Study population

A total of 1,826,456 live births were registered during the study period. Of these, 116,468 (6.4%) cases were excluded because they occurred before 2015, 51,609 (3.0%) because they were registered after March 31 of the year following the birth, and 3,277 (0.2%) because of missing data on birth year. An additional 81 (0.004%) cases were excluded that corresponded to live births to women over age 49, and 25,314 (1.5%) because of missing data on the mother’s age. This left a total of 1,629,707 live births to mothers aged 10 to 49 for the estimation of the proportion of mothers under age 18 among all mothers of reproductive age.

To evaluate the relationship between low birth weight, preterm birth, and prenatal care according to the marital status of mothers who were minors, adolescents, and young adults, the following cases were excluded: 877,926 live births to mothers aged 25 and over; 7,558 cases of multiple births; 5,544 that had missing data on ethnicity or maternal marital status; 20,823 that had missing data on birth weight or gestational age; 860 cases in which birth weight exceeded four standard deviations from the sex- and gestational age-specific mean birth weight; and 4094 cases of live births to mothers who were divorced, separated, or widowed, or who had missing data on place of residence. Some cases met more than one of the exclusion criteria. Therefore, the study population for multivariate analysis of low birth weight and preterm births included 712,902 live births to mothers from 10 to 24 years of age. For the analysis of inadequate prenatal care, 710,723 cases were considered after excluding cases with no data on prenatal care.

Measures

Marital status of mothers was categorized as follows: legally married, stable union, single, widowed/separated/divorced, and unreported. For descriptive analyses intended to determine the prevalence of child marriage, live births to widowed/separated/divorced mothers were grouped with married mothers since they had been previously married. However, due to the low number of live births to widowed/separated/divorced women and those with missing data on marital status, these groups were not included as comparison groups in multivariate analyses.

Maternal age was grouped in the following categories: 10-15, 16-17, 18-19, 20-24, and 25-49 years of age. The 25-49 age group was not included in multivariate analyses given that the focus of this study was on minors, adolescents, and young adult mothers.

Low birth weight (<2,500 g) was subdivided into very low birth weight (<1,500 g) and moderately low birth weight (1,500-2,499 g), with normal birth weight of ≥2,500 g as the reference group. Preterm birth (<37 weeks of gestation) was subdivided into very preterm (24-31 weeks) and moderately preterm (32-36 weeks), with full-term births (37 or more weeks) as the reference group.

Prenatal care was categorized as adequate or inadequate, based on the Revised Graduated Prenatal Care Utilization Index (Revised-GINDEX),²⁸ which combines information on number of prenatal care visits and gestational age.

Statistical analysis

For descriptive analyses intended to estimate the prevalence of mothers under 18 years of age by marital status and sociodemographic characteristics, proportions expressed as percentages were calculated, with all live births to mothers aged 10-49 as the denominator.

In order to evaluate the associations between maternal marital status and low birth weight, preterm birth, and prenatal care, only mothers aged 10 to 24 years were considered. Multinomial logistic regression was used to model the two levels of low birth weight (very low and moderately low) and of preterm birth (very preterm and moderately preterm), whereas binomial logistic regression was used for inadequate prenatal care (yes vs. no). To assess the interaction between age groups and maternal marital status, an interaction term for these variables was included in the models. Statistical significance of the interaction was measured with a likelihood-ratio test, which compares the model “Ln(Odds) = marital status + age group + marital status * age group” with the simpler model “Ln(Odds) = marital status + age group” with the null hypothesis that the more complex model is not more informative than the simpler model. In order to show the joint influence of marital status and maternal age on the dependent variables, the results of each model are presented in two ways: i) taking single women aged 20-24 as the only reference category; and ii) taking single women as the reference category within each age group, equivalent to a stratified analysis.²⁹

Models were adjusted for sex of newborn (except for prenatal care), literacy (yes/no), primiparity (yes/no), immigrant status (yes/no), ethnic group (Indigenous/Afro-Ecuadorian/mestiza/white/other), area of residence (rural/urban), and region of residence (Coastal/Eastern/Mountain/Islander/other) of the mother.

In order to avoid unstable and imprecise estimations, we only reported associations based on ten or more cases in the subgroups defined by the intersection between marital status and maternal age group.

Data manipulation and statistical analysis were performed using SAS and R statistical programs. Graphics were created in Excel and Prisma. Ecuador’s live birth database is public domain and therefore no Ethical Review Board approval is needed for its use.

Main findings

In the first place, this study shows that there was a drastic reduction in legal marriages of mothers under age 18 over the course of the study period, along with a 64% reduction in the number of stable unions, while the proportion of live births to single mothers under age 18 remained relatively stable, constituting the majority. Differences were also observed with respect to ethnicity, with higher proportions of live births to mothers under age 18 in Indigenous and Afro-Ecuadorian populations.

Secondly, our results confirm the association between marital status and better health outcomes among mothers aged 20-24 that has been observed in other studies,^8,9,10,26 but they do not provide evidence of a marriage advantage (or that of stable unions) among minors with respect to low birth weight and preterm birth. Being in a stable union, however, was associated with higher utilization of prenatal care services in all age groups, especially among legally married women, even among minors. Despite these differences, inadequate prenatal care was generally high in all subgroups, ranging from 47% to 69%. As other studies have pointed out, prenatal health risks are higher with decreasing maternal age.^26,30,31

Interpretation

Although the proportion of live births to minors is very high in Ecuador compared to other countries in the Americas,³¹ a steady decrease could be observed during the study period, from 10% in 2015 to 9% in 2018, to less than 8% in 2020. According to the data analyzed in this study, this decline could be observed in women who were legally married (decreasing from 0.62% in 2015 to 0.01% in 2020) and in stable unions (decreasing from 4.95% in 2015 to 1.79% in 2020), but not in live births to single mothers under age 18. The virtual disappearance of legal marriages among minors seems to be a consequence of the 2015 legislation prohibiting marriage before the age of 18, although it could also reflect a social desirability bias, whereby minor mothers may not have reported their true marital status out of fear of social stigmatization or other repercussions. In any case, this recent legislation does not contemplate stable unions, which currently represent one of the most significant challenges despite a reduction in number. With our data, it is not possible to determine whether the reduction in stable unions was an indirect consequence of the legislation - which in principle condemns all types of unions among minors - either as a real decrease or merely a decline in reporting, or if it was due to the trend toward increasing maternal age across the entire age distribution.

The higher proportion of live births to mothers under age 18 among Indigenous and Afro-Ecuadorian mothers does not seem to be caused by child marriage, but rather by early pregnancies among single mothers. The high proportions of births among mothers under age 18 in the Coastal and Eastern (Amazon) regions are correlated with the higher concentration of Afro-Ecuadorian and Indigenous population in these regions respectively.^32,33

The main finding of our study is that the relationship between marital status and perinatal outcomes depends on maternal age. We observed a marriage advantage among women aged 20 to 24, which was consistent with the literature on preterm birth, low birth weight, and other perinatal outcomes,^{8,9,10,13,14,31} but not among women under age 20, in particular those under age 18. This modification of the association of marriage with perinatal health indicators according to lower maternal age suggests that the mechanisms through which marriage influences perinatal health may be different for adult women than for minors. The marriage advantage for adults is usually understood in terms of two theories,¹⁶ which are not mutually exclusive. The first assigns a causal effect to marriage, in that it provides a context which promotes healthier behaviors (for example, lower levels of smoking and alcohol consumption) which translate to better health outcomes. The second (called the marriage selection hypothesis) assumes that marriage itself is not a cause, but rather that individuals who eventually choose to marry have sociodemographic characteristics associated with better health, such as higher income, wealth, education, and other indicators of more privileged socioeconomic backgrounds. These mechanisms may not be the same for minors. Marriage among minors may not be a protective factor for minors in the same way it may be for adult women due to the deeper gender inequalities that are present for minors, which manifest themselves in the lower age and power in comparison to their spouse,³⁴ gender stereotypes, lack of autonomy, and financial dependence.^4,5,6,7,23 Furthermore, marriage selection mechanisms may differ among age groups - and may not necessarily confer the same protective aspects to minors - including the pressure to marry by family members (driven by religious beliefs), the urgency to legitimize an unintended pregnancy, or the idea that marriage is a way to escape poverty or abuse in the family of origin.^23,32,35) To summarize, these aspects point to the complexity of the relationship between marital status, maternal age, and perinatal health, and the consequent need to employ diverse research strategies such as longitudinal and qualitative studies in order to understand the causes and consequences of these relationships.

There is little information regarding the context and circumstances surrounding births to mothers who are minors. Although child marriage (including common-law unions) is the focus of the prevention strategies in Asia and Africa - where forced or arranged marriages are more common and the majority of births to minors occur under these circumstances - in the Americas, single mothers account for the majority of cases of births to minors and the worse health outcomes among different marital arrangements.³¹ Although forced marriages are not the norm in the Americas, the possibility of family and economic coercion does exist, for example when young daughters are married off to economically independent adult men as a means to alleviate family poverty under the guise of providing better economic conditions for the minor.^23,35 In any case, strategies for the reduction of early pregnancies and their consequences should contemplate single women, who also may also have been victims of different types of gender and power inequalities or forms of abuse that violate their right to free and informed consent to engage in sexual relations.

Strengths and limitations

Among the strengths of this study, the large volume of population-based data stands out, which allowed for a detailed analysis of different age and marital status subgroups, along with the ability to distinguish between legal marriages and common-law unions, something that is not possible when analyzing birth records in the USA and Canada.^9,10 The limitations are comparable to other studies that use similar data sources.^8,9,10,31 First, due to the cross-sectional nature of data collection at the time of birth, reported marital status may have changed since the time of conception. Since data is not collected on marital status at the time of conception, it is not possible to determine if a preexisting marriage or union existed when the child was conceived. This limitation also contributes to an underestimation of early pregnancies, given that around two thirds of births to 18-year-old mothers were conceived eight or nine months prior, when many of them were still 17 years of age.³¹ Second, sociodemographic information is self-reported by mothers and may be affected by social desirability biases, for example reporting legal marriage in the case of common-law unions. In particular, pre-established marital status categories in statistical reporting related to live births may distort the complexities and specificities of marriage arrangements of Indigenous mothers, who belong to diverse groups with social practices that may not be readily understood from a Western perspective. Third, some mothers may have had more than one live birth during the study period, however due to the lack of unique maternal identifiers in birth records, it is not possible to identify live births to the same mother. Lastly, there is the possibility of residual confounding in multivariate analyses due to the exclusion of potentially confounding variables such as paternal characteristics,¹⁷ or possible measurement errors in some variables. Despite these limitations, the results of this study are relevant for other countries in Latin America.