Our analysis of the coverage of DFPSm and institutional delivery across 91 LMICs revealed distinct age patterns for the two indicators. For family planning, the prevailing patterns were either an increase in coverage with older age or an inverse U-shaped pattern, suggesting that the need for modern contraceptives among currently married/cohabiting women is less satisfied among adolescents and also among older women. For institutional delivery, on the other hand, the prevailing patterns were either similar coverage across the age groups or a decrease in coverage with age. Hence, there is no indication that young women have lower coverage of institutional delivery. Even within the same country, the age patterns for DFPSm tended to be different from that observed for institutional delivery. Of the 87 countries with both indicators available, only 9 showed similar age pattern for both indicators.
In both prevailing patterns of DFPSm, adolescents had lower coverage than women in older age groups. Region of the world, country income and national DFPSm coverage were associated with age patterns. In countries with low or intermediate DFPSm coverage, inverse U-shaped patterns prevailed, whereas for high-coverage countries increases with age were as common as the inverse U-shaped pattern. Thus, it seems that as the national coverage increases, older women are reached and the inverse U-shaped pattern is gradually replaced by higher coverage among older women, and lower coverage among adolescents compared to the other age groups. It is interesting to note that – even at the 70% or more coverage levels found in the third tertile – similar coverage for all age groups is yet to be achieved in all the countries. South Asia and Latin America & Caribbean shown a markedly pattern of increase with age. In both regions, sterilization plays an important role in modern contraceptive use. It is observed a lower use of modern contraception where sterilization is not common, given the reduction of fertility and sexual activity.
Our findings for institutional delivery were markedly different to those observed for family planning. There were also marked differences by region and country income groups, that appeared to be mediated by national coverage levels: high-coverage countries tended to have similar coverage in all age groups, whereas most of the remaining countries tended to show declines with age. These findings suggest that increases in national coverage tend to be reflected in higher access to institutional delivery by older women, who would previously deliver at home. In fact, the top tertile of institutional delivery coverage is close to universal coverage, with more than 92% of the women being reached. These findings suggest that improvements in overall coverage of institutional delivery are less vulnerable to ongoing social inequities in health relative to that seen for national level improvements in family planning coverage. Most countries from East Asia & Pacific are in the first tertile of institutional delivery coverage and have lower levels of coverage among both older women and adolescents. In addition to the resistance of older women who had previously delivered at home, this pattern can be explained by the fact that most adolescent mothers in the region are from more vulnerable subgroups, such as those from lower socioeconomic levels and who live in rural areas.
In summary, both adolescent girls and – to a lesser extent – older women tended to show lower levels of demand for modern family planning satisfied in most countries, whereas for institutional delivery older women were more likely to be left behind. Pregnancy during adolescence has a major negative impact on women’s health and education [20], and children born to adolescent girls are at greater risk for neonatal mortality [21] so that family planning is particularly important in this age group. In many settings, child marriage is very common, and in this context, marriage often means motherhood as girls need to prove their fertility [22, 23]. We must address this and other social norms that may inhibit adolescents to reach health services in tandem with increasing access to sexual and reproductive health services and to contraception. In addition, women in late reproductive age, an often-forgotten population, are at increased risk for maternal mortality and those who become pregnant may be among the most vulnerable socially and biologically, so this is a concerning situation which must be tackled by public health initiatives [24, 25]. Strategies that have been shown to increase coverage include mobilization of political and community support, increased integration of services and, in some countries, public-private partnerships [26]. However, there is no place for one-size-fits-all approaches, as strategies for reaching universal coverage for these two interventions should be different, and, in addition, need to be tailored to specific country situations. Communication campaigns and other strategies to increase uptake across particular age groups may be needed for both interventions, but the content of the strategies will be very different for the two.
We did not have an a priori assumption that results would be similar for both indicators, for a number of reasons including the fact that national coverage levels tend to be higher for institutional delivery than for family planning, women’s desire for a given number of children, and that societal and religious norms seem more likely to affect contraception than delivery care. In addition, parity may be an important factor affecting coverage with both indicators, with younger women being less concerned than older women about unplanned pregnancies, and older women with previous healthy deliveries being less inclined to give birth in a hospital.
The present results contribute to the literature on intervention coverage by woman’s age, which has so far being mostly focused on differences among adolescents and older women, with the latter including ages 20–49 years [6,7,8]. In this paper, we used seven five-year age groups, which allowed us to find five different patterns of coverage by age. By doing so, we increased the granularity of age inequalities, and thus highlighted the vulnerability of older women (40 years or more), a group whose higher risk is not evident when the broad age group of 20–49 years is treated as a single category.
Evidence on the inverse association between age at first birth and institutional delivery had already documented for several low- and middle-income countries [27]. Globally, there is increasing emphasis on adolescents and even very young adolescents (10–14 years) in terms of comprehensive sexual education [28] and access to services, yet older women may have had little opportunity to receive this information when they were young. In perimenopause, there may be confusion regarding fertility, and this can compromise women’s health seeking for contraceptive use [24, 25, 29]. There may also be less investment in women in late reproductive age giving birth due to assumptions of they are already aware of the risks of late childbearing. Another possible explanation is that women in this age range are often multiparous, and may have had uneventful prior deliveries at home, and thus do not see the need for seeking care in an institution. In particular, recognition of increased risk for maternal mortality among older women giving birth, even when compared to adolescents, seems to be inadequately recognized. Poorest and highest fertility countries are most affected, but also poorest and rural women are likely the most affected within these countries [30].
Our analyses have limitations. Although data were analyzed for a large number of countries, these were not representative of all countries in the world: data were available for 82% of all low-income countries, 72% of lower-middle and 45% of upper-middle income countries. In addition, most surveys were carried out before 2015 in order to provide data for assessing success in terms of the Millennium Development Goals. Age heaping might be an issue, but we carried out sensitivity analyses with different age categories, which showed similar patterns in the extreme groups. It is also important to note that, according to the sampling methodology of the surveys, some groups of women were excluded. For institutional delivery analyses, all women with recent deliveries, regardless of marital status, except in a few countries (Afghanistan, Bangladesh, Egypt, Jordan and Pakistan) where the sample was restricted to married women. For family planning, we opted to include only women who were married or in union because 18 countries only had information for this group; for the 73 countries that also had information on unmarried women, only 11 showed a difference greater than 3% points between DFPSm coverage among women who were married or in union, and coverage among all sexually active women (results not shown). A final limitation is reliance on visual inspection of age patterns to derive a typology. This decision was taken after extensive attempts to use statistical approaches to identify such patterns; due to the large sample sizes in some surveys, statistical tests showed significant departures from linearity or from homogeneity, even when visual inspection showed monotonic increases or declines with age, or very similar coverage in all age groups. To address this limitation, we peer reviewed each pattern and the few inconsistencies in the typologies were reviewed by a third author. We also present in appendix (Supplementary Figures 1–16) the coverage by age in each country for the two coverage indicators, so that readers can assess the patterns. Future studies should attempt to deal with the above-mentioned limitations and seek for a better understand on locally-relevant mechanisms contributing to these age-related inequalities.
Parity was not taken into account in the present set of descriptive analyses, even though it likely affects coverage of DFPSm and institutional delivery [31, 32], and in addition is associated with socioeconomic position, education levels and other cultural characteristics. Further research is needed to elucidate the role of parity in the determinations of age patterns in coverage with RMNCH interventions.
Among the strengths of our analyses, these represent the most comprehensive overview so far on how coverage with two key interventions, both of which represent SDG goals or proxies for such goals, is affected by woman’s ages in a large number of LMICs. The standardized nature of the survey questionnaires, consistent indicator definitions and analytical methods also support the robustness of the present findings. Future studies should also investigate changes over time in age patterns, and expand the set of analyses to a broader range of reproductive and maternal health indicators. Characteristics of each health intervention and how they are delivered at population level may lead to different age patterns in coverage. Age-related inequalities should be routinely assessed, as is already the case for inequalities according to family wealth, education or sex.