We found that a number of health outcomes were significantly associated with gender variables after adjusting for major covariates and the false discovery rate. Within this section we discuss the main findings across the four gender domains, the importance of women’s autonomy and empowerment, potential reasons for discrepancy between the women’s and men’s surveys, and how gender analysis findings from coverage surveys can be used to improve maternal health interventions.
Decision-making and autonomy
Variables in the decision-making domain had the most associations, both within individual and health system access outcomes. Making one’s own decisions about health, major household purchases, and daily purchases were all associated with contraception use, and a number of other health system access-related outcomes, including going to the health facility for one’s own care (as opposed to only care for family). These results suggest that women’s autonomy and independent decision-making power may significantly promote positive health behaviors.
Social norms and beliefs
Gender norms underpin many different aspects of health-seeking behaviors and health services delivery. Not only are they difficult to measure, but due to their pervasive nature, it is often difficult to establish direct causal relationships between norms and health and health systems factors and as a result, likely reflect indirect impact on health outcomes. Harmful gender norms were significantly associated with negative health outcomes, underscoring their importance in shaping health experiences and outcomes .
Access to resources
This domain was associated with access to a mobile phone and mobile banking. Access to these structures may allow women to hold money for themselves and use it for personal purchases, such as medications or fees related to childbirth or ANC. Previous studies of mobile cash transfer programs among adolescent girls and young women found increased access to resources through “survival spending,” such as for basic household necessities including food and shelter, and “self-care spending,” such as clothing or sanitary items for personal use were two major spending categories [42, 43].
Also, mobile phone access offers an expanded social network and may form a communication strategy for health care workers and facilities with women about their health . For example, many intervention studies have found that text message reminders, phone calls, and mobile payment incentivization have increased retention in health services [44,45,46,47]. Further research should investigate the potential role of different types of resource access and where gendered access to resources play the most role in impacting health outcomes.
Labor-sharing and roles
We limited our gender indicator variables to only validated metrics in a coverage survey. While labor-sharing and gender roles have been studied extensively, one of the best ways to assess the division of labor includes direct observation, which was not possible in the context of a coverage survey. Therefore, while we found few associations with labor-sharing, this area needs further research. The specific indicators we used may not correctly measure the construct of labor distribution in this context, or this may reflect a true null effect, where the other domains of gender play a larger role in health services access.
The importance of women’s autonomy and empowerment
For analytical purposes in our inferential analyses with the women’s survey, we explored gender power relations across each of the four gender domains (decision-making, roles, norms, and access to resources) separately. However, it is important to recognize that these domains intersect and relate to one another. Gender norms, for example, influence roles and behaviors, access to resources, and decision-making power; while decision-making power influences access to resources and roles and behaviors. Due to the complexity and context-specificity of gender power relations, it is difficult to explore the interplay between the different domains quantitatively. Despite this, associations were found between the different gender domains and health and health systems outcomes, highlighting the role and importance of gender power relations in shaping health.
These findings suggest that women’s actual decision-making power in particular, may have a direct relationship on their health. Previous studies of gender dynamics and health have focused on women’s empowerment [48,49,50,51], using a range of metrics, including ecological measures of women’s involvement in parliament and economic positions , DHS survey questions around decision-making [27, 53, 54], as well as emic definitions and context-specific scales developed through formative research . These studies have found associations between gender domains and health outcomes and highlight that gender-responsive and gender-transformative interventions can improve these outcomes [48, 51]. For example, one study from urban Nigeria found that women who had access to their own money had a 16% increase in the odds of using modern family planning methods . Our study compliments such findings, and expands the number of health outcomes beyond those previously associated with gender dynamics, but also to broader health outcomes, including those at least partially controlled by health facility resources.
Gendered discrepancies between men and women
In the paired analysis, there were a number of indicators without a statistically significant correlation between men and women within a household, suggesting noticeable discrepancies between responses. Most notably, women’s ability to make decisions about their own health was not significantly concordant between men’s and women’s surveys, with more women reporting they could make their own health decision and men reporting that woman could not. Women reporting that they could make decisions about their own health had significant associations with improved health outcomes (while men reporting that they could make their own health decisions did not show such associations), suggesting that women’s self-reported and own understanding of autonomy is most impactful.
Other areas with low concordance were between social norms and beliefs, such as a woman having the right to refuse sex, and whether the husband has the right to reprimand her when she refuses sex. Across all of these social norms, more women than men reported that men had the right to retaliate. Further work must be done to understand this potential effect better.
These discrepancies may be due to social acceptability bias, whereby men and women are more or less likely to endorse gender norms, based on their perception of the interviewer(s) and the most acceptable response . Such discrepancies may also be due to different perceptions of gender and gender equity between men and women. For example, men may not perceive women’s lack of decision-making power to be problematic due to gendered norms around men’s and women’s roles. Or they may think that the limited power that women do have is enough, so when asked if women are allowed to make decisions respond in the affirmative. Alternatively, they may think that women have a lot more power and autonomy than they do vis-à-vis the status quo. In our experience, we have found that those in positions of power are often blind to the status quo as their lived experiences are positively shaped by it, and that is their norm. In contrast, those who lack power are often more aware of it and its consequences.
Using gender analysis results within coverage surveys to improve maternal health interventions
We found that health surveys do not often include multiple gender variables [5, 57, 58]. This is likely because it can be difficult to see how gender power relations are directly related to health outcomes. For example, a question about whether a woman has access to resources does not in itself directly relate to health outcomes – further analysis is required to connect these associations. This is because gender inequity is a social determinant of health, a root cause of health inequity. Addressing root causes requires multi-sectoral collaboration and sustained effort over the long term.
We found that gender power relations are related to health outcomes. The purpose of a coverage survey is to explore whether those in need of an intervention received it and to identify potential barriers to accessing these interventions. Understanding how gender affects health outcomes allows us to better understand how gender power relations affect an intervention’s ability to reach its target population. By doing so, we can modify our interventions according to the context to ensure that our interventions are gender accommodative (i.e., they take into account gendered considerations or barriers), or transformative (i.e., they challenge or change harmful gender norms, roles, and relations) . By not considering the gendered context in which interventions are implemented, we ignore the inequities that shape people’s lived experiences and reality and potentially perpetuate and reinforce harmful gender norms, roles, and relations.
According to the findings presented here, sexual, reproductive, and maternal health interventions should seek ways to promote women’s autonomy and decision-making power if they are to improve health outcomes. Increasing access to resources, such as a mobile phone and mobile banking, can also help increase women’s autonomy and improve health outcomes. However, a gender-sensitive implementation is extremely important so as to ensure that interventions do not have unintended consequences which further marginalize women . Likewise, interventions should consider whether and how males are engaged . Male engagement should align with what women want and not perpetuate male decision-making power .
The study had a number of limitations. Firstly, there were time and budget constraints given that gender dynamics were not priority indicators in the parent coverage survey. We were required to use an accessible number and depth of indicators and were restricted to only previously validated measures. This resulted in a somewhat lopsided assessment, with more questions feasible in the decision-making and social norms domains, and fewer possible questions in the access to resources and labor-sharing domains. For example, validated metrics of labor-sharing include time-use surveys [61,62,63,64]. This was not possible in the context of a cross-sectional survey, and therefore only 3 questions could be asked. Additionally, the majority of the interviewers who administered the parent survey were male, which held over to the gender-specific survey as well and may potentially skew responses. For instance, women may underreport interpersonal violence measures.
In addition, analyses were limited to descriptive statistics and basic inferences in order to rigorously explore one of the first comprehensive assessments of gender dynamics in a health coverage survey. In order for reproducibility of this methodology, we used a straightforward study design. This limited the ways we could aggregate our measures, and interpretability of each individual gender dynamic and health outcome remains a challenge. Future studies will make use of more complex methods to elucidate further interactions between gender variables and other sociodemographic factors, to impact health outcomes. Finally, while these are important associative findings, we used cross-sectional data from a household coverage survey and did not undertake a formal causal analysis. Future research is needed to establish temporality and potential causal associations.