreviations for statistical terminology, including: SD–standard deviation; CI–confidence interval; OR–odds ratio; and IRR–incidence rate ratio.