Compliant Eating of Maternally Prompted Food Predicts Increased Body Mass Index z -Score Gain in Girls: Results from a Population-Based Sample

Abstract

Background:

Poorer “division of responsibility” (DoR) feeding, characterized by high parental control and reduced child food choice, may promote pediatric obesity, although population-based prospective data are lacking. We tested whether poorer DoR feeding predicts childhood overweight/obesity onset and BMI z-score gain, over 10 years in the National Longitudinal Study of Youth.

Methods:

We studied 302 girls and 316 boys, with mean ages 52.24 and 52.35 months, respectively, in 1986, who were followed for 10 years. We excluded children who were initially overweight/obese. Mothers completed three DoR feeding questions in 1986: (1) child eating compliance of prompted foods; (2) child eating compliance of initially refused foods; and (3) mother-allotted child food choice. Child BMI (kg/m²) was calculated from measured weights and heights in 1986, 1988, 1992, 1994, and 1996.

Results:

Daughters who complied with maternal food prompts [odds ratio (OR), 2.01] and those who obeyed maternal prompts to consume initially rejected foods (OR, 2.29) “most of the time” were significantly more likely than daughters who complied less frequently to become overweight/obese after 8 years. Also, more frequent eating compliance (p<0.001) and more frequent compliance of initially rejected foods (p=0.003) predicted greater BMI z-score gain in girls. These associations were not found for boys. Maternal obesity consistently predicted overweight/obesity risk in girls (ORs, 2.48–8.63) and boys (ORs, 2.27– 4.03).

Conclusions:

Teaching parents to avoid coercive feeding practices, while encouraging child self-selection of healthier foods, may help their daughters to achieve better energy balance.

Introduction

“Division of responsibility” (DoR) refers to noncoercive parent-child feeding practices, in which parents provide an array of healthy food choices and children decide if and how much to eat.^1–6 DoR is believed to foster self-regulatory eating and a healthy body weight, as children learn to recognize internal satiety signals.^7–9 Indeed, children whose food choices were restricted by parents were heavier, compared to children whose food choices were less restricted.^9,10 Restrictive feeding by mothers also was prospectively linked to disinhibited eating and excess weight gain in girls.¹¹ At the same time, mothers who were neglectful of their child's hygiene, or unaware of their intake of sweets, had children who were ∼9 times and 4.5 times more likely to become obese adults.^12,13 Additionally, children whose mothers had permissive or neglectful parenting styles were at increased risk of obesity, compared to children of authoritative mothers.¹⁴ Thus, active parental involvement in children's food choices may be protective against obesity.¹⁵ Long-term prospective data are needed to identify feeding practices that promote or protect against childhood obesity.¹⁶

In the present study, we tested whether poorer DoR feeding predicted onset of childhood overweight/obesity and greater BMI z-score gain over 10 years in a US population-based sample. DoR feeding was assessed by three mother-report questions addressing child eating compliance of maternally prompted foods, child eating compliance of previously rejected foods, and mother-allotted child food choice. We hypothesized that poorer DoR feeding, operationalized as a child's more frequently eating maternally prompted foods and reduced child food choice, would predict greater pediatric overweight/obesity risk and BMI z-score gain over 10 years.

Methods

Participants

Mothers were participants in the National Longitudinal Study of Youth (NLSY) main sample,¹⁷ which is one of several “National Labor Surveys” administered by the Bureau of Labor Statistics. The original NLSY sample consisted of a nationally representative cohort of 12,686 men and women (14–22 years of age), who were first studied in 1979. Starting in 1986, female NLSY participants who had become mothers were invited to participate in supplemental home or phone interviews addressing the health and development of their children. During these interviews, which occurred every 2 years after 1986, children's weights and heights were obtained by direct measurement (home interviews) or by maternal report (phone interviews). Only children with direct measures were included in the present report. In addition, three DoR questions were asked during these interviews (see below).

For the present secondary analysis, we examined a subsample of the NLSY families who had (1) measures of child age, sex, and race/ethnicity, (2) measures of maternal height and weight in 1986, (3) a valid score on one or more of the DoR questions in 1986, (4) directly measured child height and weight in 1986, and (5) directly measured child height and weight in one or more of the follow-up time points (i.e., 1988, 1992, 1994, and 1996). Thus, this prospective cohort design allowed us to test whether DoR questions in 1986 predict the onset of childhood overweight/obesity over 10 years, adjusting for maternal BMI and demographic factors. We did not analyze child weight and height collected in 1990 because, unlike the other years, the codebook did not specify which children had objectively measured versus parent-reported heights and weights. Also, because we were interested in testing the onset or incidence of overweight/obesity, we restricted our analyses to children with a BMI less than the 85th percentile in 1986.

Informed consent for primary data collection from participants was obtained by the original NLSY investigators. Specifically, procedures were reviewed and approved by institutional review boards (IRBs) at the following institutions under contract with the Bureau of Labor Statistics: Ohio State University and the National Opinion Research Center at the University of Chicago. The IRBs of St. Luke's-Roosevelt Hospital, the University of Pennsylvania, and University of North Carolina–Chapel Hill reviewed the present secondary data analysis and it met criteria for exemption.

Measures: Exposure Variables

Division of Responsibility feeding questions

The following DoR feeding questions were administered in 1986. Psychometric data for these questions were not provided in the source documents and hence are not presented below.

Child eating compliance of prompted foods

“When it is mealtime, how often does your child eat what you want him/her to eat?” Response options were almost never, less than half the time, half the time, more than half the time, and almost always, with higher scores on this 1–5 scale reflecting greater compliance. Children were classified either as being either “less frequent” (i.e., scores of 1, 2, 3, or 4) or “almost always” (i.e., score of 5) for logistic regression analyses. We established the cut-off point just below the most extreme score (i.e., “almost always” compliant) for conceptual reasons, rather than basing it on a median split. Specifically, we wanted to express this comparison such that “cases” (coded as 1) would be the most highly eating compliant youth, as compared against all other youth (coded as 0).

Child eating compliance of initially refused foods

“When your child doesn't eat what you want him/her to eat and you tell him/her to do so, how often does he/she obey and eat?” Response options were almost never, less than half the time, half the time, more than half the time, and almost always, with higher scores on this 1–5 scale reflecting greater compliance. Children were classified either as being “less frequent” (i.e., scores of 1, 2, 3, or 4) or “almost always” (i.e., score of 5) for logistic regression analyses.

Mother-allotted food choice

“How much choice is your child allowed in deciding what foods he/she eats at breakfast and lunch?” Response options were no choice, little choice, some choice, and a great deal of choice, with higher scores on this 1–4 scale reflecting greater choice. Children were classified either as being “lower choice” (i.e., scores of 1 or 2) or “high choice” (i.e., scores of 3 or 4) for logistic regression analyses.

Demographics

Child demographic measures were maternally reported sex and race/ethnicity, which was scored as Hispanic, African American, or Non-Hispanic/Non–African American (NHNAA) in the NLSY data set. This latter category was predominantly white children.¹⁵ Because household income was available in 1986 for only 108 girls and 114 boys (i.e., approximately one third of the total sample of children that we analyzed), we did not include it as a covariate because of the large number of cases that would have been excluded. This limitation of the database has been noted in previous reports.¹⁸

Maternal BMI

Maternal height and weight were self-reported in 1985 and 1986, respectively. BMI was computed on mothers as weight in kilograms divided by the square of height in meters. For logistic regression models, mothers at baseline were classified either as nonobese (BMI <30 kg/m²) or obese (BMI ≥30 kg/m²).

Outcome Variables

Child BMI and overweight/obesity status

Child height and weight were directly assessed by tape measure and portable bathroom scale following standardized procedures of the research protocol.¹⁷ Information on the brand of scale and tape measure, whether children were dressed or undressed, and the reliability of measurements were not provided. At each year (1986, 1988, 1992, 1994, and 1996), child weight and height measures were converted to BMI values, which, in turn, were converted to age- and sex-specific z-scores and percentiles per the CDC growth charts. Children whose BMI z-scores were <–4 or >5 were excluded from analyses, as recommended by the World Health Organization report on physical anthropometry for weight-for-height measures.¹⁹ Percentile scores were used to define overweight/obesity status for the logistic regression models (i.e., <85th percentile vs. ≥85th percentile), whereas BMI z-scores were used for mixed-model analyses that examined the full spectrum of scores (see below).

Data Analytic Plan

Descriptive statistics are presented as means and standard deviations (SDs) and percentages. Logistic regression tested whether poorer DoR feeding (i.e., almost always child eating compliance of maternally prompted foods, almost always child eating compliance of initially rejected foods, and lower mother-allotted child food choice) predicted greater overweight/obesity risk at each follow-up year. These analyses tested for incidence of overweight/obesity, because we excluded children who were overweight/obese (i.e., BMI ≥85th percentile) at the initial assessment year, 1986. Covariates were maternal obesity status, maternal age (i.e., <25 vs. ≥25 years), maternal education (i.e., <12 vs. ≥12 years, including college and professional school), child sex, child race/ethnicity (dummy coded as NHNAA vs. Hispanic/African American), and child age (<51 vs. ≥51 months—which was the median age in 1986). The outcome measure was child overweight/obesity status at the given follow-up year. We conducted separate (i.e., independent) logistic regression analyses at each follow-up time point to determine whether the hypothesized risk-factor associations would occur at all of the time points or just some. Odds ratios (ORs) and 95% confidence intervals were calculated for the individual predictors. Analyses were stratified by child sex, and separate analyses were tested for each DoR question.

We also conducted complementary mixed-model analyses to test whether 10-year changes in BMI z-score (i.e., 1986–1996) varied as a function of 1986 DoR feeding. Sex-stratified analyses were conducted for each DoR domain, with the following predictors: maternal BMI; maternal age; maternal education; child race/ethnicity (dummy coded); child age; and score on the given DoR question. A significant effect of the DoR question would mean that, controlling for covariates in the model, differences in DoR predict greater (or less) change in BMI z-score over 10 years. (This analysis did not model change scores per se. As is customary of mixed-model approaches, we modeled BMI z-scores across all time points, including the initial year of 1986.) For a child to be included in these analyses, he or she had to have a valid BMI (and its corresponding z-score) in 1986 and at least two anthropometric follow-up assessments. Alpha was set at 0.05, and two-tailed significance tests were performed for all analyses.

Results

Sample Characteristics

A total of 302 girls and 316 boys were studied. The race/ethnicity breakdown was 32% African American, 30% Hispanic, and 38% NHNAA for girls and 35% African American, 25% Hispanic, and 40% NHNAA for boys. The mean (SD) age for girls and boys in 1986 was 52.24 (SD, 16.10) and 52.35 (SD, 16.59) months, respectively (see Table 1). The range of ages when children were assessed in 1986 was 22.0–83.0 months (i.e., 1.8–6.9 years). Only 1 child was less than 24 months, and so 99% of the sample was between 2.0 and 6.9 years of age. Girls and boys did not differ in terms of mother-allotted child food choice (p=0.50), compliant eating of prompted foods (p=0.10), or compliant eating of prompted foods that were initially refused (p=0.21). Mothers of girls and boys did not significantly differ in age (p=0.32) or BMI (p=0.83). Descriptive data on children's BMI measurements, and the percent of overweight/obese children at each assessment year, are presented in Table 1.

Table 1.

Age and Anthropometric Characteristics of Study Participants by Year and Child Sex

Measure	Girls	Boys
1986
Maternal age	25.19 (2.10)	25.01 (2.16)
Maternal BMI	24.45 (4.89)	24.19 (5.01)
Maternal obesity (%)	16.2	13.6
Child age (months)	52.24 (16.10)	52.35 (16.59)
Child BMI (kg/m²)	15.04 (1.13)	15.19 (1.13)
Child BMI z-score	−0.47 (1.06)	−0.57 (1.07)
Child BMI percentile	39.56 (26.88)	36.48 (26.85)
Child overweight/obese (%)	0	0
Eating compliance	3.72 (1.1)	3.61 (1.1)
Eating obedience	4.09 (0.9)	3.95 (1.0)
Food choice	3.11 (0.7)	3.06 (0.7)
1988
Child age (months)	81.16 (16.23)	81.30 (16.67)
Child BMI (kg/m²)	15.90 (2.00)	15.91 (1.91)
Child BMI z-score	−0.00 (1.09)	−0.06 (1.17)
Child BMI percentile	51.08 (30.06)	50.13 (30.12)
Child overweight/obese (%)	17.6	14.9
1992
Child age (months)	126.25 (16.83)	128.06 (16.03)
Child BMI (kg/m²)	19.06 (3.62)	18.50 (4.38)
Child BMI z-score	0.39 (1.03)	0.18 (1.19)
Child BMI percentile	61.57 (29.14)	55.48 (31.38)
Child overweight/obese (%)	27.8	24.1
1994
Child age (months)	151.39 (9.65)	153.50 (10.31)
Child BMI (kg/m²)	19.94 (4.02)	18.94 (3.67)
Child BMI z-score	0.27 (1.10)	−0.00 (1.24)
Child BMI percentile	58.51 (30.18)	51.38 (32.44)
Child overweight/obese (%)	24.8	21.4
1996
Child age (months)	159.59 (8.08)	158.89 (8.12)
Child BMI (kg/m²)	21.39 (5.29)	20.31 (4.71)
Child BMI z-score	0.34 (1.20)	0.17 (1.22)
Child BMI percentile	60.09 (31.24)	55.09 (31.66)
Child overweight/obese (%)	31.5	22.5

All table values are means (SDs), with the exception of child overweight/obese and maternal obesity, which are percentages. There were no overweight/obese children in 1986 because children with a BMI ≥85th percentile were excluded from this analysis. Mothers were participants in the National Longitudinal Study of Youth (NLSY) cohort, which was one of several “National Labor Surveys” administered by the Bureau of Labor Statistics.

A total of 302 girls and 316 boys met the inclusion criteria, although the exact number of boys and girls assessed at each follow-up time point varied (see Table 2). The exact percentage of children classified as less frequent or almost always with respect to eating compliance of prompted foods, as well as eating compliance of initially refused foods, differed across logistic regression analyses because not all children were measured at each follow-up assessment (see Table 3). That is, we stratified logistic regression analyses by the year of follow-up (i.e., separate analyses for 1988, 1992, 1994, and 1996), such that they were formally treated as distinct prospective analyses. Similarly, the percentage of children classified as lower or high in mother-allotted food choice differed for each follow-up analysis (see Table 3).

Table 2.

Number of Girls (Boys) Assessed at Each Follow-Up Year for the Division of Responsibility Feeding Questions Asked in 1986

	Outcome year
1986 predictor variable	1988	1992	1994	1996
Eating compliance	284 (295)	302 (316)	286 (285)	124 (111)
Eating obedience	284 (293)	301 (316)	286 (284)	124 (110)
Food choice	187 (186)	185 (203)	186 (192)	70 (54)

Table 3.

Percentage of Girls (Boys) Classified as “Low” or “High” and “Less Frequent” or “Frequent” for the 1986 Division of Responsibility Feeding Questions in Logistic Regression Analyses

		Follow-up year
1986 feeding question	Category	1988	1992	1994	1996
Eating compliance	Less frequent	55.6 (59.3)	57.0 (62.3)	59.1 (61.1)	62.1 (63.1)
	Frequent	44.4 (40.7)	43.0 (37.7)	40.9 (38.9)	37.9 (36.9)
Eating obedience	Less frequent	70.4 (72.0)	72.4 (75.0)	78.0 (75.4)	79.8 (83.6)
	Frequent	29.6 (28.0)	27.6 (25.0)	22.0 (24.6)	20.2 (16.4)
Food choice	Lower choice	71.1 (77.4)	73.0 (75.9)	77.4 (78.1)	71.4 (72.2)
	High choice	28.9 (22.6)	27.0 (24.1)	22.6 (21.9)	28.6 (27.8)

Inferential Analyses for Girls

Child eating compliance of prompted foods

Daughters who reportedly complied most of the time when eating were significantly more likely to be overweight/obese after 8 years (OR, 2.01), compared to daughters who reportedly complied less frequently (see Table 4). Additionally, maternal obesity predicted increased incidence of overweight/obesity in daughters at all time points through 10-year follow-up, with ORs ranging from 2.51 to 8.63.

Table 4.

Odds Ratios (95% Confidence Intervals) for Overweight/Obesity Onset at 2, 6, 8, and 10 Years of Follow-Up in Girls (Upper Panel) and Boys (Lower Panel) Who Were Not Overweight/Obese in 1986, Based on Logistic Regression Analyses

Results for girls
	Follow-up assessment year
Predictor	1988	1992	1994	1996
Maternal obesity	2.72 (1.29–5.74)	2.51 (1.36–4.62)	8.63 (4.32–17.24)	3.84 (1.38–10.72)
Maternal age	0.85 (0.39–1.87)	1.58 (0.92–2.70)	1.10 (0.60–2.02)	1.50 (0.67–3.36)
Maternal education	1.00 (0.42–2.42)
Child race/ethnicity	0.85 (0.39–1.88)	1.21 (0.71–2.07)	1.22 (0.65–2.27)	1.29 (0.56–2.95)
Child age	0.51 (0.23–1.13)	0.82 (0.49–1.39)	0.58 (0.30–1.13)	N/A
Eating compliance	0.61 (0.25–1.50)	0.93 (0.55–1.56)	2.01 (1.10–3.67)	1.25 (0.57–2.81)
Maternal obesity	2.94 (1.40–6.18)	2.48 (1.34–4.57)	8.77 (4.40–17.52)	4.04 (1.45–11.26)
Maternal age	1.04 (0.53–2.03)	1.59 (0.93–2.72)	1.16 (0.63–2.13)	1.52 (0.68–3.41)
Maternal education	0.71 (0.35–1.43)
Child race/ethnicity	0.97 (0.50–1.88)	1.20 (0.71–2.06)	1.17 (0.63–2.19)	1.29 (0.56–2.95)
Child age	0.46 (0.23–0.91)	0.82 (0.49–1.39)	0.54 (0.27–1.05)	N/A
Eating obedience	1.75 (0.88–3.45)	0.91 (0.51–1.62)	2.29 (1.17–4.45)	1.12 (0.43–2.93)
Maternal obesity	3.06 (1.27–7.36)	2.03 (0.91–4.55)	5.97 (2.64–13.49)	1.72 (0.42–7.07)
Maternal age	0.85 (0.39–1.87)	1.60 (0.80–3.19)	1.57 (0.73–3.35)	1.47 (0.49–4.46)
Maternal education	1.00 (0.42–2.42)
Child race/ethnicity	0.85 (0.39–1.88)	1.01 (0.51–2.01)	0.89 (0.42–1.89)	1.14 (0.39–3.35)
Child age	0.51 (0.23–1.13)	0.76 (0.39–1.50)	0.72 (0.34–1.52)	N/A
Food choice	0.61 (0.25–1.50)	0.71 (0.33–1.54)	0.84 (0.34–2.03)	1.40 (0.44–4.44)

Results for boys
	Follow-up assessment year
Predictor	1988	1992	1994	1996
Maternal obesity	2.27 (1.00–5.18)	2.78 (1.47–5.25)	2.40 (1.19–4.84)	3.84 (1.38–10.72)
Maternal age	0.88 (0.45–1.74)	0.81 (0.47–1.39)	0.70 (0.39–1.25)	1.50 (0.67–3.36)
Maternal education	0.57 (0.28–1.16)
Child race/ethnicity	1.46 (0.71–3.03)	0.94 (0.54–1.63)	1.81 (0.94–3.50)	1.29 (0.59–2.95)
Child age	0.71 (0.36–1.39)	0.71 (0.42–1.21)	0.95 (0.52–1.73)	N/A
Eating compliance	0.64 (0.32–1.28)	0.95 (0.55–1.64)	0.76 (0.41–1.40)	1.25 (0.56–2.81)
Maternal obesity	2.21 (0.98–5.01)	2.78 (1.47–5.25)	2.35 (1.17–4.72)	4.03 (1.45–11.26)
Maternal age	0.89 (0.45–1.75)	0.82 (0.48–1.41)	0.70 (0.39–1.25)	1.52 (0.68–3.41)
Maternal education	0.57 (0.28–1.16)
Child race/ethnicity	1.47 (0.71–3.04)	0.92 (0.53–1.61)	1.82 (0.95–3.52)	1.29 (0.56–2.95)
Child age	0.71 (0.36–1.40)	0.67 (0.39–1.45)	0.97 (0.54–1.77)	N/A
Eating obedience	0.85 (0.39–1.83)	1.46 (0.80–2.67)	0.63 (0.31–1.31)	1.12 (0.43–2.93)
Maternal obesity	2.93 (1.05–8.18)	2.58 (1.17–5.72)	1.77 (0.76–4.09)	3.90 (0.82–18.53)
Maternal age	1.05 (0.42–2.61)	0.90 (0.46–1.77)	0.80 (0.40–1.62)	2.22 (0.47–10.57)
Maternal education	0.43 (0.17–1.06)
Child race/ethnicity	1.18 (0.42–3.13)	0.92 (0.44–1.91)	1.61 (0.72–3.61)	2.23 (0.38–12.98)
Child age	0.87 (0.35–2.14)	0.75 (0.38–1.48)	0.90 (0.45–1.82)	N/A
Food choice	0.60 (0.19–1.92)	0.80 (0.36–1.79)	0.73 (0.30–1.81)	0.30 (0.03–2.75)

Eating compliance refers to the child eating compliance of prompted foods question, eating obedience refers to the child eating compliance of initially refused foods question, and food choice refers to the mother-allotted child food choice question. All table values are odds ratios (95% confidence interval) for each predictor variable. Bold values are statistically significant. Because all girls included in the 1986–1996 logistic regression models were greater than 51 months of age in 1986, “child age” was not included as a predictor variable in these analyses. Mothers were participants in the National Longitudinal Study of Youth (NLSY) cohort, which was one of several “National Labor Surveys” administered by the Bureau of Labor Statistics.

Abbreviation: N/A, not applicable.

Mixed-model analyses revealed significant effects of time (p<0.0001), maternal obesity status (p<0.0001), and eating compliance (p<0.0001) on child BMI z-score (see Table 5). Thus, 10-year changes in BMI z-score were greater for daughters of obese, compared to nonobese, mothers and for daughters of relatively older, compared to younger, mothers. Daughters who reportedly complied more frequently with maternal prompts showed greater 10-year BMI z-score gains. Each 1-unit difference on this scale was associated with an estimated BMI z-score change of 0.13 SD over 10 years. Thus, the estimated difference in BMI z-score change among daughters who complied “almost never” and “almost always” was 0.52 SD.

Table 5.

Results of Mixed-Model Analyses Testing the Association of 1986 Predictor Variables with 10-Year Changes in BMI z-Score Among Girls and Boys Who Were Not Overweight/Obese in 1986

	Results for girls			Results for boys
	β (SE)	t	p-value	β (SE)	t	p-value
Model 1 predictors
Time	0.10 (0.01)	16.43	<0.0001	0.08 (0.01)	11.29	<0.0001
Maternal obesity	0.39 (0.10)	3.94	<0.0001	0.41 (0.11)	3.76	0.0002
Maternal age	0.03 (0.02)	1.60	0.11	0.01 (0.02)	0.52	0.61
Maternal education	0.01 (0.02)	0.66	0.51	0.01 (0.02)	0.30	0.77
Child race/ethnicity	0.03 (0.07)	0.37	0.71	0.08 (0.08)	1.08	0.28
Child age	0.00 (0.00)	1.29	0.20	0.00 (0.00)	1.19	0.23
Eating compliance	0.13 (0.03)	3.99	<0.0001	0.01 (0.03)	0.35	0.73
Model 2 predictors
Time	0.10 (0.01)	16.39	<0.0001	0.08 (0.01)	11.38	<0.0001
Maternal obesity	0.43 (0.10)	4.27	<0.0001	0.41 (0.11)	3.77	0.0002
Maternal age	0.03 (0.02)	1.61	0.11	0.01 (0.02)	0.49	0.62
Maternal education	0.02 (0.02)	0.89	0.38	0.00 (0.02)	0.24	0.81
Child race/ethnicity	0.05 (0.07)	0.70	0.48	0.08 (0.08)	1.07	0.28
Child age	0.00 (0.00)	0.79	0.43	0.00 (0.00)	1.11	0.27
Eating obedience	0.09 (0.03)	2.84	0.005	0.01 (0.03)	0.40	0.69
Model 3 predictors
Time	0.09 (0.01)	12.00	<0.0001	0.08 (0.01)	9.02	<0.0001
Maternal obesity	0.35 (0.12)	2.86	0.004	0.46 (0.13)	3.46	0.0006
Maternal age	0.02 (0.02)	0.94	0.35	0.00 (0.02)	0.18	0.86
Maternal education	0.03 (0.03)	1.04	0.30	−0.02 (0.02)	−0.68	0.50
Child race/ethnicity	0.05 (0.09)	0.53	0.59	0.07 (0.10)	0.68	0.50
Child age	0.01 (0.00)	0.94	0.35	0.01 (0.00)	2.14	0.03
Food choice	0.00 (0.06)	0.50	0.96	−0.05 (0.07)	−0.72	0.47

Eating compliance refers to the child eating compliance of prompted foods question, eating obedience refers to the child eating compliance of initially refused foods question, and food choice refers to the mother-allotted child food choice question. Bold values are statistically significant. Mothers were participants in the National Longitudinal Study of Youth (NLSY) cohort, which was one of several “National Labor Surveys” administered by the Bureau of Labor Statistics. Bold values are statistically significant. β refers to unstandardized beta weight, representing the estimated 10-year change in BMI z-score associated with a 1-unit change in the independent variable in 1986.

Abbreviation: SE, standard error.

Child eating compliance of initially refused foods

Daughters who reportedly obeyed maternal prompts to eat refused foods most of the time were significantly more likely to be overweight/obese after 8 years (OR, 2.29), compared to daughters who reportedly obeyed less frequently (see Table 4). Daughters who were older, compared to younger, at the initial assessment were less likely to become overweight/obese after 2 years (OR, 0.46). Additionally, maternal obesity predicted the incidence of overweight/obesity at all time points through 10-year follow-up, with ORs ranging from 2.48 to 8.77.

Mixed-model analyses revealed significant effects of time (p<0.0001), maternal obesity status (p<0.0001), and eating obedience (p=0.005) on child BMI z-score (see Table 5). Daughters who reportedly complied more frequently with maternal prompts to eat initially refused foods showed greater 10-year BMI z-score gains. Each 1-unit difference on this scale was associated with an estimated BMI z-score change of 0.09 SD over 10 years. Thus, the estimated difference in BMI z-score change among daughters who complied “almost never” and “almost always” was 0.36 SD.

Mother allotted-child food choice

Lower mother-allotted child food choice was not a risk factor for child overweight/obesity (p>0.05 across analyses), although daughters of obese mothers were more likely to become overweight/obesity after 2 (OR, 3.06) and 8 years (OR, 5.97; see Table 4).

Mixed-model analyses revealed significant effects of time (p<0.0001) and maternal obesity status (p=0.004) on BMI z-score, but there was no main effect for allotted food choice (p=0.96; see Table 5).

Inferential Analyses for Boys

Child eating compliance of prompted foods

Maternal obesity predicted the incidence of overweight/obesity at all follow-up time points, with ORs ranging from 2.27 to 3.84 (see Table 4). None of the other predictors, including child's eating compliance of prompted foods, was significant. Mixed-model analyses revealed significant effects of time (p<0.0001) and maternal obesity status (p<0.0002), but no effect of eating compliance (p=0.73) on child BMI z-score (see Table 5).

Child eating compliance of initially refused foods

Maternal obesity predicted the incidence of overweight/obesity after 6 (OR, 2.78), 8 (OR, 2.35), and 10 years (OR, 4.03; see Table 4). None of the other predictors, including child's eating compliance of refused foods, was significant. Mixed-model analyses revealed significant effects of time (p<0.0001) and maternal obesity status (p<0.0002), but no effect of eating compliance of refused foods (p=0.69), on child BMI z-score (see Table 5).

Mother allotted-child food choice

Maternal obesity predicted the incidence of overweight/obesity after 2 (OR, 2.93) and 6 years (OR, 2.58; see Table 4). None of the other predictors, including mother-allotted child food choice, was significant. Mixed-model analyses revealed significant effects of time (p<0.0001), maternal obesity status (p=0.0006), and child age (p=0.03), but no effect of allotted food choice (p=0.47) on child BMI z-score (see Table 5).

Discussion

In this population-based study, we found that more frequent eating compliance of maternally prompted foods, including those previously rejected by children, was associated with greater BMI z-score gain over 10 years only in girls. This association was found controlling for maternal BMI, education, and other sociodemographic variables. These are the first prospective data from a national cohort to implicate poorer DoR feeding in child energy balance and long-term BMI gain. Pressuring children to eat can negatively affect food preferences^20–22 and may interfere with the development of self-regulatory eating²³ and self-control strategies.²⁴

Teaching parents to avoid coercive feeding strategies, whereas encouraging child self-selection of healthier food choices, may help promote and sustain energy balance in girls. There also may be opportunities to teach these skills to parents in community settings, such as Head Start sites,^25,26 where there already is movement toward child self-selection of healthy foods.

Although more compliant eating predicted a greater BMI z-score increase over 10 years in girls, it was not a risk factor for overweight/obesity per se in most logistic regression analyses. It is possible that any effect of compliant eating on energy balance in girls is not sufficiently potent on its own to “cause” obesity. Also, dichotomization of DoR feeding measures may have resulted in a loss of information that reduced statistical power, compared to analyses of continuous predictors. Only after 8 years was there a significant effect, in which daughters who complied with maternal food prompts (OR, 1.86) and obeyed maternal prompts to consume initially rejected foods (OR, 2.60) “most of the time” were significantly more likely to become overweight/obese. Daughters, on average, were 12.6 years old after 8 years, corresponding to the beginning of adolescence. The effect of compliant eating on girls' overweight/obesity risk could interact with biological changes associated with puberty along with youth's increasing need for autonomy. Future studies might further explore this from a “life course” framework.²⁷

DoR feeding questions did not predict overweight/obesity risk or BMI z-score gain in boys. Mothers may have different attitudes for daughters and sons regarding “appropriate” eating practices, accounting for our sex-specific findings. Of note, an earlier review concluded that the association between restrictive feeding and child overeating and obesity may be limited to girls, although data were not conclusive.¹⁰ Whether any association between DoR feeding patterns and childhood obesity risk is sex specific warrants additional research.

Maternal obesity at baseline was the strongest predictor of childhood overweight/obesity onset in girls and boys. Coupled with previous findings from this^28–30 and other cohorts,^31–34 these findings support the critical role of maternal weight status in onset of childhood obesity. Maternal BMI reflects genetic influences,^35–37 as well as modifiable environmental influences.^38,39 Strategies to reduce maternal BMI, or environmental factors correlated with maternal BMI, should be explored for childhood obesity prevention.

The present findings must be interpreted in light of the study limitations. First, data on the validity of the three DoR questions have not been published. Additionally, there was no time frame specified for the items (e.g., within the past month or year). The strength of these questions is their face validity, direct questioning of habitual eating compliance, and brevity. Second, information on the brand of scale and tape measure, whether children were dressed or undressed, and the reliability of measurements were not provided. Third, we did not control household income in our analyses because of the large number of missing values. Fourth, none of the DoR questions referred to specific types of foods that were prompted by mothers. Fifth, there was noteworthy attrition over time. For analyses of child eating compliance of prompted foods, the proportion of girls retained in follow-up analyses was 93% (in 1988), 100% (in 1992), 94% (in 1994), and 41% (in 1996). The proportion of boys retained in follow-up analyses was 93% (in 1988), 100% (in 1992), 91% (in 1994), and 36% (in 1996). There also was differential dropout of girls included in the 10-year logistic regression analyses, with a smaller representation of Hispanics and greater representation of African Americans (i.e., 21% Hispanic, 38% African American, and 41% NHNAA), compared to baseline. In comparison, demographic characteristics of boys included in the 10-year logistic regression analyses (i.e., 25% Hispanic, 36% African American, and 39% NHNAA) were almost identical to the breakdown at baseline. Also noteworthy is that, among boys only, those who dropped out in 1996, compared to those who did not drop out, had a relatively higher mean BMI z-score [mean (SD), −0.48 (1.11) SD vs. −0.77 (1.17) SD; p=0.03] and BMI percentile [mean (SD), 39.70 (27.29) vs. 32.52 (27.30) percentile; p=0.02] in 1986. Sixth, results from the logistic regression analyses are dependent upon the specific cut points we established to create low versus high scores for each DoR feeding question. However, it was reassuring to see convergent findings for our logistic regression analyses and mixed-model analyses that treated each DoR question as a continuous variable. Finally, as our findings were based on data from 1986 to 1996, generalizability to the current “obesogenic environment” needs to be established.

Two final interpretative caveats should be noted. First, the DoR feeding questions addressing compliant eating could reflect child's inherent tendency to be more (or less) compliant overall or compliant when eating, rather than poorer feeding practices per se. That is, it may reflect more of a child attribute or “driver” of obesity risk more so than the parenting behavior being the causal agent. The wording of these questions makes it hard to discern. Stated differently, our design does not preclude reverse causal inferences in which excess child BMI gain elicits maternal feeding prompts. Future research examining the parental pressure to eat prompts and (non)compliant eating by children should better attempt to establish causal sequence, which could very well be bidirectional. Second, the null associations between “mother-allotted child food choice” and child weight status over time do not support the theory that restrictive feeding is a risk factor for pediatric obesity. To the extent that low food choice on this question reflects “restrictive feeding,” one would have expected this item to predict greater child BMI z-score gain or obesity. This was not the case.

Conclusion

In summary, girls who more frequently ate when prompted by their mothers to do so showed a significantly greater BMI z-score gain over 10 years. Mechanisms underlying this association need to be identified. Interventions that teach children to self-select healthier diets without coercion by parents, as well as child care and school food service staff, warrant investigation.⁴⁰ Additionally, maternal obesity was a strong predictor of obesity onset over 10 years in girls and boys. Long-term promotion of a maternal healthy body weight may be a strategy beneficial for childhood obesity prevention, which warrants evaluation.

Footnotes

Acknowledgments

This work was supported, in part, by the CDC through the ATPM/CDC/ATSDR Cooperative Agreement (TS 318-15/15). The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the CDC.

Author Disclosure Statement

No competing financial interests exist.

References

Satter

. Internal regulation and the evolution of normal growth as the basis for prevention of obesity in children. J Am Diet Assoc, 1996; 96:860–864.

Satter

. The feeding relationship. J Am Diet Assoc, 1986; 86:352–356.

Satter

. Children, the feeding relationship, and weight. Md Med, 2004; 5:26–28.

Satter

. The feeding relationship: Problems and interventions. J Pediatr, 1990; 117:S181–S189.

Dietz

, Stern

. American Academy of Pediatrics Guide to Your Child's Nutrition. Villard Books: New York, 1999.

Satter

. Child of Mine. Bull Publishing: Palo Alto, CA, 2000.

Birch

, Deysher

. Caloric compensation and sensory specific satiety: Evidence for self regulation of food intake by young children. Appetite, 1986; 7:323–331.

Birch

, Deysher

. Conditioned and unconditioned caloric compensation: Evidence for self-regulation of food intake by young children. Learn Motiv, 1985; 16:341–355.

Johnson

, Birch

. Parents' and children's adiposity and eating style. Pediatrics, 1994; 94:653–661.

10.

Faith

, Scanlon

, Birch

et al. Parent-child feeding strategies and their relationships to child eating and weight status. Obes Res, 2004; 12:1711–1722.

11.

Fisher

, Birch

. Eating in the absence of hunger and overweight in girls from 5 to 7 y of age. Am J Clin Nutr, 2002; 76:226–231.

12.

Lissau

, Sorensen

. Parental neglect during childhood and increased risk of obesity in young adulthood. Lancet, 1994; 343:324–327.

13.

Lissau

, Breum

, Sorensen

. Maternal attitude to sweet eating habits and risk of overweight in offspring: A ten-year prospective population study. Int J Obes Relat Metab Disord, 1993; 17:125–129.

14.

Rhee

, Lumeng

, Appugliese

et al. Parenting styles and overweight status in first grade. Pediatrics, 2006; 117:2047–2054.

15.

Faith

, Heshka

, Keller

et al. Maternal-child feeding patterns and child body weight: Findings from a population-based sample. Arch Pediatr Adolesc Med, 2003; 157:926–932.

16.

Dietz

, Gortmaker

. Preventing obesity in children and adolescents. Annu Rev Public Health, 2001; 22:337–353.

17.

US Department of Labor. NLSY79 Child and Young Adult User's Guide. Columbus, OH: CHRR, The Ohio State University, 2004.

18.

Strauss

, Pollack

. Epidemic increase in childhood overweight, 1986–1998. JAMA, 2001; 286:2845–2848.

19.

WHO Expert Committee on Physical Status. Physical Status: The Use and Interpretation of Anthropometry Physical Status. World Health Organization: Geneva, 2005.

20.

Galloway

, Fiorito

, Lee

et al.

Parental pressure, dietary patterns, and weight status among girls who are “picky eaters”

J Am Diet Assoc, 2005; 105:541–548.

21.

Galloway

, Fiorito

, Francis

et al. ‘Finish your soup’: Counterproductive effects of pressuring children to eat on intake and affect. Appetite, 2006; 46:318–323.

22.

Batsell

Jr. , Brown

, Ansfield

et al. “You will eat all of that!”: A retrospective analysis of forced consumption episodes. Appetite, 2002; 38:211–219.

23.

Birch

. Psychological influences on the childhood diet. J Nutr, 1998; 128:407S–410S.

24.

Anzman

, Birch

. Low inhibitory control and restrictive feeding practices predict weight outcomes. J Pediatr, 2009; 155:651–656.

25.

Hoerr

, Utech

, Ruth

. Child control of food choices in Head Start families. J Nutr Educ Behav, 2005; 37:185–190.

26.

Gable

, Lutz

. Nutrition socialization experiences of children in the Head Start program. J Am Diet Assoc, 2001; 101:572–577.

27.

Ben-Shlomo

, Kuh

. A life course approach to chronic disease epidemiology: Conceptual models, empirical challenges and interdisciplinary perspectives. Int J Epidemiol, 2002; 31:285–293.

28.

Wang

, Chyen

, Lee

et al. The association between body mass index in adolescence and obesity in adulthood. J Adolesc Health, 2008; 42:512–518.

29.

, Kaur

, Choi

et al. Additive interactions of maternal prepregnancy BMI and breast-feeding on childhood overweight. Obes Res, 2005; 13:362–371.

30.

, Goran

, Kaur

et al. Developmental trajectories of overweight during childhood: Role of early life factors. Obesity (Silver Spring), 2007; 15:760–771.

31.

Reilly

, Armstrong

, Dorosty

et al. Early life risk factors for obesity in childhood: Cohort study. BMJ, 2005; 330:1357.

32.

Whitaker

, Wright

, Pepe

et al. Predicting obesity in young adulthood from childhood and parental obesity. N Engl J Med, 1997; 337:869–873.

33.

Berkowitz

, Stallings

, Maislin

et al. Growth of children at high risk of obesity during the first 6 y of life: Implications for prevention. Am J Clin Nutr, 2005; 81:140–146.

34.

Agras

, Kraemer

, Berkowitz

et al. Influence of early feeding style on adiposity at 6 years of age. J Pediatr, 1990; 116:805–809.

35.

Faith

, Rhea

, Corley

et al. Genetic and shared environmental influences on children's 24-h food and beverage intake: Sex differences at age 7 y. Am J Clin Nutr, 2008; 87:903–911.

36.

Faith

, Pietrobelli

, Nunez

et al. Evidence for independent genetic influences on fat mass and body mass index in a pediatric twin sample. Pediatrics, 1999; 104:61–67.

37.

Wardle

, Carnell

, Haworth

et al. Evidence for a strong genetic influence on childhood adiposity despite the force of the obesogenic environment. Am J Clin Nutr, 2008; 87:398–404.

38.

Davison

, Birch

. Obesigenic families: Parents' physical activity and dietary intake patterns predict girls' risk of overweight. Int J Obes Relat Metab Disord, 2002; 26:1186–1193.

39.

Davison

, Birch

. Childhood overweight: A contextual model and recommendations for future research. Obes Rev, 2001; 2:159–171.

40.

Fisher

. Effects of age on children's intake of large and self-selected food portions. Obesity (Silver Spring), 2007; 15:403–412.