Early adolescent childbearing in low- and middle-income countries: associations with income inequity, human development and gender equality

Objective. Reducing unwanted adolescent childbearing is a global priority. Little is known about how national-level economic and human development indicators relate to early adolescent childbearing. This ecological study evaluates associations of Gross Domestic Product (GDP), GINI index, Human Development Index (HDI) and Gender-related Development Index (GDI; i.e. the HDI adjusted for gender disparities) with early adolescent childbearing in 27 low- and middle-income countries (LMICs) across three time periods.

Methods. Among women ages 18–24, prevalence estimates for early birth (<16 years) were calculated by nation, and weighted linear regressions evaluated associations between national indicators and early childbearing. To examine temporal trends, analyses were stratified by year groupings.

Findings. Early adolescent childbearing declined over time, with the greatest change observed in Bangladesh (31.49% in 1996/7 to 19.69% in 2011). In adjusted models, GDI was negatively associated with early childbearing, i.e. early childbearing prevalence decreased as GDI increased. In the most recent time period, relative to the lowest GDI group, the average prevalence of early childbearing was significantly lower in the middle (-12.40, P < 0.00) and upper (-10.96, P = 0.03) tertiles after adjustment for the other indicators. These other indicators showed no consistent association with early childbearing.

Conclusion. As national-level GDI increased, early adolescent childbearing declined. The GDI, which reflects human development adjusted for gender disparities in educational and economic prospects, was more consistently related to early adolescent childbearing than the absolute development prospects as given by the HDI. While creating gender equality is an important goal in and of itself, the findings emphasize the potential for improved national-level gender equitable development as a means to improve adolescents’ sexual and reproductive health.

Adolescent childbearing, human development, development, gender equality, LMIC

Key Messages

• At a national level, early adolescent childbearing prevalence declined as National-level Gender Development Index values rose

• Findings emphasize the potential role of gender equitable development in adolescent sexual and reproductive health

Introduction

Despite recent declines, the global burden of adolescent childbearing (from 10 to 19 years of age), continues (United Nations 2013; Hindin et al. 2016). Almost one in five women (19%) ages 20–24 report a live birth before age 18 (Magadi et al. 2007; Kurth et al. 2010; UNFPA 2013; United Nations 2013). This experience can impact mortality, educational attainment, earning potential, and overall well-being (WHO 2011; McQuestion 2012; UNFPA 2013). Childbearing in the earliest phases of adolescence, i.e. prior to age 16, can be particularly problematic with regard to poor birth outcomes (Phipps and Sowers 2012) yet it persists with an estimated 2 million births annually to girls under the age of 15 (UNFPA 2013). Since the 1994 International Conference on Population and Development (ICPD) (United Nations 2013), reducing adolescent childbearing has been a global priority. Two decades later, it remains on the global agenda United Nations 2013), with the adolescent birth rate (number of births per 1000 women ages 15–19), a global indicator of Millennium Development Goal 5 - Improve Maternal Health (United Nations 2008), and as an indicator in the Sustainable Development Goals (number of births per 1000 women ages 10–14 and ages 15–19) Beyond its impact on health outcomes, early adolescent childbearing may also adversely impact national-level productivity. High birth rates among adolescents can threaten the achievement of the “demographic dividend” (Gribble and Bremner 2012), i.e. the economic growth that occurs when birth rates decline and the working-age population increases such that the economically dependent proportion of the population shrinks (Bloom 2003). Delayed childbearing may facilitate economic gains and women’s participation in education and the labour force (Eloundou-Enyegue and Stokes 2004; McQuestion 2012). In the context of rich economic, educational, social, and civic participation opportunities, women may postpone family formation in favour of other aspirations (United Nations 2012).

At the national level, countries have realized the importance of young people as the future engine of growth and development. The proportion of governments with an adolescent reproductive health policy has increased from 60% in 1996 to 90% in 2013 (United Nations 2014). Beyond health-related policy, social and economic development determinants spanning economic status, gender equality, income inequality, and human development at the national level can shape adolescent childbearing.

The following four indicators that span economic and social development are useful to consider: the Gross Domestic Product (GDP), the GINI index, the Human Development Index (HDI) and the Gender -related Development Index (GDI). The GDP represents the monetary value of all goods and services produced on a national basis. The GINI index measures the extent of equity in income distribution; a Lorenz curve plots the percentages of total income received against the total recipients, and the GINI index measures the area between the curve and the hypothetical line of absolute equality. GINI index scores range from 0, representing perfect equality, to 100, which represents perfect inequality (World Bank 2014). US-based research links both poverty and income inequality with adolescent birthrates, particularly for younger teens (Gold et al. 2001). The Human Development Index (HDI) emerged in 1990 to reflect the national capacity for socioeconomic development, spanning life expectancy, adult literacy, and purchasing power (UNDP 2014). HDI is associated with both infant and maternal mortality rates (Lee 1997). Associations of low maternal education and severe maternal health outcomes are exacerbated by low HDI (Tuncalp 2014), highlighting the role of development context in maternal health. Independent of national economic wellbeing and capacity for development, women’s economic, educational, and social outlook can shape their health. The Gender-related Development Index (GDI), developed in 1995, adjusts the HDI for gender disparities in its components (UNDP 2014). The GDI, and the Gender Empowerment Measure (GEM) also released in 1995 from UNDP to approximate gender inequality in economic participation and decision-making, political participation, and power over resources, have been used to approximate national-level gender inequality ( Shah 2008; Kenyon and Buyze 2015) GEM has been linked with poor reproductive health outcomes spanning low birthweight, and infant and under 5 mortality, even after adjusting for GDP (Varkey et al. 2010), as well as other indicators such as male-female differences in smoking. We selected GDI over the GEM due to computation issues and limited data availability (Klasen 2006; Schuler 2006).

Despite the potentially reciprocal relationships between national social and economic development and adolescent childbearing, these associations have not been explored in low- and middle-income countries (LMICs). In the context of attention to adolescent health globally (WHO 2001; Blum 2012; Patton 2012), this ecological study draws on Demographic and Health Surveys (DHS) data to evaluate associations of national-level economic and social indicators (GDP, GINI, HDI, GDI) with early adolescent childbearing (<16 years) patterns over three time periods in 27 LMICs.

Methods

Data sources

National estimates for early adolescent childbearing were obtained through the Demographic and Health Surveys (DHS). The DHS are nationally representative household-based surveys conducted about every 5 years with women and men of childbearing age (typically 15-49 years) in over 90 LMICs. The DHS uses a stratified two-stage cluster sampling design (ICF International 2012; ICF International 2004-2012 ) response rates generally exceed 95% (ICF International 2012). Our outcome measure, early adolescent childbearing (i.e. before age 16), is based on the reports of women ages 18–24 at the time of the survey.

Our independent variables were chosen to reflect a range of economic and social development at the national level. GDP data were drawn from the World Bank’s World Development Indicators (1980–2013) (World Bank 2014) and were available for every survey year, for each nation. GINI data were drawn from the World Bank’s World Development Indicators (1980–2013) (World Bank 2014). The HDI and GDI were drawn from the UNDP Human Development Reports (UNDP 2014).

Sample

Our sample inclusion criteria for countries was having had a DHS survey for a minimum of three time points since 1990 with one in the past 5 years (n = 31). GINI coefficient data are released sporadically for every country over time; in those cases when the data for the survey year was unavailable, data from the closest available year is used, with preference given to the closest year after the survey year. For six countries (Ghana, Guinea, Kenya, Mozambique, Nepal, and Tanzania), a GINI coefficient was only available for two time points, and was imputed for the third survey year using regression mean imputation. While GDI and HDI data are released every year, data for every survey country is not always available for every year. In cases of where the year did not match, data from the closest available year is used. We further restricted to nations for which the national indicators of interest (GDP, GINI, GDI, and HDI) were available for a minimum of two of those time points; four countries (Benin, Haiti, Pakistan and Zimbabwe) were excluded based on availability of one or more national indicators for only one time point. The remaining 27 nations comprised a total of 1,131 603 women. A total of 275 513 (24.3%) female DHS participants were ages 18–24 and had complete data, and thus contributed data to this analysis.

Analysis

The prevalence of early adolescent childbearing was calculated for each nation, for each available survey year. To aid in interpretation, national-level indicators (GDP/capita, HDI score, GDI score, and GINI) were divided into tertiles, and a categorical variable (low, medium high) was created. We constructed weighted linear regression models inclusive of all four indicators to explore the associations between the indicators and the prevalence of early birth. The models were stratified by year groupings to explore temporal trends: Group 1 includes the oldest of the three surveys for every country and spans 1997–2002; Group 2 includes the second time point for every nation, 2003–2007; Group 3 includes the most recent survey for every nation and spans 2008–2013.

Results

Table 1 shows the prevalence of early childbearing in each nation by survey year for the three time points used in the analysis, rank ordered by the largest absolute change from the earliest time point to the most recent time point. Bangladesh exhibits the largest change in early adolescent childbearing, falling by nearly 12 percentage points between 1996/7 and 2011 (31.5–19.7%). During the 2008–2013 period, early adolescent birth is most prevalent in Niger at 19.9% (2012), followed by Bangladesh at 19.7% (2011).

Table 1.

Prevalence of early adolescent childbearing (before age 16) among women ages 18–24 by country and time period, ordered by absolute change over time periods

	Time period 1: 1997–2002		Time period 2: 2003–2007		Time period 3: 2008–2013
Country	Year	Early Childbearing %	Year	Early Childbearing %	Year	Early Childbearing %	Absolute Change Time 3 to Time 1
Bangladesh	1996/97	31.49	2007	22.68	2011	19.69	11.80
Cote d'Ivoire	1998/99	16.82	2005	12.22	2011/12	10.91	5.91
Indonesia	1997/97	6.64	2007	4.66	2012	1.36	5.28
Senegal	1997	8.44	2005	8.13	2012/13	4.94	3.50
Guinea	1999	21.31	2005	18.89	2012	17.98	3.34
Mozambique	1997	16.32	2003	16.36	2011	13.09	3.23
Burkina Faso	1999	9.00	2003	6.16	2010	6.15	2.86
Egypt	2000	5.00	2005	3.33	2008	2.58	2.42
Uganda	2000/01	13.23	2006	10.76	2011	10.85	2.39
Nigeria	1999	13.56	2003	12.02	2008	11.27	2.29
Niger	1998	17.80	2006	20.08	2012	19.91	2.11
Cameroon	1998	12.27	2004	12.26	2011	10.75	1.52
Colombia	2000	4.25	2005	5.86	2010	5.58	1.33
Nepal	2001	5.02	2006	3.65	2011	3.93	1.09
Bolivia	1998	3.93	2003	4.41	2008	4.94	1.01
Madagascar	1997	14.25	2003/04	10.46	2008/09	13.39	0.86
Cambodia	2000	1.46	2005	1.14	2010	0.62	0.84
Malawi	2000	8.10	2004	9.33	2010	8.94	0.83
Rwanda	2000	1.67	2005	0.93	2010	0.86	0.81
Ethiopia	2000	6.45	2005	9.86	2011	7.12	0.67
Ghana	1998	3.94	2003	2.98	2008	4.54	0.60
Tanzania	1999	6.67	2004/05	5.58	2010	6.10	0.57
Armenia	2000	0.36	2005	0.18	2010	0.00	0.36
Kenya	1998	7.33	2003	5.27	2008/09	7.08	0.25
Philippines	1998	1.46	2003	1.19	2008	1.29	0.18
Peru	2000	3.27	2004/06	3.34	2012	3.42	0.16
Jordan	2002	1.25	2007	1.46	2012	1.21	0.04

No statistically significant associations between the national-level indicators and early childbearing are observed in the first time period (1997–2002; Table 2). At both the second and third time periods (2003–2007 and 2008–2013, respectively), on average, the early childbearing prevalence is significantly lower in the middle and most equal tertiles of Gender Development Index (GDI), i.e. where scores indicate the greater equality in human development. Specifically, in the second time period (2003–2007), early childbearing was significantly lower in the medium tertile relative to the least equal tertile (β = −9.77, P = 0.001), as well as the most equal tertile relative to the least equal tertile (β = −14.56, P = 0.001). In the third and most recent time period (2008–2013), similar differences were identified with a significantly lower prevalence of early childbearing in the medium tertile relative to the least equal tertile (β = −12.40, P = 0.001), as well as in the most equal tertile relative to the least equal tertile (β = −10.96, P = 0.02).

Table 2.

Adjusted associations of GDP, GINI Index, Human Development Index and Gender Development Index with early adolescent childbearing by time point

	1997 – 2002 (n = 27 countries)			2003 – 2007 (n = 27 countries)			2008 – 2013 (n = 27 countries)
	β	P-value	95% CI	β	P-value	95% CI	β	p-value	95% CI
Gross Domestic Product (GDP)
Low (Reference)	–			–			–
Medium	3.92	0.30	−3.83, 11.67	−4.02	0.37	−13.21, 5.18	−4.44	0.13	−10.23, 1.36
High	7.65	0.06	−0.22, 15.52	−1.36	0.88	−19.74,17.01	−2.42	0.58	−11.52, 6.67
GINI Index (GINI)
Least equitable (Reference)	–			–			–
Medium	3.59	0.23	−2.43, 9.62	−1.75	0.56	−7.86, 4.37	−2.41	0.20	−6.21, 1.40
Most equitable	1.72	0.40	−2.48, 5.92	−1.39	0.31	−4.19, 1.41	−3.80	0.03	−7.25, −0.36
Human Development Index (HDI)
Low (Reference)	–			–			–
Medium	−4.25	0.30	−12.66, 4.17	10.16	0.06	−0.44, 20.77	9.05	0.01	2.42, 15.67
High	−6.24	0.14	−14.78, 2.30	7.16	0.41	−10.83, 25.16	4.93	0.38	−6.56, 16.42
Gender Development Index (GDI)
Least equal (Reference)	–			–			–
Medium	6.01	0.08	−0.78, 12.8	−9.77	0.001	−15.86, -3.69	−12.40	0.001	−17.81, −6.98
Most equal	−2.94	0.31	−8.88, 3.00	−14.56	0.001	−21.81, -7.31	−10.96	0.02	−20.24, −1.69
R2	0.58			0.54			0.64

*

All models are weighted.

Several additional patterns were identified in the third and most recent time period. As income inequity, assessed via the GINI index, increased, early childbearing decreased with significantly lower early adolescent childbearing prevalence identified in the most equitable tertile relative to the least equitable tertile (β = −3.8, P = 0.03; 2008–2013). For the Human Development Index (HDI), significantly higher prevalence of early childbearing was identified in the middle tertile relative to the lowest tertile (β = 9.05, P = 0.01; 2008–2013). No statistically significant associations between GDP and early childbearing are observed in any of the time points. The most variance in early birth rates was explained in the third and most recent time period (r²=0.58 for 1997–2002; r²=0.54 for 2003–2007; and r²=0.64 for 2008–2013).

Discussion

Findings from this ecological analysis suggest the relevance of some domains of national context in early adolescent childbearing. Of the social and economic development indicators explored, the Gender-related Development Index (GDI) was significantly associated with early adolescent childbearing in the two most recent of the three time periods examined, with lower prevalence of early childbearing on average as gender equality in development increases. Equitable income distribution as indicated by GINI index was also related to a lower prevalence of early adolescent childbearing in the most recent time period available (2008–2013), when comparing the most vs. the least equitable tertiles. By contrast, in the domain of human development, early adolescent childbearing was significantly higher in the middle Human Development Index (HDI) tertile relative to the lowest tertile. Our analysis demonstrates that educational and economic prospects adjusted for gender differences, indicated by the GDI, rather than the absolute levels as given by the HDI, relates more consistently to early adolescent childbearing. Over the time period studied, overall gains were observed in HDI and in equality as measured by GINI index in most nations, while GDI trends were more mixed with gains between time periods 1-2, and a mix of gains and losses from time period 2-3. Together, GDP, GDI, GINI and HDI explained 64% of the variation in early adolescent childbearing across LMICs in this study at the most recent time period available (2008–2013). Overall, these findings suggest the role of social contexts of equality, be it gender equality or equitable income distribution, in supporting conditions for reproductive health goals in the form of reducing early adolescent childbearing.

This analysis extends past research on social determinants and health, that to date have only minimally focused on adolescent health, and rarely beyond the US (Gold 2001; Eloundou-Enyegue and Stokes 2004; Hindin 2014). Current evidence linking gender equality and equitable income distribution with improvements in early adolescent childbearing demonstrates the relevance of the broader social context in shaping opportunities and sexual health even for adolescents who may have yet to complete their education or engage formally in the economy. The findings emphasize the potential role of gender equitable development in adolescent sexual and reproductive health. Young women in more equitable societies may delay childbearing or marriage in favour of education or workforce participation, given the opportunity. Societies with more equitable gender norms and gender-equitable distribution of human development gains may shape social norms that discourage early childbearing or permit greater access to contraception. Delayed childbearing may enable women to achieve economic and educational equality. The lack of association of early adolescent childbearing with GDP, and with GINI in the first and second time points, is striking. The differences observed across time periods may reflect shifts in the indicators themselves, or the ways in which they relate to early adolescent childbearing. Notably, GDI was not associated with early adolescent childbearing in the first time period, however, following relatively large changes in absolute GDI, this indicator was significantly associated with early adolescent childbearing in the second and third time periods. Future research to disentangle these multi-level complexities is needed to enable a more comprehensive understanding of national context and adolescent sexual and reproductive health.

There are several additional limitations that must be noted. Measurement of gender equality and empowerment has been long debated, and multiple indices have emerged, each with their own challenges in computation, conceptualization and interpretation (Klasen 2006; Schuler 2006). The GDI used in the current analysis faced early criticism for its life expectancy calculations, and more significantly, its earned income component. In its initial form used herein, the GDI was calculated to reflect an HDI adjusted for gender disparities in its components, not interpreted directly as a gender gap or indicator of inequality (Schuler 2006). In 2014, UNDP updated the GDI calculation to reflect the ratio of female HDI to male HDI (UNDP 2014), enabling more direct interpretation as a measure of gender equality for research and surveillance. Our analysis draws on GDI data prior to this change in 2014; nonetheless, we note this indicator evolution to aid in precision for future research. Our ecological analysis with three cross-sectional data points does not allow us to understand fully the mechanisms, causal pathways and mediators, and temporal links among the associations identified. By their nature, our indicators are national-level and thus do not reflect the local nuances and variation that may be relevant in understanding within-country variation. The analysis may also mask regional trends—with small sample sizes in some regions, we are unable to generate reliable regional estimates. Finally, the results reflect only those countries included in the analysis and cannot be generalized to high-income countries or countries omitted from our analyses.

Despite limitations, our analysis provides one of the few quantitative explorations in the role of gender equality in development in adolescent sexual and reproductive health (Kawachi 1999; Varkey 2010), and the first to focus specifically on early childbearing. The findings provide an empirical basis for further exploring the role of women’s empowerment and adolescent childbearing in a more comprehensive manner. In their 2011 Guidelines on Preventing Early Pregnancy and Poor Reproductive Outcomes Among Adolescents In Developing Countries (Guidelines) (WHO 2011), WHO notes that youth may be under pressure for early childbearing based on limited educational and economic prospects. Our analysis demonstrates that the gender disparity in those educational and economic prospects as indicated by the GDI, rather than the absolute prospects as given by the HDI, relates to adolescent childbearing. Our findings support their recommendations for maintaining and improving efforts to retain girls in school, and responds to the need for further research to determine the impact of socioeconomic improvements (WHO 2011) at a national, macro level.

The findings from this study have significant policy implications, particularly in terms of the potential women’s health gains of advancing gender equitable development. The historic Beijing Declaration and Platform for Action of 1995 set forth a global agenda for gender equality, and highlighted the role of gender equality in advancing women’s health. More recently, as we move beyond millennium development goals (MDGs) towards sustainable development goals (SDGs), the centrality of gender equality, women’s empowerment and rights have been increasingly recognized in achieving sustainable development (UN Women 2014). Current results add urgency to calls for advancing gender equitable development, and identifying meaningful, durable solutions for improvement at the national level. Future research should clarify feasible and sustainable methods for advancing gender equality in human development. Current findings demonstrate that the global gender equality agenda is aligned with reductions in early adolescent childbearing as a related and critical international health and development priority.

Acknowledgement

We thank Lale Say, MD, at the Department of Reproductive Health and Research at WHO for providing valuable input.

Funding

This study was conducted with financial support from the Department of Reproductive Health and Research including UNDP/UNFPA/UNICEF/WHO/World Bank Special Programme of Research, Development and Research Training in Human Reproduction (HRP), World Health Organization. Funders were involved in design and dissemination. The findings in this article represent the conclusions of the authors and not necessarily those of the funders.

Conflict of interest statement. None declared.