Impact of the Momentum pilot project on male involvement in maternal health and newborn care in Kinshasa, Democratic Republic of the Congo: a quasi-experimental study

Background: The World Health Organization recommends that programs that seek to improve maternal and newborn health outcomes actively involve men during pregnancy, childbirth, and postpartum. However, there is little evidence on what strategies work to increase male knowledge of and involvement in antenatal and postnatal care. This study assessed the impact of the Momentum project on male involvement in maternal health and newborn care. The project involved monthly home visits to a cohort of first-time mothers aged 15–24 recruited at six-months gestation and group education sessions for their male partners using the Program P toolkit. Participants were followed-up for 16 months. Methods: The study used a quasi-experimental design with three intervention and three comparison health zones. Baseline data were collected in 2018 and endline data in 2020. Exploratory factor analysis was used to develop scales of male involvement. We measured the causal influence of Momentum using an intent-to-treat analysis at the health-zone level and a dose–response analysis at the individual level. We used random-effects probit and linear models for outcomes measured at baseline and endline, and treatment effects models with inverse-probability weighting for outcomes measured only at endline. The impact analysis involved 1,204 male partners of first-time mothers with live births. Results: Intervention health zones were associated with an 18.1 percentage point (95% CI [(10.6, 25.6]) increase in knowledge of three or more obstetric danger signs and a 13.9 percentage point (95% CI [6.3, 21.6]) increase in knowledge of newborn danger signs. Significant increases in male involvement in antenatal care (average treatment effect (ATE) = 0.728, 95% CI [0.445, 1.010]), birth planning (ATE = 0.407, 95% CI [0.157, 0.657]), and newborn care (ATE = 0.690, 95% CI [0.359, 1.021]) were found. The magnitude of Momentum’s impact increased steadily with the number of prenatal home visits and was statistically significant for all behavioral outcomes except shared decision making. Exposure to both home visits and group education sessions during the prenatal period had a significant impact on all outcomes relative to no exposure. Conclusions: The study demonstrated the effectiveness of Momentum on male involvement in maternal health and newborn care.

Momentum was a quasi-experimental community-based pilot project conducted in three intervention health zones (Kingasani, Lemba, and Matete) and three comparison health zones (Bumbu, Masina I, and Ndjili) in Kinshasa. Health zones were selected based on similar sociodemographic characteristics that could affect project outcomes (e.g., ethnicity, socioeconomic status, education, population density, access to services), the presence of a facility-based MCH project that was funded by the same donor as Momentum to facilitate referral linkages, and the absence of other projects targeting FTMs and implementing gender-transformative interventions for young parents. The project used 150 third-year nursing students (75 males and 75 females) from 11 nursing schools (Institut Technique Médical) to test a gender-integrated model of community-based delivery, focusing on FTMs age 15–24 and their male partners. The objectives of the project were to increase postpartum FP uptake and the adoption of MNH care seeking behaviors and household practices beneficial to mother and baby and promote more gender-equitable behaviors and attitudes among FTMs and their male partners. The Momentum intervention included: (1) home visits; (2) group education sessions; and (3) social and behavioral change communication (SBCC). Home visits included not only the monitoring of healthcare seeking and health behaviors of FTMs, but also the promotion of couple communication about postpartum FP, MNH and nutrition and infant care, and shared decision making. Monthly fathers group meetings structured around Program P (an approach for engaging men in fatherhood, caregiving, and MCH) provided opportunities for reflection and dialogue during group education sessions. Program P sessions aimed at promoting male engagement in FP/MNH and nutrition and transforming gender-related attitudes, and beliefs held by male partners [20]. The sessions included the following topics: father’s expectations; father’s impact/legacy; pregnancy; birth; FP; care giving; gender; non-violence; children’s needs and rights; and dimensions of care giving. Program P sessions targeted at male partners were complemented by Program M group education sessions for FTMs that promoted awareness of human, sexual, and reproductive rights. Program M is a model for promoting young women’s awareness about gender inequities, rights, and health, and for empowering them to make decisions about different aspects of their lives [21]. The project also engaged male partners directly through video storytelling on topics related to gender norms and decision making. These videos were used to facilitate group discussions on those topics. Community dialogue sessions in the SBCC component targeted mothers, mothers-in-law, and other household influencers and aimed to influence intrahousehold dynamics in support of FTMs’ agency and equitable gender relations. Street theatre was conducted to help create an enabling environment for broad-based gender and community norm change. The project was implemented in close collaboration with Action Santé et Développement, Johns Hopkins Center for Communication Programs (the SBCC component), the Direction de l’Enseignement des Sciences de Santé (Directorate of Health Sciences Education of the Ministry of Health), the Direction de Santé des Familles et des Groupes Spécifiques (Directorate of the Health of Families and Specific Groups), and the Ministère de Genre, Famille et Enfant (Ministry of Gender, Family, and Children). Project interventions were implemented for 16 months in the intervention health zones. The project trained instructors from 11 participating nursing schools, trainers from the Ministry of Health and Ministry of Gender, Family, and Children, and nursing students not only in WHO recommendations for postpartum FP, MCH and nutrition, but also in the Program M manual for FTMs and Program P manual for male partners, which had been adapted to the DRC context by the Ministry of Health, the Ministry of Gender, Family and Children, and project partners. The training focused on gender-sensitive community-based service provision that promoted egalitarian decision-making between couples, respect for women’s rights and autonomy, and the positive role that male partners could play as partners and fathers. Representatives from both Ministries as well as nursing school instructors and project partners supervised students during monthly home visits and group education sessions. The sample size was powered to detect a minimum absolute difference of 10–15 percentage-points change in key behavioral indicators with a 99% confidence interval (margin of error = 0.01) and 99% power, assuming an attrition rate of 25%. This minimum magnitude of change in behavioral indicators was selected for practical reasons as resources for data collection would have been insufficient to measure smaller changes with adequate precision. As the project was required to report on 15 behavioral indicators related to postpartum FP, MNH and nutrition, and gender equity outcomes among FTMs and their male partners, we selected the percent of newborns to FTMs aged 15–24 years who received a postnatal care check within two days of birth, estimated at 6.5% nationwide among women younger than age 20 in the 2013–2014 DRC DHS, as the baseline value for sample size estimation. This indicator was selected because it had the lowest prevalence compared to other indicators of interest among FTMs. Based on these assumptions, and the likelihood of resorting to cluster sampling (with a design effect of 2.0) if social conditions in Kinshasa at the start of field implementation (which occurred in the four months preceding the December 2018 general election) did not permit a cohort follow-up study, the sample size at the project design stage was estimated at 1,213 FTMs aged 15–24 years and an equal number of male partners in the intervention health zones and 1,213 FTMs aged 15–24 years and an equal number of male partners in the comparison health zones [22]. Although we were able to implement a cohort study, we did not reduce the sample size, even though statistical formula suggested a smaller sample of FTMs and male partners [23], because we wanted to know whether the Momentum project was equally effective for FTMs age 15–19 and those age 20–24. Therefore, we had to estimate separate sample sizes for each age group to ensure that an adequate sample size was obtained, working within the confines of the resources available to our program evaluation efforts, and adding a little extra cushion for non-response. We considered that the sample size required to measure changes in indicators over time was larger than the sample size needed to measure an indicator at one point in time and that a slightly larger sample would enable us to detect changes in indicators if they occurred, even if we encountered both non-response and dropout. The questionnaire used for the male partner survey reflected population and health issues relevant to Momentum’s project objectives and results framework and was shaped by both Momentum’s formative (qualitative) research conducted in 2018 and input from various stakeholders from the Ministry of Health, Ministry of Gender, Family and Children, non-governmental organizations, and international donors working on FP and MCH in the DRC. The questionnaire covered a range of topics: (a) household characteristics; (b) respondent’s background; (c) reproductive history; (d) contraception and fertility desires; (e) ANC; (f) delivery and postnatal care; (g) fertility preferences; (h) gender-relations (roles, decision making, attitudes, perceived norms, and practices related to routine childcare activities; (i) intimate partner violence perpetration; and (j) exposure to the Momentum interventions. The questionnaire was translated from English into French. The baseline survey of male partners was conducted from September 5 to November 23, 2018, and the endline survey from May 25 to August 15, 2020. Survey instruments were pretested in Kinshasa before the start of data collection and data were collected via smartphones using the SurveyCTO mobile data collection application. Inclusion criteria for male partners’ enrollment in the study were: (1) willing and mentally competent to provide consent; (2) able to speak Lingala or French; (3) resides permanently (i.e., not visiting) in the intervention/comparison health zones; (4) husband/male partner of a recruited FTM who was approximately six-months pregnant at baseline; and (5) receipt of the pregnant FTM’s consent for her husband/male partner to be involved in the study. Written informed consent was provided by all survey participants. Interviewers gave a hard copy of the informed consent form to each participant, and then read the informed consent form out loud on the screen of the programmed smartphone. Once the male partner understood the voluntary nature of the study and agreed to participate, he signed the consent screen or checked the consent box on the interviewer’s smartphone which then unlocked the survey questionnaire. A total of 1,766 male partners were completely interviewed in the baseline survey, of whom 1,276 were completely interviewed in the endline survey (165 traveled or moved, 137 could not be located, 47 refused to participate in the endline survey, 131 were not at home, 5 had died, 3 postponed, and 2 interviews were partly completed). A unique quick response code assigned at baseline to the couple (FTM and male partner) permitted participants’ endline data to be linked to their baseline data. After matching, a total of 1,248 male partners (600 in intervention and 648 in comparison health zones) were retained, with the remaining 518 male partners (249 in intervention and 269 in comparison health zones) considered as lost to follow-up. Overall, there was a small insignificant difference in attrition between the two arms: 29.5% in the intervention health zones and 29.3% in the comparison health zones. The analysis presented here was based on 1,204 male partners who were completely interviewed in both the baseline and endline surveys, whose data could be linked to that of the FTMs, who had live-born babies, and who had no missing data on the variables analyzed in this study. None of the socioeconomic characteristics examined in the study differed significantly between the 44 male partners with missing data/non-live births and the 1,204 male partners with non-missing data and live births (see Table S1 in Additional file 1). The study was approved by the Tulane University Biomedical Institutional Review Board (2018–1028) and the University of Kinshasa School of Public Health Ethics Committee (ESP/CE/066/2018). Authorization to implement the Momentum pilot project was granted by the Secretary General of the Ministère de la Santé Publique (MS.1251/SG/PNSR/1358/LBE/2018) on June 11, 2018. There were four measures of intervention exposure (that is, treatment levels), the first of which was binary and indicated that the male partner resided in the intervention health zones versus the comparison health zones at baseline. The second variable measured the type of exposure to Momentum interventions that the male partner had when the FTM was pregnant: none (reference group), home visit only, group education only, and both home visits and group education. The third variable measured the total number of home visits that the male partner received when the FTM was pregnant: none (reference group), 1–2, or 3 or more. The fourth variable measured the total number of group education sessions that the male partner attended in the prenatal and postnatal periods: None (reference group), 1–2, or three or more. Control variables were measured at baseline, with two exceptions. Age was included in the analysis as reported. A household wealth index was created from housing characteristics and household possessions using principal components analysis. The index was divided into terciles (low (reference group), medium and high). We controlled for the male partner’s number of years of schooling, marital status (never married versus ever married/formally engaged (reference group)), parents’ education (i.e., whether both parents had secondary or higher schooling; no (reference group) versus yes), and whether the male partner (a) had always lived in the locality, (b) worked in the past 12 months, (c) watched television at least once a week, (d) was a first-time father, (e) had resided with his biological father up to the age of 15 (no versus yes; data not collected in the baseline survey), and (f) reported that his biological father or father figure was very involved in raising him up until age 15 (no versus yes; data not collected in the baseline survey). We controlled for the male partner’s satisfaction with his relationship with the FTM, using the Relationship Assessment Scale [25], a seven-item additive index that ranged from 7 to 35 (alpha = 0.845; KMO = 0.875). Finally, we included a control for the male partner’s perceived power in the relationship, using the power subscale of the Gender Relations Scale [26]. The power subscale comprised seven items (e.g., my partner has more say than I do about important decisions that affect us; my partner dictates who I spend my time with, etc.). For each item, responses reflecting the most power in the relationship (after reverse coding as appropriate) were assigned a value of 1 (i.e., “totally agree”) and other responses (i.e., “partially agree” and “disagree”) were assigned 0. Thereafter, items were summed to create a composite score ranging from 0 to 7, with an alpha of 0.523. Percentages and means were calculated to summarize the data. Chi-square tests were used to examine the significance of differences in sociodemographic characteristics between male partners in the intervention health zones and those in the comparison health zones. To determine whether there was a statistically significant difference in the mean of a given outcome variable at endline compared to baseline, we used McNemar’s Test for paired samples (i.e., repeated measures from the sample group) with binary outcomes (i.e., knowledge of obstetric and newborn dangers signs) and the Paired Samples T-test for continuous outcomes that were normally distributed and had a similar spread between the two groups (i.e., the GEM scale). The Mann–Whitney U Test was used to compare the differences between the intervention health zones and the comparison health zones in outcomes that were measured only at endline and not normally distributed (i.e., male involvement in ANC index, male involvement in birth planning index, male engagement in newborn care index, and MNH shared decision-making index). The analysis was conducted separately for adolescent/young male partners (age 15–24) and older male partners to determine if the interventions were equally beneficial (or not) in both age groups. For the Mann–Whitney U test, which was performed as a two-sided test, we reported p-values calculated by an exact randomization test (available when the number of observations is less than or equal to 1,000) for each age group. To measure the causal effect of the Momentum interventions on our outcomes of interest, we first conducted an intent-to-treat analysis, whereby all male partners were analyzed according to the FTM’s health zone of residence at baseline, regardless of whether the male partner received any interventions. For panel data, knowledge of danger signs and the GEM scale, we fitted random-effects probit and linear models, respectively, and conducted pairwise comparisons of average marginal effects. Each regression model controlled for baseline values of age, household wealth, education, marital status, parents’ education, lifetime residence in the area, weekly exposure to television, employment, perceived power in the relationship, being a first-time father, and relationship satisfaction. The impact of Momentum on a given outcome was expressed as the average treatment effect (ATE), which was estimated as the difference in the predicted outcome probability or linear-form outcome prediction between the intervention and comparison health zones after the project was implemented, considering the already-existing differences (at baseline) between the intervention and comparison health zones. For outcomes that were measured only at endline, we estimated project impact using treatment effects models with inverse-probability-weighting (IPW). We modeled our binary treatment variable, residence in intervention health zones versus comparison health zones, as a logistic function and our multivalued treatment variables (i.e., type of exposure to Momentum interventions when the FTM was pregnant, number of prenatal home visits, and number of group education sessions) as a multinomial logit function. As we did not randomly assign who would receive the Momentum interventions and who would not, treatment could be related to covariates that also affected our outcomes. Therefore, all estimates were adjusted for the following covariates: age, household wealth, education, marital status, parents’ education, lifetime residence in the area, weekly exposure to television, employment, perceived power in the relationship, being a first-time father, relationship satisfaction, co-residence with the biological father up to age 15, and high level of involvement of the biological father or father figure in raising the male partner up to age 15. The IPW estimator is based on three assumptions: (a) conditional independence of the treatment, which means that variables that affect both treatment level and outcomes are observable; (b) overlap, which ensures that data are available on each male partner in each treatment level; and (c) independent observations, which imply that the outcome and treatment for an individual male partner has no effect on the outcome and treatment for another male partner [27]. Our data met all three assumptions. For example, visual inspection of plots of the estimated densities of the probability of getting each treatment or exposure level (see Figure S1 and S2 in Additional file 1) showed that the estimated densities had most of their respective masses in areas in which they overlapped each other, and not around 0 or 1. Therefore, there was no evidence that the overlap assumption was violated. We also conducted tests after estimation of ATEs to check whether our covariates were balanced over treatment or exposure levels and obtained a p-value of 0.873, which signified that our treatment model balanced the covariates. The variance inflation factor was 1.23 and suggested that multicollinearity was not of concern. When examining the impact of the number of prenatal home visits on our outcomes of interest, we included a binary variable measuring participation in group education sessions while the FTM (i.e., the female partner) was pregnant among the covariates in the outcome model. Similarly, we included receipt of any prenatal or postnatal home visit from a Momentum nursing student among the covariates when estimating the impact of the number of group education sessions on the outcomes of interest. For the observational data (i.e., outcomes measured only at endline), the ATE measures the differences in average outcomes between male partners in the intervention health zones and male partners in the comparison health zones, after controlling for other factors. A positive ATE meant that the Momentum interventions increased the average predicted outcome while a negative ATE suggested that the Momentum intervention decreased the average predicted outcome. The impact analysis was conducted in Stata version 17 [27].

• gage, anastasia j.
• wood, francine e.
• woo, madeline
• gay, rianne