Early childhood development and its association with maternal parity

Abstract Background Maternal parity, which is usually measured as the number of children born to a mother, has a substantial impact on the social and environmental factors around children and their development. This paper estimates the Early Childhood Development Index (ECDI) of 3‐ and 4‐year‐old children in Bangladesh and examines the relationship between maternal parity and early childhood development. Methods The study analysed nationally representative data from the Bangladesh Multiple Indicator Cluster Survey 2019. The dataset had 9453 children aged from 36 to 59 months. The ECDI was computed following the UNICEF's approach involving psychometric computation of four domains of development: physical, literacy‐numeracy, learning and social–emotional. Since the dataset has a hierarchical structure, we used multilevel logistic regression. Results A quarter (25%) of the children were not on track in their early childhood development. Seventy‐one percent were not developmentally on track in the literacy‐numeracy domain, 27% were not in the social–emotional and smaller percentages were not in learning (9%) and physical (1%) domains. There was a significant negative association between maternal parity and ECDI (adjusted odds ratio [AOR] 0.95; 95% CI: 0.91–0.99). Attendance at early childhood education programmes was significantly associated with early childhood development (AOR 1.73; 95% CI: 1.47–2.03). Also, female children, those who were not stunted, located in rural areas, received parental stimulation activities, lived in relatively wealthy households or had mothers who had received secondary or further education were more likely than others to be on track of early childhood development. Conclusions Early childhood development is negatively correlated with maternal parity. The literacy‐numeracy domain constitutes the major developmental delay. Programmes for parental awareness should be widely expanded.


| INTRODUCTION
There is growing evidence that early childhood is the most important developmental phase throughout the lifespan. Healthy early child development, which includes the physical, social-emotional and language-cognitive domains of development, influences children's overall health and well-being and establishes their developmental trajectories and life-courses (Loizillon et al., 2017;UNICEF, WHO, & World Bank, 2016). It is now well accepted that apart from biological factors, children's development is shaped by the quality of their homes, neighbourhood environments, characteristics of their parents (Downey, 2001) and social factors (Ranjitkar et al., 2019). This growing evidence has prompted national and international calls for investment in early childhood development (Wilkinson, Marmot, & WHO, 1998) and for researchers to identify specific factors in children's social environments that facilitate or hinder early childhood development and can be modified. However, the evidence so far for the effects of social factors is mostly context-specific and varies between studies and settings (Donald et al., 2019;McDonald et al., 2016;Walker et al., 2011).
In the research undertaken to identify significant factors associated with early childhood development, maternal parity has received little attention. Maternal parity, which is usually measured as the number of children born to a mother, has a substantial impact on the social and environmental factors around children and their development because the family system and structure vary between one-child and multichild families and keep changing with each new birth (Fish & Stifter, 1993). In earlier literature, the effects of sibship sizes (i.e., number of children in the family) on intellectual skills or educational achievement received the most attention (Downey, 2001;Guo & VanWey, 1999;Lu & Treiman, 2008), while aspects such as social, emotional and physical development received inadequate attention. In recent literature, quantitative studies investigating factors that influence early childhood development were sometimes adjusted for maternal parity, but it was seldom the main focus (de la Rochebrochard & Joshi, 2013). Moreover, although a small number of studies identified maternal parity as important, the findings are inconsistent (Aronen, 2009;Hayashida & Nakatsuka, 2014). Also, the ways that the studies calculated early childhood development varied, and the developmental indicators were rarely validated appropriately, making comparisons of the findings impossible. In fact, the dearth of globally accepted indicators for child development has hampered the progress of research in this area (Loizillon et al., 2017 (Loizillon et al., 2017). This index is the first international population-based measure of early childhood development for lowand middle-income countries. Using population-representative data from the Multiple Indicator Cluster Survey (MICS) programme, this paper estimates the number of 3-and 4-year-old children who were on track in their development and the relationship between maternal parity and early childhood development of their children in Bangladesh.

| Data
We used the 2019 MICS data for Bangladesh. MICS is a nationally representative and internationally standardized household survey developed by UNICEF that captures information about children in low-and middle-income countries (Khan & Hancioglu, 2019;UNICEF, 2021). A two-stage, stratified cluster sampling was used for data collection. The urban and rural areas within each of 64 administrative districts were identified as the main sampling strata. First, within each stratum, a specified number of primary sampling units were selected systematically with probability proportional to size.

Key messages
• A quarter (25%) of the children aged 36 to 59 months were not on track in their early childhood development.
• The literacy-numeracy domain constitutes the major developmental delay.
• This study found a significant negative association between maternal parity and early childhood development.
• Early childhood education programmes, education for girls and the creation of parental awareness should be widely expanded. parental stimulation activities they had. The structural covariates were districts, communities and households. The primary sampling units were considered to be their communities.
The standards of the WHO were used for identifying children with stunting. Hence, children whose height-for-age was more than 2 standard deviations below the median of the reference population were considered short for their age and are classified as stunted (WHO, 2021). Parental stimulation was measured by asking the primary caregivers whether any household members had engaged in the following six activities with the children in the past 3 days: reading books or looking at pictures, telling stories, singing songs, taking the child outside, playing with the child and naming, counting or drawing together. A summary score was created, which varies from 0 (no paternal engagement in any stimulation activity) to 6 (paternal engagement in all stimulation activities). This approach has been used as a measure of caregiving and stimulation in several previous studies (Bornstein et al., 2015;Bornstein & Putnick, 2012;Jeong et al., 2016;Sun et al., 2016).

| Outcome variable
In the MICS dataset, information for the ECDI was collected for 3-and 4-year-old children. We computed the ECDI for following the UNICEF's approach involving psychometric computation, the details of which can be sourced elsewhere (Loizillon et al., 2017). Briefly, this index includes 10 questions covering four domains of early childhood development: language-cognitive, physical, social-emotional and approaches to learning (Table 1). Children are regarded as having healthy early childhood development if they are developmentally on track in at least three of these domains.

| Data analysis
We conducted descriptive analysis to summarize the characteristics of the children and their ECDI status, attendance at early childhood education programmes, nutritional status, households' economic status, places of residence and mothers' educational status. The dataset has a hierarchical structure: children are nested in households, households are nested in primary sampling units (i.e., communities) and these are then nested in the 64 administrative districts. Thus, we used multilevel logistic regression-both for overall ECDI and the four individual domains as outcome variables. We used a three-level multilevel model: children (level 1), community (level 2) and districts (level 3).
The main study factor, maternal parity, was included as a continuous variable in the regression models. Three models were developed.
Model 1 was the null model and did not have any predictor variables, Model 2 examined the relationship between maternal parity and ECDI of their children and Model 3 was the final model that examined the relationship while being adjusted for other covariates.
The general contextual effects have been estimated using the between-neighbourhood variance and the median odds ratio (Larsen & Merlo, 2005;Merlo et al., 2016). The median odds ratio translates the area level variance in the odds ratio scale and quantifies the variation between clusters by comparing two individuals from two Univariate analysis with the ECDI and maternal parity shows a significant negative association. Figure

| DISCUSSION
The findings of this study suggest that, overall, a quarter of the children were not on track in their early childhood development, which is much higher than is estimated in many other low-and middle-income countries (Emerson et al., 2018). The performance in literacy-numeracy was the worst of the four domains of the ECDI.
The findings also suggest that children with relatively large sibship sizes are significantly more likely not to be on track in their early childhood development. The other important finding is that female children who were not stunted, had attended early childhood education programmes, received parental stimulation or lived in relatively wealthy households or whose mothers had received secondary or further education were more likely than their counterparts, respectively, to be on track in their early childhood development.
Our results suggest that literacy-numeracy is the most challenging  organizations also offer preprimary education. However, the relatively low rate of attendance (<20%) could be because the programmes do not yet include all children aged 3 and 4 years (Tribune Desk, 2020).
The attendance at early childhood education programmes can accrue additional developmental benefits, as they also offer most of the parental stimulating activities such as reading books, telling stories, singing songs and naming, counting or drawing with the child.
Several covariates in our study, including stunting, access to early childhood education, the household's wealth status, the mother's educational status and the location of the residence, have been consistently identified as important for early childhood development (Frongillo et al., 2017;Kang et al., 2018;Sk et al., 2020). These are indeed a set of social determinants of health (Islam, 2019) and highlight that early childhood development follows a social gradient.
Unfavourable social determinants of health increase exposure among young children to biological and psychosocial risks that subsequently affect development (Amugsi et al., 2020;Walker et al., 2007). In wealthy countries, maternal parity may not have a significant impact on early childhood development because the negative effects of high parity can be offset by the positive effects of factors such as good nutrition, educated mothers, healthy environments and access to early childhood education. However, in resource-poor settings, relatively large sibship sizes can impact children's shares of base resources such as food, clothes and shelter, and they may only receive minimal supervision (Downey, 2001). As the risks accumulate, development may be compromised in a compound manner (Moore et al., 2015;Walker et al., 2007).
While improving social conditions in low-income countries takes time, at least some programmes can be established for parents from relatively low socioeconomic backgrounds, as they may not have the same access to health information as other families (Begum, 2019).
Parental knowledge of childhood development can affect attitudes and behaviours and impact their children's outcomes through a "mechanism of change" (Moran et al., 2004). Parenting programmes can target educating parents in families of lower socioeconomic status about the importance of early childhood development, their engagement, child nutrition and early childhood learning (Aboud, 2007). There could be several explanations for the negative association that was observed between the maternal parity and the ECDI. Since we had no information about the genetics of the children, and we examined only the psychosocial factors, one possible explanation is that of "resource dilution" (Downey, 2001), which refers to a lower parental investment in children that goes along with increasing parity and/or competition between siblings for finite resources (Helfrecht & Meehan, 2016). A second explanation is the omitted variable bias.
There may have some unmeasured confounding variables that were not split equally between the children who were developmentally on track and those who were not, thus producing a spurious association.
Nevertheless, there is a high chance that such an unequal distribution is affected by resource dilution. Another explanation could be that the fertility rate is relatively low for families that can afford multiple children, while the rate remains higher among the least affluent families (Bangladesh Bureau of Statistics, 2015). However, the findings of this study warrant due attention, particularly in countries like Bangladesh, which has a large population in a relatively small piece of land:  (Gil et al., 2020). There could be several reasons for this finding. First, more of the children in urban areas who were developmentally on track had also attended early childhood education programmes than the children in rural areas who were on track, and after adjustment in regression for early childhood education, this offered a net performance that was better for rural children. Second, this finding may be partly attributable to measurement bias; since many mothers in rural areas had only preprimary or even less education, some may have erroneously identified their children as being on track in the literacy-numeracy domain and therefore contributed to differential misclassification.

| LIMITATIONS
This study has several limitations. Firstly, this is a cross-sectional study. Thus, the relationship between maternal parity and early childhood development is associational only. Secondly, although we adjusted the regression models for many potential confounders, it is not unlikely that some residual confoundings remain unadjusted.
Future research should examine this relationship using a prospective cohort dataset. Lastly, the contemporary theoretical conceptualization of child development incorporates the biopsychosocial model of health, which puts importance on both nature and nurture. We were unable to account for the biological and genetic endowment of the children.

| CONCLUSION
Our study found a negative association between maternal parity and early childhood development. Among the four domains of early childhood development, the performance in literacy-numeracy was the worst. Several factors such as children's nutritional status, their attendance at early childhood education programmes, the parental stimulation activities they had, their mothers' education and their households' wealth status were significantly associated with early childhood development, and this highlights the importance of social determinants of health. While addressing all these factors may not be an achievable goal over the short term, particularly within the context of a developing country, ongoing efforts for early childhood education, education for girls and the creation of parental awareness should be widely expanded.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are publicly available on the website of UNICEF MICS upon its approval.