Rural–urban differences in body mass index and obesity-related behaviors among low-income preschoolers

Abstract

Background

With one in eight preschoolers classified as obese in the USA, childhood obesity remains a significant public health issue. This study examined rural–urban differences in low-income preschoolers’ body mass index z-scores (BMIz), eating behaviors, dietary quality, physical activity (PA) and screen time.

Methods

Pre-intervention data from 572 preschooler-parent dyads participating in a randomized, controlled obesity prevention trial in the Midwest USA were analyzed. We examined the associations among living in rural versus urban areas, child BMIz and child obesity-related behaviors, including eating behaviors, dietary quality, PA and screen time.

Results

Rural children had higher BMIz, more emotional overeating behaviors and more time spent playing outdoors compared with urban children. We found no associations between children living in rural versus urban areas and dietary quality and screen time.

Conclusions

The study found that rural–urban differences in BMIz may start as early as 3–4 years of age, if not earlier. To reverse the weight-related health disparities between rural and urban low-income preschoolers, structural changes in rural locations and family supports around coping skills may be needed.

Introduction

With one in eight preschool-aged children with obesity in the USA and even higher rates among low-income preschoolers, childhood obesity remains a significant public health issue.¹ Several studies have examined associations of living in rural versus urban locations with children’s weight status and health behaviors, such as dietary quality and physical activity (PA).^2–6 Most research has shown higher rates of overweight and obesity for rural, compared with urban children, despite higher levels of PA among rural children.^2–4,6–8 The few studies involving preschoolers have had mixed findings.^2,4,5 Among 14 332 youth aged 2–19 years participating in the US National Health and Nutrition Examination Survey,⁴ rural youth were 30% more likely to be overweight or obese than urban youth, independent of covariates. The association was moderated by age such that the effect was present in adolescents, but not among children younger than age 12 years.⁸ In a meta-analysis of 74 168 youth between 2 and 19 years across five published studies,² rural youth were 26% more likely to be obese than urban youth. However, the only study in the meta-analysis focusing on preschoolers found no differences in obesity based on location.

Characteristics of the rural environment that may contribute to a higher risk of obesity include poverty, barriers to PA, limited access to affordable healthy food and reduced access to preventative care and nutrition education.^9–13 Childhood exposure to the stressors associated with living in a rural environment may also lead to obesity, as well as obesity-promoting eating behaviors and poor dietary quality.^14–19 Specifically, financial hardship, poor maternal health and adverse life events have been associated with lower intake of fruits and vegetables among preschoolers.¹⁵ Additionally, the higher rates of poverty and psychosocial stressors associated with poverty in rural areas may contribute to higher weight status.¹⁶ Early childhood exposure to psychosocial stress has been associated with prospective increases in eating in the absence of hunger and emotional overeating.¹⁶

This study sought to examine rural–urban differences in preschoolers’ body mass index (BMI) and obesity-related behaviors, including eating behaviors, dietary quality, PA and screen time. We hypothesized that preschoolers living in rural (versus urban) environments would have a higher BMI, more obesity-promoting eating behaviors, poorer dietary quality, less PA and more screen time.

Materials and methods

Study design and sample

The study was an analysis of pre-existing data from the Growing Healthy Study, a randomized controlled obesity prevention trial among 697 preschoolers attending Head Start. For a full description of the Growing Healthy Study, see Lumeng et al.’s and Miller et al.’s work.^20,21 Parents and their preschoolers, ages 3–4 years, who attended Head Start preschool programs from 2011 to 2015 in three rural and urban Michigan counties were recruited for participation. Families who expressed an interest in the study were provided a detailed description of the study and eligibility criteria. Exclusion criteria were child living in foster care, child having a significant developmental disability or medical problem or a parent who was not fluent in English. Families who met inclusion criteria were enrolled following written informed consent. The participating parent was the primary caregiver in each household and is referred to as parent throughout the study. This study was approved by the Institutional Review Boards at the University of Michigan and Michigan State University. Data collection occurred in the home and classroom and included height and weight measurements, 24-hour dietary recalls, questionnaires and participant demographics, such age and race/ethnicity. The sample was limited to participants with data for location, weight status, eating behaviors, dietary quality, PA and screen time, resulting in a final analytic sample of N = 572. Urbanization was determined by the Census Bureau’s Urban–Rural Classifications. In this study, counties with a population of 50 000 or more were identified as urban. Rural areas contained nonmetropolitan areas, including areas identified as ‘urban clusters’ (e.g. a population at least 2500 and less than 50 000). The sample was relatively evenly divided between preschoolers living in rural (n = 273) versus urban (n = 299) areas, as determined by the location of the Head Start facility.

Measures

Anthropometry

Trained research assistants measured preschoolers’ and parents’ height and weight without shoes or heavy clothes using the Seca 213/217 portable stadiometer and Detecto portable scale. BMI (weight in kilograms divided by height in meters squared [kg/m²]) was calculated for preschoolers and parents. Preschoolers’ BMI z-scores (BMIz) and percentiles were generated based on the US Centers for Disease Control reference growth curves for age and sex.^22,23

Obesity-promoting eating behaviors

The Child Eating Behavior Questionnaire²⁴ was used to measure preschooler’s eating behaviors. This 34-item parent-report measure is based on a 5-point Likert scale (1 = never to 5 = always). We examined subscales previously reported to be associated with greater obesity risk and considered to be food approach behaviors^25,26: Food Responsiveness, e.g. ‘My child is always asking for food’ (5 items; α = 0.79), Enjoyment of Food, e.g. ‘My child loves food’ (4 items; α = 0.85), Satiety Responsiveness/Slowness in Eating, e.g. ‘My child eats slowly’ (9 items; α = 0.78) and Emotional Overeating, e.g. ‘My child eats more when worried’ (3 items [1-item was removed to increase alpha]; α = 0.73). Lower satiety responsiveness/slowness in eating scores was indicative of greater obesity risk.

Dietary quality

Three 24-hour dietary recalls of preschoolers’ intake were obtained by trained dietitians using the USDA 5-step automated multiple-pass method, which has been shown to improve the accuracy of dietary recall.²⁷ The initial 24-hour dietary recall was implemented at a home visit where parents reported their preschooler’s intake through the assistance of food models and handouts showing child-appropriate portion sizes. Two additional unannounced 24-hour dietary recalls were collected from the parents via phone. The 24-hour dietary recalls generated data on the nutrient and energy content of each food item using the Nutrition Data System for Research program.²⁸ We then estimated mean daily nutrient and food intake across all the 24-hour recalls.

Physical activity

Parental report of outdoor play is a valid estimation of PA for preschoolers.²⁹ Outdoor playtime was measured by a 2 item ‘Outdoor Time Recall’ scale (minutes per week). This recall scale collects information on daily time, in minutes, spent playing outdoors. Parents reported the amount of time their child ‘typically’ spent playing outdoors each day in the last month. All responses to this question were obtained in September or early October. Higher scores indicated that the child spent more time, in minutes, playing outdoors per week.

Screen time

Parental report of screen time included 2 questions about the child’s TV viewing and 1 question about the child’s computer use. The questions were obtained from the National Longitudinal Survey of Youth.³⁰ Parents were asked about the amount of time their child spends watching TV, including videos and time using a computer on a typical weekday and on a typical weekend day (hours per day).

Covariates

Parent demographic characteristics, household food insecurity status and parent BMI were included as covariate variables. Demographic variables included parents’ race/ethnicity, education and marital status. A majority of parents identified as non-Hispanic White or Black. Therefore, two dichotomous (1/0) variables of non-Hispanic White and Black parent racial/ethnic categories were used in the analyses. Parent education was rated on a 6-point continuum, ranging from 1 (no high school diploma) to 6 (more than a 4-year college degree). Lastly, parents were coded as single-parent if they were not married and not living with a partner. Food insecurity was measured using the US Household Food Security Questionnaire.³¹ This 18-item survey determines the level of food security (high, moderate, low or very low) and the amount of identifiable hunger in the household members. Low and very low household food security levels were coded as 1.

Statistical analyses

Univariate statistics were used to describe the sample. Independent samples t-test and Chi-square analyses were used to examine differences in sample characteristics between rural and urban families. To test the study’s hypotheses, we examined: (i) the association between geographic location and child BMIz and (ii) associations among geographic location, child obesity-related behaviors and child BMIz.

All analyses were computed in Mplus v.8 (Muthén & Muthén, 1998–2017) and controlled for relevant covariates. Model fit was assessed with the chi-square (χ²) test, comparative fit index (CFI), root mean square error of approximation (RMSEA) and standardized root mean square residual (SRMR). A non-significant χ²P-value, CFI > 0.90 and RMSEA and SRMR < 0.08 indicated good fit. Cohen’s d effect size estimates of the unstandardized regression coefficients were calculated for geographic location (rural versus urban) model paths. Full information maximum likelihood estimation was used to handle missing data.

Results

Table 1 summarizes the data between rural and urban families. Preschoolers were aged 4.08 years (SD = 0.51) on average, with about half male. Parent age ranged from 19 to 64 years (M = 29.60 years, SD = 6.85). Fifty-nine percent of parents were non-Hispanic White, 29% were Black or African American and 7% were Hispanic/Latino or other. Forty-three percent were single parents and 48% had a high-school diploma or less as the highest level of education. Difference tests indicated that more rural parents were non-Hispanic White than any other race/ethnicity. More urban parents were Black than any other race/ethnicity. Urban parents (versus rural) were more likely to be single parents.

Table 1.

Sample characteristics and assessment data by geographic location

	Total (N = 572)	Geographic location		P-value
		Rural (n = 273)	Urban (n = 299)
Sample characteristics
Child age, M (SD), in years	4.08 (0.51)	4.11 (0.53)	4.06 (0.49)	0.29
Child sex, n (%)				0.37
Male	278 (49)	138 (51)	140 (47)
Female	294 (51)	135 (49)	159 (53)
Parent age, M (SD), in years	29.60 (6.85)	29.96 (7.15)	29.27 (6.56)	0.23
Parent race/ethnicity, n (%)				<0.001
White (non-Hispanic)	337 (59)	250 (92)	87 (29)
Black or African American	166 (29)	0 (0)	166 (56)
Hispanic/Latino	39 (7)	12 (4)	27 (9)
Other or not specified	30 (5)	11 (4)	19 (6)
Parent education, n (%)				0.16
<High school	87 (15)	35 (13)	52 (17)
High school diploma or GED	183 (32)	96 (35)	87 (29)
>High school	302 (53)	142 (52)	160 (54)
Single parent, n (%)	243 (43)	82 (30)	161 (54)	<0.001
Assessment data
Food insecure, n (%)	215 (38)	103 (38)	112 (38)	0.95
Parent BMI, M (SD)	31.54 (8.65)	30.91 (8.35)	32.12 (8.89)	0.10
Child BMI z-scores, M (SD)	0.62 (1.16)	0.73 (1.17)	0.53 (1.15)	0.04
Child obesity-related behaviors, M (SD)
Obesity-promoting eating behaviors
Food responsiveness	2.76 (0.97)	2.80 (0.96)	2.73 (0.97)	0.42
Enjoyment of food	4.00 (0.79)	4.00 (0.74)	4.00 (0.84)	0.97
Satiety responsiveness/slowness in eating	2.90 (0.63)	2.93 (0.61)	2.87 (0.65)	0.28
Emotional overeating	1.58 (0.66)	1.63 (0.67)	1.54 (0.64)	0.11
Dietary quality	60.1 (10.8)	60.35 (10.86)	59.87 (10.82)	0.60
PA	1082.0 (632.1)	1192.9 (681.0)	978.1 (564.3)	<0.001
Screen time	2.63 (1.60)	2.34 (1.39)	2.90 (1.72)	<0.001

Bivariate associations between geographic location, child BMIz, obesity-related behaviors and covariate variables are presented in Table 2. Child BMIz were related to parent BMI (r = 0.20, P < 0.001), rural area (r = 0.09, P = 0.04), enjoyment of food (r = 0.14, P < 0.001), satiety responsiveness/slowness in eating (r = −0.21, P < 0.001) and PA (r = 0.10, P = 0.02). Rural area was positively associated with PA (r = 0.17, P < 0.001). Rural families reported less screen time (r = −0.18, P < 0.001). Screen time was also correlated with food responsiveness (r = 0.12, P = 0.004) and dietary quality (r = −0.18, P < 0.001).

Table 2.

Correlations among study variables and covariates

Variables	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15
1. Parent race/ethnicity (non-Hispanic White)	–
2. Parent race/ethnicity (Black)	−0.77*	–
3. Parent education	−0.03	0.05	–
4. Single parent	−0.27*	0.26*	−0.06	–
5. Food insecure	0.02	−0.04	−0.04	−0.02	–
6. Parent BMI	−0.04	0.05	00.03	0.01	0.09*	–
7. Rural area	0.63*	−0.61*	0.03	−0.24*	0.00	−0.07	–
8. Child BMI z-scores	0.06	−0.02	−0.02	0.04	0.03	0.20*	0.09*	–
9. Food responsiveness	−0.02	0.02	−0.13*	−0.02	0.12*	−0.04	0.03	0.07	–
10. Enjoyment of food	−0.06	0.09*	−0.05	0.04	0.01	−0.09*	0.00	0.14*	0.51*	–
11. Satiety responsiveness/slowness in eating	0.08†	−0.09*	0.06	−0.02	0.05	0.04	0.05	−0.21*	−0.15*	−0.49*	–
12. Emotional overeating	−0.01	0.02	−0.10*	0.06	0.10*	0.05	0.07	0.04	0.36*	0.15*	−0.03	–
13. Dietary quality	0.08†	−0.09*	0.09*	−0.02	0.01	0.00	0.02	−0.04	−0.02	0.06	−0.06	0.00	–
14. PA	0.14*	−0.14*	−0.04	0.04	0.01	0.03	0.17*	0.10*	0.03	−0.03	−0.03	0.07	0.05	–
15. Screen time	−0.19*	0.19*	−0.04	0.09*	0.06	−0.03	−0.18*	0.04	0.12*	0.03	0.01	0.06	−0.18*	−0.05	–

^† P < 0.10.

^* P < 0.05.

Parameter estimates for model paths from covariate variables and geographic location to child BMIz are shown in Table 3. The model fit statistics demonstrated a good fit to the data, χ² = 4.29, df = 3, P = 0.23, CFI = 1.00, RMSEA = 0.03 and SRMR = 0.01. In the presence of covariates, there was a significant, small-to-medium association of living in rural areas with higher BMIz (β = 0.13, P = 0.02; d = 0.26, 95% CI: 0.09, 0.42).

Table 3.

Model coefficients of the association between geographic location and children’s BMI z-scores

	Child BMI z-scores
Model path	B	SE	β	P-value
Parent race/ethnicity (non-Hispanic White)	0.18	0.17	0.08	0.29
Parent race/ethnicity (Black)	0.25	0.17	0.10	0.14
Parent education	−0.03	0.04	−0.03	00.44
Single parent	0.14	0.10	0.06	0.16
Parent BMI	0.03	0.01	0.20	<0.001
Food insecure	0.04	0.10	0.02	0.69
Rural area	0.30	0.13	0.13	0.02

Expanding on the previous model, the inclusion of children’s obesity-related behaviors paths associated with geographic location and child BMIz was simultaneously tested. The model fit statistics demonstrated a good fit to the data, χ² = 22.91, df = 17, P = 0.15, CFI = 0.99, RMSEA = 0.03 and SRMR = 0.02. Figure 1 summarizes the standardized path coefficients for the associations among geographic location, obesity-related behaviors and children’s BMIz. After adjusting for child obesity-related behaviors, the association between geographic location and child BMIz was significant, consistent to the previous model. Rural children showed higher emotional overeating (β = 0.17, P = 0.001; d = 0.34, 95% CI: 0.18, 0.51) and more PA (β = 0.14, P = 0.01; d = 0.29, 95% CI: 0.12, 0.45). Rural children also showed marginally higher food responsiveness (β = 0.10, P = 0.06; d = 0.20, 95% CI: 0.04, 0.37) and enjoyment of food (β = 0.10, P = 0.08; d = 0.20, 95% CI: 0.03, 0.36). Furthermore, satiety responsiveness/slowness in eating scores significantly related to lower BMIz (β = −0.20, P < 0.001). PA marginally related to BMIz (β = 0.07, P = 0.07).

Fig. 1 .

Summary of model fitting associations among geographic location, obesity-related behaviors and children’s BMI z-scores, controlling for parent race/ethnicity, parent education, single parent and parent BMI and food insecurity. Note. Standardized coefficients are provided. Significant (^*P < 0.05) paths and coefficients are depicted in bold. To simplify the presentation, the covariate variable paths are not depicted.

Discussion

Main findings of this study

This study found that rural, compared with urban, low-income preschoolers in the Midwest USA had higher BMIz and demonstrated more obesity-promoting eating behaviors, particularly emotional overeating. They also participated in more PA than urban preschoolers. No differences in dietary quality or screen time were found.

What is already known on this topic

Several studies in the USA have shown a higher BMI among rural, compared with urban, school-aged children.^32,33 However, similar studies on preschoolers have not been as abundant or conclusive. A meta-analysis of published studies examining associations between living in rural versus urban locations with children’s weight status found only one study focused on preschoolers.² The findings from that study showed no rural–urban differences in the weight status of preschoolers. Other published research found that rural residency was a risk factor for a high BMI that started as early as age 1–6 years and continued on a high trajectory into the adolescent years.^34,35

Previous studies examining associations between PA and BMI in preschoolers have been inconclusive.³⁶ Some studies have found an inverse association between PA and BMI in preschoolers,^37,38 and other studies have found a positive association.^39–42 Recommendations have been that further research is needed.

While associations between weight status and obesity-promoting eating behaviors among young children have been well-studied,^26,43–46 there is a paucity of literature examining differences in eating behaviors and BMI between children living in rural versus urban environments. Studies examining eating behaviors in children, regardless of geographic location, have found a positive relationship between food approach behaviors and weight gain.^26,43–48

What this study adds

The findings of this study add to the literature, which suggests that rural–urban BMI and obesogenic behavior differences are found among preschoolers, an age group that has been little studied in the examination of rural versus urban contexts to date. The findings differ from prior studies of low-income preschoolers,^2,4 which found no difference in the BMI of rural and urban preschoolers. In addition to having higher BMIz, rural preschoolers in this study reported more PA than the urban preschoolers. Similar to some other studies, this research showed a trend toward a positive relationship between PA and BMIz in preschoolers.^39,40 These results may be explained by rural–urban differences in frequency or intensity in PA engagement in a community or environmental feature unmeasured by this study (e.g. preschool) or parental misreporting due to possible recall bias. Another unique rural–urban difference in the current study was found when examining preschoolers’ eating behaviors. Rural, compared with urban, preschoolers had more emotional overeating behaviors. It is possible that emotional overeating is adopted as a way to deal with stressors within a rural environment or parental feeding practices may promote the child’s emotional overeating.^17,49,50 Rural children as young as 3–4 years old may be learning to cope with stress through emotional overeating,⁵¹ which aligns with other findings indicating that school-aged children who perceive stress may use eating as a coping mechanism.⁵² Also, such eating patterns may be related to adults’ eating habits. For example, low-income women in rural households report emotional overeating in association with their level of perceived social isolation.⁵³

Differences in nutrition education and prenatal care^4–6 may also explain the higher BMIz among rural, compared with urban, preschoolers. Nutrition education services usually address dietary quality and children’s eating behaviors. Yet, rural families are often underserved with limited access to nutrition education services.^4–6 Similarly, rural families may have less access to family support services that ameliorate family stressors,⁵⁴ which may contribute to children’s eating behaviors. Nutrition education and family supports that target rural parents of very young children may help reduce overweight and obesity early in life.

Limitations

A limitation of this study is the use of parental recalls for some of the measures, including PA (outdoor active play). Additionally, measures of environmental or community features that may contribute to eating and PA behaviors were not available. Our study results might not be generalizable to populations who are not low-income, not English-speaking, not from the specific geographical areas involved in this study, or those who may not volunteer for research trials.

Conclusions

In summary, rural–urban differences in BMIz may start as early as 3–4 years of age, if not earlier. The rural environment has characteristics that may increase the risk of overweight and obesity among children, such as high levels of poverty and limited access to nutrition education, preventative care and grocery stores.^4–6 Obesity-related eating behaviors resulting from these limitations common to rural environments may be adopted by children at very young ages.⁵¹ To reverse the weight-related health disparities between rural and urban preschoolers, structural and cultural changes, as well as parental education and family supports around coping skills and mindful eating, may be needed. Due to the limited geographical scope of this research, additional studies should be conducted in other locales to confirm the results. Further research is also needed to directly examine mechanisms in rural environments that increase the risk of obesity and obesity-promoting behaviors among low-income children aged 3–4 years.

D. A. Contreras, Senior Extension Specialist

T. L. Martoccio, Faculty Research Associate

H. E. Brophy-Herb, Professor

M. Horodynski, Professor Emerita

K. E. Peterson, Professor and Chair

A. L. Miller, Associate Professor

N. Neda Senehi, Post-doc Fellow

J. Sturza, Statistician

N. Kaciroti, Research Scientist

J. C. Lumeng, Professor

Conflict of interest

The authors declare that they have no competing interests.

Funding

This work was supported by the US Department of Agriculture, National Institute of Food and Agriculture, Agriculture and Food Research Initiative (grant number 2011–68001-30089) and the Michigan Nutrition and Obesity Research Center (grant number P30DK089503–01). N.S. was supported, in part, by the National Institute of Mental Health (grant number MH015442) and a University of Colorado Anschutz Medical Campus, Developmental Psychobiology Endowment Fund.

Authors’ contributions

DC, TM, HBH, MH, JCL ALM, KEP, NK and NS conceptualized the study and contributed to the development of the study design and drafted the manuscript. NK, TM and JS conducted the statistical analyses. All authors read and approved the final manuscript.