Childhood Self-Control and Adolescent Obesity: Evidence from Longitudinal Data on a National Cohort

Abstract

Background:

Whether self-control at school entry and changes in self-control in the early school years are predictive of BMI change and obesity onset over the next 8 years using longitudinal data on a nationally representative sample of US children.

Methods:

Data from the Early Childhood Longitudinal Study–Kindergarten Class 1998 were analyzed. The analysis sample included 7060 children with data from kindergarten (mean age = 6 years) until eighth grade (mean age = 14 years). Self-control was assessed using teacher-reported scales. Child BMI and obese status (BMI ≥95th percentile for age and gender) were computed from height and weight measurements. Weighted linear and logistic regression models were estimated, adjusting for child's cognitive ability and a rich set of child and family covariates.

Results:

High self-control in kindergarten was associated with lesser BMI increase (p < 0.05) and 43% lower odds (adjusted OR [AOR] = 0.57, 95% confidence interval [CI]: 0.38–0.86) of new onset obesity over the subsequent 8 years. The beneficial effect of high self-control in kindergarten emerged between fifth and eighth grade. Among children with low self-control in kindergarten, increase in self-control between kindergarten and fifth grade was associated with a 1.42 kg/m² (95% CI: −2.82 to −0.02) lesser increase in BMI and 66% lower odds of new obesity onset (AOR = 0.34; 95% CI: 0.14–0.83), between kindergarten and eighth grade.

Conclusions:

Low self-control at school entry is an important risk factor for unhealthy BMI increase during the transition to adolescence. Increase in self-control in the early school years may prevent unhealthy BMI increase and obesity in adolescence.

Introduction

Childhood obesity remains a top US public health concern.^1,2 While research has identified factors contributing to population-level increases in unhealthy weight (e.g., inexpensive energy-dense foods, built environment, sedentary culture),³ individual variability remains considerable even after accounting for population-level and other factors, such as race-ethnicity and socioeconomic status.⁴ Better understanding of what contributes to this individual variability is important for improving the effectiveness of childhood obesity prevention programs.

One individual-level factor, self-control, has been of longstanding scientific interest as a potentially important predictor of health behaviors and outcomes.⁵ Self-control develops between early childhood and adolescence as brain regions associated with these skills develop.⁶ It is generally viewed as a personality trait capturing one's ability to control impulses, although related, it is different, constructs have been studied across disciplines include self-regulation, delay-of-gratification, executive function, intertemporal choice, and holistic measures of children's socioemotional functioning.⁷

Seminal work by Mischel et al.^8–10 famously known as the “Marshmallow test,” showed that preschoolers who chose to receive a larger treat later instead of a smaller treat now had higher academic achievement and better socioemotional outcomes a decade later. More recently, two longitudinal cohort studies in non-US samples^11,12 found that more self-control in childhood (using a composite score based on parent-, teacher-, self-, and interviewer-reports in one study, and hypothetical monetary choices in the other) was associated with better school performance, earnings, employment, and overall health, and less criminal behavior.

Low self-control in childhood may also be an important risk factor for obesity because unhealthy diet and activity behaviors often provide more immediate gratification than healthy behaviors. Studies suggest that low self-control in childhood may predict greater weight gain and risk of obesity during adolescence. However, this literature is limited by small, unrepresentative, or non-US samples.^11–20 In particular, US data are important because, even though childhood obesity has become a world-wide phenomenon, its prevalence in the United States is among the world's highest,²¹ and its drivers may differ across countries due to institutional, cultural, and other differences.

Also, most of these studies compare average changes in weight-related outcomes across children with different levels of self-control in childhood, instead of examining changes in self-control within children. While self-control generally increases with age, change varies considerably across individuals, which leaves open whether changes in self-control can alter weight gain trajectories and offer opportunities for obesity prevention.

Finally, the link between self-control and key mediators, including diet, physical activity, and sedentary behaviors, has been studied primarily in cross-sectional and/or unrepresentative samples, and evidence from national longitudinal data is unavailable.^22,23

In this study, we use longitudinal data on a nationally representative sample of US children from the Early Childhood Longitudinal Study–Kindergarten Class of 1998–1999 (ECLS-K) to examine whether children's self-control when they start school and changes in self-control during the early school years are predictive of changes in BMI and development of obesity by the end of middle school. Our self-control measure is based on teacher ratings of children on items that capture attention control as well as control of actions and emotions in interactions. We hypothesize that greater self-control at school entry and increase in self-control in the early school years will be associated with less BMI increase and lower odds for developing obesity between kindergarten and the end of middle school. As a secondary aim, we examine whether more self-control at school entry is associated with better dietary and physical activity behaviors at the end of middle school.

Methods

Data and Sample

We analyzed data from the Early Childhood Longitudinal Study–Kindergarten Class of 1998–1999 (ECLS-K). The ECLS-K used a multistage probability sample design to select a nationally representative sample of children attending kindergarten in the United States in 1998–99. The primary sampling units (PSUs) were geographic areas consisting of counties or groups of counties. The second-stage units were public and private schools within sampled PSUs. The third and final stage units were students within schools. Two independent sampling strata were formed within each school, one containing Asian or Pacific Islander students and the second containing all other students. Within each stratum, students were selected using equal probability systematic sampling. In general, the target number of children sampled at any one school was 24. A total of 1277 schools were sampled, of which 1018 participated during the kindergarten school year. Additional details about the ECLS-K study design are available elsewhere.²⁴ Data were collected in spring of kindergarten (1999), first, third, fifth, and eighth (2007) grades on children's cognitive, health, and developmental outcomes from parents, teachers, and schools.

In a longitudinal study, sample attrition due to nonresponse or ineligibility is expected. The sample of respondents decreases with each round of data collection. In the ECLS-K first-grade to fifth-grade data collections, subsampling of movers was used to reduce data collection costs. Specifically, in the first-, third-, and fifth-grade waves, a subsample of children who moved out of their original (kindergarten) sample schools were flagged to not be followed into their new schools, resulting in a decrease in the sample. In first- and third-grade waves, 21,292 kindergarten respondents were eligible to participate. In the fifth-grade wave, the number of eligible respondents reduced to 12,129 as remaining children excluded due to mover subsampling or ineligibility (respondent died or moved out of the country). The sample design for the final wave (eighth grade) called for including all 12,129 children eligible after fifth grade (regardless of their fifth-grade response status) and following all movers without any subsampling. In eighth grade, there were 9360 children with any child assessment data, which is 77% of the kindergarten respondents who were eligible for eighth grade data collection.

Tourangeau et al.²⁵ conducted an examination of the potential for nonresponse bias (due to children who were respondents in the base year but were nonrespondents in the eighth-grade data collection) using three methods: (1) comparison of respondents and nonrespondents using the available sample frame, (2) multivariate analysis to identify the characteristics of cases most likely to respond, and (3) analysis of attrition bias applicable to longitudinal studies. They found that nonresponse bias of the estimates from the eighth grade was present but small. In most cases, the use of a mover status category in the fifth-grade nonresponse adjustment weighting helped reduce the bias, and the sample-based raking to the characteristics of the base-year children further reduced the nonresponse bias and variance of the estimates.

Therefore, the proper use of the ECLS-K weights in data analysis is needed to minimize the effect of nonresponse bias. The ECLS-K weights are designed to compensate for differential probabilities of selection at each sampling stage and to adjust for the effects of nonresponse. Nonresponse adjustment uses both school (i.e., school affiliation, locale, census region, school enrollment classified into size category) and child (i.e., year of birth, sex, and race/ethnicity) characteristics. Meanwhile, mover adjustment uses variables that include response status for child assessments and parent interviews in previous rounds as well as school type, household type, language minority status, and whether the child was homeschooled.

Our analysis sample includes 7060 children with longitudinal data from kindergarten (mean age = 6 years) until eighth grade (mean age = 14 years). Descriptive statistics for the sample, weighted to adjust for the multistage sampling design, attrition, and nonresponse, are reported in Table 1. Additional sample details and assessment of nonresponse bias in our analysis sample are provided in the Supplementary Data.

Table 1.

Summary Statistics for Selected Measures

	Unweighted N	Mean (95% CI)
BMI and obesity
K
BMI (kg/m²)	8446	16.5 (16.4–16.6)
BMI z-score	8446	0.42 (0.38–0.47)
Overweight^a	8446	14.3% (12.9–15.8)
Obese	8446	12.5% (11.3–13.7)
Fifth grade
BMI (kg/m²)	8306	20.7 (20.5–20.8)
BMI z-score	8306	0.62 (0.57–0.67)
Overweight^a	8306	18.9% (16.9–20.8)
Obese	8306	20.8% (19.4–22.2)
Eighth grade
BMI (kg/m²)	7997	23.1 (22.9–23.3)
BMI z-score	7997	0.64 (0.59–0.68)
Overweight^a	7997	16.9% (15.6–18.2)
Obese	7997	19.5% (17.9–21.0)
BMI change
Between K-5	8284	4.2 (4.0–4.3)
Between K-8	7985	6.6 (6.4–6.8)
New onset obesity among not obese in kindergarten
Between K-5	8338	11.8% (10.7–13.1)
Between K-8	8061	11.9% (10.5–13.4)
Self-control scores (range 1–4)
Score in K	8160	3.2 (3.16–3.21)
Score in fifth grade	7238	3.2 (3.18–3.24)
Diet and activity behaviors in eighth grade
Times per week ate
Fast food	7937	2.6 (2.4–2.7)
Sugar sweetened beverages	8072	5.6 (5.3–5.9)
Fruits and vegetables	7946	16.7 (16.2–17.3)
Hours per week
Watched television	8216	24.5 (23.6–25.4)
Played videogames	8238	12.4 (11.6–13.1)
Vigorous physical activity (days per week)	8262	4.5 (4.4–4.7)
Selected kindergarten covariates
Male	8477	51.9% (49.8–54.0)
White	8477	57.6% (54.2–60.9)
Black	8477	17.0% (14.8–19.3)
Hispanic	8477	18.1% (15.5–20.6)
Asian	8477	2.9% (2.4–3.5)
Other	8477	4.3% (2.5–6.2)
Living in single parent household	8477	21.2% (19.2–23.2)
Living in city	8477	36.8% (32.7–40.8)
Living in town	8477	42.0% (35.7–48.3)
Living in rural area	8477	21.2% (16.1–26.3)

Overweight is defined as age- and sex-adjusted BMI percentile ≥85th and <95th percentile. All estimates are weighted and nationally representative of 3.1 million kindergarteners entering school in 1998–1999. Other covariates not summarized here include child's age, cognitive ability, household socioeconomic status, mother's age, number of siblings, school lunch participation, urbanicity, and census region.

K, kindergarten.

Exposure and Outcome

Self-control

Children's self-control was measured only from kindergarten through fifth grade via survey items administered to teachers. Our primary measure is the four-item self-control scale, adapted from the Social Skills Rating System,²⁶ that captures attention control and control of emotions and actions. The items assess the extent to which the child can: control his or her temper, respect others' property, accept his or her peers' ideas, and handle peer pressure.

Responses to the four items were on a four-point Likert-metric (never, sometimes, often, very often), with higher scores indicating that the child was engaged in those behaviors more frequently (and therefore exhibited more self-control). The scores were averaged across items in the scale to construct a scale score that ranged from 1 (lowest) to 4 (highest) points with mean (standard deviation [SD]) of 3.2 (0.6) in kindergarten. The scale has high internal consistency—alpha reliability coefficients ranging from 0.79 to 0.91 across waves (0.80 in kindergarten).²⁴ Self-control in kindergarten was first measured as a continuous variable and then converted into a categorical variable based on the distribution: <3 (low), 3 to <4 (medium) and 4 (high). Change in self-control between kindergarten and fifth grade was measured as a three-category indicator derived from the continuous variable—score remained unchanged, score increased by any amount, and score decreased by any amount.

BMI and obesity

Trained ECLS-K staff measured children's height and weight in each wave using the Shorr Board (accuracy = 0.01 cm) and a Seca digital bathroom scale (model 840; accuracy = 0.1 kg), respectively. Height and weight were each measured twice to minimize error. Composite height and weight were computed by ECLS-K staff from the two readings, which were then used to compute the BMI. We classified children as obese if their BMI was at least the 95th percentile based on age- and gender-specific growth charts issued by the Centers for Disease Control and Prevention. Our primary outcome measures were raw BMI in eighth grade, an indicator for obese status in eighth grade, and change in raw BMI between kindergarten and eighth grade. Changes in raw BMI are considered better measures of adiposity change in children and adolescents than changes in percentiles or z-scores.^27–30

Diet and activity behaviors

Dietary intake in eighth grade was assessed via food frequency questionnaire administered to the child. Measures included the numbers of times during the last 7 days that the child reported consuming fruits and vegetables, fast food, and sugar-sweetened beverages (SSBs). Physical activity was measured as child-reported days per week of at least 20 minutes of vigorous exercise. Sedentary behavior was measured as child-reported numbers of hours per week of television viewing and playing videogames.

Statistical Analysis

First, we estimated linear and logistic regression models to examine whether self-control in kindergarten was associated with children's BMI change and obesity onset between kindergarten and fifth grade (K-5)as well as between kindergarten and eighth grade (K-8). For BMI change, we used the entire analytic sample. For obesity onset, we used the subsample (87.7%) of children who were not already obese in kindergarten. We examined fifth and eighth grades as alternate end points, since these grade levels captured the last wave of elementary school and middle school for our cohort, respectively. We estimated both unadjusted and adjusted regression models. The adjusted models controlled for covariates in kindergarten spanning individual-, family-, and geographic-level risk factors for obesity, including child's race-ethnicity, gender, household socioeconomic scale quintiles (details in Supplementary Data), indicator for single-parent household, number of siblings in the household, whether the child usually ate a school lunch, and indicators for urbanicity (urban, suburban, rural) and Census region. We also controlled for child's cognitive ability using scores on standardized tests (details in Supplementary Data) administered to all children.

Next, we estimated the relationship between change in self-control between K-5 and change in BMI between K-8 using linear regressions, separately for children with low, medium, or high self-control in kindergarten. In each of the three subsamples, we examined whether children whose self-control increased (or decreased) experienced significantly different BMI change between K-8 relative to children whose self-control did not change. For obesity onset, we estimated corresponding logistic regression models. For time-varying covariates, we included data from both kindergarten and eighth grade.

Finally, we examined whether self-control in kindergarten was associated with obesogenic behaviors in eighth grade. We estimated negative binomial regression models for the dietary and sedentary outcomes (since there was evidence of overdispersion) and Poisson regression models for physical activity that adjusted for BMI and other covariates in kindergarten. Since diet and activity measures were not consistently measured across wave and were not available in the early grades (diet measures), we could not examine changes in these behaviors between K-8.

We tested all models for interaction by gender, but report combined results because effect modification was not statistically significant.

All analyses were conducted in STATA version 14.1 (College Station, TX). Data were weighted to adjust for multistage sampling design, attrition, and nonresponse and were representative of 3.1 million US children entering kindergarten in 1998–1999. The study involved secondary analyses of existing data and was approved by the University of Southern California's IRB (UP-15-00246).

Results

Descriptive Results

In kindergarten, mean BMI was 16.5 kg/m², and 12.5% (95% confidence interval [CI]: 11.3–13.7) were obese (Table 1). Between K-8, the mean BMI change was +6.6 kg/m² (95% CI: 6.4–6.8) and obesity prevalence increased to 19.5%.

In kindergarten, the mean (SD) of self-control scores was 3.2 (0.6). Kindergarten self-control scores were below 3 (low) for 29.9%, at least 3 but below 4 (medium) for 54.2%, and 4 (high) for 15.9% of the cohort (Table 2). Between K-5, self-control scores increased among 45.3% of children, decreased among 40.6% and remained stable among the rest.

Table 2.

Change in Self-Control Scores between K-5, by Self-Control in Kindergarten

	Total	Score increased between K-5		Score decreased between K-5		No change between K-5
Self-control in K	% (95% CI)	% (95% CI)	Mean change in score (95% CI)	% (95% CI)	Mean change in score (95% CI)	% (95% CI)
Low	29.9 (27.8 to 32.1)	21.5 (19.7 to 23.4)	0.83 (0.80 to 0.87)	5.5 (4.6 to 6.6)	−0.48 (−0.53 to 0.42)	2.9 (2.2 to 3.6)
Medium	54.2 (52.0 to 56.4)	23.8 (22.0 to 25.7)	0.46 (0.45 to 0.48)	22.8 (21.2 to 24.5)	−0.64 (−0.67 to 0.60)	7.6 (6.8 to 8.5)
High	15.9 (14.6 to 17.4)	—	—	12.3 (11.1 to 13.7)	−0.74 (−0.79 to 0.69)	3.6 (2.9 to 4.4)
Total	100	45.3		40.6		14.1

All estimates are weighted and representative of 3.1 million kindergarteners entering school in 1998–1999.

Figure 1A shows obesity prevalence in kindergarten, first, third, fifth, and eighth grades, by self-control in kindergarten. Higher self-control in kindergarten was associated with lower obesity in each subsequent wave, and the association became statistically significant in eighth grade. While obesity increased significantly between K-5 for all children, there was a decline in obesity from fifth to eighth grade among all children, particularly in those with high self-control in kindergarten, although the decline was never statistically significant. Figure 1B shows the same graph for the subsample of children who were not obese in kindergarten (i.e., obesity prevalence in grades 1, 3, 5, and 8 represents onset of obesity after kindergarten). Again, we find that the negative association between kindergarten self-control and subsequent onset of obesity starts to emerge in fifth grade, but becomes significant only in eighth grade.

Figure 1.

Obesity prevalence from Kindergarten through eighth grade, by self-control in kindergarten (K). (A) Full sample. (B) Sample not obese in K.

Self-Control in Kindergarten and BMI Change and Obesity Onset between K-5 and K-8

Between K-5 (Table 3, Columns 1 and 2), self-control in kindergarten was not associated with BMI change overall. There was, however, some association with obesity onset among children not obese in kindergarten (Row A)—medium self-control in kindergarten, but not high self-control, was associated with significantly lower odds of obesity onset (adjusted odds ratio [AOR] = 0.64; 95% CI: 0.47–0.86). Models that used continuous self-control scores also suggested no association with BMI change but a significant association with obesity onset (Row B).

Table 3.

Estimated Association between Self-Control in Kindergarten and BMI Change and Obesity Onset between K-5 and K-8

	K-5		K-8
	Change in BMI^a	Obese in fifth grade	Change in BMI^a	Obese in eighth grade
	(1)	(2)	(3)	(4)
	b (95% CI)	AOR (95% CI)	b (95% CI)	AOR (95% CI)
A. Self-control in kindergarten (categorical)
Low (score <3) (reference)	0.00	1.00	0.00	1.00
Medium (3 ≤ score <4)	−0.28 (−0.63 to 0.06)	0.63^*** (0.47 to 0.86)	−0.61^** (−1.09 to 0.13)	0.71 (0.46 to 1.07)
High ( = 4)	−0.09 (−0.50 to 0.32)	0.73 (0.48 to 1.11)	−0.72^** (−1.27 to 0.17)	0.57^*** (0.38 to 0.86)
B. Self-control in kindergarten (continuous)	−0.07 (−0.30 to 0.16)	0.78^** (0.62 to 0.98)	−0.39^** (−0.74 to 0.05)	0.71^*** (0.55 to 0.92)
Unweighted N	7344	6489	7058	6227
Sample	All	Not obese in K	All	Not obese in K

All models are weighted and control for child's race-ethnicity, gender, birth order, birth weight, and kindergarten-specific covariates, including age, cognitive ability, household socioeconomic status, single parent household status, mother's age, number of siblings, school lunch participation, urbanicity, and census region.

These models also control for child's BMI in kindergarten.

***

p < 0.01, ^**p < 0.05, ^*p < 0.1.

Between K-8 (Table 3, Columns 3 and 4), children with medium and high self-control in kindergarten had significantly lesser increase in BMI: −0.61 kg/m² (95% CI: −1.09 to −0.13) and −0.72 kg/m² (95% CI: −1.27 to −0.17), respectively—than children with low self-control in kindergarten (Row A). Moreover, among children not obese in kindergarten, those with high self-control had 43% lower odds of becoming obese (AOR = 0.57, 95% CI: 0.38–0.86). Models that used continuous self-control scores suggested associations with K-8 BMI change and with K-5 and K-8 obesity onset (Row B).

Full regression results from unadjusted and adjusted models are reported in Supplementary Tables S1 and S2.

Changes in Self-Control between K-5 and Changes in BMI and Obesity Onset between K-8

Among children with low self-control in kindergarten, those whose self-control increased experienced 1.42 kg/m² (95% CI: −2.82 to −0.02) lesser increase in BMI, and had 66% lower odds of becoming obese (AOR = 0.34; 95% CI: 0.14–0.83), between K-8 compared with those with no change in self-control (Table 4).

Table 4.

Association between Change in Self-Control between K-5 and Changes in BMI and Obesity Onset between K-8, by Self-Control in Kindergarten

Dependent variable->	Change in BMI between K-8 b (95% CI)			Became obese between K-8^a AOR (95% CI)
Change in self-control between K-5
No change in score (reference)	0.0	0.0	0.0	1.0	1.0	1.0
Score increased	−1.42^** (−2.82 to 0.02)	−0.47 (−1.23 to 0.30)	—^b	0.34^** (0.14 to 0.83)	0.78 (0.41 to 1.47)	—^b
Score decreased	−1.10 (−2.72 to 0.52)	−0.24 (−0.98 to 0.50)	0.58^** (0.01 to 1.16)	0.52 (0.19 to 1.41)	0.85 (0.46 to 1.56)	3.04^** (1.11 to 8.36)
Unweighted N	1428	3417	1145	1253	3018	1029
Sample: self-control category in kindergarten	Low	Medium	High	Low	Medium	High

Each column represents a separate regression. All estimates are from weighted and adjusted models. The following time-varying covariates for kindergarten and eighth grade are included as controls in all the models: child's cognitive ability, age, household socioeconomic status, single parent household status, mother's age, number of siblings, school meal participation, urbanicity, and census region. In addition, time-invariant covariates, including child's race-ethnicity, sex, birth weight were added in all models.

Sample is restricted to those not obese in kindergarten.

Since high self-control is defined as score = 4, by definition the score cannot increase and is therefore not estimable.

***

p < 0.01, ^**p < 0.05, ^*p < 0.1.

b = coefficient estimate from ordinary least squares regression.

AOR, adjusted odds ratio; CI, confidence interval.

Among children with medium self-control in kindergarten, there was no significant relationship between an increase or decrease in self-control and BMI change or obesity onset.

Among children with high self-control in kindergarten, those whose self-control decreased experienced a 0.58 kg/m² (95% CI: 0.01–1.16) greater increase in BMI between kindergarten and eighth grade, compared with those with no change in self-control. These children also had almost three times higher odds of becoming obese by eighth grade (AOR = 3.04; 95% CI: 1.11–8.36) compared to children who continued to have high self-control.

In results not shown, we also estimated models that used a continuous measure of change in self-control between K-5 but did not find a significant linear relationship between self-control change between K-5 and BMI change or obesity onset between K-8.

Self-Control in Kindergarten and Diet and Activity Behaviors in Eighth Grade

Relative to children with low self-control in kindergarten, those with high self-control had lower weekly consumption of fast food and SSBs (incidence rate ratio [IRR] = 0.82; 95% CI: 0.72–0.93), played fewer hours per week of videogames (IRR = 0.76, 95% CI: 0.65–0.89), and did vigorous physical activity on more days per week (Table 5). Self-control was not associated with television watching or fruit and vegetable consumption. Children with medium self-control in kindergarten played slightly fewer hours per week of videogames (IRR = 0.90; 95% CI: 0.80–1.01) than children with low self-control, but there was no association with other diet or activity behaviors in eighth grade. Models that used continuous self-control yielded similar results.

Table 5.

Association between Self-Control in Kindergarten and Diet and Activity Behaviors in Eighth Grade

	IRR (95% CI)
	Fast food (times/week)	SSBs (times/week)	Fruits and vegetables (times/week)	Television (hours/week)	Videogames (Hours/week)	Vigorous activity (days/week)
A. Self-control in K (categorical)
Low (score <3) (reference)	1.00	1.00	1.00	1.00	1.00	1.00
Medium (3 ≤ score <4)	0.94 (0.82–1.07)	0.91 (0.80–1.03)	0.99 (0.90–1.10)	0.98 (0.92–1.06)	0.90^* (0.80–1.01)	1.01 (0.97–1.06)
High ( = 4)	0.82^*** (0.72–0.93)	0.82^*** (0.71–0.94)	0.99 (0.88–1.11)	0.98 (0.88–1.09)	0.76^*** (0.65–0.89)	1.06^*** (1.02–1.11)
B. Self-control in K (continuous)	0.90^** (0.83–0.99)	0.88^*** (0.81–0.96)	0.98 (0.91–1.05)	0.98 (0.92–1.04)	0.87^*** (0.80–0.93)	1.03^** (1.005–1.06)
Unweighted N	7046	7170	7060	7300	7310	7330

All estimates are from weighted and covariate-adjusted models. Covariates include child's race-ethnicity, gender, and birth weight as well as kindergarten measures of child's cognitive ability, age, household socioeconomic status, single parent household status, mother's age, number of siblings, school meal participation, urbanicity, and census region.

***

p < 0.01, ^**p < 0.05, ^*p < 0.1.

CI, confidence interval; IRR, incidence rate ratio; SSBs, sugar-sweetened beverages.

Discussion

This is the first study to use longitudinal data on a nationally representative cohort of US children to examine links between self-control in early childhood and children's BMI change, obesity onset, and obesogenic behaviors in early adolescence. Our results show that children with higher self-control in kindergarten have lesser BMI increase and lower odds of developing obesity over the next 8 years.

Our results, which show more consistent associations between early self-control and BMI at eighth grade than fifth, suggest that obesity-related benefits of higher self-control in kindergarten may not fully emerge until middle school, when children are transitioning to adolescence. When children are starting to make health behavior decisions independently of their parents, self-control is likely to play an increasingly important role. Growing evidence suggests that habit is a potential mechanism through which self-control translates into healthy behaviors.³¹ The extent to which healthy habits initiated at younger ages solidify by early adolescence may explain these associations.

Our results also suggest an association between self-control in kindergarten and obesity onset, but not mean of BMI, at fifth grade. This suggests that the protective effects of self-control on unhealthy BMI increase may first manifest in fifth grade in children who are at higher risk for obesity and may subsequently benefit children even with lower obesity risk. Future research should explore reasons for these differences.

We also find that self-control often changes during elementary school, and that these changes may influence downstream weight change. An increase in self-control was associated with lower BMI increase and lower likelihood of developing obesity by eighth grade.

Our findings are consistent with evidence from neurobiological and behavioral research on brain development during childhood.³² This research has found that preschool years are a renaissance period in the development and interconnection of regions of the brain that support skills important for self-control (e.g., following rules, reasoning, suppressing impulses, decision making, attention).⁶ Therefore, self-control at school entry might be an important benchmark. Moreover, transition to adolescence is a critical period of relative plasticity before adulthood—if improving self-control in childhood can confer protection through adolescence, long-term benefits might be profound. Our findings are similar to those of Moffitt et al.¹¹ who found that children who became more self-controlled from childhood to young adulthood had better nonobesity-related outcomes by age 32.

Finally, we also find that early self-control is related to adolescent dietary and activity behaviors, including lower consumption of fast food and SSBs, less time spent playing videogames, and more days per week of vigorous physical activity in eighth grade. These findings are consistent with the limited prior literature^22,33 and help connect potential dots between low self-control and unhealthy weight gain.

Our findings have important implications for obesity prevention efforts. Given that overweight and obese adolescents have a significantly higher risk of becoming obese adults,^34–36 our findings suggest an important opportunity for early identification of high-risk children. Moreover, our findings add to the growing evidence that self-control in childhood is malleable³⁷ and might be a useful target for interventions. If improving self-control has multiple benefits with respect to health and academic achievement beyond obesity prevention, as some research suggests,^11,12 such interventions might be extremely cost-effective. Several interventions seeking to improve self-control in children in educational, home, and clinical settings show promise.^38–41 Whether such interventions will ultimately impact obesogenic behaviors and unhealthy weight gain in childhood and beyond should be an important area for future research.

Limitations

Our study's limitations are as follows. First, self-control measures in the ECLS-K were based on teacher assessments, which may be less objective than observation of specific cognitive tasks but may capture a more holistic measure of self-control. Second, self-control measures were only collected between kindergarten and fifth grade and do not account for changes in self-control after fifth grade. Third, there are no clinical or established thresholds for categorizing self-control scores, and therefore, our categorization into high, medium, and low may be somewhat arbitrary. This might contribute to some of the results we found. In particular, medium self-control in kindergarten was significantly associated with odds of obesity onset by fifth grade but not by eighth grade. Also, medium self-control in kindergarten, but not high self-control, was significantly associated with odds of obesity onset in fifth grade. Fourth, diet and activity behaviors in eighth grade were self-reported. Fifth, although BMI was based on measurements, it can be a limited measure of adiposity.⁴² Sixth, there is potential for omitted variables; the ECLS-K did not collect information on relevant factors such as puberty and sleep, which may be linked with both obesity and self-regulation.⁴³ Finally, while our data are longitudinal, the study design is observational and therefore cannot establish causality.

Conclusion

Low self-control in children in kindergarten is a significant predictor of unhealthy BMI change and development of obesity during their transition to adolescence. Increase in self-control during the early school years might be protective against unhealthy BMI change. Obesity prevention efforts may benefit from tracking children's self-control in the early school years and developing interventions that promote self-control in childhood.

Footnotes

Acknowledgment

None.

Author Disclosure Statement

No competing financial interests exist.

References

Ogden

, Carroll

, Kit

, Flegal

. Prevalence of childhood and adult obesity in the United States, 2011–2012. JAMA, 2014; 311:806–814.

Datar

, Chung

. Changes in socioeconomic, racial/ethnic, and sex disparities in childhood obesity at school entry in the United States. JAMA Pediatr, 2015; 169:696–697.

Anderson

, Butcher

. Childhood obesity: Trends and potential causes. Future Child, 2006; 16:19–45.

Courtemanche

, Pinkston

, Ruhm

, Wehby

. Can changing economic factors explain the rise in obesity?. South Econ J, 2016; 82:1266–1310.

Lawless

, Drichoutis

, Nayga

. Time preferences and health behaviour: A review. Agric Food Econ, 2013; 1:1–19.

Tarullo

, Obradovic

, Gunnar

. Self-control and the developing brain. Zero to Three, 2009; 29:31–37.

Duckworth

, Kern

. A meta-analysis of the convergent validity of self-control measures. J Res Pers, 2011; 45:259–268.

Mischel

, Shoda

, Peake

. The nature of adolescent competencies predicted by preschool delay of gratification. J Pers Soc Psychol, 1988; 54:687–696.

Mischel

, Shoda

, Rodriguez

. Delay of gratification in children. Science, 1989; 244:933–938.

10.

Shoda

, Mischel

, Peake

. Predicting adolescent cognitive and self-regulatory competencies from preschool delay of gratification: Identifying diagnostic conditions. Dev Psychol, 1990; 26:978–986.

11.

Moffitt

, Arseneault

, Belsky

, et al. A gradient of childhood self-control predicts health, wealth, and public safety. Proc Natl Acad Sci U S A, 2011; 108:2693–2698.

12.

Golsteyn

BHH

, Grönqvist

, Lindahl

. Adolescent time preferences predict lifetime outcomes. Econ J, 2014; 124:F739–F761.

13.

Schlam

, Wilson

, Shoda

, Mischel

, Ayduk

. Preschoolers' delay of gratification predicts their body mass 30 years later. J Pediatr, 2013; 162:90–93.

14.

Graziano

, Kelleher

, Calkins

, et al. Predicting weight outcomes in preadolescence: The role of toddlers' self-regulation skills and the temperament dimension of pleasure. Int J Obes (Lond), 2013; 37:937–942.

15.

Duckworth

, Tsukayama

, Geier

. Self-controlled children stay leaner in the transition to adolescence. Appetite, 2010; 54:304–308.

16.

Tsukayama

, Toomey

, Faith

, Duckworth

. Self-control protects against overweight status in the transition from childhood to adolescence. Arch Pediatr Adolesc Med, 2010; 164:631–635.

17.

Seeyave

, Coleman

, Appugliese

, et al. Ability to delay gratification at age 4 years and risk of overweight at age 11 years. Arch Pediatr Adolesc Med, 2009; 163:303–308.

18.

Francis

, Susman

. Self-regulation and rapid weight gain in children from age 3 to 12 years. Arch Pediatr Adolesc Med, 2009; 163:297–302.

19.

Sutter

, Kocher

, Glätzle-Rüetzler

, Trautmann

. Impatience and uncertainty: Experimental decisions predict adolescents' field behavior. Am Econ Rev, 2013; 103:510–531.

20.

Koike

, Hardy

, Richards

. Adolescent self-control behavior predicts body weight through the life course: A prospective birth cohort study. Int J Obes (Lond), 2016; 40:71–76.

21.

OECD. Obesity update: 2017. Organization for Economic Cooperation and Development. Paris, France. 2017. Available at www.oecd.org/health/obesity-update.htm (last accessed March 1, 2018).

22.

Wills

, Isasi

, Mendoza

, Ainette

. Self-control constructs related to measures of dietary intake and physical activity in adolescents. J Adolesc Health, 2007; 41:551–558.

23.

Liang

, Matheson

, Kaye

, Boutelle

. Neurocognitive correlates of obesity and obesity-related behaviors in children and adolescents. Int J Obes (Lond), 2014; 38:494–506.

24.

Tourangeau

, Nord

, Lê

, et al. Early Childhood Longitudinal Study, Kindergarten Class of 1998–99 (ECLS-K), Combined User's Manual for the ECLS-K Eighth-Grade and K-8 Full Sample Data Files and Electronic Codebooks. National Center for Education Statistics, Institute of Education Sciences, U.S. Department of Education, Washington, DC, 2009.

25.

Tourangeau

, Lê

, Nord

, Sorongon

. Early Childhood Longitudinal Study, Kindergarten Class of 1998–99 (ECLS-K), Eighth-Grade Methodology Report. (NCES 2009–003) National Center for Education Statistics, Institute of Education Sciences, US Department of Education. Washington, DC, 2009.

26.

Gresham

, Elliott

. The Social Skills Rating System. Circle Pines, MN: American Guidance Service, 1990.

27.

Boyer

, Nelson

, Holub

. Childhood body mass index trajectories predicting cardiovascular risk in adolescence. J Adolesc Health, 2015; 56:599–605.

28.

Berkey

, Colditz

. Adiposity in adolescents: Change in actual BMI works better than change in BMI z score for longitudinal studies. Ann Epidemiol, 2007; 17:44–50.

29.

Cole

, Faith

, Pietrobelli

, Heo

. What is the best measure of adiposity change in growing children: BMI, BMI%, BMI z-score or BMI centile?. Eur J Clin Nutr, 2005; 59:419–425.

30.

Schwartz

, Pollak

, Bailey-Davis

, et al. Antibiotic use and childhood body mass index trajectory. Int J Obes (Lond), 2016; 40:615–621.

31.

Gillebaart

, Adriaanse

. Self-control predicts exercise behavior by force of habit, a conceptual replication of Adriaanse et al. (2014). Front Psychol, 2017; 8:190.

32.

Zelazo

, Carlson

. Hot and cool executive function in childhood and adolescence: Development and plasticity. Child Dev Perspect, 2012; 6:354–360.

33.

Riggs

, Chou

C-P

, Spruijt-Metz

, Pentz

. Executive cognitive function as a correlate and predictor of child food intake and physical activity. Child Neuropsychol, 2010; 16:279–292.

34.

Serdula

, Ivery

, Coates

, et al. Do obese children become obese adults? A review of the literature. Prev Med, 1993; 22:167–177.

35.

McTigue

, Garrett

, Popkin

. The natural history of the development of obesity in a cohort of young

U.S

. adults between 1981 and 1998. Ann Intern Med, 2002; 136:857–864.

36.

The

, Suchindran

, North

, et al. Association of adolescent obesity with risk of severe obesity in adulthood. JAMA, 2010; 304:2042–2047.

37.

Roberts

, Walton

, Viechtbauer

. Patterns of mean-level change in personality traits across the life course: A meta-analysis of longitudinal studies. Psychol Bull, 2006; 132:1–25.

38.

Diamond

, Barnett

, Thomas

, Munro

. Preschool program improves cognitive control. Science, 2007; 318:1387–1388.

39.

Rueda

, Posner

, Rothbart

. The development of executive attention: Contributions to the emergence of self-regulation. Dev Neuropsychol, 2005; 28:573–594.

40.

Traverso

, Viterbori

, Usai

. Improving executive function in childhood: Evaluation of a training intervention for 5-year-old children. Front Psychol, 2015; 6:525.

41.

Hayes

, Eichen

, Barch

, Wilfley

. Executive function in childhood obesity: Promising intervention strategies to optimize treatment outcomes. Appetite, 2018; 124:10–23.

42.

Freedman

, Sherry

. The validity of BMI as an indicator of body fatness and risk among children. Pediatrics, 2009; 124(Supplement 1):S23–S34.

43.

Carlo

, Crockett

, Wolff

, Beal

. The role of emotional reactivity, self-regulation, and puberty in adolescents' prosocial behaviors. Soc Dev, 2012; 21:667–685.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.09 MB