2.1 Study context

This analysis used data from the LBWSAT, a non-blinded, cluster randomized controlled trial conducted in the Dhanusha and Mahottari districts of Nepal. The LBWSAT assessed the impact of community-based participatory learning and action (PLA) women’s groups, with and without food or cash transfers, on birth weight and infant weight-for-age z (WAZ)-scores. All married women and girls aged 10–49 years across 80 village clusters were invited to take part in the trial, permitted they or their husband had not undergone surgical family planning (Saville et al., 2018). About 63 308 participants consented to menstrual monitoring from 80 randomized village development committee (VDC) clusters, and 25 090 pregnancies were detected between December 2013 and February 2015 (Saville et al., 2018). VDC clusters were randomly assigned to one of four interventions: behavioral change PLA groups, PLA groups + cash given to pregnant participants, PLA groups + “Super Cereal” supplement given to pregnant participants, or existing government programmes. The detailed study protocol and primary outcome results have been published previously (Saville et al., 2016, 2018; Style et al., 2017).

Informed written consent was taken from all participants in this study, and their guardians if they were aged <18 years. Research ethics approval was obtained from the Nepal Health Research Council (NHRC) (108/2012) and the University College London (UCL) Ethical Review Committee (4198/001) for primary data collection, and from the NHRC (292/2018) and the UCL Ethical Review Committee (0326/015) for the secondary analyses presented in this article.

2.2 Variable selection

The outcome variable for this analysis was the occurrence of preterm delivery (<37 weeks/259 days gestation). Gestation length (GL) was calculated as the time between last menstrual period (LMP) and date of delivery. LMP was determined using maternal recall for 18 306 pregnancies and using ultrasound for 1321 pregnancies where available. Agreement analysis between GL determined by ultrasound and maternal recall was undertaken by calculating Pearson’s correlation coefficient and Cronbach’s alpha for GL categories. The binary preterm delivery variable was selected rather than a continuous GL variable to align with other research and to minimize the influence of inaccuracies in maternal recall of LMP.

The primary exposure variables for this analysis were the participant’s age at marriage and age at first pregnancy. In this trial’s setting, most people count their age in running years rather than completed years. Therefore, age at marriage and first pregnancy were collected as integer values in running years and converted to completed years (running years −1) for analysis. To investigate the importance of policy aiming to reduce childhood marriage and motherhood, and based on the distribution and pattern of data in our sample, age at marriage was coded into four groups: ≤14 years, 15 years, 16–17 years, and ≥18 years, while age at first pregnancy was coded into three groups: ≤15 years, 16–17 years, and ≥18 years.

Using a priori knowledge, a directed acyclic graph (DAG) was constructed using DAGitty to identify the minimal sufficient adjustment variables to account for confounding (Textor et al., 2011; Supplemental Figures 1–4). The arrows in the DAG were entered to represent the hypothesized direct causal effect of one variable on another. Variables that are hypothesized to be directly antecedent to the exposure (age at marriage or age at pregnancy) and outcome (preterm birth) are indicated in pink and identified as a confounder (Hernán et al., 2002). This approach identified maternal education, maternal caste group and material household assets as confounders for the association between age at marriage and preterm delivery, as these factors are hypothesized to have a causal effect on both exposure and outcome (Supplementary Figures 1 and 2). For the association between age at first pregnancy and preterm delivery, the same factors plus age at marriage were identified as confounders (Supplementary Figures 3 and 4).

Maternal education was coded into four levels based on the Nepali education system and the distribution of data in our sample: no formal education, primary (1–5 years), lower secondary (6–8 years) and secondary or higher (≥9 years). The household asset score was determined using principal component analysis, reflecting ownership of consumer goods such as a color television, motorbike, or computer, land ownership and household infrastructure (Saville et al., 2018). Maternal caste was grouped into three groups: disadvantaged comprising of Dalit and Muslim, middle comprising of Janajati and other Terai castes, and advantaged comprising of Yadav and Brahmin.

Maternal weight and mid-upper arm circumference (MUAC) were measured in early pregnancy (8–30 weeks). Maternal height was measured in early pregnancy, and endpoint where earlier height was missing. Body mass index (BMI) was subsequently calculated using early pregnancy weight and height collected at either timepoint, and adjusted for gestational age. However, these measurements were only available for a sub-set of women. Season of birth and sex of the baby was determined at birth (within 42 days). Maternal age at current pregnancy was also collected at baseline for additional analysis in multigravida women, categorized into six categories: <18 years, 18–20 years, 21–23 years, 24–26 years, 27–29 years, and ≥30 years.

2.3 Statistical methods

The characteristics of primigravida and multigravida participants were described and compared using chi-squared tests. The characteristics of participants with missing and complete GL data were also compared using chi-squared tests. The relationship between age at first pregnancy and age at marriage was explored visually by generating heat tables in Microsoft Excel (Microsoft Corporation, 2018) and a correspondence analysis plot (“mcaplot” command in Stata). Correspondence analysis represents contingency tables as a low-dimension geometric map of the association, representing the closeness between rows (age at first pregnancy) and columns (age at marriage; Greenacre, 1992).

Mixed-effects logistic regression models were fitted to assess associations of the exposures with preterm delivery, with the VDC cluster included as a random effect term on the intercept (“xtlogit” function in Stata). Minimally adjusted models adjusted for study design, including the random effect for cluster and fixed effects for study arm and strata. Fully adjusted models adjusted for study design plus additional confounders identified using the DAG. Missingness patterns of variables were identified and multivariate imputation using chained equations (MICE) was applied to deal with missing data (“mi” function in Stata; 30 completed datasets; van Buuren & Groothuis-Oudshoorn, 2011). To account for the multi-level nature of the data, the imputation was executed stratifying for each study arm. Models were also run on un-imputed data as a sensitivity analysis. All models analyzed the association between exposures and preterm delivery for primigravida participants and multigravida participants separately.

To determine the potential for mediation by age at pregnancy in the association between age at marriage and preterm delivery, additional models were constructed. These included age at current pregnancy as a covariate in the model for the association between age at marriage and preterm delivery, and removed age at marriage as a covariate in the model for the association between age at first pregnancy and preterm delivery. Further models were also constructed on the subsample of women with data on maternal height, BMI and MUAC to investigate the role of markers of maternal nutritional status. As rates of child marriage are gradually declining over time, additional models including only participants marrying within the last 5 years were constructed to rule out the potential bias of mothers marrying young being older.

All models report adjusted odds ratios (aOR) and 95% confidence intervals (CI). Statistical analyses were performed in Stata IC 16.1 (StataCorp, 2019).

5.1 Psychological stress

We propose several potential explanations for the association observed between a young age at marriage and preterm delivery. First, while marriage is considered a successful outcome in Nepal, the transition to married life is often a time of psychological stress. Young brides move from their parental to their marital home, where they must adapt and conform to the expectations and requests of their new husband’s family (Davis, 2009; Gram, 2018). As a result, they are at risk of becoming overworked, under-nourished, and vulnerable to physical and emotional abuse (Gram, 2018). This transition is distressing, particularly during early adolescence when their young age contributes to their sense of disempowerment (Jensen & Thornton, 2003; Parsons et al., 2015). Also, younger girls have a stronger emotional attachment to their own families, which may make the transition more difficult and impact on relations with their in-laws (Moretti & Peled, 2004; Palriwala, 1993). While large studies on the emotional consequences of child marriage are lacking, there is increasing evidence that adolescents marrying young experience negative mental health consequences (Burgess et al., 2021; Kidman, 2017; Kumar et al., 2018; Oshiro et al., 2011; Raj et al., 2018; Sezgin & Punamäki, 2020). However, the differences in social and cultural contexts between studies limits comparison, and research on the mental health consequences of child marriage in Nepal is limited (Chaulagain et al., 2019). Furthermore, adolescence is a time of critical development with regards to one’s mental health, likely amplified while adapting to the practical, cultural and religious expectations of marriage during adolescence (Powers et al., 1989; Rose-Clarke et al., 2021). Gender norms which drive the practice of child marriage may also play a role in normalizing physical and emotional abuse (Le Strat et al., 2011; Nour, 2009). Girls marrying young are at a higher risk of experiencing IPV and, due to their reliance on their husband’s family, are less able to speak out against it (Kanesathasan et al., 2008). Emotional stress is a well-known risk factor for preterm delivery, mediated by neuroendocrine and inflammatory mechanisms and changes in health behaviors (Dunkel Schetter, 2011; Gennaro & Hennessy, 2003), therefore it represents a plausible explanation for the association we observed between child marriage and preterm delivery. Future research into the negative reproductive health sequelae of child marriage would benefit from research into the consequences associated with mental health and stress.

5.2 Undernutrition

Young brides are also commonly undernourished (Goli et al., 2015). Markers of undernutrition were included in our DAG and did not emerge as key confounders (Supplemental Figures 1–4), however the DAG is a simplification of reality and undernutrition remains a plausible mechanism for the association between child marriage and preterm delivery. While undernutrition is prevalent within Nepali children generally, the intrahousehold allocation of food discriminates against pregnant women and girls in their marital home in particular (Chalise et al., 2018; Harris-Fry et al., 2018; Palriwala, 1993). In addition, their mental health may impact on their appetite (Chorghade et al., 2006). A shorter stature, a marker of undernutrition in childhood, has indeed been associated with a young age at marriage and first pregnancy (Marphatia, Saville, Manandhar, Amable, et al., 2021; Marphatia, Saville, Manandhar, Cortina-Borja, et al., 2021). Nutritional status has been associated with adverse pregnancy outcomes, including preterm delivery (Bloomfield, 2011; Kramer et al., 1992; Nsereko et al., 2020; Shah et al., 2014). This may be compounded by the increased risk of undernourishment observed in adolescent pregnancies, as mother and fetus compete for nutrients, undermining the growth of each (Goli et al., 2015; Marphatia, Saville, Manandhar, Cortina-Borja, et al., 2021). Furthermore, seasonal patterns in preterm deliveries exist within this population, which may be mediated by seasonal variability in food availability and nutrition (Hughes et al., 2014; Saville et al., 2021). We undertook some analysis adjusting models for nutritional status (height, BMI and MUAC; Supplemental Tables 3 and 4), finding that a significant association remained between child marriage and preterm delivery. However, data on nutritional status was only available for a small subset of our sample (between 17% and 78% of our study population), limiting our conclusions from this analysis. The potential contribution of nutritional status to the association between early marriage and preterm delivery merits further research.

5.3 Education

Our models adjusted for a girl’s level of formal education, however there is a wider education which occurs during adolescence as a part of their transition to womanhood, which is not accounted for in formal education. Girls develop a sense of their identity and place in society during adolescence, undergoing vast emotional and psychosocial development (Marphatia et al., 2017). They also gain an improved understanding of their reproductive and sexual health through interactions with their peer groups (Mathur et al., 2003). As a girl’s husband and in-laws generally determine their participation in society, childhood marriage tends to isolate girls from their peers and wider society, reducing their social capital and networks (Parsons et al., 2015). Therefore, girls marrying at younger ages miss out on an informal education from their peers, which could have consequences for their reproductive health (Nour, 2009). This may contribute to the reduced knowledge of sexual health and uptake of contraception in girls marrying young and the increased risk of developing an sexually transmitted infection (STI; Nour, 2006; Raj et al., 2018; Raj & Boehmere, 2013). While the risk of contracting an STI is lower within marriage, an unmet need for contraception remains, particularly within couples where one is a migrant worker, among whom sexual risk taking is more common (Mehata et al., 2020; Ministry of Health and Population (MOHP), 2017; Simkhada et al., 2017). As STIs are associated with an increased risk of preterm delivery, this may contribute to the association observed within the LBWSAT cohort (Johnson et al., 2011; Mann et al., 2010; Moodley & Sturm, 2000).

5.4 Compounding vulnerabilities

Socioeconomic vulnerabilities are often cited as a possible cause of the association between child marriage and adverse maternal health outcomes (Ganchimeg et al., 2014; United Nations Population Fund, 2012). Indeed, analysis of the LBWSAT cohort has found that a woman’s age at marriage determines their access to different forms of capital (Marphatia, Saville, Manandhar, Amable, et al., 2021). However, our models adjusted for caste group, material household assets and maternal education level, and still found a significant association for marrying young outside of these socioeconomic factors. More broadly, the social context for child marriage in Nepal is complex, with different studies finding contrasting associations between socioeconomic status and child marriage (Aryal, 2007; Pandey, 2017; Raj et al., 2014). However, as maternal health care utilization is lower among those married during childhood, it is likely that compounding social vulnerabilities negatively impact health care access following child marriage (Sekine & Carter, 2019).

5.5 Age at pregnancy

Previous research has focused on the negative reproductive health consequences of child marriage within the context of early childbearing (Gurung et al., 2020; Rahman et al., 2018). While the significance of our observed association between age at marriage and preterm delivery is reduced when adjusting for age at pregnancy (p = .054), the change in significance is small and the effect size remains constant. This suggests that the effect of age at marriage is not mediated by age at pregnancy, and that an independent effect of age at marriage exists. The two exposures are closely intwined; child brides are less likely to use contraception prior to their first pregnancy, often forcing them into early motherhood (Godha et al., 2013; Raj, 2010). In these circumstances, there are a range of consequences of girls entering into pregnancy while they are physiologically immature themselves, due to the nutritional competition between mother and fetus (Gibbs et al., 2012; King, 2003). However, 40% of our primigravida participants marrying at age ≤14 years gave birth aged 18 years or older. In the bio plots in Figure 2 we observe a closer relationship between marriage at ages ≤14 years and first pregnancy at 10–15 years for multigravida participants, whereas this age at marriage group among primigravida participants is more evenly distributed between all three categories of age at pregnancy. This suggest there is a developing trend for married couples to delay their first birth, despite the proportion of women using modern contraceptives not changing from 2006 to 2016 (Diamond-Smith et al., 2020; Ministry of Health and Population (MOHP), 2007, 2017). Furthermore, no significant association was observed directly between age at pregnancy and preterm delivery. Future research should include investigation of both age at marriage and pregnancy, particularly in regions with high rates of both, to determine where to focus efforts to improve the health and wellbeing of mothers and their babies.

Our finding that there is no increased risk of preterm delivery among those entering into pregnancy at a very young age contrasts with previous research (Althabe et al., 2015; Ganchimeg et al., 2014; Gurung et al., 2020; Kumar et al., 2018; Stewart et al., 2007). The assumption that adolescent pregnancy is detrimental to a woman’s reproductive health may not apply to all birth outcomes. In fact, some research suggests that young adolescents deliver better than older adolescents, with a lower risk of obstetrical intervention (Robillard et al., 2019). While adolescent pregnancy has been associated with reduced linear growth, depletion of fat stores and increased nutritional vulnerability due to the nutritional competition between the mother and fetus, our findings suggest that this vulnerability does not increase the risk of delivering preterm (Rah et al., 2008; Wells et al., 2012; Wells et al., 2021). Furthermore, previous research refers to the social risks of adolescent pregnancy, such as increased isolation and exclusion from healthcare facilities (Ganchimeg et al., 2014; Leftwich & Alves, 2017). However, early childbearing in South Asia is generally considered a successful outcome, with extended family members supporting both mother and baby, negating some of the social and environmental problems associated with adolescent pregnancy in other countries (Choe et al., 2005). It is also possible that publication bias exists within the existing literature on associations between adolescent pregnancy and preterm delivery, in which negative findings are withheld from publication (Joober et al., 2012).

While age at first pregnancy did not have a significant association with preterm delivery, gynecological age, which represents the time since menarche rather than chronological age, may still play a role. In the two-to-three years following menarche, nutritional competition diminishes, cervix length increases, vaginal pH becomes more acidic, and menstrual frequency increases, all of which contribute to reproductive maturity and are associated with a reduced risk of preterm delivery (Brabin et al., 2005; D’Agostini et al., 2013; Stevens-Simon et al., 2002). As menarche has been reported to start later in girls with a poor nutritional status, girls may only begin menstruation during their late teens and still be gynecologically immature at age 18 (in the reference category in our models; Gibbs et al., 2012; Stevens-Simon et al., 2002). Girls experiencing menarche during later adolescence may also contribute to the disconnect between age at marriage and pregnancy. We were not able to assess the association between gynecological maturity and preterm delivery as the LBWSAT did not collect data on age at menarche. As there is a considerable gap in the data on girls’ age at menarche and how it is associated with pregnancy outcomes, improved data collection on age at menarche would better inform our understanding of this topic and help determine when women are physiologically ready for pregnancy (Leone & Brown, 2020).

5.6 Strengths and limitations

There are several strengths of this study. The high proportion (90%) of participants marrying under the age of 18 years in this study has enabled us to shed new light on this issue; we were able to observe associations with preterm delivery according to the specific age at which girls were married during childhood, whereas previous studies have categorized participants as marrying either under or over the age of 18 years (Huang et al., 2021; Pandya & Bhanderi, 2015; Rahman et al., 2018). Also, our large sample size enabled a thorough analysis of the consequences of marrying and falling pregnant at specific ages during adolescence. The richness of the data collected also enabled successful adjustment for confounders. More broadly, by disentangling the consequences of early marriage and early first pregnancy, we have shed new light on the interpretation of existing studies in this field and highlighted the importance of considering the social and physiological and determinants of health.

The inferences from our research are limited by the constraints of the study. Firstly, gestation length was determined from maternal recall of LMP, which is prone to error. To account for this, we conducted analysis based on the binary outcome of deliveries being term or preterm, but in doing so, we lost information with regard to the extent of prematurity (extremely-, very- or late-preterm deliveries, or even early term). Furthermore, a substantial proportion of the study population had implausible GL data and were therefore excluded from analysis. Second, due to unforeseen difficulties experienced by the research team during data collection, there was a relatively high proportion of loss to follow-up resulting in missing gestational age data. Significant differences between those with missing or excluded GL data may have introduced bias into our findings. However, results including and excluding multivariate imputation methods to account for missing data found similar results. Third, the lack of an indicator for psychological stress limits the conclusions of this work, and future work should include such variables in models. Despite these limitations, our data are well suited to provide new insights into the consequences of child marriage and the mechanisms leading to preterm delivery.