Menu
Menu
Menu
Menu
Background: Ensuring women have information, support and access to family planning (FP) services will allow women to exercise their reproductive autonomy and reduce maternal mortality, which remains high in countries such as Madagascar. Research shows that women’s social networks – their ties with partners, family members, friends, and providers – affect their contraceptive use. Few studies have considered the role of men’s social networks on women’s contraceptive use. Insofar as women’s contraceptive use may be influenced by their male partners, women’s contraceptive use may also be affected by their partner’s social networks. Men may differ by the types of ties they rely on for information and advice about FP. It is unknown whether differences in the composition of men’s FP networks matter for couples’ contraceptive use. This study assessed the association between men’s FP networks and couples’ contraceptive use. Methods: This egocentric network study was conducted among married/partnered men (n = 178) in rural Madagascar. Study participants listed who they relied on for FP information and advice, including health providers and social ties. They provided ties’ gender, age, relationship, and perceived support of contraceptive use. The primary outcome was couples’ contraceptive use, and explanatory variables included FP networks and their composition (no FP network, social-only network, provider-only network, and mixed network of social and provider ties). Analyses used generalized linear models specifying a Poisson distribution, with covariate adjustment and cluster robust standard errors. Results: Men who had FP networks were 1.9 times more likely to use modern contraception as a couple compared to men with no FP network (95% confidence interval [CI] 1.64–2.52; p ≤ 0.001). Compared to men with no FP network, men were more likely to use modern contraception if they had a social-only network, relative risk (RR) = 2.10 (95% CI, 1.65–2.68; p ≤ 0.001); a provider-only network, RR = 1.80 (95% CI, 1.54–2.11; p ≤ 0.001); or a mixed network, RR = 2.35 (95% CI, 1.97–2.80; p ≤ 0.001). Conclusions: Whether men have a FP network, be it provider or social ties, distinguishes if couples are using contraception. Interventions should focus on reaching men not only through providers but also through their social ties to foster communication and support for contraceptive use.
The study sample was recruited from 27 villages randomly selected from the 80 villages within a 5 km radius of the semi-urban town of Ambalavao in Fianarantsoa Province, Madagascar. Data collectors enumerated all households by village and then used random number draws to identify households for study participation. The total sample chosen from each village was proportionate to the number of households in the village, with a minimum of 5 households per village. Within each selected household, the data collector enumerated all men ages 18 years and older who were willing to answer questions about FP, and then randomly selected one man. Surveys were conducted in the local language (Malagasy) by male data collectors in order to facilitate study recruitment and elicit candid answers about reproductive health topics. Prior to data collection, written informed consent was obtained, with verbal informed consent permitted for participants who could not read and write. All men who were randomly chosen within a household consented to participate, yielding a study sample of 208 men. Men who were not in partnership (that is, who were either divorced, widowed, or single; n = 17) and men whose partners were pregnant (n = 13) were excluded from this study. Thus, the final analytical sample included 178 study participants. Participants were provided with a small gift for their time and participation in the form of a lamba, a cloth used locally for multiple purposes. The study received approval from the National Ethics Committee in Madagascar and the University of California San Francisco Institutional Review Board. Cross-sectional data were collected from June to July 2019. The primary outcome was a binary measure of whether the participant reported that their partner is currently using modern contraception. We hereafter interpret this variable as the couple using modern contraception though it is understood that the measure is based on the male partner’s perception of his partner’s contraceptive use. We defined modern contraception to include injectables, the pill (oral contraceptives), implants, intra-uterine devices (IUD) and external condoms (i.e., male condoms). Participants were also asked if they knew anyone else using these modern methods, which was recorded as a dichotomous yes or no. Male and female sterilization were excluded as forms of modern contraception because of data validity issues. In our data, 15% of men reported using sterilization. For reference, in national surveys among reproductive age women in Madagascar, 4% report using female sterilization and 0.3% report that men were sterilized [45]. This discrepancy was likely due to a misunderstanding of the question, so we excluded reports of sterilization as a form of modern contraception. Participants were also asked about their desire for pregnancy as a couple in the last 4 weeks (with answer categories for wanted to become pregnant, did not want to, or was not certain). Participants also reported who, within the couple, made decisions about contraceptive use (participant, his partner, or together as a couple). The number of births that the female partner had experienced was also elicited from participants. The number of births was used to reflect fertility, given high infant and child mortality in Madagascar [46]. Other data recorded included age, whether the participant was married and/or living with the partner or has partner but not living together, number of household members, whether attended school and highest grade attained, current occupation, estimated household earnings in the past month, and whether the household had electricity. Missingness for these variables is < 3% except for pregnancy desires (12%), contraceptive use decision-making (20%), and number of births (11%). To collect network data, we adapted the people network survey developed by Brunson (2013) [47, 48] to ask about FP. We asked participants (egos) to list the individuals (alters) from whom they obtain information, advice, and/or guidance about FP.1 The question was worded as follows: “The purpose of these next questions is to gather accurate information about your people network; in other words, the people from whom you obtain information, advice and/or guidance about your use of family planning. Please take a moment to think about who these people might be. Please provide the first names of the people who have influenced your use of family planning.” Interviewers then prompted participants to review if they had forgotten any alters, prompting them to consider whether they wished to include their partner(s), siblings, parents, CHWs, and other health providers to ensure completeness. The maximum number of alters elicited from any study participant was 3. For each alter that was listed, name interpreter questions elicited alter age and gender, nature of the ego’s relationship with that alter, and the ego’s perception of that alter’s support of modern contraceptive use (categorized as supportive of modern contraceptive use vs. not supportive). Several network measures were used as explanatory variables in separate analyses. The first explanatory variable was a dichotomous variable for whether the study participant named at least one individual in his FP network. The second was a categorical variable that included four mutually exclusive categories to represent network composition: 1) having no FP ties and thus no FP network, 2) all FP ties were composed of social relations only, including partner, other family members, and/or friends (hereafter referred to as social-only FP network); 3) all FP ties were composed of providers only, including CHWs, health educators, nurses, mid-wives or doctors (hereafter referred to as provider-only FP network); and 4) FP ties were composed of a combination of social and provider ties (hereafter referred to as a mixed FP network). The size of the FP network was categorized as 0 alters, 1 alter, or ≥ 2 alters. There were no differences in contraceptive use or network composition by missingness of baseline covariates. If observations had missing covariates, then they were not included in the analysis. However, missingness of covariates was significant at p = 0.051 level, for differences in FP network composition. Men with no FP network were more likely to have missing covariates and therefore were not included in the analysis: 26% of men with missing covariates had no FP network compared to 16% of men with non-missing covariates. We fitted a generalized linear model specifying a Poisson distribution with robust standard errors, interpreting the estimated incidence rate ratios as relative risk ratios (RR) [49]. In all analyses, the individual-level characteristics included age, number of births by partner, living together status, primary school completion, and household earnings. To avoid model over-specification, other individual characteristics were not included because of limited variation (electricity, occupation as farmer). Household size was included instead of number of births because of higher missingness rates (results are similar when births is included instead). Models also included fixed effects by sub-districts (the Malagasy Fokontany, which represents groups of villages). Analyses were conducted at the ego-level using robust standard errors clustered by sub-district. Additional analyses were conducted at the alter-level only using data from men with a FP network to examine the association between alter characteristics (gender, age, perceived support for contraceptive use, and type of relationship (e.g. CHW) and couple’s use of contraception, with robust standard errors clustered by sub-district. Alternative specifications included adjusting for: 1) couples’ pregnancy desire and 2) couples’ contraceptive decision-making. These covariates were not included in the main model because of higher missingness rates. In other specifications, we explored: 3) including both network composition and knowledge of contraceptive users variables, and 4) excluding partners as sources of advice in the network since the outcome simultaneously measures partners’ use of contraception. We also conducted an e-value sensitivity analysis to estimate how large the relative risk ratio of an unobserved confounder would need to be associated with both contraceptive use and men’s social networks in order to completely explain the associations in the study [50, 51].
N/A
