Anthropometric measures at birth and early childhood are associated with neurodevelopmental outcomes among Bangladeshi children aged 2–3 years

Abstract

Among a cohort of children located in rural areas of Bangladesh affected by high levels of exposure to environmental metals, we investigated the associations between anthropometric measures, growth trajectory, and neurodevelopment at age 20–40 months. Our study population included mothers and their children who participated in a longitudinal birth cohort study that took in place in the Pabna and Sirajdikhan areas of Bangladesh. Anthropometric measures including weight, length, and head circumference were measured at birth, age 12 months, and age 20–40 months. Neurodevelopment was assessed using Bayley Scales of Infant and Toddler Development—Third Edition (BSID-III) multi-scale at age 20–40 months. A total of 777 mother-child pairs were included. Higher anthropometric measures at 20–40 months were associated with higher cognitive, language, and motor scores on BSID-III. For example, a 1-kg increment in birthweight was associated with an increase of 2.11 for cognitive score (p<0.0001), 1.63 for language score (p=0.006), and 0.89 for motor scores (p=0.03). Greater positive changes in growth parameters, or growth trajectory, between birth and 20–40 months were also associated with higher BSID-III scores. These associations remained significant after adjusting for potential confounders and prenatal exposure to environmental metals. These findings suggest that even when taking into account high environmental metal exposures, prenatal and early childhood growth have strong associations with neurodevelopmental test scores in early childhood.

Graphical Abstract

1. Introduction

Anthropometric measures at birth and early in childhood are associated with health outcomes across infancy, childhood, adolescence, and adulthood.^{1, 2} Low birth weight is associated with increased risk of developing hypertension,^{3, 4} diabetes mellitus,⁴ and cardiovascular disease⁵ in late childhood and adulthood. Similarly, early childhood stunting, defined as height-for-age more than two standard deviations below the median of World Health Organization Child Growth Standards,⁶ is a strong predictor of many adverse health outcomes including obesity⁷ and metabolic disorders.⁸ In addition to increasing risk of cardiovascular and metabolic diseases, low birthweight and stunting are associated with higher risk of long-term neurodevelopmental impairment,^9–11 and a growing body of literature suggests that the associations between physical growth and neurodevelopment are also found among infants within the normal ranges of anthropometric measures.^12–14

Early life exposure to chemicals in the environment may further increase the risk of neurodevelopmental impairment.¹⁵ In particular, metals such as arsenic, manganese, and lead have been found to be associated with poorer performance on neurodevelopmental assessments, and these associations have been found across a wide range of exposure levels.^16–18 Chemical exposures are more prevalent and are found at higher levels in low- and middle-income countries,^19–23 many of which also have high rates of nutritional and other risk factors that contribute to poor physical growth.

In Bangladesh, where the current study takes place, an estimated 70 million people have been chronically exposed to high levels of arsenic via contaminated drinking water following the installation of approximately 19 million hand-pumped wells in the beginning of the 1980s.^{19, 21, 24} The wells were intended to provide pathogen-free groundwater for the prevention of waterborne diseases, but have since raised concerns regarding high levels of environmental arsenic found in the water. Studies of children exposed to arsenic in utero demonstrate that prenatal exposure to arsenic is associated with poorer performance on neurodevelopmental assessments in children.^25–31 Our previous study reported the associations between exposure to environmental arsenic with decreased cognitive scores among Bangladeshi children,¹⁸ and our observations are consistent with other studies conducted in this country.²⁹ Recent studies have also shown that children in areas of Bangladesh have high exposures to other environmental metals, including high levels of manganese and lead,^{19, 32} both of which are recognized to be associated with adverse neurodevelopmental outcomes.

Growing evidence has described associations between environmental metal exposures and adverse birth and growth outcomes.^32–35 We have shown that the arsenic concentration in wells used by mothers during pregnancy is associated with decreased birth weight in infants.³⁴ In addition, our group demonstrated that environmental lead exposure was associated with stunting among 2- to 3-year-old Bangladeshi children.³² Despite the growing recognition that environmental chemicals influence prenatal and early childhood growth, most studies in Bangladesh and other countries with high levels of chemical exposures do not incorporate important exposure to environmental chemicals when evaluating the associations between growth and neurodevelopment; this omission may lead to significant issues with confounding.^36–39

The primary purpose of this study was to investigate anthropometric measures at birth and in early childhood to identify which physical measures, and at which particular time points are associated with neurodevelopment at age 2–3 years. We further examined whether these associations changed after adjustment for exposure to environmental arsenic, manganese, and lead. Because our study population was exposed to high concentrations of environmental toxicants, secondary aims included assessing whether there was an independent effect of anthropometric measures on neurodevelopmental measures as well as identifying critical windows for growth on neurodevelopment.

2. Materials and Methods

2.1 Study Sample

The children in this study were participants in a prospective birth cohort study recruited in the Sirajdikhan and Pabna regions of Bangladesh between 2008 and 2011. The design, inclusion criteria, and recruitment strategy have been described previously.¹⁸ Briefly, pregnant women (≤16 weeks of gestational age) were recruited between 2008 and 2011 from clinics associated with Dhaka Community Hospital in the Sirajdikhan and Pabna Sadar Upazilas of Bangladesh. Gestational age was determined by first trimester ultrasound. When children were aged 12–40 months, families were re-contacted and invited to participate in this current study. This study included two visits for mothers and their children, one at approximately 12 months of age, and one at 20 to 40 months of age. Anthropometric measures of the children, including weight, height (length) and head circumference, were assessed at three time points, which were at birth, at 12 months of age, and at 20–40 months of age.

The Human Research Committees at the Harvard T.H. Chan School of Public Health and Dhaka Community Hospital approved this study. Parents provided written informed consent for their children. The Institutional Review Board at Boston Children’s Hospital formally ceded review of this study to the Harvard T.H. Chan School of Public Health.

2.2 Measurement of Growth Parameters

Weight, length, and head circumference of newborns were assessed at the time of delivery by trained field staff from the Dhaka Community Hospital clinics. Birthweight was recorded to the nearest 0.1 kg using a calibrated digital infant scale. Birth length was measured to the nearest 0.1 cm with extended legs and heels against the measuring board using an infantometer. Head circumference was measured to the closest 0.1 cm at the maximal occipital-frontal circumference using a standard measuring tape. Weight, length, and head circumference were also measured at the 12 month and 20–40 month visits.

2.3 Assessment of Neurodevelopmental Status

Neurodevelopment at 20–40 months was assessed using a translated and culturally adapted version of the Bayley Scales of Infant and Toddler Development Third Edition (BSID-III) administered by trained staff. The BSID-III is a well-established psychometric instrument for measuring neurological development in children 1–42 months of age^{18, 40, 41} and includes subscales that assess cognitive, language (receptive and expressive), and motor (fine and gross) development. Receptive language and expressive language scores were combined to compute an overall language score.^42–44 Similarly, the fine and gross motor scores were summed to compute an overall motor score.^{42, 44} To ensure the quality of assessments and scores, a trained pediatrician (OS) and pediatric neurologist (MM) frequently observed study staff during field visits, and approximately 5% of the sessions were reviewed by a senior pediatric neuropsychologist (DCB).

2.4 Umbilical Cord Blood Metal Concentrations

At time of delivery, umbilical cord blood samples were collected using trace metal-free 6 ml capacity Royal Blue Top Vacutainer (Becton Dickinson) tubes with ETDA acid as an anticoagulant (reference number 368380, Becton Dickinson, Franklin Lakes, New Jersey, USA). Cord blood samples were refrigerated immediately at 4 C and shipped in batches from study sites in Bangladesh to the Harvard T.H. Chan School of Public Health Trace Metals Laboratory. Detailed methods for the analysis of cord blood metal concentrations have been described previously.⁴⁵ Briefly, metal concentrations were analyzed by inductively coupled plasma mass spectrometry following an acid digestion method.^{45, 46} The mean limit of detection was 0.03 μg/dL for arsenic, 0.02 μg/dL for manganese, and 0.02 μg/dL for lead. None of the samples had concentrations below the limit of detection.

2.5 Measurements of Covariates

During the enrollment and follow-up visits, trained interviewers used structured questionnaires to collect sociodemographic, medical, family and environmental information. Covariates were chosen through review of the literature based on their relationship with exposures and/or outcomes.^{12, 13, 18, 32, 34, 47–53} Covariates included child sex, child age, gestational age, maternal age, maternal education (primary education or less vs. secondary education or greater), maternal protein intake (low vs. medium. vs. high), exposure to environmental tobacco smoke (smoking environment vs. non-smoking environment), Home Observation for Measurement of the Environment score (HOME), child’s hematocrit level, and study site (Sirajdikhan vs. Pabna). Gestational age was determined by first-trimester ultrasound. Mother’s age and maternal education were assessed at the time of enrollment via questionnaires. Educational attainment of parents was categorized as follows: illiterate, able to write, primary education, secondary education, higher secondary education, college/graduate, and post-graduate. Protein intake (low, medium, high) of the mothers was assessed at third trimester by a food frequency questionnaire.³² At birth, information regarding child’s sex and mother’s exposure to environmental tobacco smoke was collected. At the 20–40 month visit, the child’s age and hematocrit levels were measured. At the 20–40 month visit, a translated and culturally appropriate version of the HOME Inventory was administered.⁵⁴

2.6 Statistical Analysis

All of the variables were evaluated for normality and potential outliers. A natural logarithm transformation was applied to metal concentrations in blood to normalize their skewed distributions. Independent t-tests for continuous variables and Chi-square test for categorical variables were used to assess potential differences in characteristics between included and excluded participants. Pearson correlation coefficients were used to explore the correlations between anthropometric measures (body weight, length, and head circumference measured at birth, at 20–40 month of age as well as changes in growth) and neurodevelopmental test scores. Multivariable-adjusted linear regression analyses were performed to assess the relationships between various anthropometric measure and neurodevelopmental test scores.

Separate models were constructed for each anthropometric measure (i.e., weight, length, and head circumference assessed at birth or 20–40 month visit) and each outcome measure (i.e., cognitive, language, and motor scores of the BSID-III assessed at 20–40 month visit). Models were adjusted for child’s age, child’s sex, gestational age, maternal age, maternal education (primary education or less vs. secondary education or greater), maternal protein intake (low vs. medium. vs. high), exposure to environmental tobacco smoke (smoking environment vs. non-smoking environment), HOME score, child’s hematocrit levels, and study site. When changes in physical parameters were used as predictors, anthropometric measures at birth were additionally adjusted in the models. For instance, when we assessed the association between weight change and neurodevelopmental status, baseline weight was added in the regression models.

In additional analyses, prenatal exposures to arsenic, manganese, and lead, as estimated by cord blood concentrations, were also tested as potential confounders in models. The β-coefficients from the linear regression models were based on each unit increment in growth parameters at baseline or follow-up, or each unit increase in changes of growth parameters between baseline and follow-up. We further tested for the interactions between physical parameters and child sex using cross-product terms. For example, we included the cross-product term of birth weight by child sex in the multivariable-adjusted regression model to assess the effect modification of child sex on the relationship between birth weight and neurodevelopmental score. The p-value of the interaction term less than 0.05 was used to evaluate whether the sex-interaction is statistically significant in the model. For the models with significant sex-interaction term (p<0.05), we further evaluated for potential interactions by sex-stratified analyses. The variance inflation factors from each regression model were computed to examine the collinearity among variables.

Generalized additive models were also constructed to assess whether additional terms, such as a quadratic term, would be appropriate in the regression models. Furthermore, we assessed the relationships of the changes in anthropometric measures between 1) birth and 12 months of age and 2) 12 and 20–40 months of age, with neurodevelopmental status at 20–40 months, using multivariable-adjusted linear regression models with identical covariate adjustment as indicated above.

We did not further adjust our analyses for multiple testing because the primary purpose of this investigation was hypothesis generating (i.e., to explore the potential associations between birth and growth parameters with neurodevelopment status). A two-tailed p<0.05 was considered statistically significant. All analyses were performed with SAS version 9.4 (SAS Institute, Cary, NC).

3. Results

3.1 Descriptive Characteristics

Of the 825 children who participated in follow-up visit at 20–40 months, a total of 777 (94.2 percent) were included in this study. Specifically, one child had missing anthropometric measures at 20–40 months, 10 had missing BSID-III scores, and participants missing data on covariates, including gestational age (n=3), exposure to environmental tobacco smoke (n=1), HOME score (n=6), and hematocrit level (n=31). A comparison of the characteristics of the study sample and those excluded or not available revealed no significant differences. Several exceptions were noted where the mean gestational age and HOME score were higher among those who were excluded from the study; and the mean birth length, length and head circumference at 20–40 months, head circumference change between birth and 20–40 months, and BSID-III scores were lower among those who were excluded from the study.

A total of 327 participants were available for the analysis involving anthropometric measures at 12 months. Due to the small sample size, models that included 12 month measures were considered to be sensitivity analyses only.

Table 1 summarizes the maternal and child’s characteristics at enrollment, birth, and 20–40 month visit. The mean hematocrit level of children was slightly above the cutoff of 33% for iron deficiency anemia, which suggests that a large number of children in our cohort experienced iron deficiency. The average (standard deviation) anthropometric measurements at 12 month visit were 8.6 kg (1.0 kg) for weight, 69.9 cm (3.2 cm) for length, and 32.1 cm (1.7 cm) for head circumference.

Table 1.

Characteristics of Study Population

Characteristics	Overall Participants (N=777)
Maternal Characteristics
Age at Enrollment (Years)	22.9 (4.2)
Education (%)
Primary or Less	46.9 (364)
Secondary or Greater	53.2 (413)
Secondhand Smoke Exposure (%)
Yes	42.6 (331)
No	57.4 (446)
Protein Intake
Low	25.4 (197)
Medium	51.6 (401)
High	23.0 (179)
HOME Score	42.6 (2.6)
Study Site
Sirajdikhan	50.2 (390)
Pabna	49.8 (387)
Child Characteristics
Gestational Age (Months)	37.9 (1.9)
Age at 20–40 Month Visit (Months)	27.7 (2.9)
Female (%)	49.0 (381)
Hematocrit Levels at 20–40 Month Visit (%)	34.5 (3.5)
Birth Parameters
Weight (kg)	2.9 (0.4)
Length (cm)	46.6 (2.5)
Head Circumference (cm)	32.7 (1.2)
Growth Parameters at 20–40 Month Visit
Weight (kg)	11.1 (1.4)
Length (cm)	82.9 (4.2)
Head Circumference (cm)	45.5 (1.7)
BSID-III at 20–40 Month Visita
Cognitive Score	59.9 (4.8)
Language Score	52.7 (7.3)
Motor Score	92.8 (5.1)

Data are shown as means (standard deviations) for continuous variables or percentages (counts) for categorical variables.

^aBSID-III raw scores were used. Receptive language and expressive language scores were combined to compute an overall language score. Fine and gross motor scores were combined to compute an overall motor score.

Abbreviations: BSID-III, Bayley Scales of Infant and Toddler Development Third Edition; HOME, Home Observation for Measurement of the Environment.

3.2 Distribution of Arsenic, Manganese, and Lead Concentrations

Distributions of the arsenic, manganese, and lead concentrations in the cord blood at birth are presented in Supplemental Table 1. These levels of metal exposure are much higher than those reported in most other populations.^{55, 56}

3.3 Associations between Physical Parameters and Neurodevelopmental Status

Child’s age- and sex-adjusted Pearson correlation coefficients are shown in Supplemental Table 2. In general, higher weight, length, and head circumference were associated with higher cognitive, language, and motor scores, regardless of the assessment time point (r ranging from 0.05 to 0.24). Tables 2, 3 and 4 present the associations between physical parameters and BSID-III cognitive, language, and motor scores, respectively. Overall, higher values of anthropometric measures at birth, at 20–40 months, and greater positive changes in anthropometric measures between these time points were consistently associated with higher BSID-III scores. For example, 1 kg increment in birthweight was associated with increase of 2.11 for cognitive score (p<0.0001); 1.63 for language score (p=0.006), and 0.89 for motor scores (p=0.03), based on multivariable-adjusted models (Tables 2–4). For each additional 1 cm increase in head circumference from birth to 20–40 months of age, cognitive, language, and motor scores increased by 0.70 (p<0.0001), 0.70 (p<0.0001), and 0.61 (p<0.0001), respectively (Tables 2–4).

Table 2.

Associations between Birth and Growth Parameters with Bayley Scales of Infant Development-III Cognitive Score

Parameters	Modelsa	Overall Participants (n=777)
		β (Standard Error)	95% Confidence Interval	p-Value
Birth Parameters
Weight	MV	2.11 (0.38)	1.37, 2.85	<0.0001
	MV + Metals	2.00 (0.38)	1.26, 2.74	<0.0001
Length	MV	0.40 (0.06)	0.28, 0.51	<0.0001
	MV + Metals	0.38 (0.06)	0.26, 0.49	<0.0001
Head Circumference	MV	0.35 (0.12)	0.12, 0.59	0.004
	MV + Metals	0.36 (0.12)	0.12, 0.60	0.004
Growth Parameters at 20–40 Month Visit
Weight	MV	0.35 (0.11)	0.14, 0.56	0.001
	MV + Metals	0.34 (0.11)	0.13, 0.55	0.002
Length	MV	0.11 (0.04)	0.02, 0.19	0.01
	MV + Metals	0.10 (0.04)	0.02, 0.19	0.02
Head Circumference	MV	0.73 (0.10)	0.54, 0.93	<0.0001
	MV + Metals	0.71 (0.10)	0.51, 0.90	<0.0001
Changes between Birth and 20–40 Month Visitb
Weight	MV	0.24 (0.11)	0.03, 0.46	0.03
	MV + Metals	0.24 (0.11)	0.02, 0.45	0.03
Length	MV	0.07 (0.04)	−0.009, 0.16	0.08
	MV + Metals	0.07 (0.04)	−0.01, 0.16	0.09
Head Circumference	MV	0.70 (0.10)	0.50, 0.90	<0.0001
	MV + Metals	0.67 (0.10)	0.47, 0.88	<0.0001

Data are shown as β estimates (standard errors). The results show the association with Bayley cognitive score for each additional unit increment in growth parameters.

^aMultivariable (MV) model was adjusted for child sex, child age, gestational age, maternal age, maternal education (primary education or less vs. secondary education or greater), maternal protein intake (low vs. medium. vs. high), exposure to environmental tobacco smoke (smoking environment vs. non-smoking environment), Home Observation for Measurement of the Environment score, child’s hematocrit level, and study site (Sirajdikhan vs. Pabna). MV + Metals model was additionally adjusted for natural logarithmically transformed cord blood arsenic, manganese, and lead levels. Sample size for the MV + Metals models were limited to 771 due to missing blood metal values.

^bRespective birth measures were additionally adjusted in the MV and MV + Metals models. For example, birthweight was additionally adjusted in the weight change model.

Table 3.

Associations between Birth and Growth Parameters and Bayley Scales of Infant Development-III Language Score

Parameters	Modelsa	Overall Participants (n=777)
		β (Standard Error)	95% Confidence Interval	p-Value
Birth Parameters
Weight	MV	1.63 (0.59)	0.39, 2.72	0.006
	MV + Metals	1.48 (0.60)	0.31, 2.66	0.01
Length	MV	0.38 (0.09)	0.20, 0.55	<0.0001
	MV + Metals	0.37 (0.09)	0.19, 0.55	<0.0001
Head Circumference	MV	0.30 (0.19)	−0.07, 0.67	0.11
	MV + Metals	0.33 (0.19)	−0.05, 0.71	0.09
Growth Parameters at 20–40 Month Visit
Weight	MV	0.34 (0.17)	0.01, 0.67	0.04
	MV + Metals	0.33 (0.17)	−0.001, 0.66	0.051
Length	MV	0.19 (0.07)	0.06, 0.32	0.005
	MV + Metals	0.19 (0.07)	0.05, 0.32	0.006
Head Circumference	MV	0.72 (0.16)	0.41, 1.03	<0.0001
	MV + Metals	0.71 (0.16)	0.40, 1.03	<0.0001
Changes between Birth and 20–40 Month Visitb
Weight	MV	0.26 (0.17)	−0.07, 0.60	0.12
	MV + Metals	0.26 (0.17)	−0.07, 0.59	0.13
Length	MV	0.16 (0.07)	0.03, 0.29	0.02
	MV + Metals	0.16 (0.07)	0.03, 0.29	0.02
Head Circumference	MV	0.70 (0.16)	0.38, 1.02	<0.0001
	MV + Metals	0.69 (0.16)	0.37, 1.01	<0.0001

Data are shown as β estimates (standard errors). The results show the association with Bayley language score for each additional unit increment in growth parameters.

^aMultivariable (MV) model was adjusted for child sex, child age, gestational age, maternal age, maternal education (primary education or less vs. secondary education or greater), maternal protein intake (low vs. medium. vs. high), exposure to environmental tobacco smoke (smoking environment vs. non-smoking environment), Home Observation for Measurement of the Environment score, child’s hematocrit level, and study site (Sirajdikhan vs. Pabna). MV + Metals model was additionally adjusted for natural logarithmically transformed cord blood arsenic, manganese, and lead levels. Sample size for the MV + Metals models were limited to 771 due to missing blood metal values.

^bRespective birth measures were additionally adjusted in the MV and MV + Metals models. For example, birthweight was additionally adjusted in the weight change model.

Table 4.

Associations between Birth and Growth Parameters and Bayley Scales of Infant Development-III Motor Score

Parameters	Modelsa	Overall Participants (n=777)
		β (Standard Error)	95% Confidence Interval	p-Value
Birth Parameters
Weight	MV	0.89 (0.40)	0.11, 1.68	0.03
	MV + Metals	0.86 (0.40)	0.07, 1.64	0.03
Length	MV	0.34 (0.06)	0.22, 0.46	<0.0001
	MV + Metals	0.33 (0.06)	0.21, 0.45	<0.0001
Head Circumference	MV	0.38 (0.13)	0.14, 0.63	0.003
	MV + Metals	0.40 (0.13)	0.15, 0.66	0.002
Growth Parameters at 20–40 Month Visit
Weight	MV	0.21 (0.11)	−0.007, 0.44	0.058
	MV + Metals	0.20 (0.11)	−0.02, 0.42	0.08
Length	MV	0.23 (0.04)	0.15, 0.32	<0.0001
	MV + Metals	0.23 (0.04)	0.14, 0.31	<0.0001
Head Circumference	MV	0.66 (0.11)	0.45, 0.86	<0.0001
	MV + Metals	0.64 (0.11)	0.43, 0.85	<0.0001
Changes between Birth and 20–40 Month Visitb
Weight	MV	0.17 (0.11)	−0.05, 0.40	0.14
	MV + Metals	0.16 (0.11)	−0.07, 0.38	0.17
Length	MV	0.21 (0.04)	0.12, 0.29	<0.0001
	MV + Metals	0.20 (0.04)	0.11, 0.29	<0.0001
Head Circumference	MV	0.61 (0.11)	0.40, 0.83	<0.0001
	MV + Metals	0.59 (0.11)	0.38, 0.81	<0.0001

Data are shown as β estimates (standard errors). The results show the association with Bayley motor score for each additional unit increment in growth parameters.

^aMultivariable (MV) model was adjusted for child sex, child age, gestational age, maternal age, maternal education (primary education or less vs. secondary education or greater), maternal protein intake (low vs. medium. vs. high), exposure to environmental tobacco smoke (smoking environment vs. non-smoking environment), Home Observation for Measurement of the Environment score, child’s hematocrit level, and study site (Sirajdikhan vs. Pabna). MV + Metals model was additionally adjusted for natural logarithmically transformed cord blood arsenic, manganese, and lead levels. Sample size for the MV + Metals models were limited to 771 due to missing blood metal values.

^bRespective birth measures were additionally adjusted in the MV and MV + Metals models. For example, birthweight was additionally adjusted in the weight change model.

The additional adjustment of cord blood arsenic, manganese, and lead did not change the direction or materially change the magnitude of these associations. Only one result showed estimate change more than 10% (i.e., the association of birth weight and language score) after adjusting for cord blood metals. The β decreased from 1.68 to 1.48 after additionally adjusting the model for cord blood metals (13.5% decrease in β) (Table 3).

Significant sex-interactions were observed between birth length and cognitive score (p=0.002) as well as birth length and language score (p=0.005). The magnitude of the associations was stronger in girls, as opposed to boys. All other p-values for the sex-interactions were ≥0.16. The variance inflation factors ranged from 1.0 to 4.5, suggesting the degree of multicollinearity is acceptable based on the less than 10 threshold value.⁵⁷

3.4 Sensitivity Analysis

To identify the critical windows of growth that are associated with neurodevelopmental test scores, we investigated the associations between changes in anthropometric measures (weight, length, and head circumference) between 1) birth and 12 months of age and 2) 12 months and 20–40 months of age. For changes between birth and 12 months, none of the anthropometric measures were associated with cognitive (all p≥0.15), language (all p≥0.14), or motor scores (all p≥0.71) (Supplemental Table 3). Similarly, between 12 months and 20–40 months, none of the anthropometric measures were related to BSID-III cognitive (all p≥0.30), language (all p≥0.87), and motor scores (all p≥0.69). One notable exception was observed between length change and motor score where one-unit increase in the length change was associated with a 0.13 increase in motor score (standard error=0.05, p=0.01). Among the associations assessed between the physical measures at 12 months and neurodevelopmental outcomes, weight was positively related to motor score (p=0.03) (Supplemental Table 3).

Child sex did not appear to be an important effect modifier. One notable exception was that an association between weight change in the second year of life and language score, where a stronger association between weight change in the second year and language score was seen for girls than for boys (β for interaction term=1.26, standard error=0.60, 95% confidence interval=0.09, 2.43, p=0.04; stratified analysis for girls β=0.34, standard error=0.47, 95% confidence interval=−0.59, 1.28, p=0.72; stratified analysis for boys β= −0.54, standard error=0.45, 95% confidence interval=−1.41, 0.34, p=0.23).

Use of generalized additive models did not suggest the presence of non-linear associations between the physical birth and growth parameters and neurodevelopmental scores.

4. Discussion

We explored the associations between anthropometric measures at birth and early childhood and neurodevelopmental test scores among Bangladeshi children assessed at age 20–40 months using the BSID-III. Our principal findings are fourfold. First, anthropometric measures at birth, including birthweight, length, and head circumference, were associated with cognitive, language, and motor scores. Second, at 20–40 months, anthropometric measures were positively associated with concurrent neurodevelopmental status. Third, increases in weight, length, and head growth between birth and 20–40 months were associated with higher neurodevelopmental scores. Finally, in general, prenatal exposure to environmental arsenic, manganese, and lead did not significantly affect the overall associations between anthropometric measures and neurodevelopmental test scores. These findings suggest that when taking into account high environmental metal exposures, prenatal and early childhood growth have strong associations with neurodevelopmental test scores in early childhood, possibly above and beyond the contribution of metal exposures.

Our findings are consistent with many previous investigations that have reported the associations between lower birthweight and head circumference and adverse neurodevelopmental outcomes in early childhood.^{12–14, 58–61} Whereas the evidence linking anthropometric measures and neurodevelopment is abundant for preterm infants,^{9–11, 58–61} the literature for infants with term-born or standard ranges of these physical measures, such as the ranges found in our population, is relatively sparse.^12–14 Most relevant to our investigation, a publication from the Avon Longitudinal Study of Parents and Children (ALSPAC), a population-based cohort study of children born in Avon, England, reported that among term-born children, larger head circumference at birth as well as greater increases in head circumference during the first year of age were associated with higher cognitive scores at 8 years of age.¹⁴ Importantly, the increase in head circumference between 1 and 4 years of age and between 4 and 8 years of age were not associated with cognitive test scores, supporting the hypothesis that in utero head growth may be a particularly sensitive developmental window for later neurodevelopment.¹⁴ Our study found that head circumference at birth, 20–40 months as well as growth trajectory between birth and 20–40 months were associated with neurodevelopmental test scores assessed at 20–40 months. We extend the existing literature by studying a vulnerable population that is exposed to high concentrations of environmental toxicants^{19, 24, 62} and highlighted the importance of considering early childhood growth in understanding the determinants of neurodevelopment.

In addition to anthropometric measured at specific time points, our study suggests that the rate of growth, or growth trajectory, in the early childhood period may be an important contributor to neurodevelopmental outcomes. Physical growth of the children is rapid from birth to 1 or 2 years of age. Between birth and 12 months of age, body weight triples, body length increases more than 50%, and brain increase to 75% of its adult size.⁶³ Between 12 months to 24 months of age, body weight and length increase more than 20% and head circumference increase about 5%.⁶⁴ Our study investigated whether growth parameters measured at 12 and 20–40 months as well as changes between birth and 12 months and between 12 and 20–40 months are associated with neurodevelopmental outcomes. Although we found that the relation of physical growth between both birth and 12 months as well as between 12 and 20–40 months, were not statistically significant, the magnitude of some of the beta coefficients was relatively large [e.g., β=0.68, standard error=0.36 for the association between weight change from birth to 12 months of age and language score (Supplemental Table 3)]. We interpret this to mean that there may be an important association between growth trajectory and neurodevelopmental test scores, but that our analyses lacked power to reach statistical significance due to small sample size.

There are many possible explanations for the association between early childhood growth trajectory and neurodevelopmental test scores. The early post-natal period may be an important window during which physical growth affects neurodevelopment. Many domains of brain development are completed within the early years of life.^{65, 66} These processes encompass neurogenesis, proliferation, synaptogenesis, with synaptic pruning and myelination occurring later in childhood and adolescence.⁶⁵ Between 7 and 9 months of age, substantial growth in head circumference is accompanied by rapid cerebral growth, myelination of the limbic system, enhanced associative pathways, and improved inhibitory control.⁶⁷ Poor physical growth may be a marker for disruption of these important steps in brain development.¹⁵

Poor prenatal growth and measures of growth trajectory in early childhood may also serve as indirect indicators of poor nutrition, poverty, or low socioeconomic status,^{52, 68} all of which are factors that affect brain development.^69–71 In our study population, 29.5% of our sample (229 out of 777 children) had a mean hematocrit level below the cutoff of 33% for iron deficiency anemia. Iron is one of the most important trace elements that is critical for proper brain growth and development in children due to its functional properties that are necessary for basic neuronal processes (e.g., myelination and neurotransmitter production).⁷² This observations suggests that iron status is a crucial covariate in the associations between growth and neurodevelopment. Accordingly, in this present study, child’s hematocrit level as well as maternal protein intake were adjusted in the statistical models to evaluate whether the associations between growth trajectory and neurodevelopmental test scores were dependent on these nutritional factors. The results from our multivariable-adjusted models confirmed that growth parameters are strongly related to neurodevelopment after accounting for these measures of maternal and child nutritional status.

Exposure to inorganic arsenic and other chemicals are major public health concerns as these toxicants are linked with numerous adverse health outcomes.^{19, 34, 73, 74} Anthropometric measures and neurodevelopmental test scores must be evaluated in the context of multiple chemicals in the environment, especially when the levels of these exposures are high. In our study, additional adjustments of cord blood arsenic, manganese, and lead did not materially change the associations between anthropometric parameters and neurodevelopmental test scores. However, one of the results, the association between birth weight and language score, showed 13.5% decrease in the β estimate after adjusting for cord blood metals. This finding suggests that substantial portion of the relationship between weight at birth and language score assessed at 2–3 years of age were explained by the exposure to environmental metals during pregnancy, suggesting that prenatal exposure to metals may be an important confounder.

We attempted to evaluate whether child sex modified the associations between anthropometric measures and neurodevelopmental test scores, as it is increasingly recognized growth trajectories differ by sex.^75–78 The strongest evidence for effect modification by child sex came from a stratified analysis between birth length and cognitive score. In stratified analyses, the relationship between birth length and language score was significant among girls (β=0.63, 95% confidence interval=0.38, 0.88, p<0.0001 for girls; β=0.22, 95% confidence interval=−0.03, 0.47, p=0.09 for boys).

Our study has several strengths. First, our study population provides a unique opportunity to study health outcomes in the context of high environmental metal exposures. Second, our prospective design allowed us to understand whether particular windows during which physical growth is more strongly associated with neurodevelopment. Finally, we were able to use individual measures of growth, biomarkers of environmental and nutritional exposures as well as individual level measures of important covariates, including exposure to environmental toxicants.

Some limitations warrant mention. As with all observational studies, we are limited in our ability to draw causal inferences between anthropometric measures and neurodevelopmental outcomes, and there may be residual confounding by unmeasured potential confounders (e.g., socioeconomic status or other nutritional measures not captured by our biomarkers or questionnaires). We were unable to evaluate whether growth was a mediator of the previously observed relationships between environmental metal exposures and neurodevelopmental test scores due to the small sample size. Mediation analysis would be an important next step to understand the pathways linking these important influences on neurodevelopment. Nevertheless, our findings underscored that anthropometric measures at birth as well as early childhood growth, are associated with neurodevelopmental status of 2- to 3-year-old children in the context of high environmental toxicant exposures. Further studies are needed to elucidate the pathways of neurodevelopment involving growth parameters and environmental chemicals.

5. Conclusions

Anthropometric measures at birth as well as growth trajectory are positively associated with neurodevelopmental test scores of 20–40-month-old children, even after accounting for environmental metal exposures. Studies evaluating mediation effects are needed to further understand these associations.

Highlights.

• Children affected by high levels of exposure to environmental metals were included

• Associations of anthropometric measures and neurodevelopmental status were assessed.

• Birth and growth parameters were associated with neurodevelopmental outcomes.

• These remained associated after accounting for metal exposures, such as arsenic.

Abbreviations

BSID-III

Bayley Scales of Infant and Toddler Development Third Edition

HOME

Home Observation for Measurement of the Environment