The final study sample is comprised of 381 057 women of reproductive age, although the population included in the analyses of each opportunity varies depending on the baseline population and the missing values. The missing values for each opportunity represent less than 5% of the observations of almost all country samples, except for Comoros and Sierra Leone in antenatal care visits and Zimbabwe in not having anaemia, that represent less than 15% of the observations, and Namibia in antenatal care visits that represent less than 25%.

The average HOIs and coverage rates (across countries, unweighted by population) of the opportunities included in the study are shown in Figure 2 and the numerical values can be found in Table S2 in Online Supplementary Document. The distance between them represents the penalty for inequality of opportunity (which is P = C × D, using equations (1) and (2) (Figure 1). The lowest average HOI corresponds to the maternity care package (26.08, 95% confidence interval (CI) = 19.28-32.88), followed by the met need for family planning (37.82%, 95% CI = 32.66-42.98) and currently attending school among older adolescent girls (39.46%, 95% CI = 33.55-45.36). Exclusive breastfeeding among children aged 0-6 months has the highest average coverage rate (80.14%, 95% CI = 76.03-84.25) and average HOI (76.67%, 95% CI = 72.26-81.08). The lowest D-indices (inequality) correspond to not having anaemia (3.40%, 95% CI = 2.45-4.35) and exclusive breastfeeding (4.54%, 95% CI = 3.77-5.31). Except for the last two indicators, on average, the opportunities are low and unequally distributed across groups of women with different circumstances.

Figure 2.  Average HOIs and coverage rates. Note: HOI – Human Opportunity Index, BMI – body mass index, HIV – human immunodeficiency virus.

The Shapley decomposition results (Figure 3, unweighted averages across countries) show that family wealth status, education and area of residence are the factors that contribute the most to inequality for the majority of the opportunities for women of reproductive age. Age is an important contributor to inequality in having the recommended BMI (ranked just after family wealth status), which is an indicator for overall health status of a woman. The findings show that older women tend to display inadequate BMIs, generally high. Marital status stands as an important contributor to inequalities (D-index) for specific opportunities, with married women having the advantage over unmarried women for some indicators (eg, prophylaxis during pregnancy) and vice versa for others (eg, not having anaemia or met need for family planning).

Figure 3.  Average circumstances’ contributions to the D-index by opportunity. Note: D-index – Dissimilarity Index, BMI – body mass index, HIV – human immunodeficiency virus.

Figure 3 shows as well that among older adolescent girls, marital status is the major source of inequalities. The contribution of marital status is particularly high for the indicator of having never been pregnant, accounting for two thirds of the D-index. After marital status, wealth, area, and occupation are important contributors to inequality of opportunity among adolescent girls. For school attendance, occupation is the second most important source of inequality, which reflects that girls who work outside home are less likely to be in school. The D-index values for all the opportunities can be found in Table S2 in Online Supplementary Document.

The relatively large association of marital status and adolescents’ access to opportunities might have the effect of muting the contributions of the other circumstances on the D-index. To investigate further the role of these other circumstances, the sample is split into “in union” adolescents – married, divorced, living with a partner, or widowed – and “never in union”. Family wealth is the most important contributor to inequality in all opportunities for adolescents, except for the case of those currently attending school among “never in union” adolescents, where occupation is the main contributor to inequalities (or to the D-index) (Figure 4). The HOIs and D-indices of the split analysis by marital status can be found in Figure S1 in Online Supplementary Document.

Figure 4.  Older adolescents’ opportunities by marital status: average circumstances’ contributions to the D-index. Note: D-index – Dissimilarity Index.

The weighted multi-country pooled analysis is consistent with the results shown in this article; coverage rates and HOIs are low while inequalities are high, and wealth, education and area of residence are the main contributors to inequalities [14]. The only fact that needs to be highlighted is the presence of religion as a relevant contributor to inequalities for some indicators (anaemia and malaria prophylaxis). This finding could suggest that religion is a confounding factor of the association between these indicators and geographical location.

This study offers a novel approach to understand inequality of opportunities in reproductive, maternal, newborn, child and adolescent health by simultaneously analysing all possible factors for which data are available (which includes wealth, education, place of residence, religion, marital status and age), that matter for inequality in access to services among different groups, and the relative contribution of each factor to these inequalities.

This study highlights the challenges that women in SSA still face when seeking reproductive and maternal health services. On average, the HOIs obtained in the analysis appear to be low while inequalities are high. Notably, coverage is lower, and inequalities higher, for those interventions that require higher provider-patient interaction (eg, antenatal care or delivery attended by skilled personnel) than for interventions that can be delivered through strategies outside the health system (eg, exclusive breastfeeding or HIV information and testing), thus highlighting the importance of strengthening health systems to ensure equitable access. The indicator for (not having) anaemia is notable for showing low coverage and low inequality, whereas low coverage in other cases is almost always accompanied by high inequality. It is important to remark that differences in anaemia incidence, as it happens for all health outcomes, could be due also to other factors than inequality. Thus incidence of anaemia among women in SSA is high on the average and tends to affect all socio-economic groups within countries at nearly similar rates. This might suggest that anaemia has less to do with differential access to health services than other systemic country or region level factors (including exposure to specific diseases), identifying which will require more in-depth research. In contrast, inequality is higher for the opportunity of having the recommended BMI, which suggests large differences in general health between different groups of women in SSA countries, where family wealth status and age of the woman seem to matter the most for these difference.

The findings suggest that wealth, educational level and area of residence (urban/rural) are the three main circumstances associated with inequality of access to health care by women. Importantly, marital status appears as the main contributor to inequality among older adolescent girls, and only once marital status is controlled for, wealth becomes most important factor in the generation of inequality, similar to what is observed for adult women. Notably, all of these socio-economic barriers are often interlinked – women in poor families are also likely to be less educated and more likely to be living in rural areas. But the fact that education and location contribute a lot to inequality even after the role of economic status (family wealth) is accounted for suggests that these circumstances are important in their own right, net of the effect of economic status.

In order to interpret all these results, a couple of important considerations should be taken into account. The relationships above presented should not be interpreted as causal, but rather as providing information about associations between individual and household circumstances and inequality; more context-specific research tracing causal relationships will thus be needed to identify the determinants of inequalities in order to design the most appropriate interventions to address the observed inequities. Regarding the methodology used in the analysis, the HOIs and the D-indices are always upper and lower boundaries respectively, as they depend on the set of circumstances included in the analysis. The results for D-index are thus best interpreted as the most conservative estimates of inequality between groups – and the associated HOIs as the most liberal estimates of inequality-discounted coverage – which will only increase if other important yet unobserved circumstances could be included [6,7].

The main strengths of the study are listed as follows: the large number of countries and observations included in the analysis provides strong statistical power to the study; the DHS samples used are representative of the populations they include; and the average results of the 29 countries included in the study allow for drawing policy implications that can be generalised for SSA.

The main limitation of the study is the fact that the opportunities analysed here are likely to be affected by factors we cannot control for such as individual efforts and decisions or biological factors in the case of health outcomes, and therefore may not represent “opportunities” in the strict economic sense. Specifically regarding individual decisions, we have argued that inequalities between groups with different circumstances are still important to quantify, since the circumstances that drive these inequalities also effectively constrain the choices available to a woman in SSA. The inequalities we observe, therefore, can be interpreted as reflecting the association of circumstances with both unequal (physical) access and constrained (behavioural) choices. While our analysis cannot distinguish between these two channels, examining their relative roles in linking circumstances to outcomes should be an important topic for future research, so that inequities in health service delivery can be addressed by policies that are appropriately designed to address the underlying causes.

Another limitation of the methodology used is the fact that it does not give information about the direction of inequality. The HOI and the D-index highlight the presence of inequalities, but their direction has to be analysed in more detail in further analyses. However, this fact does not undervalue this research since the quantification of inequalities by itself sheds light on the maternal and reproductive health aspects that deserve more attention.

Finally, by using the same set of circumstances for the analyses in all countries, the specific country estimates could not be as optimal and adjusted as desired since drivers of inequality are very site specific. However, to be able to perform a multi-country analysis, and to compute comparable HOIs for policymaking in the region, it was necessary to select a single set of circumstances. The authors considered that it reflects well the majority of the country specific circumstances that drive inequality across SSA.

Despite notable progress in the last decade, reproductive and maternal health opportunities for women and girls in SSA are scarce, with half of women and girls not receiving essential services. Accelerated progress towards the improvement of women’s and adolescents’ health and well-being can only be achieved by expanding coverage and reducing inequalities. In the current context of progressive universalization of basic health care in many contexts, the application of an equity principle that prioritizes the expansion of services among underserved or excluded populations can have important implications for policy choices.

Ensuring progressive and equitable expansion of health coverage should be the cornerstone of efforts as it is key to achieving the SDG3 health-related targets – including the ones addressing reproductive, maternal, newborn and child health, communicable diseases and UHC – and beyond, particularly SDG4 (Inclusive and equitable education), SDG5 (Gender equality) and SDG10 (Reduced inequalities). Research has a key role to play to further ascertain the levels and causes of inequalities, bridge the existing data gaps for specific subgroups of vulnerable women and girls, as well as for monitoring progress and accountability towards realizing the 2030 goals.