This cohort study investigates the association between self-regulation among toddlers and risk of obesity at 5.5 years of age and whether there is a difference in this association between boys and girls.
Key Points
Questions
How are levels of toddler self-regulation associated with obesity at kindergarten age, and do patterns differ between boys and girls?
Findings
In this cohort study of a nationally representative sample of more than 6000 US children, the pattern of the association between self-regulation and the risk of obesity differed between boys and girls. Among boys, obesity prevalence was lower for those with the most self-regulation, whereas among girls, obesity prevalence was highest for those with the most and least self-regulation and lowest for those with intermediate levels of self-regulation.
Meaning
Obesity prevention efforts aimed at improving self-regulation may have different results for girls and boys.
Abstract
Importance
Poor self-regulation in childhood is associated with increased risk of obesity. However, studies have assumed that greater self-regulation is associated with a lower obesity risk and have rarely examined differences in the association by sex.
Objectives
To examine how different levels of toddler self-regulation are associated with the prevalence of obesity at kindergarten age and whether the pattern of association is different between boys and girls.
Design, Setting, and Participants
This was a prospective cohort study using data from the Early Childhood Longitudinal Study, Birth Cohort, a nationally representative sample of 10 700 US children born in 2001 and followed up through kindergarten entry (2006-2007). The analytic sample included 6400 children with observed toddler self-regulation. Data collection occurred in children’s homes and consisted of a parent interview and direct assessment of the child. Data analysis took place between May 2016 and March 2018.
Exposures
During a standardized, in-home, developmental assessment at 24 months of age, observers scored 4 dimensions of children’s self-regulation: adaptability, attention, persistence, and frustration tolerance. Self-regulation scores ranging from a low of 4 to a high of 20 were grouped into quartiles.
Main Outcomes and Measures
With use of measured heights and weights at 5.5 years, obesity was defined as a body mass index for age in the 95th percentile or greater.
Results
The analytic sample consisted of 6400 children (3250 boys [50.6%, weighted]), with a median age of 24.1 months (interquartile range, 23.4-24.7 months) and 64.5 months (interquartile range, 61.7-67.6 months) at the self-regulation and body mass index assessments, respectively. Self-regulation scores were lower for boys than for girls (mean, 13.7 [95% CI, 13.4-13.9] vs 14.9 [95% CI, 14.7-15.1]), and the lowest self-regulation quartile comprised more boys than girls (weighted percentages, 66.5% vs 33.5%). The prevalence of obesity at 5.5 years was 19.2% among boys and 16.5% among girls. The pattern of association between toddler self-regulation and obesity at 5.5 years was different for boys and girls (P = .008 for interaction). Among boys, the adjusted prevalence of obesity was 19.7%, 18.3%, 20.3%, and 15.9% from lowest to highest quartile of self-regulation. In contrast, among girls, there was a U-shaped association (adjusted prevalence of obesity from lowest to highest self-regulation quartile, 17.0%, 10.3%, 10.7%, and 15.0%).
Conclusions and Relevance
In a large national cohort of US children, there were differences between boys and girls in the pattern of the association between self-regulation at 24 months and obesity at 5.5 years of age. Obesity prevention efforts aimed at improving self-regulation may have different results for girls and boys.
Introduction
Self-regulation is a multidimensional construct involving the integration and modulation of cognitive and emotional processes to support goal-directed and cooperative social behavior.1,2 Poor self-regulation during childhood is associated with multiple long-term consequences for adult health and functioning,1,3,4 including obesity,5,6,7,8 and self-regulation is likely to be shaped early in brain development.1,9 The hypothesized mechanisms connecting poor self-regulation to obesity involve physiologic and behavioral responses to environmental demands. This response affects appetite, food intake, sleep, and energy expenditure.10 Childhood obesity prevention efforts have begun to focus on enhancing self-regulation11 because self-regulation may be influenced by early intervention12 and could have benefits beyond obesity prevention.7
Multiple prospective studies have found an association between poor self-regulation assessed in early childhood and risk of future obesity.5,13,14,15,16,17 In one study, delay of gratification at preschool age was associated with meaningful differences in self-reported body mass index (BMI) 3 decades later,5 but the sample was small (n = 164) and selected. However, evidence from population-based studies3,14,15,16 supports the generalizability of the findings. The Dunedin Study3 reported associations between childhood self-regulation and a composite measure of adult metabolic health that included obesity. Analyses from the US Study of Early Childcare and Youth Development showed that the inability to delay gratification at preschool age was associated with being overweight at age 11 years15 and greater childhood weight gain.14 In the UK Millennium Cohort Study, 3-year-old children’s poorer emotional self-regulation—as reported by parents—was associated with obesity at age 11 years.16 Graziano and colleagues17 used a laboratory protocol to assess multiple aspects of toddlers’ self-regulation and found that greater self-regulation was associated with lower BMI at age 10 years. Notwithstanding these findings from observational studies, a recent randomized clinical trial of an intervention among low-income preschool-aged children showed improvements in self-regulation but no effects on obesity.11
Despite the potential relevance of early self-regulation for later obesity, to date, only 1 large population-based cohort has produced data on observed self-regulation and measured height and weight.14,15 The children studied were aged 3 years when self-regulation was first assessed, and the analyses used self-regulation as a binary variable, comparing those with and without low self-regulation.14,15 This analytic approach may not reveal patterns of association between self-regulation and obesity that could be relevant for developing interventions. Furthermore, the association between self-regulation and obesity may differ between boys and girls. In the Dunedin Study, adult metabolic health was examined separately for men and women across the entire range of childhood self-regulation, and similar graded associations were found for each sex.3 However, sex differences in young children’s self-regulation have been observed,18 and it has been suggested that boys and girls potentially differ in how self-regulation is associated with obesity.14,19,20
Using data from the Early Childhood Longitudinal Study, Birth Cohort (ECLS-B), we examined how different levels of toddler self-regulation were associated with the prevalence of obesity at kindergarten age. We also examined whether patterns of associations differed between boys and girls.
Methods
Study Population
The ECLS-B was a large, nationally representative, prospective cohort study conducted by the National Center for Education Statistics (NCES) between 2001 and 2007. The design of the study has been published.21 In brief, a probability sample of 14 000 US births in 2001 was selected using a clustered list-frame design. Children born to mothers younger than 15 years or who died or were adopted before 9 months of age were excluded. Oversampling of some population subgroups (eg, twins and children with Chinese or American Indian/Native Alaskan ethnicity) was used, and analytic sampling weights allow inference to the US population. The final cohort of 10 700 was formed when the children were approximately 9 months of age, with follow-up assessments at 24 months of age (n = 9850) and preschool age (n = 8750). For budgetary reasons, an 85% random sample was selected for follow-up at 5.5 years of age. Data collection occurred in children’s homes and consisted of an interview with the child’s mother (or in a few cases, the father or other guardian) and direct assessment of the child. The ECLS-B interviewers conducted assessments in English and Spanish ( ~ 7% of visits). If families spoke other languages ( ~ 3% of visits), an interpreter was used.21 All procedures were approved by the NCES Ethics Review Board, and parents provided written informed consent. The Ohio State University has a restricted-use data agreement with the NCES, which requires that unweighted sample sizes be rounded to the nearest 50 children.
Exposure Variable: Self-regulation at 24 Months of Age
During the 24-month assessment in the child’s home, ECLS-B researchers administered the Bayley Short Form–Research Edition,22 which is a modified version of the Bayley Scales of Infant Development, 2nd edition.23 The Bayley Short Form–Research Edition was used to assess children’s cognitive and motor development, but this structured assessment, lasting approximately 30 minutes, also allowed observation of the children’s behavior.22 At the completion of the approximately 2-hour home visit, interviewers rated the children’s behavior during the Bayley Short Form–Research Edition using 11 of the 30 items of the Behavior Rating Scales of the Bayley Scales of Infant Development, 2nd edition.22 The developers of the ECLS-B chose these items to cover a range of behaviors; items were not chosen to form a scale to measure self-regulation or any other construct.22 However, we selected 4 of these 11 observer-rated aspects of children’s behavior that had face validity as measures of toddler self-regulation. Each used a 5-level (1-5) response scale anchored by verbal descriptions. For example, 1 item asked the observer to rate the child’s frustration tolerance: “rate the degree to which [CHILD] became frustrated when [he/she] was unable to understand or complete a task” with response options of (1) consistently becomes frustrated, (2) typically becomes frustrated, (3) occasionally becomes frustrated, (4) rarely becomes frustrated, and (5) never becomes frustrated.22 The other items that we selected asked the observer to rate adaptability (“ability to repeatedly relinquish material used for one Bayley item and accept the material for the next Bayley item”), persistence (“persisted at tasks in attempting to complete them”), and attention (“remained focused on the tasks presented by the interviewer”).22 The internal reliability Cronbach α of these 4 items was 0.85. We created a composite measure of observed toddler self-regulation by summing the ratings. Thus, the lowest self-regulation score was 4 (indicating a child rated as consistently lacking persistence, becoming frustrated, resisting requests to relinquish materials and/or refusing to accept new materials, and being inattentive and consistently off task), and the highest score was 20 (indicating a child who displayed persistence, did not become frustrated, readily exchanged materials, and remained focused). Because the distribution of this self-regulation score was skewed, with few children having very low scores, we grouped children into self-regulation levels based on quartiles. Quartile cut points were determined from the overall distribution to ensure comparability by gender (eg, boys and girls whose self-regulation score was ≤11 were defined as being in the lowest quartile).
Outcome Variable: Obesity at 5.5 Years of Age
With use of a standardized protocol, children’s heights were measured with a portable stadiometer and children’s weights were measured with a digital scale while they were wearing light clothing and no shoes.24 We defined children’s obesity status at age 5.5 years based on the 2000 US growth reference.25 We calculated BMI (weight in kilograms divided by height in meters squared) and defined obesity as a BMI for age at or above the 95th percentile.
Covariates
Additional variables were selected to describe children’s sociodemographic context and to control for potential confounding. Children’s birth weight (grams), twin status (yes or no), and sex were abstracted from birth certificates and confirmed in the 9-month interview with the mother. We use the term gender because the primary influences that we examined are more likely to be social (gender) than biologic (sex).26 At the 9-month interview, mothers were asked whether they had ever breastfed the child and the child’s age at weaning. Children’s race/ethnicity was identified by mothers and categorized as described elsewhere.27 Maternal smoking status was self-reported and categorized as current smoker or not. Maternal obesity was defined as a BMI at or above 30. Mothers reported their height at 9 months and were weighed on a digital scale following a standardized protocol. If maternal weight was missing at 9 months but available at a later assessment, we used the earliest available value. Maternal educational level, household income, and household size were reported at 24 months. The ratio of household income to the 2003 Census Bureau poverty thresholds was calculated.
Statistical Analyses
Data analysis took place between May 2016 and March 2018. Our analytic sample included 6400 children with information available on self-regulation at 24 months of age and measured height and weight at 5.5 years of age. Of the 6950 children assessed at 5.5 years of age, approximately 400 children were excluded for missing self-regulation data, and approximately 200 additional children were excluded for missing data on measured height or weight (eFigure 1 in the Supplement). We compared characteristics of children excluded from the analytic sample with those of children included in the analytic sample, and we assessed group differences using a design-corrected χ2 test. All analyses used ECLS-B longitudinal sampling weights, which adjust for disproportionate sampling, potential under coverage, and nonresponse. Percentages are weighted to be representative of US births in 2001. Variance estimates account for the complex clustered design of ECLS-B using the SURVEY procedures in SAS, version 9.3 (SAS Institute Inc). We tested whether the pattern of association between self-regulation and obesity was different in boys and girls by using a multivariate F test for the interaction of gender with quartile of self-regulation at 24 months (as a categorical variable) in a logistic regression model. With evidence of statistically significant interaction, we subsequently stratified our analyses by gender. Unadjusted and covariate-adjusted models were estimated, and we report the prevalence (95% CI) of obesity at 5.5 years relative to level of self-regulation at 24 months for boys and girls. All statistical tests were 2-sided, and statistical significance was set at .05.
Results
The 6400 children (3250 [50.6%] boys, 3200 [49.4] girls) in the analytic sample had a median age of 24.1 months (interquartile range, 23.4-24.7 months) and 64.5 months (interquartile range, 61.7-67.6 months) at the self-regulation and BMI assessments, respectively; length of follow-up ranged from 30.3 to 49.1 months (mean, 40.3 months). Sociodemographic characteristics of the sample are described in Table 1. Children excluded because of missing information were more likely to be boys and were from households with lower socioeconomic status (eTable 1 in the Supplement). The mean self-regulation score was 13.7 (95% CI, 13.4-13.9) among boys and 14.9 (95% CI, 14.7-15.1) among girls. The distribution of self-regulation scores differed for boys and girls (Figure 1). Boys were more likely to have low levels of self-regulation, and girls were more likely to have high levels of self-regulation (66.5% of children in the lowest self-regulation quartile were boys, whereas 41.8% of those in the highest self-regulation quartile were boys) (Table 1). The prevalence of obesity at 5.5 years was higher among boys than among girls (19.2% vs 16.5%). Socioeconomic status was associated in a graded manner with self-regulation and obesity; lower socioeconomic status was associated with lower levels of self-regulation (Table 1) and more obesity (Table 2). Maternal smoking, maternal obesity, and nonwhite race/ethnicity were also associated with poorer self-regulation and obesity. Twins and children born at very low birth weights were more likely to have poor self-regulation (Table 1), but these children were less likely to be obese (Table 2). Relationships of covariates with self-regulation and obesity were similar in both boys and girls (eTables 2-5 in the Supplement).
Table 1. Sociodemographic Characteristics by Toddler Self-regulationa.
| Characteristic | No. (%) | Observed Self-regulation at 24 mo,b % (95% CI) | |||
|---|---|---|---|---|---|
| Q1 | Q2 | Q3 | Q4 | ||
| All | 6400 (100) | 19.9 (17.9-21.9) | 26.6 (24.6-28.5) | 27.4 (25.6-29.2) | 26.1 (23.8-28.4) |
| Child characteristic | |||||
| Gender | |||||
| Boys | 3250 (50.6) | 66.5 (63.5-69.5) | 51.6 (48.4-54.9) | 46.6 (43.3-49.8) | 41.8 (38.6-44.9) |
| Girls | 3200 (49.4) | 33.5 (30.5-36.5) | 48.4 (45.1-51.6) | 53.4 (50.2-56.7) | 58.2 ((55.1-61.4) |
| Race/ethnicity | |||||
| Non-Hispanic white | 2650 (53.8) | 43.7 (39.2-48.3) | 51.2 (47.8-54.6) | 58.2 (55.4-61.0) | 59.4 (56.2-62.6) |
| Non-Hispanic black | 1100 (15.5) | 19.3 (16.7-21.9) | 15.9 (13.6-18.1) | 15.5 (13.5-17.5) | 12.1 (10.1-14.1) |
| Hispanic | 1200 (24.2) | 28.3 (24.2-32.4) | 27.4 (24.5-30.3) | 20.0 (17.8-22.3) | 22.3 (19.6-24.9) |
| Other, non-Hispanic | 1500 (6.5) | 8.7 (6.8-10.6) | 5.5 (4.5-6.5) | 6.3 (5.0-7.5) | 6.2 (5.2-7.3) |
| Twin status | |||||
| Yes | 1100 (2.9) | 3.8 (3.3-4.4) | 3.1 (2.7-3.6) | 2.9 (2.5-3.3) | 1.8 (1.5-2.1) |
| No | 5300 (97.1) | 96.2 (95.6-96.7) | 96.9 (96.4-97.3) | 97.1 (96.7-97.5) | 98.2 (97.9-98.5) |
| Birth weight,c g | |||||
| >4000 | 400 (9.6) | 9.2 (6.9-11.5) | 9.0 (7.0-11.1) | 10.8 (8.9-12.7) | 9.1 (7.1-11.1) |
| >2500 to 4000 | 4400 (83.0) | 81.0 (78.2-83.8) | 83.1 (81.0-85.2) | 81.6 (79.7-83.5) | 85.9 (83.8-88.1) |
| 1500-2500 | 950 (6.2) | 7.4 (6.2-8.5) | 6.5 (5.6-7.5) | 6.8 (5.9-7.7) | 4.4 (3.7-5.1) |
| <1500 | 650 (1.2) | 2.4 (2.1-2.7) | 1.4 (1.2-1.6) | 0.9 (0.7-1.1) | 0.5 (0.4-0.7) |
| Breastfeeding durationc | |||||
| Never | 850 (14.0) | 38.2 (33.7-42.8) | 29.5 (25.8-33.3) | 28.9 (25.8-31.9) | 25.5 (22.5-28.6) |
| <2 mo | 1500 (22.1) | 28.3 (24.0-32.6) | 32.9 (29.6-36.1) | 35.9 (32.5-39.2) | 37.2 (33.9-40.6) |
| 2 to <6 mo | 2050 (33.9) | 23.7 (20.3-27.2) | 23.9 (21.0-26.9) | 20.1 (17.1-23.0) | 20.9 (17.9-24.0) |
| ≥6 mo | 1950 (30.0) | 9.7 (7.1-12.3) | 13.7 (11.2-16.1) | 15.2 (12.7-17.6) | 16.3 (13.6-19.0) |
| Parent characteristic | |||||
| Maternal educational levelc | |||||
| Bachelor’s degree or higher | 1850 (25.1) | 16.2 (12.8-19.6) | 21.7 (18.9-24.5) | 29.3 (26.4-32.2) | 30.7 (27.8-33.7) |
| Some college/technical degree | 1800 (28.3) | 27.1 (23.7-30.5) | 28.2 (24.8-31.5) | 26.7 (23.6-29.8) | 31.1 (28.3-34.0) |
| High school diploma/GED | 1750 (29.7) | 33.6 (30.0-37.2) | 32.4 (29.4-35.3) | 27.5 (24.9-30.1) | 26.2 (23.4-29.1) |
| Less than high school | 1000 (16.9) | 23.1 (19.9-26.2) | 17.8 (15.4-20.1) | 16.5 (14.0-18.9) | 11.9 (9.7-14.1) |
| Household income-to-poverty ratio | |||||
| >3.00 | 1650 (25.5) | 18.9 (15.7-22.1) | 22.9 (19.2-26.7) | 26.9 (23.7-30.2) | 31.7 (28.4-35.0) |
| 1.86-3.00 | 1800 (29.3) | 25.8 (22.1-29.5) | 29.8 (26.4-33.3) | 30.7 (27.4-34.0) | 30.1 (26.2-34.0) |
| 1.00-1.85 | 1450 (22.6) | 26.6 (23.4-29.8) | 25.4 (22.3-28.4) | 18.8 (15.7-22.0) | 20.6 (18.2-23.1) |
| <1.00 | 1500 (22.6) | 28.7 (24.7-32.7) | 21.9 (19.3-24.4) | 23.6 (19.8-27.3) | 17.6 (14.7-20.5) |
| Maternal obesityc | |||||
| BMI≥30 | 1750 (27.9) | 32.0 (28.1-36.0) | 28.4 (25.2-31.6) | 25.3 (22.2-28.5) | 26.9 (23.6-30.3) |
| BMI<30 | 4450 (72.1) | 68.0 (64.0-71.9) | 71.6 (68.4-74.8) | 74.7 (71.5-77.8) | 73.1 (69.7-76.4) |
| Maternal smokingc | |||||
| Yes | 1150 (18.4) | 23.1 (19.9-26.3) | 21.5 (18.5-24.6) | 16.1 (13.4-18.8) | 14.2 (11.7-16.7) |
| No | 5250 (81.6) | 76.9 (73.7-80.1) | 78.5 (75.4-81.5) | 83.9 (81.2-86.6) | 85.8 (83.3-88.3) |
Abbreviations: BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); GED, General Equivalency Diploma; Q, quartile.
Early Childhood Longitudinal Study–Birth Cohort, restricted to children with information on self-regulation at 24 months and measured height and weight at 5.5 years. Sample sizes are rounded to the nearest 50 participants as required by a restricted-use license agreement and may not total because of rounding. Percentages are weighted to be representative of US births in 2001. Variance estimates account for the complex sample design.
Self-regulation from interviewer ratings of 4 aspects (adaptability, attention, persistence, and frustration tolerance) of child’s behavior during administration of the Bayley Short-form–Research Edition at 24 months; each aspect was rated on a 5-point scale (with lower score indicating less self-regulation and a higher score indicating greater self-regulation). Composite score (Cronbach α = 0.85) was created as the sum of ratings (minimum score of 4 and maximum score of 20); quartile (Q) definitions were based on the overall distribution of scores (Q1 = 4-11, Q2 = 12-14, Q3 = 15-16, and Q4 = 17-20). All sociodemographic characteristics tabulated were statistically significantly (P < .05) associated with toddler self-regulation (Rao-Scott design-corrected χ2). Column percentages are presented.
Number of participants with missing information for birth weight, maternal educational level, and maternal smoking was less than 50; 50 participants had missing information for breastfeeding duration; and 200 participants had missing information for maternal obesity.
Figure 1. Distribution of Observed Toddler Self-regulation Scores by Gender.
Violin plots indicate self-regulation at toddler age (24-month assessment) from interviewer ratings of 4 aspects (adaptability, attention, persistence, and frustration tolerance) of the child’s behavior during administration of the revised research edition of the Bayley Short-form–Research Edition; each aspect was rated on a 5-point scale (with a lower score indicating less self-regulation and a higher score indicating greater self-regulation). Composite score (Cronbach α = 0.85) was created as the sum of ratings (minimum score of 4 and maximum score of 20). The heavy line indicates median score; darker shading, interquartile range. P < .001 for difference in means by gender.
Table 2. Sociodemographic Characteristics by Obesity Status at 5.5 Years of Agea.
| Characteristic | No. (%) | BMI for Age at 5.5 y,b % (95% CI) | |
|---|---|---|---|
| <95th Percentile | ≥95th Percentile | ||
| All | 6400 (100) | 82.2 (80.9-83.5) | 17.8 (16.5-19.1) |
| Child characteristic | |||
| Gender | |||
| Boys | 3250 (50.6) | 49.8 (48.8-50.8) | 54.4 (50.6-58.3) |
| Girls | 3200 (49.4) | 50.2 (49.2-51.2) | 45.6 (41.7-49.4) |
| Race/ethnicity | |||
| Non-Hispanic white | 2650 (53.8) | 56.3 (54.4-58.2) | 41.9 (38.3-45.5) |
| Non-Hispanic black | 1100 (15.5) | 15.0 (14.0-16.1) | 17.5 (14.8-20.3) |
| Hispanic | 1200 (24.2) | 22.3 (20.7-23.9) | 33.1 (29.4-36.8) |
| Other, non-Hispanic | 1500 (6.5) | 6.3 (5.6-7.0) | 7.5 (5.7-9.2) |
| Twin status | |||
| Yes | 1100 (2.9) | 3.1 (2.9-3.3) | 1.9 (1.5-2.4) |
| No | 5300 (97.1) | 96.9 (96.7-97.1) | 98.1 (97.6-98.5) |
| Birth weight,c g | |||
| >4000 | 400 (9.6) | 8.3 (7.3-9.4) | 15.2 (12.2-18.3) |
| >2500 to 4000 | 4400 (83.0) | 83.7 (82.6-84.8) | 79.8 (76.5-83.2) |
| 1500-2500 | 950 (6.2) | 6.6 (6.4-6.9) | 4.3 (3.3-5.3) |
| <1500 | 650 (1.2) | 1.4 (1.3-1.4) | 0.6 (0.4-0.8) |
| Breastfeeding durationc | |||
| Never | 850 (14.0) | 28.4 (26.5-30.3) | 37.5 (33.3-41.8) |
| <2 mo | 1500 (22.1) | 34.6 (32.6-36.6) | 30.7 (26.7-34.7) |
| 2 to <6 mo | 2050 (33.9) | 22.2 (20.7-23.7) | 21.5 (17.7-25.2) |
| ≥6 mo | 1950 (30.0) | 14.8 (13.6-15.9) | 10.3 (8.0-12.6) |
| Parent characteristic | |||
| Maternal educational levelc | |||
| Bachelor’s degree or higher | 1850 (25.1) | 26.8 (25.8-27.8) | 17.0 (14.2-19.8) |
| Some college/technical degree | 1800 (28.3) | 28.4 (26.9-29.9) | 28.0 (24.0-32.1) |
| High school diploma/GED | 1750 (29.7) | 28.9 (27.5-30.4) | 33.0 (29.2-36.8) |
| Less than high school | 1000 (16.9) | 15.8 (14.7-17.0) | 21.9 (18.7-25.2) |
| Household income-to-poverty ratio | |||
| >3.00 | 1650 (25.5) | 26.6 (24.7-28.6) | 20.2 (17.3-23.2) |
| 1.86-3.00 | 1800 (29.3) | 29.7 (27.6-31.8) | 27.6 (23.8-31.3) |
| 1.00-1.85 | 1450 (22.6) | 22.1 (20.5-23.7) | 24.8 (21.4-28.2) |
| <1.00 | 1500 (22.6) | 21.5 (19.5-23.5) | 27.4 (23.7-31.1) |
| Maternal obesityc | |||
| BMI≥30 | 1750 (27.9) | 23.9 (22.1-25.7) | 46.7 (42.7-50.6) |
| BMI<30 | 4450 (72.1) | 76.1 (74.3-77.9) | 53.3 (49.4-57.3) |
| Maternal smokingc | |||
| Yes | 1150 (18.4) | 17.5 (16.1-18.9) | 22.8 (18.9-26.8) |
| No | 5250 (81.6) | 82.5 (81.1-83.9) | 77.2 (73.2-81.1) |
Abbreviations: BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); GED, General Equivalency Diploma.
Early Childhood Longitudinal Study–Birth Cohort, restricted to children with information on self-regulation at 24 months and measured height and weight at 5.5 years. Sample sizes are rounded to the nearest 50 participants as required by a restricted-use license agreement and may not total because of rounding. Percentages are weighted to be representative of US births in 2001. Variance estimates account for the complex sample design.
Height and weight at age 5.5 years were measured. Obesity was defined as 95th percentile or higher of the Centers for Disease Control and Prevention BMI for age growth reference. All sociodemographic characteristics tabulated were statistically significantly (P < .05) associated with obesity at 5.5 years (Rao-Scott design-corrected χ2 test).
Number of participants with missing information for birth weight, maternal educational level, and maternal smoking was less than 50; 50 participants were missing data for breastfeeding duration; and 200 participants were missing data for maternal obesity.
The pattern of association between self-regulation and obesity was different for boys and girls (P = .008 for test of the interaction between gender and self-regulation quartile). Given this evidence of effect modification, our results were stratified by gender. For boys, the prevalence of obesity was lower among those in the highest quartile of self-regulation compared with those in the other 3 quartiles, but this difference was not statistically significant (prevalence of obesity in the covariate-adjusted model: 19.7% [95% CI, 15.4%-25.0%] in Q1, 18.3% [95% CI, 13.3%-24.8%] in Q2, 20.3% [95% CI, 15.4%-26.3%] in Q3, and 15.9% [95% CI, 11.7%-21.2%] in Q4; P = .34) (Table 3). In contrast, among girls, there was a U-shaped association between self-regulation and obesity in girls, with high obesity prevalence in the least (Q1) and most (Q4) self-regulated quartiles and lower prevalence in the middle (Q2, Q3) quartiles (prevalence of obesity in the covariate-adjusted model: 17.0% (95% CI, 12.5%-22.5%) in Q1, 10.3% (95% CI, 7.5%-13.9%) in Q2, 10.7% (95% CI, 7.5%-15.0%) in Q3, 15.0% (95% CI, 10.9%-20.4%) in Q4; P < .001) (Figure 2 and Table 3).
Table 3. Association of Toddler Self-regulation at 24 Months in Relation to Risk of Obesity at 5.5 Years of Age Among Boys and Girlsa.
| Toddler Self-regulationb | Unadjusted (n = 6400) | Covariate Adjusted (n = 6150)c | ||
|---|---|---|---|---|
| No. | Obesity Prevalence at 5.5 y,d % (95% CI) | No. | Predicted Obesity Prevalence at 5.5 y,d % (95% CI) | |
| Boyse | ||||
| Q1 | 950 | 21.9 (18.1-26.4) | 900 | 19.7 (15.4-25.0) |
| Q2 | 900 | 19.4 (15.7-23.8) | 900 | 18.3 (13.3-24.8) |
| Q3 | 750 | 19.3 (15.4-23.9) | 750 | 20.3 (15.4-26.3) |
| Q4 | 600 | 15.3 (12.0-19.3) | 600 | 15.9 (11.7-21.2) |
| P valuef | .09 | .34 | ||
| Girlse | ||||
| Q1 | 550 | 22.4 (18.3-27.1) | 500 | 17.0 (12.5-22.5) |
| Q2 | 800 | 13.2 (10.7-16.2) | 800 | 10.3 (7.5-13.9) |
| Q3 | 950 | 13.9 (11.1-17.4) | 900 | 10.7 (7.5-15.0) |
| Q4 | 900 | 19.1 (15.7-22.9) | 850 | 15.0 (10.9-20.4) |
| P valuef | <.001 | <.001 | ||
Sample sizes are rounded to the nearest 50 participants as required by a restricted-use license agreement and may not total because of rounding. Percentages are weighted to be representative of US births in 2001. Variance estimates account for the complex sample design.
Information on scoring of self-regulation is given in Table 1.
Adjusted for child age (months) at 24-month assessment, birth weight (<1500 g, 1500-2500 g, >2500 to 4000 g, or >4000 g), twin status (yes or no), maternal obesity (body mass index [calculated as weight in kilograms divided by height in meters squared]≥30 or <30), maternal smoking (yes or no), breastfeeding (never, <2 months, 2 months to <6 months, or ≥6 months), child race/ethnicity (non-Hispanic white, non-Hispanic black, Hispanic, or other non-Hispanic), maternal educational level (less than high school, high school, some college, or college graduate or greater), household income to poverty ratio (<0.50, 0.50-0.99, 1.00-1.85, 1.86-3.00, >3.00) (sample size was reduced by 250 because of missing covariates). Estimates are least squares means.
Obesity was defined as 95th percentile or higher of the Centers for Disease Control and Prevention body mass index for age growth reference based on measured height and weight at age 5.5 years. Estimates are least squares means.
Models were stratified by gender.
P value for effect of self-regulation (Wald χ2 test).
Figure 2. Association Between Toddler Self-regulation and Prevalence of Obesity at 5.5 Years of Age in Boys and Girls.
Estimates are least squares means from gender-stratified, covariate-adjusted models (n = 6150). Error bars are ±1 SE self-regulation at toddler age (24-month assessment) from interviewer ratings of 4 aspects (adaptability, attention, persistence, and frustration tolerance) of the child’s behavior during administration of the Bayley Short-form–Research Edition; each aspect was rated on a 5-point scale (with a lower score indicating less self-regulation and a higher score indicating greater self-regulation). The y-axis is the predicted prevalence of obesity at age 5.5 years. The x-axis is the quartile of self-regulation. Composite score (Cronbach α = 0.85) was created as the sum of ratings (minimum score of 4 and maximum score of 20). Quartile definitions were based on the overall distribution of scores (Q1 = 4-11, Q2 = 12-14, Q3 = 15-16, and Q4 = 17-20).
Results were robust to alternative analytic specifications. Findings were similar whether using gender-specific values to define quartiles of self-regulation (eTable 6 in the Supplement) or imputing missing covariate data (eTable 7 in the Supplement). The distinctive patterns of association between self-regulation and obesity seen for boys and girls were also present when we analyzed each aspect of self-regulation (adaptability, attention, persistence, and frustration tolerance) (eFigure 2 in the Supplement).
Discussion
We provide evidence from a large, nationally representative cohort of US children that the pattern of association between early self-regulation and later obesity may differ among boys and girls. Self-regulation was observed in a standardized manner at 24 months of age, and height and weight were measured at 5.5 years. We hypothesized that, for both boys and girls, poorer self-regulation would be associated with a higher prevalence of obesity, whereas better self-regulation would be associated with a lower prevalence of obesity. We found that, for girls, the highest prevalence of obesity occurred among girls with the lowest and highest self-regulation scores, whereas the lowest prevalence of obesity occurred among those with scores in the middle. Among boys, obesity prevalence was similar across the lowest and middle quartiles of self-regulation and was only somewhat lower among those in the highest quartile. These findings suggest the possibility that interventions to increase self-regulation in young children might have different results with respect to obesity for girls and boys.
Similar to other researchers,5,14,15,16,17 we found some evidence that children with the lowest levels of self-regulation were more likely to be obese, but the pattern of association between self-regulation and obesity was different between boys and girls. Neither pattern conforms to a clear stepwise decrease in the prevalence of obesity associated with increasing self-regulation. Our most unexpected finding was the high prevalence of obesity among girls at the highest levels of self-regulation. To our knowledge, this U-shaped pattern has not previously been reported and warrants further investigation. In contrast, the pattern among boys suggests little association between the prevalence of obesity in the first 3 quartiles of self-regulation and indicates that obesity prevalence is only somewhat lower among boys with the highest levels of self-regulation. Direct comparison of our findings with the Dunedin Study3 is not possible because of the considerable differences in the age of assessment and the manner in which exposures and outcomes were measured. The ECLS-B is large and nationally representative, and it includes children across a full spectrum of self-regulation capabilities. Compared with more homogeneous samples, the wide range of self-regulation in this study may have permitted detection of the U-shaped pattern that we observed among girls.
Self-regulation is modifiable,12 and interventions to improve young children’s self-regulation have been studied in the context of obesity prevention.11 In a rigorously designed, cluster-randomized clinical trial conducted in preschool classrooms with children from low-income households, Lumeng and colleagues11 compared the effect of an obesity prevention curriculum with and without the addition of an intervention (Incredible Years) designed to improve children’s self-regulation. The trial did not demonstrate effects on the children’s weight, activity, or diet.11 However, it was conducted over a short time frame and was not designed to assess associations between obesity and the Incredible Years intervention alone. Furthermore, although the intervention was successful in increasing children’s self-regulation, it is possible that teachers’ reporting was influenced by their involvement in delivering the intervention. Thus, the negative results should not preclude additional investigations.
We found in this national US sample that the distribution of self-regulation differed between girls and boys (Figure 1), with girls more likely to exhibit behaviors characteristic of good self-regulation and boys more likely to exhibit behaviors characteristic of poor self-regulation. In a smaller sample, other investigators showed similar differences in self-regulation between toddler boys and girls observed in a laboratory setting across a variety of self-regulation measures.18 It is not possible to determine from our data whether observers’ appraisals were influenced by gendered expectations of behavior in young children. Nevertheless, boys and girls received ratings across the full distribution of self-regulation scores, and the mean scores differed by only 1.2 units.
Our findings raise questions about whether and how gender is associated with the development of self-regulation. In the United States, many parents and caregivers, whether consciously or not, expect and value different behaviors from boys and girls. Behavior associated with poor self-regulation, such as inattention and expressed frustration, may be accepted more readily in a toddler boy than a toddler girl. Similarly, girls may be rewarded more than boys for behaviors characteristic of better self-regulation. Compared with boys, young girls may be under greater social expectations to exhibit better self-regulation even if the biologic potential for self-regulation is similar between the sexes at this age. When social expectations exceed their capabilities, some girls may experience stress. This stress might result in differences in energy balance and metabolism between girls and boys, especially in the group observed to have high self-regulation. Our results should provide some caution against the assumption that the highest levels of self-regulation lead to optimal outcomes for both genders or in all contexts.
Strengths and Limitations
There are both strengths and weaknesses to our assessment of self-regulation in this study that reflect the challenges and trade-offs faced by researchers trying to measure this construct. Self-regulation was observed and rated during a standardized, in-home assessment of children’s cognitive and motor development. Interviewers were trained and demonstrated reliability,22 and their observations may be more objective than parent report. Our composite measure of self-regulation had good internal reliability. However, this measure has not been validated. For example, it is not known how this measure relates to laboratory assessments of delay of gratification or other aspects of self-regulation or to caregiver reports of children’s self-regulation.
Our study has other limitations. First, this is an observational study, and no inferences should be drawn regarding causality. Second, children’s behavior may not have been typical on the day that self-regulation was observed. Third, to address potential confounding, we adjusted our analyses for variables related to self-regulation and obesity, but we cannot exclude the possibility of residual confounding. It is also possible, however, that we overadjusted by controlling for socioeconomic factors that influence self-regulation and also influence obesity, in part, through self-regulation. Finally, the children in this study were born in 2001, and it is unknown whether our findings would apply to children born more recently.
Conclusions
In a nationally representative US cohort of young children, the pattern of association between level of self-regulation and the risk of obesity differed between boys and girls. More research is needed to replicate this finding and understand the cause of this difference, particularly as it may relate to different social norms for behavior in girls and boys. Obesity prevention efforts aimed at improving self-regulation may have different results for girls and boys.
eTable 1: Sociodemographic characteristics of Early Childhood Longitudinal Study–Birth Cohort participants included in analytic sample compared to those not included in analytic sample
eTable 2: Sociodemographic characteristics by toddler self-regulation: boys
eTable 3: Sociodemographic characteristics by toddler self-regulation: girls
eTable 4: Sociodemographic characteristics by obesity status at 5.5 years: boys
eTable 5: Sociodemographic characteristics by obesity status at 5.5 years: girls
eTable 6: Association of toddler self-regulation in relation to risk for obesity at 5.5 years in boys and girls: alternative analytic specification using gender-specific self-regulation quartile definitions
eTable 7: Association of toddler self-regulation in relation to risk for obesity at 5.5 years in boys and girls: alternative analytic specification using multiple imputation
eFigure 1: Participant flow and analytic sample definition
eFigure 2: Association between toddler adaptability, persistence, attention, and frustration tolerance and obesity prevalence at 5.5 years for boys and girls
References
- 1.Murray DW, Rosanbalm K, Christopoulos C, Hamoudi A. Self-Regulation and Toxic Stress: Foundations for Understanding Self-Regulation from an Applied Developmental Perspective. Washington, DC: Office of Planning, Research and Evaluation, Administration for Children and Families, US Dept of Health and Human Services; 2015. [Google Scholar]
- 2.Blair C, Ursache A. A bidirectional model of executive functions and self-regulation In: Vohs KD, Baumeister RF, eds. Handbook of Self-Regulation: Research, Theory, and Applications. 2nd ed New York, NY: The Guilford Press; 2011:300-320. [Google Scholar]
- 3.Moffitt TE, Arseneault L, Belsky D, et al. A gradient of childhood self-control predicts health, wealth, and public safety. Proc Natl Acad Sci U S A. 2011;108(7):2693-2698. doi: 10.1073/pnas.1010076108 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Mischel W, Ayduk O, Berman MG, et al. “Willpower” over the life span: decomposing self-regulation. Soc Cogn Affect Neurosci. 2011;6(2):252-256. doi: 10.1093/scan/nsq081 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Schlam TR, Wilson NL, Shoda Y, Mischel W, Ayduk O. Preschoolers’ delay of gratification predicts their body mass 30 years later. J Pediatr. 2013;162(1):90-93. doi: 10.1016/j.jpeds.2012.06.049 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Epstein LH, Anzman-Frasca S. The promise of early childhood self-regulation for obesity prevention. Pediatrics. 2017;139(5):e20170389. doi: 10.1542/peds.2017-0389 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Whitaker RC, Gooze RA. Self-regulation and obesity prevention: a valuable intersection between developmental psychology and pediatrics. Arch Pediatr Adolesc Med. 2009;163(4):386-387. doi: 10.1001/archpediatrics.2008.576 [DOI] [PubMed] [Google Scholar]
- 8.Golan M. Good-enough parenting, self-regulation, and the management of weight-related problems In: Bagchi D, ed. Global Perspectives on Childhood Obesity: Current Status, Consequences and Prevention. Cambridge, MA: Academic Press; 2010:43-55. [Google Scholar]
- 9.Kopp CB. Antecedents of self-regulation: a developmental perspective. Dev Psychol. 1982;18(2):199-214. doi: 10.1037/0012-1649.18.2.199 [DOI] [Google Scholar]
- 10.Aparicio E, Canals J, Arija V, De Henauw S, Michels N. The role of emotion regulation in childhood obesity: implications for prevention and treatment. Nutr Res Rev. 2016;29(1):17-29. doi: 10.1017/S0954422415000153 [DOI] [PubMed] [Google Scholar]
- 11.Lumeng JC, Miller AL, Horodynski MA, et al. Improving self-regulation for obesity prevention in head start: a randomized controlled trial. Pediatrics. 2017;139(5):e20162047. doi: 10.1542/peds.2016-2047 [DOI] [PubMed] [Google Scholar]
- 12.Murray DW, Rosanbalm K, Christopoulos C. Self-Regulation and Toxic Stress Report 3: A Comprehensive Review of Self-Regulation Interventions From Birth Through Young Adulthood. Washington, DC: Office of Planning, Research and Evaluation, Administration for Children and Families, US Dept of Health and Human Services; 2016. [Google Scholar]
- 13.Anderson SE, Keim SA. Parent-child interaction, self-regulation, and obesity prevention in early childhood. Curr Obes Rep. 2016;5(2):192-200. doi: 10.1007/s13679-016-0208-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Francis LA, Susman EJ. Self-regulation and rapid weight gain in children from age 3 to 12 years. Arch Pediatr Adolesc Med. 2009;163(4):297-302. doi: 10.1001/archpediatrics.2008.579 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Seeyave DM, Coleman S, Appugliese D, et al. Ability to delay gratification at age 4 years and risk of overweight at age 11 years. Arch Pediatr Adolesc Med. 2009;163(4):303-308. doi: 10.1001/archpediatrics.2009.12 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Anderson SE, Sacker A, Whitaker RC, Kelly Y. Self-regulation and household routines at age three and obesity at age eleven: longitudinal analysis of the UK Millennium Cohort Study. Int J Obes (Lond). 2017;41(10):1459-1466. doi: 10.1038/ijo.2017.94 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Graziano PA, Kelleher R, Calkins SD, Keane SP, Brien MO. Predicting weight outcomes in preadolescence: the role of toddlers’ self-regulation skills and the temperament dimension of pleasure. Int J Obes (Lond). 2013;37(7):937-942. doi: 10.1038/ijo.2012.165 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Kochanska G, Murray KT, Harlan ET. Effortful control in early childhood: continuity and change, antecedents, and implications for social development. Dev Psychol. 2000;36(2):220-232. doi: 10.1037/0012-1649.36.2.220 [DOI] [PubMed] [Google Scholar]
- 19.Connell LE, Francis LA. Positive parenting mitigates the effects of poor self-regulation on body mass index trajectories from ages 4-15 years. Health Psychol. 2014;33(8):757-764. doi: 10.1037/hea0000014 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Levitan RD, Rivera J, Silveira PP, et al. ; MAVAN Study Team . Gender differences in the association between stop-signal reaction times, body mass indices and/or spontaneous food intake in pre-school children: an early model of compromised inhibitory control and obesity. Int J Obes (Lond). 2015;39(4):614-619. doi: 10.1038/ijo.2014.207 [DOI] [PubMed] [Google Scholar]
- 21.Bethel J, Green JL, Kalton G, Nord C. Early Childhood Longitudinal Study, Birth Cohort (ECLS-B), Sampling: Volume 2 of the ECLS-B Methodology Report for the 9-Month Data Collection, 2001-02 (NCES 2005-147). Washington, DC: National Center for Education Statistics, Institute of Education Sciences, US Dept of Education; 2005. [Google Scholar]
- 22.Andreassen C, Fletcher P. Early Childhood Longitudinal Study, Birth Cohort (ECLS-B): Psychometric Report for the 2-Year Data Collection (NCES 2007-084). Washington, DC: National Center for Education Statistics, Institute of Education Sciences, US Dept of Education; 2007. [Google Scholar]
- 23.Bayley N. Bayley Scales of Infant Development. 2nd ed San Antonio, TX: Psychological Corporation; 1993. [Google Scholar]
- 24.Najarian M, Snow K, Lennon J, Kinsey S. Early Childhood Logitudinal Study, Birth Cohort (ECLS), Preschool-Kindergarten 2007 Psychometric Report (NCES 2010-009). Washington, DC: National Center for Education Statistics, Institute for Education Sciences, US Dept of Education; 2010. [Google Scholar]
- 25.Kuczmarski RJ, Ogden CL, Guo SS, et al. 2000 CDC growth charts for the United States: methods and development. Vital Health Stat 11. 2002;(246):1-190. [PubMed] [Google Scholar]
- 26.Clayton JA, Tannenbaum C. Reporting sex, gender, or both in clinical research? JAMA. 2016;316(18):1863-1864. doi: 10.1001/jama.2016.16405 [DOI] [PubMed] [Google Scholar]
- 27.Anderson SE, Whitaker RC. Prevalence of obesity among US preschool children in different racial and ethnic groups. Arch Pediatr Adolesc Med. 2009;163(4):344-348. doi: 10.1001/archpediatrics.2009.18 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
eTable 1: Sociodemographic characteristics of Early Childhood Longitudinal Study–Birth Cohort participants included in analytic sample compared to those not included in analytic sample
eTable 2: Sociodemographic characteristics by toddler self-regulation: boys
eTable 3: Sociodemographic characteristics by toddler self-regulation: girls
eTable 4: Sociodemographic characteristics by obesity status at 5.5 years: boys
eTable 5: Sociodemographic characteristics by obesity status at 5.5 years: girls
eTable 6: Association of toddler self-regulation in relation to risk for obesity at 5.5 years in boys and girls: alternative analytic specification using gender-specific self-regulation quartile definitions
eTable 7: Association of toddler self-regulation in relation to risk for obesity at 5.5 years in boys and girls: alternative analytic specification using multiple imputation
eFigure 1: Participant flow and analytic sample definition
eFigure 2: Association between toddler adaptability, persistence, attention, and frustration tolerance and obesity prevalence at 5.5 years for boys and girls