Fast food intake and excess weight gain over a 1-year period among preschool-age children

Summary

Background

Fast food is cross-sectionally associated with having overweight and obesity in young children.

Objectives

To examine whether fast food intake independently contributes to the development of overweight and obesity among preschool-age children.

Methods

Prospective cohort of 3- to 5-year-old children (n = 541) followed for year. Children’s height and weight were objectively measured at baseline and study end. Parents reported their child’s fast food intake frequency in the past week from chain fast food restaurants in six online follow-up surveys, completed approximately 8 weeks apart. Poisson regression with robust standard errors modelled the risk of a child increasing in weight status (ie, transitioning from a having a healthy weight to having overweight or from having overweight to having obesity) over the study period in relation to their average weekly fast food intake, adjusted for sociodemographics, child obesogenic behaviours, and parent weight status.

Results

At baseline, 18.1% of children had overweight and 9.8% had obesity; 8.1% of children transitioned to a greater weight status over the 1-year period. Mean fast food intake frequency among consumers was 2.1 (SD: 1.4) times per week. The risk of increasing in weight status increased linearly with each additional time fast food was consumed in an average week over the study year (RR: 1.38; 95% CI, 1.13–1.67; P < .01).

Conclusions

Greater fast food intake over 1 year was associated with increasing weight status during that time in this preschool-age cohort.

1 |. INTRODUCTION

Nearly one-quarter of 2- to 5-year-old children in the United States have overweight or obesity defined as at or above the 85th and 95th percentiles, respectively, for age- and sex-adjusted body mass index (BMI) per the US Centers for Disease Control and Prevention growth charts.¹ In the United States, 8.9% specifically have obesity.² Overweight and obesity in childhood increase the risk for numerous physical and psychosocial problems during childhood,^3–6, including non-alcoholic fatty liver disease, type 2 diabetes, and depression.⁷ Additionally, excess weight gain during the preschool years may specifically impact lifelong health: Children with overweight by the age of 5 years face a four to five times increased risk of having overweight as adolescents⁸ and an increased risk of having obesity as adults.⁹ Efforts to reduce excess weight gain in early childhood are thus paramount.

Fast food intake is a likely and eminently modifiable risk factor for excess weight gain.¹⁰ Fast food meals are characterized by large portion sizes, high saturated fat content, and added sugar.^11,12 Fast food intake is common among children: One-third of US children consume fast food per day.^13–15 Fast food intake is associated with greater total caloric, fat, and added sugar intakes among 4- to 8-year-old¹⁴ and 2- to 11-year-old children¹⁵ and is associated in cross-sectional studies with a greater BMI among children^12,16–18 including those as young as 6 years old.^16,19 However, few prospective studies have assessed associations between fast food intake and weight gain specifically during the formative preschool years,²⁰ and it is not clear whether fast food consumption independently contributes to excess weight gain at such a young age. Prospective studies in young children are needed to properly inform evidence-based guidelines and policies regarding childhood obesity prevention. Therefore, we examined the relationship between weekly fast food intake frequency, measured at multiple occasions over 1 year, and the risk of developing overweight or obesity over that same time among preschool-age children.

2 |. METHODS

Data were from a community-based, prospective cohort study of preschool-age children to identify factors associated with preschool children’s dietary intake.²¹ Participants were recruited March 2014 to October 2015 from community sites in two New Hampshire, US cities; Facebook and participant referrals were also used. Eligible children were aged 3 to 5 years with no health condition impacting food intake and who lived with the enrolled parent at least 3 days a week or alternate weeks. Eligibility criteria for parents (91.4% of whom were mothers) included English literacy, residing within 1 hour’s drive of the recruitment site and no plans to relocate within 1 year. If parents had multiple age-eligible children, the child at the recruitment site was selected; if two or more age-eligible children were present, one was randomly selected. Among the 667 parent-child dyads screened and eligible for the study, 624 (93.6%) enrolled.

Study activities included clinic visits at baseline and approximately 1 year later to measure child and parent height and weight. Parents completed a baseline survey and then six follow-up online surveys. The mean time to the first follow-up survey after baseline was 6.4 (SD: 2.3) weeks, and each follow-up survey was spaced approximately 8 weeks apart. The sixth follow-up survey was completed, on average, 46.6 (SD: 2.8) weeks after baseline. Surveys were pretested with a demographically comparable sample for comprehension, face validity, and completion time. Parents received up to $150 in gift cards, one $25 card after each follow-up survey. Children received two toys for participation, one toy provided after the baseline and a second toy provided after the year 1 clinic visit. Parents provided written, signed informed consent and the Dartmouth College Committee for the Protection of Human Subjects approved the study.

2.1 |. Exposure: children’s weekly fast food intake frequency

Parents responded to the following question at each follow-up survey: In the past 7 days, how many times did your child have something to eat or drink from the following fast food restaurants? A list of 11 popular, chain fast food restaurants located in the study region was provided, including three burger (McDonald’s, Burger King, and Wendy’s), two pizza (Domino’s and Pizza Hut), two chicken (Chick-Fil-A and Kentucky Fried Chicken), one coffee/ doughnut (Dunkin’ Donuts), one sandwich (Subway), one meal/ice- cream (Dairy Queen), and one taco (Taco Bell) restaurant. The response options for each restaurant were 0, 1, 2, 3, 4, or 5 or more times per week. Responses were summed over the 11 restaurants (five or more times was counted as 5) to compute the total number of times each child consumed fast food in the past week. Values for each follow-up survey were averaged per child to compute a mean frequency of fast food intake per week over the study period.

2.2 |. Outcome: children’s weight trajectory

Height and weight measurements were collected at baseline and study end using a standardized protocol. Mean time between the two measurements was 53.1 (SD, SD: 4.7) weeks. Age- and sex- adjusted BMI percentiles were computed using US Centers for Disease Control and Prevention formulas.²² Overweight was defined as an age- and sex-adjusted BMI ≥85th and <95th percentile and obesity was defined as an age- and sex-adjusted BMI ≥95th percentile. Ten children who had underweight (≤5th percentile) at baseline were included with the healthy weight group. Increasing in weight status was defined as moving from a healthy weight to having overweight, or having overweight to having obesity, from baseline to study end; one child moved from healthy weight to having obesity and was included in the latter group. Similarly, decreasing in weight status was defined as moving from having overweight to healthy weight, or having obesity to having overweight, over the study. Each reference group included children whose weight status did not change.

2.3 |. Covariates

2.3.1 |. Screen time

Parents reported their child’s usual TV (regular, cable, or satellite) and other screen time (DVDs/VHS, streaming, apps, Internet use, or electronic games) at baseline as the number of days per week and hours per day spent on each activity; responses were multiplied to compute hours per week.²³ Parents also reported if their child watched advertisement-supported children’s TV (yes vs no),²¹ and if the child had a TV in his/her bedroom (yes vs no).

2.3.2 |. Physical activity

Parents reported the time their child spent playing outside on a typical weekday and a typical Saturday (<1 hour, 1 to 2 hours, 3 to 4 hours, 5 to 6 hours, and >6 hours per day). A weighted average was computed to approximate usual hours of outdoor play per week. Children’s outdoor play time as reported with one week and one weekend day is a valid proxy for increased physical activity among children aged 3 to 12 years.²⁴

2.3.3 |. Nighttime sleep

Parents reported their child’s usual bedtime and wake time. Information on child naps was not collected. Nighttime sleep was dichotomized as <10 vs ≥10 hours per night based on the American Academy of Pediatrics recommendations for adequate sleep for 3 to 5-year-old children.²⁵

2.3.4 |. Diet quality

Adequate fruit and vegetable intake is an essential component of a healthy diet.^26,27 Thus, children’s daily fruit and vegetable intake was used to approximate diet quality. Parents reported on their child’s usual frequency of consuming fruit (fresh, frozen, or canned) and vegetables (fresh, frozen, canned, or salad); the responses for each were ordinal and were collapsed into one dichotomous measure of less than, vs at least, two servings of each per day. That threshold roughly approximates the USDA dietary recommendations for 3- to 5-year- old children of 2- to 3½-cup servings of fruit and 2- to 3½-cup servings of vegetables per day.²⁷

2.3.5 |. Family meals

Parents reported how many days a week their family usually ate dinner together (0, 1 to 2, 3 to 4, 5 to 6, and 7 nights per week). Few (n = 10) parents reported 2 or less nights per week, thus responses were analysed as less than, vs at least, 5 nights per week. More frequent family meals may be protective against developing obesity among preschool-age children.²⁸

2.3.6 |. Children’s usual fast food intake pre-baseline

Parents reported their child’s usual fast food intake frequency at the baseline visit (never, less than once a month, less than once a week, 1 to 2 times per week, 3 to 4 times per week, and 5 or more times per week). Responses were categorized as less than once a month, at least monthly to less than once a week, or at least weekly.

2.3.7 |. Sociodemographic, household, and parent characteristics

Parents reported child characteristics (age, gender, race/ethnicity, participation in the Women, Infants and Children supplemental nutrition program [WIC]), parent characteristics (age, education level, and cohabitation with a spouse or partner), and annual household income. Parent weight status at the baseline visit was classified as healthy weight (<25 kg/m²), overweight (25.0 to 29.9 kg/m²), or obese (≥30 kg/m²). Weight status was objectively measured for 407 (75.2%) parents and self-report for 108 (20.0%) parents who declined the objective measurements. Parent BMI was missing for another 26 (4.8%) parents who refused both assessments, and a missing category was included to retain participants’ missing weight status in the analysis.

2.4 |. Statistical analyses

Children with objectively measured weight status at both time points were included in the analysis (n = 541). Bivariate associations between baseline characteristics and the child having overweight or obesity (combined) were completed using chi-square tests or two-sample t tests as appropriate. Children’s mean frequency of fast food intake per week was compared across all combinations of weight trajectories over the study using one-way analysis of variance (ANOVA) with Tukey’s honestly significant difference (HSD) post-hoc tests to assess between group comparisons.

To assess adjusted associations between weekly fast food intake frequency and increased weight status over the study, ordinal regression was first used to fit child weight status (healthy weight, overweight, or obese) at 1 year on mean frequency of fast food intake per week during the study, adjusted for baseline weight status, the time between the baseline and 1-year clinic visit, and covariates. Child pre-baseline frequency of fast food intake was not included in the model because the goal was to assess the effect of fast food intake on subsequent weight gain, given children’s weight status at baseline. The proportional odds assumption as assessed using Brant’s test²⁹ was not violated in the ordinal regression model. Therefore, to simplify the model presentation and to calculate relative risks, adjusted Poisson regression with robust standard errors was used to model the risk of a child increasing in weight status from baseline to the study end on children’s mean frequency of fast food intake per week over the study. A model treating weekly fast food intake as ordinal (0, 0.1 to 1.0, 1.1 to 2.0, 2.1 to 3.0, or 3.1 or more times per week) supported a linear dose-response relationship. Thus, weekly fast food intake was treated as continuous in the final regression model. The final model was fit for boys and girls, separately, for comparison to previous stud- ies.^20,30 A series of sensitivity models further adjusted for children’s pre-baseline fast food intake frequency and modelled change in BMI- percentile as a linear outcome on mean frequency of weekly fast food intake using linear regression adjusted for the same covariates and children’s BMI-percentile at baseline. In an additional sensitivity analysis, we assessed whether children’s baseline BMI-percentile modified the association between their weekly fast food intake frequency and change in BMI-percentile by including an interaction term between those two predictors in the model; statistical significance was determined with a Wald test. The threshold for statistical significance was P < .05 (two-sided tests), and analyses were completed with the R Language for Statistical Computing and Stata 15.0. The study was powered at 90% to detect a 5% change in the rates of having overweight or obesity by quartiles of an exposure category based on an expected sample size of 670.

3 |. RESULTS

3.1 |. Analytic sample

Children excluded because of missing anthropometrics (n = 83) were more likely to be a racial/ethnic minority, view advertisement- supported TV, have a TV in the bedroom, have more weekly screen time, and their parents had lower levels of education, lower annual household income, and were more likely to live without a partner (all P-values < .05) than children included in the analysis (n = 541). Baseline rates of having overweight/obesity (P = .26) and pre-baseline fast food intake frequency (P = .34) did not differ between the analytic and excluded subsets. Data on past-week fast food intake were available for 73 children excluded from the analysis, and mean frequency of fast food intake per week was greater for those 73 children vs those included in the analysis (mean [SD]: 2.4 [2.5] vs 1.6 [1.2] times per week; P < .01).

Study participants in the analytic sample were socioeconomically diverse based on parent education and annual household income but less diverse based on ethnic or racial minority (Table 1). At baseline, 27.9% of children had overweight or obesity (18.1% had overweight and 9.8% had obesity). Several measures were statistically associated with a child having overweight or obesity at baseline, including more frequent pre-baseline fast food intake (P < .01) and parent weight status (P < .01).

TABLE 1.

Distribution of child, parent, and household characteristics and unadjusted associations with child’s weight status at baseline

		Baseline weight statusa
	Overall, n	Healthy weight, n (row %)	Overweight or obese, n (row %)	P valueb
Overall	541	390 (72.1)	151 (27.9)	-
Child characteristics
Age
3 y	221	158 (71.5)	63 (28.5)	.20
4 y	204	141 (69.1)	63 (30.9)
5 y	116	91 (78.5)	25 (21.6)
Gender
Female	303	230 (75.9)	73 (24.1)	.03
Male	238	160 (67.2)	78 (32.8)
Race/ethnicity
White non-Hispanic	470	348 (74.0)	122 (26.0)	.01
Racial or ethnic minority	71	42 (59.2)	29 (40.9)
Receives WIC benefits
No	474	348 (73.4)	126 (26.6)	.09
Yes	67	42 (62.7)	25 (37.3)
Watches ad-supported children’s TV
No	388	290 (74.7)	98 (25.3)	.04
Yes	153	100 (65.4)	53 (34.6)
Has a TV in the bedroom
No	448	331 (73.9)	117 (26.1)	.06
Yes	93	59 (63.4)	34 (36.6)
Weekly screen time, mean (SD)	17.7 (14.7)	16.9 (13.1)	24.9 (19.3)	.01
Hours of outside play in the past month
<14 hours per week	205	146 (71.2)	59 (28.8)	.24
14 to 21 hours per week	147	100 (68.0)	47 (32.0)
21 or more hours per week	189	144 (76.2)	45 (23.8)
Nighttime sleep
<10 hours per night	46	30 (65.2)	16 (34.8)	.36
≥10 hours per night	495	360 (72.7)	135 (27.3)
Fruit and vegetable intake
<2 times per day for each	354	246 (69.5)	108 (30.5)	.08
≥2 times per day for each	187	144 (77.0)	43 (23.0)
Frequency of family cooking dinner at home
<5 nights per week	69	41 (59.4)	28 (40.6)	.02
≥5 nights per week	472	349 (73.9)	123 (26.1)
Pre-baseline usual frequency of eating fast food
Less than once a month	190	156 (82.1)	34 (17.9)	<.001
At least monthly to less than once a week	202	135 (66.8)	67 (33.2)
At least weekly	149	99 (66.4)	50 (33.6)
Parent characteristics
Age
20–29 y	100	72 (72.0)	28 (28.0)	.99
30–39 y	353	255 (72.2)	98 (27.8)
40 y or older	88	63 (71.6)	25 (28.4)
Education
Some high school or less	118	78 (66.1)	40 (33.9)	.18
Associate’s or technical degree	92	66 (71.7)	26 (28.3)
Bachelor’s degree	180	128 (71.1)	52 (28.9)
Graduate degree	151	118 (78.2)	33 (21.9)
Annual household income
Less than $25 000	38	22 (57.9)	16 (42.1)	.02
$25 001–$75 000	183	124 (67.8)	59 (32.2)
$75 001–$125 000	208	153 (73.6)	55 (26.4)
More than $125 000	112	91 (81.3)	21 (18.8)
Lives with spouse or partner
No	69	41 (59.4)	28 (40.6)	.02
Yes	472	349 (73.9)	123 (26.1)
Parent weight statusc
Healthy weight (<25 kg/m2)	204	171 (83.8)	33 (16.2)	<.001
Overweight (25–29 kg/m2)	145	107 (73.8)	38 (26.2)
Obese (≥30 kg/m2)	166	100 (60.2)	66 (39.8)

Note: Among 541 children enrolled in a 1-year, prospective cohort.

^aOverweight or obese defined as having an age- and sex-adjusted BMI percentile at or above the 85th percentile.

^bP-values from chi-square tests comparing weight status (normal vs overweight/obese) between levels of each characteristic, except for weekly screen time, in which mean values were compared by weight status using a two sample t test.

^cWeight status was missing for n = 26 parents. Missing values were included as separate category in the regression models to enable a complete-case analysis.

^*P < .05

^**P < .01.

Most parents (85.6%) completed all six online follow-up surveys; 97.8% completed four or more. The percent of children who ate fast food in the past week ranged from 70.9% to 74.6% at each follow-up survey and was 73.2% when averaged over all visits (Table 2). Children most frequently consumed items from a coffee/doughnut shop (47.2%) or a burger restaurant (43.1%). Mean frequency of fast food intake per week ranged from 0 to 9.7 times per week and averaged 1.6 (SD: 1.5) times per week in all children and 2.1 (SD: 1.4) times per week among children who ate fast food.

TABLE 2.

Percent of children who had something to eat or drink from select fast-food restaurants in the past week and frequency of intake in the past week overall and among consumers only

		Children with any fast food intakea	Mean frequency of fast food intake per weekb
	Restaurants included	All children	All children	Consumers only
	n	% (Range)	Mean (SD)	Mean (SD)
Any fast food restaurant	11	73.2 (70.9, 74.6)	1.6 (1.5)	2.1 (1.4)
By restaurant type
Coffee/doughnut shop	1	47.2 (44.0, 49.9)	0.7 (0.9)	1.5 (0.8)
Burger	3	43.1 (41.3, 44.7)	0.6 (0.8)	1.3 (0.7)
Pizza	2	9.1 (7.9, 11.6)	0.1 (0.3)	1.1 (0.4)
Chicken	2	5.4 (4.6, 6.8)	0.1 (0.2)	1.1 (0.2)
Sandwich	1	4.7 (2.7, 6.8)	0.1 (0.2)	1.0 (0.2)
Meals/ice-cream shop	1	4.3 (2.7, 5.3)	0.1 (0.2)	1.1 (0.4)
Taco	1	2.3 (1.2, 3.2)	0.1 (0.2)	1.1 (0.3)

Note: Among 541 children enrolled in a 1-year, prospective cohort.

^aValues are the percent of children who ate fast food in the past week, where the percent is averaged over each of the six follow-up surveys. The range of percents computed at each of the six follow-up surveys is also presented.

^bValues are the mean frequency of intake in the past week (times per week), averaged over all six follow-up surveys. Values are presented for all children and subset to children who consumed any fast food from each restaurant type.

Most children (n = 448, 82.8%) maintained the same weight status over the study (Table 3); 44 (8.1%) increased in weight status and 49 (9.1%) decreased in weight status. Mean frequency of fast food intake per week was lowest for children with a healthy weight at baseline and the 1-year follow-up or who transitioned from having overweight to a healthy weight. Mean intake frequency was greatest among children who increased in weight status or who had obesity at both time points.

TABLE 3.

Child mean frequency of fast food intake per week by child’s weight trajectory over the study

				Mean frequency of fast food intake per weeka
Weight trajectory	Weight status at baselineb	Weight status at year 1b,c	n	Mean (SD)
Weight stable	Healthy	Healthy	358	1.4 (1.1)1,2,3
	Overweight	Overweight	52	1.8 (1.3)
	Obese	Obese	38	2.1 (1.1)2
Increased in weight status	Healthy	Overweight	32	2.1 (1.6)1
	Overweight	Obese	12	2.8 (2.5)3,4
Decreased in weight status	Overweight	Healthy weight	34	1.4 (1.0)4
	Obese	Overweight	15	1.8 (1.2)

Notes: Among 541 children enrolled in a 1-year, prospective cohort.

^1,2,3,4Values that share the same superscript are statistically different from each other at the P < .05 level based on Tukey’s HSD post-hoc tests, completed after a one-way ANOVA. All between group means were compared with each other and only statistically significant differences are noted with superscripts.

^aValues are the mean frequency of intake in the past week (times per week), averaged over all six follow-up surveys. Values are presented for all children and are not limited to consumers only.

^bHealthy weight defined as having an age- and sex-adjusted BMI percentile below the 85th percentile, overweight defined as having an age- and sex-adjusted BMI percentile at or above the 85th percentile to below the 95th percentile, and obese defined as having an age- and sex-adjusted BMI percentile at or above the 95th percentile.

^cMean time from baseline to study end was 53.1 (SD: 4.7) weeks.

There was a positive, linear relationship between mean frequency of fast food intake per week and the risk of increased weight status over the study (Table 4). In the fully adjusted model, the risk of transitioning to a greater weight status increased 38% (RR: 1.38; 95% CI, 1.13–1.67; P = .001) with each additional fast food intake per week, on average, over the study. In the same fully adjusted model, a 1-hour increase in weekly screen time was associated with a 2% increased risk of transitioning to a greater weight status over the study period (RR: 1.02; 95% CI, 1.01–1.04; P < .01) (Table S1), and those with an annual household income <$25 000 had an elevated risk of increasing in weight status compared with those with an annual household income of $75 001 to $125 000 (RR: 2.90; 95% CI, 1.12–7.29; P = .03). No other factor was associated with increased weight status in the fully adjusted model. Findings were similar when analysed by gender (P-for-interaction = .81) and when further adjusted for children’s pre-baseline fast food intake frequency.

TABLE 4.

Adjusted associations between children’s mean frequency of fast food intake per week and the risk of increasing in weight status over the study

		Outcome: increased weight status from baselinea (n = 44)
		RR (95% CI)	RR (95% CI)
		Unadjusted	Adjustedb
Exposure modelled as ordinal
Child mean frequency of fast food intake per weekc
0	(n = 32)	0.39 (0.05, 2.88)	0.57 (0.08, 4.05)
≤1.0	(n = 177)	0.58 (0.25, 1.33)	0.65 (0.28, 1.48)
>1.1 to 2.0	(n = 185)	1.00 (reference)	1.00 (reference)
>2.1 to 3.0	(n = 93)	1.46 (0.70, 3.03)	1.29 (0.58, 2.88)
3.1 or more	(n = 54)	2.02 (0.95, 4.32)	2.17 (0.96, 4.92)
P for linear trendd		P < .01	P = .02
Exposure modelled as continuous
Child mean frequency of fast food intake per weekc		1.37 (1.22, 1.53)*	1.38 (1.13, 1.67)*

Note: Among 541 children enrolled in a 1-year, prospective cohort.

Relative risks were computed using Poisson regression with robust standard errors; each model was adjusted for the time between baseline and study end per child.

Abbreviation: RR, relative risk.

^aIncreased weight status was defined as moving from a healthy weight at baseline to having overweight at study end, or moving from having overweight at baseline to having obese at study end.

^bCovariates were child age, gender, race/ethnicity; child watches ad-supported TV; child has a TV in his/her bedroom; child weekly screen time (hours per week), outside play time in past month (hours per week), nighttime sleep, daily fruit and vegetable intake, and frequency of dinners cooked at home; annual household income; parent living with spouse or partner; and parent weight status.

^cFast food intake frequency in the past week is the average over all six follow-up surveys per child; mean weekly intake ranged from 0 to 9.7 times per week.

^dSimple linear regression used to test for a linear trend in the relative risks over the increasing categories of fast food intake.

^*P ≤ .001.

Finally, children’s mean frequency of weekly fast food intake over the study year was not statistically associated with a linear change in BMI-percentile by the end of the study period in an adjusted linear regression model (beta for the increase in BMI-percentile for each additional time fast food was consumed per week: 0.50; 95% CI, −0.51–1.51; P = .14). However, there was a statistically significant positive interaction between children’s BMI-percentile at baseline and mean frequency of weekly fast food intake in relation to BMI percentile at the study end (P-for-interaction = .04). Specifically, a statistically significant increase in BMI-percentile over the study period was observed in children with both a higher baseline BMI-percentile and a higher mean frequency of weekly fast food intake (Figure S1).

4 |. DISCUSSION

In this prospective study of 541 preschool-age children, children who ate fast food more frequently over a 1-year period were more likely to increase their weight status during that time. Importantly, that association was independent of a comprehensive set of other obesity risk factors, including parent BMI. We also observed a linear dose- response relationship between mean frequency of fast food intake per week and increased weight status, supporting a biologically plausible effect. In a sensitivity analysis modelling change in BMI-percentile as a continuous outcome, children with the highest BMI-percentile at baseline were most susceptible to excess weight gain related to more frequent fast food intake. Together, findings suggest that a more frequent fast food intake may be most problematic for children at risk of developing, or currently with, overweight or obesity. Our findings indicate that fast food intake during preschool years may have meaningful, independent impact on obesity risk and health.

Fast food is typically energy-dense and nutritionally poor, and serving sizes often exceed recommended meal and snack caloric limits.^15,31,32 Fast food intake has been prospectively associated with excess weight gain among adults.¹⁰ However, there are limited prospective data supporting associations among young children. In a French cohort of 883 preschool-age children,²⁰ a processed dietary pattern, marked in part by a greater fast food intake, at age 2 years was not related to percent body fat nor BMI at age 5 years adjusted for screen time and outdoor play time; notably, the rate of having overweight or obesity at age 5 years in that sample was low at 7.6%. In a US prospective birth cohort,³⁰ eating fast food more than weekly (as measured annually) related to a greater BMI z-score among girls as they aged from 4 to 11 years, while the effect for boys was not significant. Our study contributes to the literature by demonstrating that fast food intake, measured at multiple times over 1 year, affected subsequent risk of having overweight or obesity equally for boys and girls during the preschool years.

Our study also extends previous research among preschoolers specifically by controlling for parental (primarily maternal) weight status, which was mostly objectively measured. Parent and child weight status are highly correlated,³³ and that effect is likely mediated by multiple factors including genetics, a shared lifestyle, and shared environmental factors. By controlling for parental weight status, our analysis in part adjusted for such unmeasured confounders. Additionally, our results suggest that reducing or eliminating fast food intake in the preschool years may reduce the risk of having overweight or obesity independently of those factors.

This study considered 11 popular chain fast food restaurants located in the study region. Other sit-down, fast-casual, or local restaurants were not included. Consuming meals away from home, even at non-fast food restaurants, is associated with a greater caloric intake than consuming meals prepared at home.³⁴ We focused on children’s intake from fast food restaurants because of the great reach and popularity of these restaurants among children and their parents.^35,36 However, because we did not include meals from other restaurant types, our findings are not generalizable to all meals consumed away from home. Furthermore, menus in fast food restaurants are standardized, which facilitates policy efforts to decrease the serving sizes, lower the caloric content, and improve the nutritional quality of children’s meals, snacks, and drinks.^32,37,38 Our data also documented that children frequently consumed items from a coffee/doughnut shop. This finding is important because these restaurants are often overlooked in efforts to improve the healthfulness of menu items for children. Notably, the coffee/doughnut shop assessed here is not currently a member of the National Restaurant Association’s Kids Live Well Program to improve the nutritional quality of children’s meals.³⁸

Strengths of this study include a high rate of follow-up, objective measures of height and weight, and adjustment for multiple obesity risk factors. Study limitations include the regional study sample and the low level of racial and ethnic diversity. Although a valid proxy,²⁴ our assessment of children’s physical activity was fairly crude: We focused specifically on outdoor play for children, and our response categories did not include half hour increments. We did not measure specific items that children consumed from the fast food restaurants, and we did not assess other aspects of children’s diet, which may have been correlated with fast food intake³⁹ and contributed to the observed shifts in weight status. Analyses did not include meals away from home from other restaurant types, and it is possible that the observed findings may be due to residual confounding because of eating other meals away from home. However, such residual confounding would likely have been captured by our question about the frequency in which the family cooks dinner at home. Differential drop-out may have affected the generalizability of our findings. Specifically, the characteristics of children not included in the analysis were characteristics associated with having overweight or obesity at baseline. Our overall follow-up rate was high, but it is important to note that the children excluded may have been most at risk of excess weight gain and consumed fast food more frequently over the study. It is uncertain if the association between more frequent fast food intake and excess weight gain would have been stronger or attenuated with that higher-risk subset included.

5 |. CONCLUSIONS

Fast food intake is a likely obesity risk factor among young children, yet longitudinal studies documenting temporal relationships have been lacking. Our study fills that gap and indicates that a greater fast food intake, measured as children’s mean frequency of fast food intake per week over a 1-year period, is associated with an increased risk of transitioning to a greater weight status. These findings should be used to inform guidelines and policies regarding childhood obesity prevention programmes and policies.