Abstract
Despite obesity being a major concern for both children and adults in the United States today, there are few successful childhood interventions that curb obesity later in life. The objective of the current study is to identify childhood predictors of adult obesity at multiple levels in a large longitudinal sample of participants from an economically disadvantaged childhood cohort. 1,065 participants (93% Black) from the Chicago Longitudinal Study were interviewed as part of a 30-year follow-up between 2012 and 2017. Parent involvement, school quality, neighborhood human capital, socioemotional learning skills, and achievement motivation assessed before age 12 years were examined as predictors of Body Mass Index (BMI) at age 35 years. Child neighborhood human capital and socioemotional learning skills predicted a lower BMI in adulthood and a decreased likelihood of being classified as obese; when separately analyzed by sex, both neighborhood human capital and higher socioemotional learning skills predicted a decreased likelihood of obesity for males and females. Being female and higher birthweight were associated with larger adult BMI. Socioemotional learning and neighborhood human capital in childhood consistently predict a decreased likelihood of being obese at age 35 in this predominately Black sample. Future obesity intervention/prevention programs should aim to bolster childhood socioemotional learning resources and neighborhood capital.
Keywords: obesity, socioemotional learning, neighborhood context
INTRODUCTION
The prevalence of adult obesity across the United States has increased over the last three decades. This is of special concern for Black American adults, who have consistently had one of the highest prevalence rates in the U.S. (today it is 47%, with that for women being 20 points higher than for men).1 This epidemic extends to young children, as there has been a 38% increase in obesity rates for children ages 2 to 5 years over the last ten years.2 These high obesity rates are a public health concern, since people with obesity have increased risks for cardiovascular disease, metabolic syndrome, discrimination, and psychopathology.3–6
Given the risks associated with obesity across the lifespan, especially for Black Americans, identifying childhood predictors is a high priority for prevention. There are multiple reasons why Black children and adults suffer from higher obesity rates than most other racial groups in the United States. These include discrimination and structural racism, which are associated with undesirable mental and physical health outcomes7; differences in built environments (e.g., less green space in neighborhoods)8; and decreased availability of healthy foods in neighborhoods9. Although various childhood interventions have been implemented to reduce obesity, a review of early childhood programs found under half were effective, with mixed results at follow-up.10 Additionally, research has demonstrated that children classified as overweight or obese frequently from preschool until age 11 are more likely to be classified as such beginning in adolescence (at age 12), indicating the importance of targeting this time period for research and intervention.11
Recent research has begun to study ecological influences on obesity to address the limited scope of individual, physiological, and parenting predictors in prior studies. Utilizing Bronfenbrenner’s ecological systems theory (EST),12,13 the importance of examining family, school, and neighborhood influences on obesity in childhood and beyond is emphasized.14,15 Though more studies are implementing EST to understand pathways to obesity, few studies to date have examined multiple ecological influences from childhood to adulthood. Potential ecological predictors of adult obesity are family/school context, cognitive and socioemotional skills, and neighborhood (dis)advantage. Family/school context factors (e.g., parent involvement) and behavioral and cognitive strategy implementation have been linked to reduced obesity in childhood.16 Increasingly more work is examining the neighborhood (dis)advantage impact on childhood obesity, though the findings have been mixed.17 Most current longitudinal studies are limited in that they only use medical and sociodemographic data, under-measure behavioral and psychological factors, and/or use a predominately white sample.18–20 Given the higher rates of obesity prevalence for Black adults, especially females,1 identifying protective childhood predictors of later physical health in these populations is a priority.
Previous findings in the Chicago Longitudinal Study (CLS), which has tracked the life-course development of a low-income, predominately Black sample since early childhood in the 1980s, indicate that factors such as school quality, parental school involvement, and early socioemotional skills predict educational attainment, health compromising behavior, economic well-being, and mental health in young adulthood.21–23 However, no CLS studies to date have examined physical health outcomes, neighborhood factors as a focal predictor, or midlife health status.
This study examines how childhood indicators (before age 12) predict obesity at age 35 by addressing the following questions: (1) Do family/school contexts, neighborhood in childhood, and psychosocial skills predict BMI in adulthood? and (2) Do these predictive patterns vary by sex or by school-level poverty? We predicted better school quality and parent involvement would predict a healthier BMI in adulthood, as would growing up in an advantaged neighborhood; additionally, we predicted both socioemotional learning and achievement motivation would predict better BMI outcomes in adulthood, with differences varying by sex and school-level poverty subgroups.
METHODS
Sample and Design
The CLS is a prospective longitudinal study with 1,539 participants born between 1979 and 1980 (93% Black, 7% Hispanic) who grew up in low-income neighborhoods in Chicago and attended school district-led early childhood programs in preschool and/or kindergarten (1983–1986)24,25. These participants have been followed into early midlife. The present study includes 1,065 (69%) from the original 1,539 participants interviewed for a 30-year follow-up. 704 participants (66.1%) in the current sample attended the Child-Parent Center (CPC) program beginning in preschool; the remaining 361 are from a demographically-matched comparison group. CPC spans preschool through 3rd grade and provides education and family support services, but is not a focus of the current study26,27.
Analyses comparing background characteristics of those who either did not provide enough data to calculate a BMI in adulthood or who did not complete the adult survey (n = 474) evinced that non-participants were more likely to be male and not participate in CPC (Appendix A). Non-participants were also more likely to have lower parent involvement, achievement motivation, and socioemotional learning scores.
Follow-up at Age 35 Years
Health data were collected as part of a larger interview at the 30-year follow-up. The interview included 130 questions about well-being, health, and life history. Participants completed the survey between 2012 and 2017. The majority of respondents (81%) completed the interview via telephone through the University of Minnesota and/or Northern Illinois University. Study approval was granted by IRBs at University of Minnesota and Northern Illinois University.
Body Mass Index (BMI)
Participants’ body mass indices (BMI) were assessed via self-report of height and weight at the 30-year follow-up. BMI is a body fat estimate based on height (in meters) divided by weight (in kilograms) squared. Overweight is defined as a BMI as 25 m/kg2 or higher, and obesity as 30 m/kg2 or higher28. This measure was highly correlated (r = 0.85) with an in-person BMI measurement conducted with a participant subsample. A full description of this measure is in Appendix C.
Family and School Contexts
Family and school contexts were assessed via parent involvement and school quality. Parent involvement was measured by teacher-reported ratings of “parent’s participation in school” on a 5-point scale from 1 (poor/not at all) to 5 (excellent/much). Answers to this question were averaged over first through third grade, resulting in a score between 1 and 529.
School quality was operationalized dichotomously as whether or not a student attended magnet school in grades 4–8 or if they attended a school where 40% or more students were above grade level in reading in 5th grade30.
Neighborhood Human Capital (NHC)
An indicator of age 10 neighborhood human capital– percent of persons 25 years or older with at least a 4-year degree by birth census tract data – was included to address the emerging focus in the health literature on the influence of neighborhood contexts31. The percentages were multiplied by 100 to yield a possible range of 0 to 100. Fifty-one cases were mean-imputed by sex and whether or not the participant attended CPC.
Psychosocial Skills
Two psychosocial measures were utilized. Socioemotional learning (SEL) skills were assessed via teacher ratings of classroom adjustment from grades 4–6 (ages 10 to 12) on six items: 1) concentrates on work, 2) follows direction, 3) is self-confident, 4) participates in group discussion, 5) gets along well with others, and 6) takes responsibility for actions. Each item was rated on a 5-point scale from 1 (poor/not at all) to 5 (excellent/very much). Ratings were summed and averaged across grades, ranging from 6 to 3023.
Achievement motivation was assessed using a score obtained from teacher questionnaires. Motivation was assessed both in kindergarten and in first grade; z-scores were created for both years and then averaged across the two years. If students did not have a kindergarten score, their first-grade score was used32.
Early Childhood, Contextual, and School-age Covariates
Many child and family covariates were included in the model, measured from birth and school records, and parent and participant report. A full list of these covariates and their descriptions are in Appendix D. Descriptive statistics of these variables are shown in Table 1. Additionally, subgroup analyses by school-level poverty (SLP) groups were also conducted using a dichotomous variable indicating whether or not the participant attended a school that had a 60% or higher school attendance area poverty rate.
Table 1.
Descriptive statistics for the CLS sample included in BMI analysis (N = 1065)
| Study Variables | N, Percent | M (SD) |
|---|---|---|
| Early Childhood Dichotomous Risk Variables (Ages 0 to 5) | ||
| Female | 566, 53.1% | -- |
| Black | 993, 93.2% | -- |
| Participation in the CPC program in preschool | 704, 66.1% | -- |
| Participation in CPC program beyond preschool | 608, 57.1% | |
| Mother under age 18 years at time of child’s birtha | 163, 15.3% | -- |
| Mother did not complete high schoola | 558, 52.4% | -- |
| Mother had some college education | 133, 12.5% | -- |
| Child in single-parent householda | 804, 75.5% | -- |
| Four or more children in householda | 178, 15.7% | -- |
| Participation in public aid (AFDC)a | 648, 60.8% | -- |
| Child eligible for free school luncha | 883, 82.9% | -- |
| Mother not employed full- or part-timea | 695, 65.3% | -- |
| Reported family history of circulatory diseases | 364, 34.2% | -- |
| Reported family history of respiratory diseases | 98, 9.2% | -- |
| Birth weight (lbs.; continuous) | -- | 6.80 (1.26) |
| Contextual Childhood Variables (Ages 0 to 12) | ||
| Neighborhood disadvantage at birth | -- | 0.39 (0.12) |
| ACEs age 6 to 10 | -- | 0.92 (1.11) |
| Recorded child case protection history before preschool | 187, 17.6% | -- |
| Reported chronic health issue (ages 0 to 10) | 34, 3.2% | -- |
| Reported asthma diagnosis (ages 0 and 12) | 148, 13.9% | -- |
| School-Related Covariates (Ages 3 to 12) | ||
| Kindergarten readiness (Iowa Tests of Basic Skills) | -- | 48.01 (8.80) |
| Year 3 reading comprehension (ages 8 to 9) | -- | 98.11 (15.85) |
| Parent expectations (years of education expected for child) | -- | 14.41 (1.62) |
| Special education enrollment (ages 6 to 14) | 147, 13.8% | -- |
| School commitment (ages 11 to 12) | -- | 50.62 (5.39) |
| Focal Study Variables | ||
| Parent involvement (ages 6 to 9) | -- | 2.64 (0.99) |
| School quality (ages 10 to 14) | 128, 12.0% | -- |
| Childhood neighborhood human capital (age 10) | -- | 5.83 (6.76) |
| Socioemotional learning skills (ages 10 to 12) | -- | 18.94 (4.51) |
| Achievement motivation (z-scored; ages 5 to 6) | -- | .081 (0.98) |
| Subgroup Analyses Variable | ||
| School-level poverty | 801, 75.2% | -- |
CLS = Chicago Longitudinal Study. CPC = Child-Parent Centers. AFDC = Aid to Families with Dependent Children. If a variable is dichotomous, the N for how many participants received a “1” on that variable are presented, along with the percentage of the sample. If a variable is continuous, the means and standard deviations are presented.
Included in family risk index, all measured from ages 0 to 3 years old. In-depth descriptions of all measures can be found in Appendix C. Participants completed the survey in 2012 (11.9%), 2013 (12.9%), 2014 (3.6%), 2015 (38.9%) 2016 (24.7%), or 2017 (6.0%).
Statistical Analysis
Analyses were conducted using IBM SPSS Statistics, Version 25 and the R software/environment. Imputation processes were conducted for a few focal variables; missing data ranged from 0% to 20%, and the imputation process primarily involved multiple imputation via Expectation Maximization (EM) algorithm in LISREL33. The data analytic plan was to use both a continuous BMI score and a dichotomous obesity variable (BMI ≥ 30) as the outcome variables. Our sample size of 1,065 was more than sufficient (minimum = 700 to 800) to detect an effect size (standardized coefficient) of .05 with 80% power. Linear and logistic regression and analyses were conducted to examine the associations of adult BMI overall and by sex and SLP with parent involvement, school quality, NHC, SEL, and achievement motivation. Variables were included in five steps according to age order.
RESULTS
Table 2 shows the descriptive statistics for BMI and obesity for the overall sample and by sex and school-level poverty (SLP) subgroups. Females had a higher rate of adult obesity (51.2%) than males (38.7%), and those in the non-SLP group had moderately lower obesity rates than those in the SLP group.
Table 2.
BMI descriptive statistics and obesity frequencies by demographic subgroups
| Continuous BMI | Obesity | |
|---|---|---|
| Study Subgroup | M (SD) | N, Percent |
| Total sample (N = 1065) | 30.49 (6.89) | 483, 45.4% |
| Sex | ||
| Female (n = 566) | 31.70 (7.71) | 290, 51.2% |
| Male (n = 499) | 29.11 (5.51) | 193, 38.7% |
| School-level Poverty | ||
| 60% or higher poverty rate in school attendance area (n = 801) | 30.49 (6.93) | 356, 44.4% |
| Less than 60% poverty rate in school attendance area (n = 264) | 30.47 (6.78) | 127, 48.1% |
The range of BMIs for the entire sample was 17.34 to 68.14. For females, the range was 17.34 to 68.14; for males, the range was 17.97 to 55.08. For those in a high school-level poverty area, the range of BMIs was 17.34 to 68.14; for non-school-level poverty areas, the range was 17.36 to 64.7. Obesity is defined as a BMI of greater than or equal to 30.
Continuous BMI at Age 35
Family and school contexts.
Table 3 shows the unstandardized coefficients, standard errors, and standardized coefficients for the linear regression. Model 4 introduces both parent involvement and school quality; neither predicts a higher or lower age 35 BMI after accounting for all other covariates. Of the covariates, being female, higher birthweight, family history of respiratory diseases, and a childhood asthma diagnosis are associated with a higher adult BMI.
Table 3.
Linear regression with standardized and unstandardized coefficients predicting age 35 BMI (N = 1,065)
| Model 1 | Model 2 | Model 3 | Model 4 | Model 5 | ||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Variables | b and SE | B | b and SE | B | b and SE | B | b and SE | B | b and SE | B |
| Female | 2.59** (0.43) | 0.187 | 2.78** (0.43) | 0.201 | 2.74** (0.44) | 0.198 | 2.71** (0.44) | 0.196 | 2.91** (0.45) | 0.211 |
| Black | 0.48 (0.87) | 0.017 | 0.40 (0.87) | 0.015 | 0.52 (0.88) | 0.019 | 0.50 (0.89) | 0.018 | 0.20 (0.89) | 0.007 |
| CPC participation in preschool | −0.79 (0.48) | −0.054 | −0.76 (0.48) | −0.052 | −0.66 (0.50) | −0.045 | −0.66 (0.50) | −0.045 | −0.65 (0.50) | −0.045 |
| CPC participation beyond preschool | −0.14 (0.45) | −0.010 | −0.23 (0.45) | −0.016 | −0.25 (0.46) | −0.018 | −0.29 (0.46) | −0.021 | −0.28 (0.46) | −0.020 |
| Birthweight (lbs.) | 0.39* (0.17) | 0.071 | 0.39* (0.17) | 0.071 | 0.40** (0.17) | 0.073 | 0.40* (0.17) | 0.073 | 0.39* (0.17) | 0.071 |
| Mother less than 18 when child is born | 0.41 (0.63) | 0.022 | 0.36 (0.63) | 0.019 | 0.35 (0.63) | 0.018 | 0.35 (0.64) | 0.018 | 0.41 (0.64) | 0.022 |
| Mother did not complete high school | −0.60 (0.49) | −0.044 | −0.59 (0.49) | −0.042 | −0.58 (0.49) | −0.042 | −0.56 (0.49) | −0.040 | −0.60 (0.49) | −0.043 |
| Mother had some college education | −0.36 (0.70) | −0.017 | −0.18 (0.70) | −0.009 | −0.12 (0.70) | −0.006 | −0.16 (0.71) | −0.008 | −0.17 (0.71) | −0.008 |
| Child in single parent household | −0.14 (0.53) | −0.009 | −0.17 (0.53) | −0.011 | −0.18 (0.53) | −0.012 | −0.17 (0.53) | −0.011 | −0.19 (0.53) | −0.012 |
| Four or more children in household | 0.94 (0.58) | 0.051 | 0.72 (0.58) | 0.039 | 0.73 (0.58) | 0.040 | 0.79 (0.58) | 0.043 | 0.76 (0.58) | 0.041 |
| Participation in public aid | −0.32 (0.61) | −0.023 | −0.31 (0.60) | −0.022 | −0.31 (0.60) | −0.022 | −0.29 (0.60) | −0.020 | −0.31 (0.60) | −0.022 |
| Child eligible for free school lunch | 0.23 (0.60) | 0.013 | 0.27 (0.59) | 0.015 | 0.24 (0.60) | 0.013 | 0.27 (0.60) | 0.015 | 0.26 (0.60) | 0.014 |
| Mother not employed full or part time | 0.25 (0.60) | 0.017 | 0.17 (0.59) | 0.012 | 0.17 (0.59) | 0.012 | 0.19 (0.60) | 0.013 | 0.19 (0.60) | 0.013 |
| Family history of circulatory diseases | 0.74 (0.45) | 0.051 | 0.72 (0.45) | 0.050 | 0.72 (0.45) | 0.050 | 0.70 (0.45) | 0.048 | 0.77 (0.45) | 0.053 |
| Family history of respiratory diseases | 2.36** (0.72) | 0.099 | 2.00** (0.73) | 0.084 | 1.98** (0.73) | 0.083 | 2.01** (0.73) | 0.085 | 1.95** (0.73) | 0.082 |
| Neighborhood disadvantage at birth | −1.88 (2.14) | −0.031 | −1.76 (2.15) | −0.029 | −1.83 (2.16) | −0.031 | −1.72 (2.16) | −0.029 | ||
| ACEs (ages 6 to 10) | −0.03 (0.20) | −0.005 | −0.01 (0.20) | −0.002 | −0.00 (0.20) | 0.000 | −0.01 (0.20) | −0.001 | ||
| Neighborhood human capital (age 10) | −0.12** (0.03) | −0.117 | −0.12** (0.03) | −0.117 | −0.12** (0.04) | −0.118 | −0.12** (0.04) | −0.118 | ||
| Any child case protection history after CPC | 0.02 (0.57) | 0.001 | −0.02 (0.58) | −0.001 | −0.02 (0.58) | −0.001 | −0.03 (0.58) | −0.002 | ||
| Reported chronic health issues (ages 0 to 10) | −1.72 (1.21) | −0.044 | −1.69 (1.22) | −0.043 | −1.75 (1.22) | −0.045 | −1.73 (1.22) | −0.044 | ||
| Reported asthma diagnosis (ages 0 to 12) | 2.10** (0.62) | 0.106 | 2.09** (0.62) | 0.105 | 2.10** (0.63) | 0.106 | 2.14** (0.63) | 0.108 | ||
| Kindergarten readiness (ages 5 to 6) | −0.03 (0.03) | −0.036 | −0.03 (0.03) | −0.040 | −0.03 (0.03) | −0.035 | ||||
| Reading comprehension (ages 8 to 9) | 0.00 (0.02) | 0.010 | 0.00 (0.02) | 0.008 | 0.02 (0.02) | 0.038 | ||||
| Achievement motivation (ages 5 to 6) | 0.09 (0.25) | 0.012 | 0.09 (0.26) | 0.012 | 0.26 (0.27) | 0.038 | ||||
| Parent involvement (ages 6 to 9) | −0.01 (0.24) | −0.001 | 0.12 (0.24) | 0.018 | ||||||
| Parent expectations (years of education) | 0.09 (0.13) | 0.022 | 0.13 (0.13) | 0.030 | ||||||
| School quality (ages 10 to 14) | 0.34 (0.69) | 0.016 | 0.25 (0.69) | 0.012 | ||||||
| Special education enrollment (ages 6 to 14) | 0.11 (0.64) | 0.006 | 0.08 (0.64) | 0.004 | ||||||
| Socioemotional learning (ages 10 to 12) | −0.18** (0.06) −0.117 | |||||||||
| School commitment (ages 11 to 12) | 0.02 (0.04) 0.013 | |||||||||
| Constant | 26.22** (1.68) | 27.54** (1.86) | 28.21** (2.46) | 27.09** (3.08) | 27.55** (3.47) | |||||
| R2 | 0.062 | 0.084 | 0.085 | 0.086 | 0.093 | |||||
| F Statistic | 4.62 | 4.55 | 4.02 | 3.46 | 3.53 | |||||
Note: Values in parentheses represent standard errors.
p <.01
p <.05. Table partially created with stargazer R package40. CPC = Child-Parent Centers.
Childhood neighborhood human capital.
Childhood NHC was introduced in model 2 and predicts a lower BMI; a one-percentage point increase in NHC is related to a 0.12 BMI reduction in adulthood, accounting for all other variables in the model.
Psychosocial skills.
Socioemotional learning (SEL) is introduced in Model 5 and predicts a lower BMI in adulthood; a one-point increase in SEL is related to a 0.18 reduction in adult BMI, accounting for all other variables in the model. Achievement motivation, introduced in Model 3, is not associated with BMI.
Obesity at Age 35
Family and school contexts.
Table 4 shows the odds ratios (ORs), confidence intervals, and model fit statistics for the logistic regression predicting obesity in adulthood. Parent involvement and school quality are both introduced in Model 4. Neither are related to adult obesity. Of the covariates, being female, a higher birthweight, and a childhood asthma diagnosis are all related to increased odds of having obesity in adulthood.
Table 4.
Logistic regression odds ratio and confidence intervals predicting age 35 obesity (BMI > 30; N = 1,065)
| Models | |||||
|---|---|---|---|---|---|
| Variables | 1 | 2 | 3 | 4 | 5 |
| Female | 1.68** (1.31, 2.17) | 1.75** (1.35, 2.27) | 1.69** (1.30, 2.20) | 1.68** (1.28, 2.20) | 1.80** (1.37, 2.37) |
| Black | 1.21 (0.72, 2.05) | 1.20 (0.70, 2.06) | 1.25 (0.73, 2.17) | 1.25 (0.73, 2.16) | 1.13 (0.66, 1.97) |
| CPC participation in preschool | 0.83 (0.63, 1.10) | 0.83 (0.62, 1.11) | 0.83 (0.62, 1.13) | 0.83 (0.62, 1.13) | 0.83 (0.61, 1.13) |
| CPC participation beyond preschool | 0.96 (0.73, 1.26) | 0.93 (0.71, 1.23) | 0.91 (0.69, 1.20) | 0.90 (0.68, 1.18) | 0.90 (0.68, 1.19) |
| Birthweight (lbs.) | 1.13* (1.02, 1.25) | 1.13* (1.02, 1.25) | 1.13* (1.02, 1.25) | 1.13* (1.02, 1.25) | 1.12* (1.01, 1.25) |
| Mother less than 18 when child is born | 1.05 (0.72, 1.53) | 1.05 (0.72, 1.54) | 1.03 (0.70, 1.52) | 1.04 (0.71, 1.52) | 1.06 (0.72, 1.56) |
| Mother did not complete high school | 0.92 (0.69, 1.23) | 0.94 (0.70, 1.26) | 0.95 (0.71, 1.28) | 0.95 (0.71, 1.29) | 0.94 (0.70, 1.27) |
| Mother had some college education | 0.93 (0.61, 1.42) | 0.96 (0.62, 1.46) | 0.95 (0.62, 1.46) | 0.98 (0.63, 1.50) | 0.97 (0.63, 1.50) |
| Child in single parent household | 0.96 (0.70, 1.32) | 0.97 (0.70, 1.33) | 0.96 (0.70, 1.33) | 0.96 (0.69, 1.32) | 0.95 (0.69, 1.31) |
| Four or more children in household | 1.18 (0.84, 1.67) | 1.11 (0.78, 1.57) | 1.12 (0.79, 1.59) | 1.13 (0.79, 1.61) | 1.12 (0.79, 1.60) |
| Participation in public aid | 0.81 (0.57, 1.17) | 0.81 (0.56, 1.17) | 0.82 (0.57, 1.18) | 0.82 (0.57, 1.18) | 0.81 (0.56, 1.17) |
| Child eligible for free school lunch | 1.15 (0.81, 1.65) | 1.17 (0.82, 1.69) | 1.18 (0.82, 1.71) | 1.19 (0.83, 1.72) | 1.19 (0.83, 1.72) |
| Mother not employed full or part time | 1.24 (0.87, 1.78) | 1.23 (0.86, 1.77) | 1.25 (0.87, 1.79) | 1.26 (0.88, 1.82) | 1.27 (0.88, 1.84) |
| Family history of circulatory diseases | 1.20 (0.92, 1.57) | 1.22 (0.93, 1.60) | 1.21 (0.93, 1.59) | 1.20 (0.92, 1.58) | 1.23 (0.93, 1.62) |
| Family history of respiratory diseases | 1.44 (0.94, 2.22) | 1.35 (0.87, 2.10) | 1.32 (0.85, 2.06) | 1.36 (0.87, 2.12) | 1.34 (0.86, 2.10) |
| Neighborhood disadvantage at birth | 0.68 (0.18, 2.57) | 0.73 (0.19, 2.78) | 0.67 (0.17, 2.55) | 0.68 (0.17, 2.62) | |
| ACEs (ages 6 to 10) | 0.92 (0.82, 1.04) | 0.92 (0.81, 1.04) | 0.92 (0.81, 1.03) | 0.91 (0.81, 1.03) | |
| Neighborhood human capital (age 10) | 0.96** (0.94, 0.99) | 0.96** (0.94, 0.99) | 0.96** (0.93, 0.99) | 0.96** (0.93, 0.99) | |
| Any child case protection history after CPC | 0.85 (0.59, 1.20) | 0.86 (0.60, 1.22) | 0.85 (0.59, 1.21) | 0.84 (0.59, 1.20) | |
| Reported chronic health issues (ages 0 to 10) | 0.66 (0.31, 1.39) | 0.66 (0.31, 1.39) | 0.64 (0.29, 1.35) | 0.64 (0.29, 1.36) | |
| Reported asthma diagnosis (ages 0 to 12) | 1.66** (1.14, 2.43) | 1.65** (1.13, 2.42) | 1.68** (1.15, 2.47) | 1.70** (1.16, 2.50) | |
| Kindergarten readiness (ages 5 to 6) | 0.99 (0.97, 1.01) | 0.99 (0.97, 1.01) | 0.99 (0.97, 1.01) | ||
| Reading comprehension (ages 8 to 9) | 1.01 (1.00, 1.02) | 1.01 (1.00, 1.02) | 1.01 (1.00, 1.02) | ||
| Achievement motivation (ages 5 to 6) | 1.01 (0.87, 1.18) | 1.04 (0.89, 1.22) | 1.10 (0.94, 1.30) | ||
| Parent involvement (ages 6 to 9) | 0.92 (0.80, 1.07) | 0.96 (0.83, 1.12) | |||
| Parent expectations (years of education) | 0.99 (0.91, 1.07) | 1.00 (0.92, 1.08) | |||
| School quality (ages 10 to 14) | 1.32 (0.87, 2.00) | 1.28 (0.84, 1.95) | |||
| Special education enrollment (ages 6 to 14) | 0.99 (0.67, 1.47) | 0.98 (0.66, 1.46) | |||
| Socioemotional learning (ages 10 to 12) | 0.94** (0.90, 0.98) | ||||
| School commitment (ages 11 to 12) | 1.01 (0.98, 1.04) | ||||
| Constant | 0.22*** (0.08, 0.60) | 0.33* (0.10, 1.05) | 0.23* (0.05, 1.04) | 0.34 (0.05, 2.25) | 0.36 (0.04, 3.04) |
| −2 Log Likelihood | 1432.65 | 1411.52 | 1409.14 | 1406.25 | 1396.23 |
| Nagelkerke R2 | .043 | .068 | .071 | .074 | .086 |
Note:
p <.01
p <.05. Any confidence intervals that do not cross 1.00 are considered statistically significant. Table partially created with stargazer R package40. CPC = Child-Parent Centers.
Childhood neighborhood human capital.
Model 2 introduces childhood NHC to the model, which is related to a lower likelihood of being obese in adulthood – a one-point increase in age 10 NHC is associated with a 4% decrease in the odds of being obese in adulthood (Model 5 OR = 0.96, 95% CI = 0.93, 0.99).
Psychosocial skills.
Achievement motivation is not associated with obesity in any of the models. SEL is related to obesity; a one-point increase in SEL scores between ages 10 and 12 is associated with a 6% decrease in the odds of being obese in adulthood (OR = 0.94, 95% CI = 0.90, 0.98).
Subgroup Analyses
Different patterns emerged for males and females. The linear regression analyses (Appendix E) indicate higher SEL and NHC for males, but only NHC for females, predict lower BMI in adulthood. Attending a higher quality school was associated with a 1.82 higher BMI for males. Appendix F shows the logistic regression predicting obesity for both males and females. A one-point change in SEL skills is associated with 6% decreased odds of being obese at age 35 for both sexes, and NHC is also associated with 5% decreased odds for males and 3% decreased odds for females of being obese.
Differences by attending a school with a high school attendance poverty rate were found (60% or more students in that school were in poverty; Appendix G). For the school-level poverty group (SLP), higher SEL skills predict a lower adult BMI. For non-SLP participants, neighborhood human capital predicts a lower BMI. Appendix H shows the logistic regression by SLP status. For the SLP group, higher NHC and SEL skills predict a lower likelihood of being obese in adulthood (4% and 5%, respectively), while the non-SLP group is more likely to not have obesity with a higher NHC and more parent involvement.
Robustness Analyses
Three robustness analyses were conducted. The first model added an inverse propensity score weight to the regression using early child risk factors to predict the probability of being in the final sample (Appendix I). The second model included a composite risk score of age 0 to 3 SES indicators instead of separate risk indicators (see Table 1). The third robustness analysis included age at time of survey completion, because participants were between ages 31 and 37 at the time of completing the survey. Age was not significant. The patterns of findings from those models did not differ, which support the robustness of these findings.
DISCUSSION
This study broadened the examination of childhood factors that predict adult physical health by studying associations between childhood psychosocial skills, family/school contexts, and neighborhood human capital and adult BMI and obesity in a large urban cohort. Rather than focusing on few childhood factors and their concurrent association with childhood obesity, the longitudinal design enabled us to examine relations between child predictors to adult health. While past studies have examined the relation between these factors and childhood obesity today, this is one of the first studies to combine childhood family, school, neighborhood, and psychosocial skills in one model to predict adult obesity.
This study utilized Bronfenbrenner’s EST model12,13 and hypothesized higher parent involvement, school quality, SEL, achievement motivation, and NHC in childhood would predict a lower BMI and decreased likelihood of being obese at age 35 years. This study focused on childhood factors due to the high association between childhood and adult BMI11. Our findings indicate SEL and NHC were the strongest predictors of adult obesity. Even when accounting for other childhood measures such as achievement motivation and childhood test scores, a one-standard-deviation increase in SEL predicted a 24% reduced likelihood of adult obesity. This is consistent with previous work conducted with the CLS, where higher SEL was linked to lower depressive symptoms23 and healthier behaviors21 in young adulthood. The findings from the present study are the first to extend the relation between SEL and health into middle adulthood, highlighting the need to examine the mechanisms through which SEL in childhood can extend to health through the lifespan.
Building on the literature examining the impact of neighborhoods on obesity17, our findings consistently indicated neighborhood human capital is a protective factor for obesity in middle adulthood. Our NHC measure indicated the education level of the child’s neighborhood at age 10. Its consistent prediction of reduced adult obesity is notable – a higher childhood NHC predicted a decreased likelihood of being obese for participants that attended schools in low-poverty areas, indicating the protective nature of NHC. While previous work has examined neighborhood correlates of child obesity or early person-centered predictors of adult obesity17–20, our study extends the literature by utilizing a longitudinal design to examine the association between NHC and adult obesity. This finding suggests a broader multilevel perspective is needed to understand the alterable childhood correlates of adult obesity.
The subgroup analyses evinced different predictive patterns, depending both on the group of interest and the outcome. For continuous BMI, NHC and SEL predicted a difference in BMI for males; for females, NHC and SEL predicted a decreased likelihood of being obese, but only NHC predicted a lower BMI. This discrepancy between predictors for BMI and obesity could be explained by differences in obesity rates by sex; over half the females in our sample were classified as obese, while only 39% of the males were. Examining the predictive differences for children attending school in high- or low-poverty areas was an innovation of this study. SEL skills predicted a better BMI and decreased likelihood of obesity for children attending schools in neighborhoods with more poverty, but NHC only predicted a decreased likelihood of being obese, not a lower BMI, for this group. Additionally, higher levels of parent involvement predicted a lower chance of being obese for students who did not attend schools in a low-poverty area.
Though SEL and NHC consistently predicted healthier weight in adulthood, elementary school parent involvement, school quality, and achievement motivation did not. Parent involvement did predict a reduced risk of obesity for participants in non-low poverty schools; parents in these schools might face fewer economic barriers, so they are able to be more involved. Our measure examined parent involvement in early elementary school; other age periods might have more salience. Attending a higher quality school was associated with a higher BMI for males. Given the lower rates of obesity in the male sample, attending a higher quality school could be associated with access to more nutrition. Finally, teacher-assessed achievement motivation may not be relevant to health behaviors beyond school-age. Though these predictors link to many young adulthood outcomes in the CLS34, this is the first study with an outcome in middle adulthood; therefore, the impact of these childhood factors may be reduced. Examination of alternative measures of these and other predictors at different ages is warranted.
Limitations
While the contributions of the current study are unique, the study is not without limitations. BMI is a commonly used measure of body fat mass, but more precise indicators such as body composition and waist circumference would provide fuller coverage35. Nevertheless, in our follow up in-person health exam for a subsample of participants (N = 267), we found sizable correlations between BMI and body composition (r = .80) and waist circumference (r = .87). This study is an examination of childhood predictors of adult obesity – there are pathways through which these childhood factors interact with and are mediated by adult factors, which future studies conducted with the CLS will examine. Our measure of neighborhood human capital is only one element of the larger neighborhood context in which a child is raised. Finally, our sample was predominately Black and low-income, so these results do not generalize to the entire U.S. population. However, given the increased risk of heart disease for Black Americans, the focus on this population is high priority for the field.
Additionally, the findings of this correlational study should be interpreted cautiously and do not infer causality. However, these rarely investigated predictors address alterable elements in childhood that warrant corroboration in future research. The overall variance explained by the models is also not large (~ 9%). Since we did not include later life variables or well-established health contributors (e.g., parent weight, caloric intake/expenditure), this low model R2 is expected. However, the magnitude of the significant predictors indicate there are many childhood correlates of adult obesity. The consistency of findings across model specifications further strengthens confidence in the results, and will be a focus of future CLS research.
Implications
Our findings that childhood experiences can predict adult weight support the need for prevention efforts and future research. Although obesity can develop at point in life, individuals who are obese or overweight during childhood are more likely to have unhealthy weights as adults36. Childhood education and family support programs have shown promise in reducing obesity risk37. The World Health Organization emphasizes that obesity prevention needs to take a life-course approach and not only focus on healthy weight, but also healthy psychosocial and cognitive growth development37; our findings that higher socioemotional learning skills in childhood are associated with a better BMI in adulthood support this assertion.
In support of the validity of our identified predictors of adult weight, the SEL and NHC contributions are consistent with previous CLS studies. Socioemotional learning has been a consistent predictor of related life course outcomes21,23, and appears to be independent of educational performance. Childhood neighborhood human capital provides a larger context for supporting socioemotional and academic success as well as peer and family supports necessary for later physical health. Along with other factors associated with higher obesity rates in Black American children and adults (e.g., discrimination, built environment, etc.), this finding further stresses the need for more research on neighborhood influences that are protective and alterable. These contributing factors may also be mechanisms through which childhood and family experiences promote optimal health and well-being over time, especially during the transition to adulthood and economic independence; extant CLS findings provide support to this view38,39. Since the proportion of U.S. adults with healthy body mass declines dramatically during early to middle adulthood1, childhood experiences at home and in school may help establish positive behaviors, attitudes, and routines that persist over the life course.
CONCLUSION
In this longitudinal study of childhood predictors of adult obesity in a predominately Black sample, socioemotional learning and neighborhood human capital consistently predicted a lower BMI and decreased risk of obesity. Future obesity prevention and intervention programs should target the childhood age range to strengthen socioemotional learning and bolster neighborhood resources for children to help curb the obesity epidemic.
Supplementary Material
Highlights for Childhood Predictors of Adult Obesity in the Chicago Longitudinal Study.
Childhood neighborhood human capital was associated with a lower BMI in adulthood.
Child socioemotional learning skills were positively associated with lower adult BMI.
These both predicted a lower BMI for males and females.
ACKNOWLEDGEMENTS
Preparation of this manuscript was supported, in part, by the National Institute of Child Health and Human Development HD034294. We thank the Chicago Public School District, Illinois Departments of Human Services and Child & Family Services, and the Chapin Hall Center for Children at the University of Chicago for cooperation in data collection and processing. The authors have no financial relationships relevant to this article to disclose.
Abbreviations:
- CLS
Chicago Longitudinal Study
- CPC
Child-Parent Center
- NHC
Neighborhood human capital (at age 10)
- SEL
Socioemotional learning
- SLP
School-level poverty
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
REFERENCES
- 1.Hales CM, Carroll MD, Fryar CD, Ogden CL. Prevalence of Obesity Among Adults and Youth: United States, 2015–2016. NCHS data brief, no. 288; Hyattsville, MD; 2017. [PubMed] [Google Scholar]
- 2.Hales CM, Fryar CD, Carroll MD, Freedman DS, Ogden CL. Trends in Obesity and Severe Obesity Prevalence in US Youth and Adults by Sex and Age, 2007–2008 to 2015–2016. JAMA. 2018;319(16):1723–1725. doi: 10.1001/jama.2018.3060 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Erermis S, Cetin N, Tamar M, Bukusoglu N, Akdeniz F, Goksen D. Is obesity a risk factor for psychopathology among adolescents? Pediatr Int. 2004;46(3):296–301. doi: 10.1111/j.1442-200x.2004.01882.x [DOI] [PubMed] [Google Scholar]
- 4.Dietz WH. Health Consequences of Obesity in Youth: Childhood Predictors of Adult Disease. Pediatrics. 1997;Supplement. [PubMed] [Google Scholar]
- 5.Folsom AR, Shah AM, Lutsey PL, et al. American Heart Association’s Life’s Simple 7: Avoiding Heart Failure and Preserving Cardiac Structure and Function. Am J Med. 2015;128:970–976.e2. doi: 10.1016/j.amjmed.2015.03.027 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Lloyd-Jones DM, Hong Y, Labarthe D, et al. Defining and Setting National Goals for Cardiovascular Health Promotion and Disease Reduction: The American Heart Association’s Strategic Impact Goal Through 2020 and Beyond. Circulation. 2010;121:586–613. doi: 10.1161/CIRCULATIONAHA.109.192703 [DOI] [PubMed] [Google Scholar]
- 7.Williams DR, Mohammed SA. Discrimination and racial disparities in health: Evidence and needed research. J Behav Med. 2009;32:20–47. doi: 10.1007/s10865-008-9185-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Alexander DS, Huber LRB, Piper CR, Tanner AE. The association between recreational parks, facilities and childhood obesity: A cross-sectional study of the 2007 National Survey of Children’s Health. J Epidemiol Community Health. 2013;67(5):427–431. doi: 10.1136/jech-2012-201301 [DOI] [PubMed] [Google Scholar]
- 9.Powell LM, Chaloupka FJ, Bao Y. The Availability of Fast-Food and Full-Service Restaurants in the United States. Associations with Neighborhood Characteristics. Am J Prev Med. 2007;33(4S):240–245. doi: 10.1016/j.amepre.2007.07.005 [DOI] [PubMed] [Google Scholar]
- 10.Sisson SB, Krampe M, Anundson K, Castle S. Obesity prevention and obesogenic behavior interventions in child care: A systematic review. Prev Med (Baltim). 2016;87:5769. doi: 10.1016/j.ypmed.2016.02.016 [DOI] [PubMed] [Google Scholar]
- 11.Nader PR, O’Brien M, Houts R, et al. Identifying Risk for Obesity in Early Childhood. Pediatrics. 2006;118:e594–e601. doi: 10.1542/peds.2005-2801 [DOI] [PubMed] [Google Scholar]
- 12.Bronfenbrenner U The Ecology of Human Development. Cambridge, MA: Harvard University Press; 1979. [Google Scholar]
- 13.Bronfenbrenner U Ecological systems theory. Ann Child Dev. 1989;6:187–249. [Google Scholar]
- 14.Penhollow TM, Rhoads KE. Preventing Obesity and Promoting Fitness: An Ecological Perspective. Am J Lifestyle Med. 2013;8(1):21–24. doi: 10.1177/1559827613507413 [DOI] [Google Scholar]
- 15.Boonpleng W, Park CG, Gallo AM, Corte C, Mccreary L, Bergren MD. Ecological Influences of Early Childhood Obesity: A Multilevel Analysis. West J Nurs Res. 2013;35(6):742–759. doi: 10.1177/0193945913480275 [DOI] [PubMed] [Google Scholar]
- 16.Sobol-Goldberg S, Rabinowitz J, Gross R. School-based obesity prevention programs: A meta-analysis of randomized controlled trials. Obesity. 2013;21(12):2422–2428. doi: 10.1002/oby.20515 [DOI] [PubMed] [Google Scholar]
- 17.Kim Y, Cubbin C, Oh S. A systematic review of neighbourhood economic context on child obesity and obesity-related behaviours. Obes Rev. 2019;20(3):420–431. doi: 10.1111/obr.12792 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Parsons TJ, Power C, Logan S, Summerbell CD. Childhood predictors of adult obesity: a systematic review. Int J Obes. 1999;23:1–107. http://www.stockton-press.co.uk/ijo. Accessed February 10, 2019. [PubMed] [Google Scholar]
- 19.Rooney BL, Mathiason MA, Schauberger CW, Rooney BL, Mathiason MA, Schauberger CW. Predictors of Obesity in Childhood, Adolescence, and Adulthood in a Birth Cohort. Matern Child Health J. 2011;15:1166–1175. doi: 10.1007/s10995-010-0689-1 [DOI] [PubMed] [Google Scholar]
- 20.Brisbois TD, Farmer AP, Mccargar LJ. Early markers of adult obesity: a review. Obes Rev. 2012;13:347–367. doi: 10.1111/j.1467-789X.2011.00965.x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Topitzes J, Godes O, Mersky JP, Ceglarek S, Reynolds AJ. Educational Success and Adult Health: Findings from the Chicago Longitudinal Study. Prev Sci. 2009;10:175–195. doi: 10.1007/s11121-009-0121-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Ou SR, Reynolds AJ. Predictors of Educational Attainment in the Chicago Longitudinal Study. Sch Psychol Q. 2008;23(2):199–229. doi: 10.1037/1045-3830.23.2.199 [DOI] [Google Scholar]
- 23.Mondi CF, Reynolds AJ, Ou SR. Predictors of depressive symptoms in emerging adulthood in a low-income urban cohort. J Appl Dev Psychol. 2017;50:45–59. doi: 10.1016/j.appdev.2017.03.009 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Reynolds AJ, Ou S-R. Early childhood to young adulthood: An introduction to the special issue. Child Youth Serv Rev. 2010;32:1045–1053. doi: 10.1016/j.childyouth.2010.03.024 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Reynolds AJ, Temple JA, Ou S-R, et al. Effects of a School-Based, Early Childhood Intervention on Adult Health and Well-being. Arch Pediatr Adolesc Med. 2007;161(8):730. doi: 10.1001/archpedi.161.8.730 [DOI] [PubMed] [Google Scholar]
- 26.Reynolds AJ, Temple JA, Ou SR. School-Based Early Intervention and Child Well-Being in the Chicago Longitudinal Study. Child Welfare. 2003;82(5):633–656. [PubMed] [Google Scholar]
- 27.Reynolds AJ. Success in Early Intervention: The Chicago Child-Parent Centers. Lincoln, NE: University of Nebraska Press; 2000. [Google Scholar]
- 28.World Health Organization. Obesity and overweight fact sheet. https://www.who.int/en/news-room/fact-sheets/detail/obesity-and-overweight. Published 2018.
- 29.Hayakawa M, Englund MM, Warner-Richter M, Reynolds AJ. Early Parent Involvement and School Achievement: A Longitudinal Path Analysis. NHSA Dialog. 2013;16(1):200–204. https://journals.uncc.edu/dialog/article/viewFile/54/123. Accessed August 26, 2019. [PMC free article] [PubMed] [Google Scholar]
- 30.Arteaga I, Chen CC, Reynolds AJ. Childhood predictors of adult substance abuse. Child Youth Serv Rev. 2010;32:1108–1120. doi: 10.1016/j.childyouth.2010.04.025 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.United States Census Bureau. 1990 Census; 2002. [Google Scholar]
- 32.Reynolds AJ, Englund MM, Ou S-R, Schweinhart LJ, Campbell FA. Reynolds A, Englund M, Ou S, Schweinhart L, & Campbell F (2010). Paths of Effects of Preschool Participation to Educational Attainment at Age 21: A Three-Study Analysis In: Reynolds AJ, Rolnick A, Englund MM, Temple JA, eds. Childhood Programs and Practices in the First Decade of Life: A Human Capital Integration. Cambridge: Cambridge University Press; 2010:415–452. doi: 10.1017/CBO9780511762666.022 [DOI] [Google Scholar]
- 33.Schafer JL. Analysis of Incomplete Multivariate Data. London: Chapman & Hall; 1997. [Google Scholar]
- 34.Reynolds AJ, Ou S-R, Topitzes JW. Paths of Effects of Early Childhood Intervention on Educational Attainment and Delinquency: A Confirmatory Analysis of the Chicago Child-Parent Centers. Child Dev. 2004;75(5):1299–1328. https://onlinelibrary.wiley.com/doi/pdf/10.1111/j.1467-8624.2004.00742.x. [DOI] [PubMed] [Google Scholar]
- 35.Müller MJ, Bosy-Westphal A, Krawczak M. Genetic studies of common types of obesity: a critique of the current use of phenotypes. Obes Rev. 2010;11:612–618. doi: 10.1111/j.1467-789X.2010.00734.x [DOI] [PubMed] [Google Scholar]
- 36.Simmonds M, Llewellyn A, Owen CG, Woolacott N. Predicting adult obesity from childhood obesity: A systematic review and meta-analysis. Obes Rev. 2016;17(2):95–107. doi: 10.1111/obr.12334 [DOI] [PubMed] [Google Scholar]
- 37.World Health Organization. Consideration of the evidence on childhood obesity for the Commission on Ending Childhood Obesity. World Heal Organ; 2016:219. doi:ISBN 978 92 4 151006 6 [Google Scholar]
- 38.Reynolds AJ, Ou SR. Paths of Effects From Preschool to Adult Well-Being: A Confirmatory Analysis of the Child-Parent Center Program. Child Dev. 2011;82(2):555582. doi: 10.1111/j.1467-8624.2010.01562.x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Hayakawa M, Giovanelli A, Englund MM, Reynolds AJ. Not Just Academics: Paths of Longitudinal Effects From Parent Involvement to Substance Abuse in Emerging Adulthood. J Adolesc Heal. 2016;58:433–439. doi: 10.1016/j.jadohealth.2015.11.007 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Hlavac M stargazer: Well-Formatted Regression and Summary Statistics Tables. R package version 5.2.1. 2018. https://cran.r-project.org/package=stargazer.
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.