Urban-Rural Differences in Childhood and Adolescent Obesity in the United States: A Systematic Review and Meta-Analysis

Abstract

Background:

A systematic literature review and subsequent meta-analysis were performed to investigate differences in childhood obesity between urban and rural areas in the United States.

Methods:

A search of published studies comparing childhood obesity in urban and rural settings was undertaken by probing PubMed and Cumulative Index to Nursing and Allied Health Literature (CINAHL) for articles that met predetermined inclusion criteria. A subsequent meta-analysis was conducted to determine the combined effect size and significance of differences in childhood obesity between urban and rural areas.

Results:

Ten studies were identified for systematic review, five of which contributed to the meta-analysis. All but one study suggested that residence in rural areas was associated with higher prevalence or increased odds of childhood obesity, compared to children living in urban areas. A meta-analysis of 74,168 pooled participants ages 2–19 found that rural children have 26% greater odds of obesity, compared to urban children (odds ratio=1.26; 95% confidence interval, 1.21–1.32).

Conclusions:

Obesity rates are higher among rural children than urban children in the United States. To ensure successful targeted interventions and effective resource allocation, practitioners and policy makers alike should be cognizant of this disparity in childhood obesity.

Introduction

Childhood obesity is one of the leading public health concerns in the United States.^1–4 Obesity rates in children have more than tripled over the past three decades.^1,3 As a result, the United States has some of the highest childhood obesity rates in the world.⁵ Approximately 17% of children in the United States are obese.³ Recent reviews suggest that the percentage of obese adolescents that go on to become overweight or obese adults is anywhere from 24% to 90%.^6,7 Obesity in adulthood is associated with increased risk of developing a number of diseases, including cardiovascular disease, hypertension, type 2 diabetes, obstructive sleep apnea, asthma, and certain cancers.⁸ In addition to contributing to poorer health outcomes, obesity has vast economic implications as well. According to one estimate, obesity-related healthcare expenses cost the US healthcare system an estimated $190 billion dollars annually.⁹

Childhood obesity is caused by a complex interaction of factors¹⁰ and has been investigated by scholars in various fields. Recent epidemiological studies have suggested several behavioral and socioeconomic characteristics as important determinants of childhood obesity.^11–13 For instance, dietary behaviors, such as lack of breastfeeding, watching television while having a meal, and high sweetened beverage intake have been found to increase the likelihood of childhood obesity.¹³ Further, lower socioeconomic status (SES), rural residence, and racial/ethnic minority status are associated with increased odds of childhood obesity.^11,12

Although several studies^14–21 indicate high rates of childhood obesity in rural areas, they lack direct comparisons between urban and rural children. Among the few studies that do compare urban and rural areas, most report childhood obesity in terms of adjusted odds ratio (OR) and have considerable variation in control variables. Despite evidence suggesting that rural children have greater odds of being obese than urban children, directly comparing results across studies should be approached with caution. Thus, the aim of this study is to provide a systematic review and meta-analysis of studies that directly compare urban and rural obesity in US children and determine the nature and magnitude of differences.

Methods

Definition of Childhood and Adolescent Obesity

A commonly used index for measuring obesity is BMI, which is weight/height (kg/m²). The CDC recommends the use of BMI to measure obesity for adults, adolescents, and children.²² One caveat of using BMI is that it does not distinguish between fat and muscle tissue. Unlike adult obesity, which is defined by the numerical BMI value, a statistical definition of obesity is used for children and adolescents in the United States. Children and adolescent BMI varies with age and sex and is compared to a reference population. The reference population is often the 2000 CDC growth charts,²² and age-sex specific BMI is reported as a percentile. The CDC defines obesity for children and adolescents as a BMI ≥95th percentile for their age and sex.²²

Definitions of Urban and Rural

Defining urban and rural areas has remained problematic for researchers and government agencies alike.²³ Metropolitan/urban areas can be defined using several criteria. Once a definition is established, nonmetropolitan/rural areas are then typically defined by exclusion (any areas not metropolitan/urban are classified as nonmetropolitan/rural).^23,24 How these criteria are determined impacts resulting delineations of urban and rural and variations in classification inevitably occur. No universally recognized classification scheme or definition for urban/rural areas exists. Commonly used definitions come from the Bureau of the Census (US Census), Office of Management and Budget (OMB), USDA, and the CDC.^23,25 The OMB and US Census generally designate areas in the United States as metropolitan/nonmetropolitan and urban/rural, respectively. Inclusion into one of these delineations is based on population density, social and economic integration with core metropolitan areas, or a combination thereof.^26–28 Dichotomous definitions of urban/rural or metropolitan/nonmetropolitan often fail to account for variations that exit within an urban or rural area, for instance, suburban areas. Further, broad generalizations about urban or rural conditions will not necessarily be representative for all subsets of individuals within those areas.²³ The USDA and CDC attempt to overcome these limitations by further dividing metropolitan/nonmetropolitan or urban/rural categories into multitier classification schemes. Three commonly used USDA classification schemes have been derived from OMB and US Census definitions. These include Urban Influence Codes (UIC), Rural-Urban Continuum Codes (RUCC), and Rural-Urban Commuting Area (RUCA).²⁵ The CDC developed the National Center for Health Statistics (NCHS) Urban-Rural Classification Scheme for Counties, which divides OMB metropolitan/nonmetropolitan areas into six urbanization levels.²⁹ Further details about specific definitions and classification schemes are presented in Table 1. Additionally, the particular classification schemes each study used are presented in Table 2.

Table 1.

Urban-Rural Classification Schemes

Classification scheme	Description
US Census Bureau Classification^a	US Census Bureau delineates areas at the census-track level into three primary categories: Urban Area (UA), Urban Cluster (UC), and Rural. UAs are continuously built-up areas with a population of 50,000 or more. UCs are any urban places outside of UAs. The US Census Bureau uses the term urban to refer to all areas located in UAs and UCs. Any areas not classified as urban are considered to be rural.
Office of Management and Budget (OMB) Metropolitan Area Standards^b	OMB delineates areas at the county level into three primary categories: Metropolitan Statistical Areas (MSAs), Micropolitan Statistical Areas (McSAs), and non-Core Based Statistical Areas (non-CBSAs). The OMB generally refers to all counties located within an MSA as metropolitan, and all other counties (McSA and non-CBSA) are designated nonmetropolitan. Inclusion into one of these delineations is based on population density as reported by the Census Bureau, in addition to social and economic integration with the core MSA county.
Urban Influence Codes (UIC)^c	UIC is a 12-teir classification scheme that subdivides OMB metropolitan and nonmetropolitan categories into two metro and 10 nonmetro categories. Metro counties are designated by the population size of their metropolitan area and nonmetro counties by size of the largest city or town within the county and proximity to MSAs or McSAs.
Rural-Urban Continuum Codes (RUCC)^d	RUCC is a nine-teir classification scheme that subdivides OMB metropolitan and nonmetropolitan into three metropolitan and six nonmetropolitan categories. Metropolitan counties are designated by the population size of their metro area and nonmetropolitan counties by the degree of urbanization and adjacency to a metro area.
Rural-Urban Commuting Areas (RUCA)^e	RUCA is a 10-teir classification scheme census-track classification scheme that designates areas by combining Census Bureau UA, UC, and Rural census-track designations with employment commuting information.
NCHS Urban-Rural Classification Scheme for Counties^f	NCHS Urban-Rural Classification Scheme for Counties is a six-teir classification scheme based on OMB standards for defining MSAs and groups all counties into six urbanization levels on a continuum ranging from most urban to most rural.

2000 United States Bureau of the Census Geographic Areas Reference Manual.

2003 Office of Management and Budget Standards for Defining Metropolitan and Micropolitan Statistical Areas.

2013 United States Department of Agriculture Economic Research Service Urban Influence Code.

2003 United States Department of Agriculture Economic Research Service Rural-Urban Continuum Codes Documentation.

2010 United States Department of Agriculture Economic Research Service Rural-Urban Commuting Areas.

2006 National Center for Health Statistics Urban-Rural Classification Scheme for Counties.

NCHS, National Center for Health Statistics.

Table 2.

Characteristics of Included Studies

Author, publication date, citation no.	Inclusion for meta-analysis	Study design	Data sources	Urban-rural classification scheme and Urban-rural categories	Participants	BMI percentile and anthropometry
Davis, 2008³⁷	No: Not enough information provided to calculate odds ratio	Cross-sectional	Survey of fifth graders from four elementary schools in Kansas	US Census urban; rural	n=138 Age: 9–11 years Location: Kansas	≥95 measured
Joens-Matre 2008³⁶	No: Not enough information provided to calculate odds ratio	Cross-sectional	State-wide survey conducted in Iowa	RUCC urban; small cities; rural	n=3416 Age: 8–12 years Location: Iowa	≥95 measured
Liu 2008¹¹	No: This study used the same population and survey as Singh (2008)³³ and is not deemed statistically independent.	Cross-sectional	NSCH^b	UIC urban; rural	n=45,082 Age: 10–17 years Location: US	≥95 parent-reported
Lutfiyya 2007³⁵	No: Not enough information provided to calculate odds ratio	Cross-sectional	NSCH^b	NCHS Urban-Rural Classification scheme for counties metropolitan; rural	n=46,396 Age: 5–17 years Location: US	≥95 parent-reported
Salois 2012³⁸	No: Not enough information provided to calculate odds ratio	Cross-sectional	PedNSS^c	RUCC Metropolitan; nonmetropolitan	n=2192 counties The unit of analysis was country level. Location: US	≥95 Measured

Studies included in meta-analysis
Davis 2011¹²	Yes	Cross-sectional	NHANES^a 2003–2006	UIC urban; rural	n=7882 Age: 2–18 years Location: US	≥95 measured
Liu 2012³⁴	Yes	Cross-sectional	NHANES^a 1999–2006	RUCA urban; rural	n=14,332 Age: 2–19 years Location: US	≥95 measured
Lewis 2006³¹	Yes	Cross-sectional	State-wide survey conducted in Georgia	Georgia Census urban; suburban; rural growth; rural decline	n=3114 Age: 9–19 years Location: Georgia	≥95 measured
McMurray 1999³²	Yes	Cross-sectional	State-wide intervention study conducted in North Carolina. This study used baseline data.	US Census urban; rural	n=2133 Age: 8–10 years Location: North Carolina	≥90 measured
Singh 2008³³	Yes	Cross-sectional	NSCH^b	NCHS urban-rural classification scheme for counties metropolitan; nonmetropolitan	n=46,707 Age: 10–17 years Location: US	≥95 parent-reported

National Health and Nutrition Examination Survey.

National Survey of Children's Health.

Pediatric Nutrition Surveillance System.

RUCC, Rural-Urban Continuum Codes; UIC, Urban Influence Codes; NCHS, National Center for Health Statistics RUCA, Rural-Urban Commuting Areas.

Literature Search Strategy and Study Inclusion Criteria

The databases PubMed (1966–December 2013) and Cumulative Index to Nursing and Allied Health Literature (CINAHL; 1977–December 2013) were searched for published literature that met predetermined inclusion criteria. The keywords or MeSH headings adolescent, childhood, obesity, overweight, body mass index (BMI), rural, urban, place of residence, and built environment were initially used to identify relevant studies. Additional studies were identified through the related articles function in PubMed and searching the reference sections of relevant studies. The databases were searched for original peer-reviewed studies, regardless of design, that reported a comparison of childhood obesity between urban and rural areas in the United States. Studies were included if they were published in a refereed scientific journal; original quantitative research; published before December 31, 2013 in English language; conducted in the United States; defined obesity as either a BMI ≥90th percentile or BMI ≥95th percentile for the individuals' age and sex; the study sample included children or adolescents from 2 to 19 years of age; and comparable information was provided about both rural and urban childhood obesity. Although the currently accepted definition of childhood obesity is an age- and sex-specific BMI ≥95th percentile,³⁰ studies published before 2000 occasionally defined childhood obesity as age- and sex-specific BMI ≥90th percentile. The definition of obesity was expanded in the inclusion criteria to be as exhaustive as possible. Studies were not considered for review if they provided information about only rural or urban childhood obesity without including comparable information about the other.

A total of 346 potentially eligible studies were identified. We subsequently excluded 336 studies for the following reasons: 187 were conducted outside of the United States; 121 did not provide comparable information about both rural and urban childhood obesity; 11 were originally published in a language other than English; 10 were reviews; two were duplicates; two were commentaries; one was a field report; one did not define obesity as either BMI ≥90th percentile or BMI ≥95th percentile for the individuals' age and sex; and one had problems with the accuracy of reported findings. The review therefore included 10 studies, five of which contributed to the meta-analysis. Figure 1 provides a flow diagram of the inclusion/exclusion process.

Figure 1.

Flow diagram of included and excluded studies.

Meta-Analysis

The meta-analysis was conducted using the statistical software Comprehensive Meta-Analysis (CMA) Version 2. Of the 10 studies included for review, five^12,31–34 contributed to the meta-analysis. Studies were excluded for the following reasons: four studies^35–38 did not provide enough information to calculate an odds ratio, and one study¹¹ could not be determined to be statistically independent from another included study. Further details about excluded studies are presented in Table 2. Included studies varied in adjusting for control variables. Some reported the control variables for adjusted ORs, whereas others did not. Therefore, we calculated unadjusted ORs using reported sample size, exposure (living in rural areas), and disease (childhood obesity). Four^12,32–34 of the five studies included for meta-analysis defined urban/rural or metropolitan/nonmetropolitan based on one of the established classification schemes illustrated in Table 1. One study³¹ created a novel four-tier classification scheme derived from Georgia census data, which included the following categories: urban, suburban, rural growth, and rural decline. This classification scheme was conceptually grounded in US Census definitions of urban and rural. Therefore, this study was similar enough for comparison and was standardized by recoding the four-tier classification scheme into dichotomous urban and rural categories. Urban and suburban were recoded as urban, whereas rural growth and rural decline were recoded as rural. The conceptual foundations of the classification schemes used in these five studies were similar enough to be included for meta-analysis. In order to test the assumption of homogeneity, we used the I² index and Cochran's Q statistic to determine whether a fixed-effects or random-effects model was warranted.

Main Data Sources

Half of the studies^{11,12,33–35} included in the review used data from either the National Survey of Children's Health (NSCH) or the National Health and Nutrition Examination Survey (NHANES). Both NSCH and NHANES are nationally representative surveys conducted by the NCHS at the CDC. The particular data sources each study used are presented in Table 2.

NSCH is a telephone survey conducted from 2003 to 2004 that surveys aspects of children's lives, including physical and mental health, access to quality healthcare, physical and social environment, and family dynamics. A random-digit-dial sample of approximately 2000 households with children under 18 years of age was selected from each of the 50 states and the District of Columbia. One child was randomly selected from all children in each identified household to be the subject of the survey. The respondent was the parent or guardian who knew the most about the child's health and healthcare. BMI was calculated using height and weight as reported by the respondent (parent or guardian).³⁹

NHANES is a continuous survey that began in 1999, designed to assess the health and nutritional status of adults and children in the United States. The survey uses in-person interviews and physical examinations for data collection. Interviews are conducted in the respondent's home, and physical examinations are conducted in mobile clinics. NHANES surveys approximately 5000 individuals of all ages each year. Using a multistage probability sampling design, NHANES oversamples certain subgroups of particular interest, one of which is adolescents 12–19 years of age. BMI is calculated using anthropometric measures collected during physical examinations.⁴⁰

Neither NSCH nor NHANES directly reports information regarding participants' urban or rural status. However, they do report the county in which participants reside. All studies that used NSCH or NHANES merged these data with one of the urban-rural classification schemes found in Table 1.

Results

Systematic Review

A total of 10^{11,12,31–38} studies met inclusion criteria for systematic review. Table 2 presents study characteristics and Table 3 summarizes the major finds for each study. Several studies refer to having an age- and sex-specific BMI ≥95th percentile in terms of overweight as opposed to obese. This apparent discrepancy in definition likely reflects changes in terminology referring to children with an age- and sex-specific BMI ≥95th percentile. In 2005, the Institute of Medicine recommended children 2–18 years of age with an age- and sex-specific BMI ≥95th percentile be defined as obese,⁴¹ whereas, before this recommendation, the terminology overweight was commonly used.³⁰ For the purpose of this study, the terms obese or obesity are used in reference to an age- and sex-specific BMI ≥95th percentile, regardless of the terminology used in the original study. Nine studies^{11,12,32–38} used one of the established classification schemes presented in Table 1 to define urban and rural areas. One study³¹ conducted a state-wide survey in Georgia and formulated a novel four-tier classification scheme based on data from the Georgia census. All studies defined childhood obesity as age- and sex-specific BMI ≥95th percentile with one exception. This study³² defined childhood obesity as age- and sex-specific BMI ≥90th percentile, as opposed to the more standard definition. It is also the only study to be published before 2000, which may explain the inconsistency in defining obesity. Anthropometry was directly measured in seven studies,^{12,31,32,34,36–38} whereas three studies^11,33,35 calculated BMI based on parent-reported measurements of height and weight. Across studies, the age of participants ranged from 2 to 19 years of age and participant ages varied among studies. The majority of studies^{11,12,33–35,38} used nationally representative data, whereas four studies^31,32,36,37 were state specific. All 10 studies had cross-sectional study designs.

Table 3.

Summary of Findings

Author, publication date, citation no.	Principal findings^a	AOR [CI]
Davis 2008³⁷	In Kansas, the prevalence of obesity was higher among children attending school in rural areas (25.7%) than urban areas (20.0%).	AOR not reported
Davis 2011¹²	The prevalence of obesity was higher among rural children (21.8%) than urban children (16.9%).	AOR not reported because the study was stratified by rural/urban
Joens-Matre 2008³⁶	In Iowa, the prevalence of obesity was higher among rural children (25%) than children from urban areas (19%) and small cites (17%). Children living in rural areas had 47% greater odds of being obese, compared to children living in small cites (OR=1.47; 95% CI: not reported).	Rural is compared to small cities: AOR=1.47 [CI not reported]
Lewis 2006³¹	In Georgia, the prevalence of obesity was higher among rural children (rural growth=23.7% and rural decline=23.0%) than suburban (20.7%) and urban children (17.4%). Children living in rural growth and rural decline areas had 82% greater odds of being obese, compared to children in urban areas (OR=1.82; 95% CI: 1.42, 2.28). Children living in rural decline areas had 43% greater odds of being obese, compared to children in urban areas (OR=1.43; 95% CI: 1.11, 1.85).	Rural growth and rural decline are compared to urban: Rural growth: AOR=1.82 [1.45, 2.28] Rural decline: AOR=1.43 [1.11, 1.85]
Liu 2008¹¹	The prevalence of obesity was higher among rural children (16.5%) than urban children (14.3%). Children living in rural areas had 13% greater odds of being obese, compared to children living in urban areas (OR=1.13; 95% CI: 1.01, 1.25).	Rural is compared to urban: AOR=1.13 [1.01, 1.25]
Liu 2012³⁴	The prevalence of obesity was higher among rural children (18.6%) than urban children (15.1%). Rural children ages 12–19 years had 32% greater odds of being obese, compared to children living in urban areas (OR=1.32; 95% CI: 1.05, 1.67).	Rural is compared to urban: Ages 12–19: AOR=1.32 [1.05, 1.67]
Lutfiyya 2007³⁵	Children living in rural areas had 25.2% greater odds of being be overweight (≥85th percentile BMI <95th percentile) or obese (≥95th percentile BMI), compared to children in urban areas (OR=1.252; 95% CI: 1.248, 1.256).	Rural is compared to metropolitan: AOR=1.252 [1.248, 1.256]
McMurray 1999³²	In North Carolina, the prevalence of obesity (BMI ≥90th percentile) was higher among rural children (29.5%) than urban children (21.7%). Children living in rural areas had 54.7% greater odds of being obese, compared to children in urban areas (OR=1.547; 95% CI: 1.26, 1.90).	Rural is compared to urban: AOR=1.55 [1.26, 1.90]
Singh 2008³³	The prevalence of obesity was higher among rural children (16.9%) than urban children (14.2%). Children living in rural areas had 32% greater odds of being obese, compared to children living in urban areas (OR=1.32; 95% CI: 1.19, 1.46).	Rural is compared to urban: AOR=1.32 [1.19, 1.46]
Salois 2012³⁸	There is no difference in the prevalence of obesity among rural low-income preschool children (14.15%) and urban low-income preschool children (14.16%).	AOR not reported because the study was stratified by rural/urban

Reported findings may not be reported in the language used in the original study.

AOR, adjusted odds ratio; CI, confidence interval.

All but one study³⁸ found higher prevalence or increased odds of obesity among children living in rural areas, compared urban areas. Eight studies^{11,12,31–34,36,37} identified the prevalence of obesity as higher among rural children than urban children. Seven studies^11,31–36 found that children living in rural areas had greater odds of being obese, compared to children living in urban areas. Adjusted ORs varied from 1.13 to 1.82 across studies and are summarized in Table 3. Only one study³⁸ found no difference between urban and rural childhood obesity. This study identified no difference in the prevalence of obesity among low-income, preschool children (2–4 years of age) between metropolitan (14.15%) and nonmetropolitan (14.16%) counties. This is likely because study participants were from 2 to 4 years of age, an age range that is not comparable to other studies.

Seven studies^{11,12,32–36} included physical activity (PA) in their analyses and examined whether there were differences in PA between rural and urban children. Five studies^{11,12,33,34,36} found increased PA for children living in rural areas, compared to children living in urban areas, one study³⁵ reported rural obese children to be less active than urban obese children, and one study³² found no difference in PA between the two groups.

Variability in effect sizes and control variables were considerable across studies. To a lesser degree, studies also differed in the way rural and urban residence was conceptualized. Therefore, a meta-analysis was conducted to further elucidate the magnitude and statistical significance of the combined effects of rural residency on childhood obesity.

Meta-Analysis

A total of five studies^12,31–34 were included for meta-analysis with a pooled sample of 74,168 participants. Four studies^31–34 used urban-rural residence as an independent variable, whereas one study¹² stratified their analytical models by urban-rural status. Two studies used NHANES data; one¹² used NHANES data from 2003 through 2006 and the other³⁴ used NHANES data from 1999 through 2006. Because there is some overlap in the years used by these two studies, a sensitivity analysis was conducted by including and excluding the two studies in various fixed-effects models. Differences among fixed-effects models were negligible and heterogeneity remained low; thus, both studies contributed to the meta-analysis.

The meta-analysis generated a pooled odds ratio of 1.26 (OR=1.26; 95% confidence interval [CI], 1.21–1.32). Children living in rural areas have 1.26 times greater odds of being obese, compared to children living in urban areas. This fixed-effect estimate was characterized by low heterogeneity (I²=20.43; Q-value=5.03; df (Q)=4; Tau²=0.001). The relative weights for each study in the fixed-effects model ranged from 4.55% to 66.87%. The study with the narrowest CI had the highest weight in determining the pooled effect. Meta-analysis characteristics and a forest plot of results are presented in Figure 2.

Figure 2.

Pooled odds ratio for the effect of rural residency on childhood obesity. CI, confidence interval.

Discussion

The aim of this study was to investigate differences in childhood obesity between urban and rural areas through systematic review and determine the nature and magnitude of differences through meta-analysis. Across studies, we found a higher prevalence of childhood obesity in rural areas, compared to urban areas. Additionally, children living in rural areas have greater odds of being obese than urban children. Results from the meta-analysis support these findings. Pooled ORs across studies indicate that children living in rural areas have 26% greater odds of being obese, compared to children living in urban areas. In fact, only one study³⁸ found no difference in obesity between urban and rural children. This particular study sampled participants from 2 to 4 years of age who were from low-SES households, whereas all other studies included children from older age groups. Although the results of this study are not directly comparable to other studies because of differences among participant ages, it does suggest that urban and rural disparities in childhood obesity may manifest as children grow older. Results from other studies^12,34 support this idea and found that children between ages 10 and 18 had a higher prevalence of obesity, when compared to their younger counterparts.

The pathways that lead to differences in obesity between urban and rural children are not well understood. However, more than half of the reviewed studies^{11,12,31,33–37} investigated the effect of PA on childhood obesity. Several studies reported that rural children engage in more PA than their urban counterparts.^{11,33,34,36,37} In fact, only one study³⁵ found rural obese children to be less physically active, compared to urban obese children. This is likely a result of how the analysis was conducted. This study differed from the others by comparing PA between rural and urban obese children, as opposed to comparing PA between all rural and urban children. Rural children having higher obesity rates despite being more physically activity suggests existing confounding effects in the environment that have not been explored. However, studies differed substantially in conceptualizing and measuring physical activities, which makes drawing meaningful conclusions about the relationship between PA and urban/rural differences in childhood obesity onerous. This limited our ability to accurately measure mediating pathways for differences in childhood obesity between urban and rural areas.

Whereas the majority of studies focused on individual factors, such as PA, one study³⁸ examined associations among environmental characteristics and obesity. Urban and rural community characteristics where compared, and aspects of the food environment appeared to contribute to differences in adult obesity between rural and urban areas. However, this is the same study that found no difference in obesity between rural and urban children. This finding suggests that longer exposure to certain physical and social environments may contribute to differences in urban and rural childhood obesity. It is likely that certain environmental aspects of rural areas promote obesity in children and warrants further study. Future research combining individual- and area-level data could further elucidate potential environmental factors that act as pathways leading to such differences between urban and rural areas.

Although our study establishes the existence and magnitude of disparity in childhood obesity between urban and rural areas, no study included in the review definitively explained why this disparity exists. We suspect that the common use of dichotomous urban/rural delineations may contribute to the elusiveness of such explanation. For example, suburban children would be misclassified as either urban or rural. Such misclassification likely distorts findings and weakens the internal validity of studies investigating differences in urban and rural childhood obesity. Thus, future research is needed that includes suburban children and the behavioral, socioeconomic, and demographic characteristics affecting them as a third comparison group. The development of an urban/suburban/rural classification scheme that more accurately differentiates these areas would provide for more robust study of geographical differences in health outcomes.

The systematic review was exhaustive and likely captured all relevant studies. Because many studies involving childhood obesity in rural areas provide no comparison group, contextually interpreting such results remains limited. This review aggregates studies that do include direct comparison of childhood obesity between rural and urban areas. Further, the meta-analysis determined the magnitude of identified differences across studies. The studies included for meta-analysis were methodologically similar and were characterized by low heterogeneity. This allowed for the use of a fixed-effects model providing for a more accurate estimate. However, our study has several limitations. All studies included for review and meta-analysis were cross-sectional surveys. Cross-sectional studies are not able to capture temporal changes and this could potentially mask causality. Three studies^11,33,35 calculated BMI based on parent-reported measurement of height and weight. One of these studies³³ was also included in the meta-analysis. Using parent-reported measurements to calculate age- and sex-specific BMI tends to overestimate obesity, compared to using directly-measured anthropometrics, specifically among younger children.⁴² Parent-reported height and weight is more or less problematic, depending upon the age range of the children, and is of particular concern for children below 10 years of age. For older children, parent-reported and directly measured height and weight are less divergent.⁴² Because the study³³ contributing to the meta-analysis included 10- to 17-year-olds, the bias introduced by using parent-reported measurements should be minimal. Nonetheless, BMI based on parent-reported height and weight potentially introduces bias and directly measured anthropometric BMI is a more valid measure.⁴³ Owing to the limited number of studies, we were not able to subgroup studies based on important covariates, such as sex, race, and age. This also limited our ability to perform funnel plot or regression-based assessments to identify publication bias. Thus, the possibility of existing publication bias cannot be ruled out. Last, included studies used different urban/rural classifications schemes, contributing to variation in urban and rural delineations. In order to enhance the overall validity of future research, longitudinal study designs that include directly measured anthropometric assessments of obesity and more discriminating urban/suburban/rural classification schemes would be a substantial contribution to understanding childhood and adolescence obesity.

Conclusions

This systematic review and meta-analysis provides further evidence that disparities in childhood obesity exist between urban and rural areas. Although research investigating differences in childhood obesity between urban and rural areas remains limited, most existing literature suggests that children living in rural areas have a higher prevalence and greater odds of obesity. To ensure successful targeted interventions and effective resource allocation, practitioners and policy makers alike would benefit from being cognizant of this disparity in childhood obesity.

Footnotes

Author Disclosure Statement

No competing financial interests exist.

References

Miech

, Kumanyika

, Stettler

, et al. Trends in the association of poverty with overweight among US adolescents, 1971–2004. JAMA, 2006; 295:2385–2393.

National Center for Health Statistics. Health, United States 2006: With chartbook on trends in the health of Americans. 2007 Nov; DHHS Pub No. 2006-1232:1–543.

Ogden

, Carroll

, Kit

, et al. Prevalence of obesity and trends in body mass index in US children and adolescents, 1999–2010. JAMA, 2012; 307:483–490.

Wang

, Gortmaker

, Taveras

. Trends and racial/ethnic disparities in severe obesity among US children and adolescents, 1976–2006. Int J Pediatr Obes, 2011; 6:12–20.

Wang

. Cross-national comparison of childhood obesity: The epidemic and the relationship between obesity and socioeconomic status. Int J Epidemiol, 2001; 30:1129–1136.

Wang

, Beydoun

. The obesity epidemic in the United States—Gender, age, socioeconomic, racial/ethnic, and geographic characteristics: A systematic review and meta-regression analysis. Epidemiol Rev, 2007; 29:6–28.

Singh

, Mulder

, Twisk

, et al. Tracking of childhood overweight into adulthood: A systematic review of the literature. Obes Rev, 2008; 9:474–488.

Dietz

. Health consequences of obesity in youth: Childhood predictors of adult disease. Pediatrics, 1998; 101:518–525.

Cawley

, Meyerhoefer

. The medical care costs of obesity: An instrumental variables approach. J Health Econ, 2012; 31:219–230.

10.

Han

, Lawlor

, Kimm

. Childhood obesity. Lancet, 2010; 375:1737–1748.

11.

Liu

, Bennett

, Harun

, et al. Urban‐rural differences in overweight status and physical inactivity among US children aged 10–17 years. J Rural Health, 2008; 24:407–415.

12.

Davis

, Bennett

, Befort

, et al. Obesity and related health behaviors among urban and rural children in the United States: Data from the national health and nutrition examination survey 2003–2004 and 2005–2006. J Pediatr Psychol, 2011; 36:669–676.

13.

Moreno

, Rodraguez

. Dietary risk factors for development of childhood obesity. Curr Opin Clin Nutr Metab Care, 2007; 10:336–341.

14.

Crooks

. Food consumption, activity, and overweight among elementary school children in an Appalachian Kentucky community. Am J Phys Anthropol, 2000; 112:159–170.

15.

Davy

, Harrell

, Stewart

, et al. Body weight status, dietary habits, and physical activity levels of middle school-aged children in rural Mississippi. South Med J, 2004; 97:571–577.

16.

Felton

, Pate

, Parsons

, et al. Health risk behaviors of rural sixth graders. Res Nurs Health, 1998; 21:475–485.

17.

Gustafson-Larson

, Terry

. Weight-related behaviors and concerns of fourth-grade children. J Am Diet Assoc, 1992; 92:818–822.

18.

Gauthier

, Hickner

, Noel

. High prevalence of overweight children in Michigan primary care practices: An UPRNet study. J Fam Pract, 2000; 49:73–76.

19.

Neal

, Demerath

, Gonzales

, et al. Coronary artery risk detection in Appalachian communities (CARDIAC): Preliminary findings. W V Med J, 2000; 97:102–105.

20.

Davis

, Lambert

. Body image and weight concerns among southwestern American Indian preadolescent school children. Ethn Dis, 1999; 10:184–194.

21.

Lacar

, Soto

, Riley

. Adolescent obesity in a low-income Mexican American district in south Texas. Arch Pediatr Adolesc Med, 2000; 154:837–840.

22.

Kuczmarski

, Ogden

, Guo

, et al. 2000 CDC growth charts for the United States: Methods and development. Vital Health Stat, 11 2002; 246:1–190.

23.

Hall

, Kaufman

, Ricketts

. Defining urban and rural areas in US epidemiologic studies. J Urban Health, 2006; 83:162–175.

24.

United States General Accounting Office. Fact sheet for congressional requesters, rural development, profile of rural areas. 1993;B-261242.

25.

Hart

, Larson

, Lishner

. Rural definitions for health policy and research. Am J Public Health, 2005; 95:1149–1155.

26.

Office of Management and Budget. Final report and recommendations from the metropolitan area standards review committee to the office of management and budget concerning changes to the standards for defining metropolitan areas. Federal Register, 2000; 65:51059–51077.

27.

Office of Management and Budget. Standards for delineating metropolitan and micropolitan statistical areas. Federal Register, 2010; 75:37246–37252.

28.

Bureau of the Census. Geographic Areas Reference Manual. US Department of Commerce: Washington, DC, 1994.

29.

Ingram

, Franco

. NCHS urban-rural classification scheme for counties. Vital Health Stat, 2 2012; 154:1–65.

30.

Krebs

, Himes

, Jacobson

, et al. Assessment of child and adolescent overweight and obesity. Pediatrics, 2007; 120(Suppl 4):S193–S228.

31.

Lewis

, Meyer

, Lehman

, et al. Prevalence and degree of childhood and adolescent overweight in rural, urban, and suburban Georgia. J Sch Health, 2006; 76:126–132.

32.

McMurray

, Harrell

, Bangdiwala

, et al. Cardiovascular disease risk factors and obesity of rural and urban elementary school children. J Rural Health, 1999; 15:365–374.

33.

Singh

, Kogan

, Van Dyck

, et al. Racial/ethnic, socioeconomic, and behavioral determinants of childhood and adolescent obesity in the United States: Analyzing independent and joint associations. Ann Epidemiol, 2008; 18:682–695.

34.

Liu

, Jones

, Sun

, et al. Diet, physical activity, and sedentary behaviors as risk factors for childhood obesity: An urban and rural comparison. Child Obes, 2012; 8:440–448.

35.

Lutfiyya

, Lipsky

, Wisdom-Behounek

, et al. Is rural residency a risk factor for overweight and obesity for US children?. Obesity, 2007; 15:2348–2356.

36.

Joens‐Matre

, Welk

, Calabro

, et al. Rural-urban differences in physical activity, physical fitness, and overweight prevalence of children. J Rural Health, 2008; 24:49–54.

37.

Davis

, Boles

, James

, et al. Health behaviors and weight status among urban and rural children. Rural Remote Health, 2008; 8:810–821.

38.

Salois

. The built environment and obesity among low-income preschool children. Health Place, 2012; 18:520–527.

39.

Blumberg

, Olson

, Frankel

, et al. Design and operation of the National Survey of Children's Health, 2003. Vital Health Stat, 1 2005; 43:1–131.

40.

Zipf

, Chiappa

, Porter

, et al. National Health and Nutrition Examination Survey: Plan and operations, 1999–2010. Vital Health Stat, 1 2013; 56:1–37.

41.

Koplan

, Koplan

, Liverman

, et al. Preventing Childhood Obesity: Health in the Balance. National Academies Press: Washington, DC, 2005.

42.

Akinbami

, Ogden

. Childhood overweight prevalence in the United States: The impact of parent‐reported height and weight. Obesity, 2009; 17:1574–1580.

43.

Goodman

, Hinden

, Khandelwal

. Accuracy of teen and parental reports of obesity and body mass index. Pediatrics, 2000; 106:52–58.