Neighbourhood socio‐economic environment predicts adiposity and obesity risk in children under two

Summary Background Neighbourhood socio‐economic environment (SEE) is associated with obesity in older children and adults, but little is known about this relationship in younger children. Breastfeeding is an important preventative of adiposity in childhood, but its relationship with neighbourhood SEE is unknown. Aims We assessed differences in adiposity and obesity in children before age two by neighbourhood SEE, controlling for family socio‐demographics and breastfeeding duration. Materials and Methods Family socio‐demographics, child body mass index z scores (BMIz), and breastfeeding duration were collected at periodic study visits from participants in PREVAIL (n = 245), a birth cohort in Cincinnati, OH. Addresses were assigned a Deprivation Index score, a validated measure of SEE, and dichotomized into highest SEE (least deprived quartile of scores) and not highest SEE (remaining quartiles). Longitudinal and Poisson models assessed differences in BMIz by SEE over the second year of life and obesity risk at age two, respectively (highest SEE, reference), while attenuation of obesity risk by breastfeeding duration was tested in mediation models. Results Residing outside of the highest SEE neighbourhoods was associated with an increased BMIz of 0.04 (95%CI 0.02, 0.06) per month of life and increased obesity risk at age two (aRR: 3.7, 95%CI 1.2, 16.2), controlling for family socio‐demographics. Breastfeeding duration attenuated >9% of the obesity risk attributable to SEE (mediated RR: 3.4, 95%CI 1.1, 14.8). Discussion In the PREVAIL Cohort, residing outside of the highest SEE neighbourhoods predicted a significant increase in BMIz and obesity risk in children before age two, a relationship that was partially mediated by breastfeeding duration. Conclusion Breastfeeding support may play an important role in reducing obesity rates in children in lower SEE neighbourhoods.


| INTRODUCTION
Nearly 14% of American children aged 2-5 have obesity, a rate that has more than doubled in the last 50 years. [1][2][3] In the United States (U.S.), children from low-and middle-income families are more likely to have obesity than their higher-income peers, and Black children have higher obesity rates than non-Hispanic White children in all age groups. 1,3,4 Reducing childhood obesity is a public health priority, as early obesity is a strong predictor of obesity in adolescence and adulthood and increases the risk of a wide range of metabolic, vascular, and endocrine disorders throughout life. [5][6][7] Evidence shows that weight status and obesity in school age children, adolescents, and adults are intimately connected to a complex network of social, structural, racial, and economic differences in neighbourhood environments. [8][9][10][11][12] While this relationship in children under two is understudied, socio-demographic measures associated with differences in child obesity risk, namely, race and family income, are correlated with one's residential neighbourhood 13,14 and factors that influence parent and sibling weight status also affect the weight status of younger children in the household. 15,16 Duration of breastfeeding, particularly exclusive breastfeeding, has been associated with reduced risk of obesity in childhood. 17,18 Disparities in breastfeeding initiation, duration, and exclusivity have been found by maternal race, income, and education level 19,20 and may, at least partially, explain differences in obesity rates by these socio-demographic factors.
While disparities in weight status have been found by race and income in older children and adults and these same factors predict both breastfeeding behaviours and neighbourhood socio-economic environment (SEE), no study in the U.S. has examined these relationships in early childhood as predictors of adiposity and obesity risk.
This project seeks to address this knowledge gap using a validated measure of neighbourhood SEE and family-level factors as predictors of adiposity and obesity, including how breastfeeding behaviours affect these relationships, in a well-characterized birth cohort. Identifying neighbourhood SEE as a predictor of early obesity could provide public health practitioners with a simple metric for focusing services and interventions in neighbourhoods with the greatest concentration of risk, potentially reducing health disparities through the lifespan.

| MATERIALS AND METHODS
This project is a secondary analysis using data from the Pediatric Respiratory and Enteric Virus Acquisition and Immunogenesis Longitudinal Cohort (PREVAIL), a prospective two-year birth cohort study in Cincinnati, OH funded by the U.S. Centers for Disease Control and Prevention (CDC). PREVAIL was approved by the Institutional Review Boards at the CDC and Cincinnati Children's Hospital Medical Center.
Recruitment, enrollment, and study methods have previously been described 21 ; methods relevant to this work are described here.
Women 18 years of age and older were provisionally enrolled in the third trimester of pregnancy, with final eligibility determined at a two-week post-partum home visit. Inclusion criteria for study mothers included delivery of a healthy, live-born, singleton infant at one of two urban study hospitals. Exclusion criteria included living more than 20 miles from the birth hospital, illicit drug use during pregnancy, major congenital anomaly in the infant, gestational age under 35 weeks, or HIV infection. Subject enrollment began in April of 2017, and data collection was completed in October 2020. No additional exclusion or inclusion criteria were applied for this analysis.
For this study, we focused on SEE, defined as aggregate socioeconomic indicators of an environment or neighbourhood, such as average income, education, or house value. Family residential address was reported at the baseline visit and updated, if changed, in REDCap data management software. 22 All PREVAIL subject addresses were geocoded to the census tract-level, generally aligned with urban neighbourhoods, and assigned a Deprivation Index score 23 using DeGAUSS software. 24 The Deprivation Index is a validated measure of SEE using six American Community Survey 25 census tract-level variables: (1) percentage of vacant homes, (2) median home value, and the percentage of adult population who (3) are without a highest school diploma, (4) have used any government social-services or income support within 12 months, (5) lack health insurance, and (6) meet the federal definition of poverty. 25 Values within each census tract were combined using a principal components analysis and standardized into a score between 0 and 1, with increasing scores representing increasing neighbourhood deprivation. 23 Differences in child health-related outcomes have been found by Deprivation Index score, including allcause hospitalization during the first year of life, emergency room use, and asthma incidence. 23,26,27 This study is novel in comparing early childhood obesity rates using this measure. Body mass index z scores (BMIz) from each study visit were calculated using the parameters provided by the CDC National Center for Health Statistics and the LMS method outlined by Flegal et al., 28,29 specific to child sex and age in months at the time of measurement. Imputation for erroneous entries (<0.5% of all entries) was performed by averaging the BMIz score from the two proximal visits. Sensitivity analysis found no significant differences in outcomes after removing these imputations. Obesity was defined as a BMIz score at or above the 95th percentile (BMIz ≥ 1.65) at age two per CDC obesity criteria. 1 For international comparison purposes, obesity rates at age two were also reported as defined by the International Obesity Task Force (IOTF) and the World Health Organization (WHO) 30,31 ; all analysis was performed using CDC criteria.

Quartiles of Deprivation
Breastfeeding duration and exclusivity (dates of cessation and formula introduction, respectively) were self-reported by the mother on quarterly study questionnaires. Duration of any breastfeeding was calculated in weeks from birth to the maternal reported date of breastfeeding cessation, censored at age two. Duration of exclusive breastfeeding was calculated in weeks from birth until the mother's reported date of the first formula introduction, censored at 6 months of age.

| Statistical analysis
Power for each outcome, calculated post-hoc, was based on the smallest sample size at each study visit for any outcome. Our study had at least 80% power to detect an effect size of ≥30% using any of our analytic methods.
To characterize the study population, family-level sociodemographics by categorical and binary SEE were compared using Fisher's exact test. Spearman correlations between categories of SEE (ranked 1-4, lowest to highest) and each family-level characteristic were calculated. Variables with a strong correlation (r ≥ 0.70) were considered collinear, and not considered for inclusion in multivariable models. Family-level variables that met non-collinearity criteria were used in all subsequent models unless otherwise indicated.
Our analytic approach was designed to identify and track differences in BMIz by SEE in early childhood and subsequent risk of obesity at age two. In line with published CDC findings of increased obesity risk for low and middle income compared to high-income young children, 4  analysis includes an American study population, all assessments of obesity risk were made using CDC standards. Relative risk of obesity at age two was calculated using Poisson regression in univariable and multivariable analysis, using the same variable selection approach previously described.
As lack of breastfeeding, particularly exclusive breastfeeding, has been associated with risk of obesity, 17,18 mediation analysis was undertaken to examine the effects of lack of breastfeeding, not as a confounder, but as a potential causal pathway of obesity. First, . All measures of BMIz were calculated using parameters associated with the child's age in months, and GEE models added child's age in months at the time of measurement as an interaction term. However, as differences associated with SEE appeared to increase over time, sensitivity analysis was performed by excluding this subset of subjects from analysis and re-running the linear regression models of BMIz, logistic models of obesity, and mediation models at age two. As no differences were found after these removals, results are reported using all subjects available at the specific study time-  38 In addition, the implications of these findings extend beyond childhood, as early obesity is associated with increased risk of obesity and co-morbid health conditions in adolescence and adulthood, 5,7 conditions that disproportionately affect populations of colour. 39,40 The significant differences in family-level sociodemographics by census tract, and the significant correlations with level of neighbourhood deprivation, provide a literal roadmap for identifying clusters of children at increased risk of obesity and other health disparities, while the finding that increased breastfeeding duration decreased risk of obesity in these neighbourhoods provides a method to at least partially address it. Thus, using the Deprivation Index to identify neighbourhoods for breastfeeding promotion and support may have consequential and long-lasting benefits for the communities most at-risk for a wide array of health disparities, including, but not limited to, a reduction in obesity rates.
Our analysis has some limitations. Our study focused on the predictive power of SEE regarding adiposity and obesity, as well as the influence of breastfeeding on obesity risk, ideally for use as a tool to identify loci and methods for health promotion. Thus, we did not test for interaction between SEE and the family-level predictors, which may provide more insight into causality. Furthermore, physical environment and domestic stressors, such as neighbourhood crime, financial hardship and food security have been significantly associated with increased risk of obesity in older children. 9 As a secondary analysis of PREVAIL, which aimed to characterize the natural history of respiratory and enteric viral illness in young children, potentially important variables for understanding differences in breastfeeding behaviours, such as timing of maternal return-to-work, availability of lactation support, or employment type, were not available and not included in any of the breastfeeding models. As this and other research shows that improving breastfeeding duration may reduce risk of obesity, studies to identify barriers to breastfeeding in these populations are needed to strategize best practices for improving these outcomes.
Finally, while PREVAIL's final enrollment was representative of the birth hospitals from which we recruited, 21 our cohort was overrepresented by low-income families compared to our region as a whole. 43 This both widened the quartile ranges for Deprivation Index strata and drove the cut-points for the categories down, resulting in more racial and economic diversity in the higher SEE quartiles than is representative of our area. Despite this possible misclassification of lower SEE neighbourhoods into our higher SEE categories, we found a strong relationship between SEE and child adiposity and obesity risk.
Studies including a larger, more representative sample from a wider geographical area are needed to explore these relationships more fully.

| CONCLUSIONS
Our approach showed that increased early childhood adiposity and obesity risk are associated with residing outside of the most affluent neighbourhoods as measured by level of material neighbourhood deprivation. We identified shorter breastfeeding durations outside of the highest SEE neighbourhoods as a partial mediator for obesity disparities. Our findings suggest that focusing breastfeeding support and promotion in the highest deprivation neighbourhoods may be an important first step to mitigate early childhood obesity and reduce significant health disparities in at-risk populations, both in childhood and later in life.