For Manuscript Submission, Check or Review Login please go to Submission Websites List.
For the academic login, please select your country in the dropdown list. You will be redirected to verify your credentials.
Socioeconomic Inequalities in Childhood Obesity in the United Kingdom: A Systematic Review of the LiteratureEl-Sayed A.M.a,b,c · Scarborough P.a · Galea S.b
aDepartment of Public Health, University of Oxford, Oxford, UK, bDepartment of Epidemiology, Columbia University, New York, NY, cCollege of Physicians and Surgeons, Columbia University, New York, NY, USA Corresponding Author
Abdulrahman M. El-Sayed
Department of Epidemiology, Mailman School of Public Health
722 W. 168th Street, Room 521, New York, NY 10032 (USA)
Childhood obesity is a major public health challenge worldwide. There is a growing literature documenting socioeconomic inequalities in childhood obesity risk. Here we draw inference from the literature about inequalities in childhood obesity risk in the UK. We summarize and appraise the extant peer-reviewed literature about socioeconomic inequalities in childhood obesity in the UK. Common area-level indices of socioeconomic position, including the Carstairs Deprivation Index, the Index of Multiple Deprivation and the Townsend Deprivation Index, as well as common household and individual-level metrics of childhood socioeconomic position, including head-of-household social class and maternal education, were generally inversely associated with childhood obesity in the UK. We summarize key methodological limitations to the extant literature and suggest avenues for future research.
© 2012 S. Karger GmbH, Freiburg
Childhood obesity is a growing worldwide epidemic . Recent research about trends in global childhood overweight and obesity over the last several decades have demonstrated increases in almost all countries for which data is available [2,3]. According to the World Health Organization (WHO), 42 million children under the age of 5 were overweight in the year 2010 worldwide , and available evidence suggests that up to 79% of these children will progress to overweight and obesity in adulthood [5,6,7]. Obesity has important social and psychological consequences among children and adolescents, as it has been shown to predict abuse from peers among children and depression among adolescents [8,9]. Moreover, as an important determinant of chronic disease risk and mortality in childhood and adulthood, childhood obesity is a looming threat to the public’s health [10,11,12,13,14,15,16,17,18,19,20].
Socioeconomic position (SEP) is a measure of the structural location of individuals with respect to access to resources relative to others in a society . In their work on fundamental causes, Link and Phelan  contend that lower SEP will always predict worse health, as SEP portends access to health-promoting resources, even as those resources may change over time. Poor SEP is a well-known predictor of adverse health outcomes among diverse populations.
There is a large literature that has demonstrated inequalities in obesity in both high- and low-income contexts worldwide – however, the direction of this association differs by economic context [23,24,25]. In high-income contexts, like the UK, there is a strong inverse association between metrics of SEP and obesity, whereas SEP and obesity are directly associated in low-income countries . Moreover, with time, and as low-income countries have developed, there have been concomitant increases in obesity prevalence , as, for example, is taking place in China . Research about socioeconomic inequalities in obesity among children worldwide has demonstrated similar trends [4,25].
Already among the highest in the world, risk for childhood obesity is increasing in the UK [28,29]. Between 1995 and 2007, the rate of obesity in children aged 2–15 years in England increased from 12 to 17% . Rising obesity rates among children have received considerable attention among public health officials. For example, they have substantiated recommendations to ban marketing toward children of unhealthy foods, and to allocate greater than 3 h/week for physical activity among children in schools in the UK [31,32].
Several studies about health inequalities in the UK have demonstrated socioeconomic differences in self-rated health, heart disease, chronic bronchitis, smoking, diet, exercise, and overall mortality [33,34]. A recent review about health inequalities in England highlighted SEP inequalities in morbidity, self-reported health, psychopathology, accidental injury, and mortality . Studies also suggest widening inequalities in several important population health metrics in the UK, such as life expectancy and mortality rates between the early 1980s and 2000s [35,36].
Several studies have suggested socioeconomic disparities in childhood obesity in the UK, with poor children at higher risk for obesity [37,38,39,40,41,42,43]. A recent report from the UK’s National Obesity Observatory showed consistent inequalities in obesity among children at both reception year and year 6 by area-level deprivation, with some evidence of increasing inequalities among boys . However, we know of no attempts to systematically appraise or synthesize findings regarding socioeconomic inequalities in obesity in this context.
Understanding socioeconomic inequalities in childhood obesity in the UK may yield important inferences about SEP inequalities in childhood obesity more broadly for several reasons. First, with a nationalized health system, public health research and policy in the UK are relatively centralized and unified. Therefore, the UK features several high-quality government-sponsored datasets that include information about childhood obesity, including the annual Health Survey for England, the Scottish Health Survey and the National Childhood Measurement Programme, as well as others. Moreover, UK health policy has been explicit about both addressing inequalities in health as well as childhood obesity, featuring national targets to reduce levels of overweight and obesity among children to below year 2000 levels by the year 2020 , and to eradicate childhood poverty by the year 2020 . Moreover, in 2007, the National Obesity Observatory was established to monitor obesity in the UK, complementing several other initiatives in place to combat the childhood obesity epidemic, including the ‘Food in Schools’ program, the ‘School Fruit and Vegetable Scheme’, and the ‘Physical Education, School Sport and Club Links Programme’ , although these programs are under threat following recently established budget cuts. Second, despite the substantial attention to both childhood obesity as well as childhood poverty on the part of the UK government, there remain strong associations between metrics of SEP and childhood obesity in the UK [37,38,39,40,41,42,43]. Therefore, the UK presents an ideal opportunity to examine etiological mechanisms relating SEP and obesity among children, including the roles of ethnic minority socioeconomic segregation, individual versus area-level production of socioeconomic inequalities, and the production of inequalities throughout the childhood life course.
Here, we review the extant peer-reviewed literature published in the past 30 years about socioeconomic disparities in childhood obesity. We summarize important differences in the prevalence and determinants of obesity by socioeconomic metrics in the UK, attempting to identify and isolate key indicators in the SEP environment likely to influence childhood obesity risk. Additionally, we comment on generalizable themes in this area of research, considering methodological limitations to the extant literature.
We reviewed the literature about socioeconomic inequalities in obesity in the UK so as to understand how SEP influenced obesity risk in the UK. We limited our review to the UK for several reasons. First, we were interested in ascertaining mechanisms that maintained SEP inequalities in the UK, and these mechanisms may plausibly differ across countries. Second, because national health systems may differ with regard to the relation between SEP and access to healthcare services as well as the focus placed on prevention within systems, generalizing across countries with regard to the relation between SEP and childhood obesity may not be warranted. Third, because of the correlation between ethnicity and SEP in high-income countries and because ethnic minority groups may have differential risk for childhood obesity than whites [48,49,50,51,52], countries with different proportions of ethnic minority groups may show different relations between SEP and obesity, precluding generalization across countries.
This review encompassed the peer-reviewed literature published between January 1, 1980 and March 8, 2010. We limited our review to these years so as to reflect current thinking about the relation between SEP and health. We identified the literature reviewed through the MEDLINE database using the ‘pubmed.gov’ interface, and it included papers that included any empirical assessment of the relation between metrics of SEP and metrics of obesity. We used MeSH search terms ‘Obesity’ and ‘Great Britain’ for English-language articles published in the peer-reviewed literature. All queries were carried out by the primary author during the month of March, 2010.
Our original search yielded 1,189 articles, 233 of which were judged to consider the relation between SEP and obesity in the UK after screening by title. Upon screening by abstract for empirical articles set in the UK, we were left with 102 articles. After reading the remaining articles, another 54 were discarded because they did not meet the following criteria: i) considered differences in outcomes by at least one defined metric of SEP and described attribution of SEP metrics among respondents; ii) described the method used to define obesity, including metric of interest, and threshold for overweight or obesity utilized in analysis; and iii) conducted a direct empiric analysis of differences in obesity outcome by metric of SEP. Reference lists from these articles were searched, and yielded a further 10 articles which fulfilled the inclusion criteria, yielding a total of 58 articles. Finally, 35 articles that did not include outcome measures among respondents under the age of 18 years were excluded, yielding a total of 23 articles reviewed here. A diagram of the search strategy employed in the present article is shown in figure 1.
For each of these papers, the primary author extracted the following information: SEP metric(s); definition of obesity; population and setting; sample and methods; findings and conclusions. The use of 20 different area-level and individual/household-level metrics to measure SEP as well as of 17 different metrics for obesity in the studies reviewed here precluded meta-analysis of the results.
The studies considered in this review utilized two empirical study designs: 5 studies included in this review were longitudinal analyses, while the remaining 18 were cross-sectional in nature. There were 20 different metrics used to ascertain child SEP as well as 17 different metrics used to measure childhood obesity. Only three studies included socioeconomic metrics collected at multiple levels (area-level, household/individual-level), and none of these studies utilized multilevel or complex systems approaches during analysis. None of the studies reviewed assessed mechanisms that mediated socioeconomic inequalities in obesity risk in the UK.
Eight of the studies we report on here used data from regional datasets from localities throughout the UK (London, Peterborough, etc.). The remaining studies reported on representative data from at least one country in the UK. Five studies reported findings from Northern Ireland; there were 7 studies that reported on data from Wales; 10 reported on data from Scotland; and 18 studies reported on data from England (table 1).
Ten studies assessed the relation between area-level indices of SEP and obesity among children in the UK. The majority (7) of studies assessing area-level SEP metrics and prevalence of obesity among children found that higher area-level deprivation was positively associated with obesity prevalence [37,38,39,40,41,42,43], although 3 studies found no association [53,54,55]. Three studies used the Indix of Multiple Deprivation, 3 studies used the Townsend Deprivation Index , 2 studies used the Carstairs Deprivation Index , 1 study used a school-based metric, and 1 used a neighborhood condition metric as indices of area-level of SEP.
Of the 3 studies that assessed area-level deprivation via the Townsend Deprivation Index, two found significant associations between deprivation and obesity [39,43], and one found no association . Kinra and colleagues  studied the relation between 1991 enumeration district Townsend material deprivation scores and obesity (BMI > 98th percentile of UK 1990 reference curves ) and found a direct relationship between deprivation and obesity risk, as girls and boys in the highest deprivation quartile had 1.39 (95% CI 1.08–1.80) and 1.29 (95% CI 1.00–1.65) higher odds of obesity, respectively, than their least deprived counterparts. Wardle and colleagues  found similar results among 5,863 students from 36 schools in London. Brunt and colleagues  studied the relation between 2001 residence lower super output area Townsend material deprivation scores  and overweight (BMI 17.9–19.6 kg/m2 among boys and 17.6–19.4 kg/m2 among girls) and obesity (BMI ≥19.7 kg/m2 among boys and ≥19.5 kg/m2 among girls) among 21,301 children aged 3 years old in Swansea, Neath, and Port Talbot in the South of Wales between 1995 and 2005. Although they found no significant relationship between deprivation scores and obesity risk, they found that there was a non-significant decrease in obesity among the least deprived, and a similar increase among the most deprived between 1995 and 2005.
Among the three studies that assessed area-level deprivation via the Index of Multiple Deprivation, two found significant associations between deprivation and obesity [40,41], and one found no association . Emerson  assessed the relation between area-level Index of Multiple Deprivation and risk for obesity (using International Obesity Taskforce (IOTF) BMI cutoffs ) among 48,819 children in the Millennium Cohort Study. He found that high area-level deprivation was associated with increased obesity prevalence. Rutter  found similar results in a data briefing about 2006–2007 data from the UK-wide National Child Measurement Programme. Dummer and colleagues  used the Index of Multiple Deprivation to assess SEP disparities in overweight and obesity (using IOTF BMI cutoffs ) among over 15,000 9- and 10-year-old children in Liverpool. They found no association between area-level deprivation and overweight or obesity among boys or girls.
Two studies utilized the Carstairs-Morris Deprivation Index to assess SEP disparities in childhood obesity, and both found that higher deprivation was predictive of higher obesity risk [37,42]. Armstrong and colleagues  assessed the relation between Carstairs Deprivation Index and prevalence of overweight (BMI > 95% percentile relative to UK 1990 reference data) or obesity (BMI > 98% percentile relative to UK 1990 reference data ) among a representative sample of 74,500 children aged 39–42 months in Scotland. In their sample, 4.7% of children were obese in the highest deprivation category compared to 3.7% in the lowest category. They found that after adjusting for child birth weight, those in the highest deprivation category had 1.43 higher odds (95% CI 1.16–1.77) of obesity compared to those with the lowest category. Sweeting and colleagues  assessed the relation between Carstairs-Morris postcode-level deprivation and obesity (BMI > 95% of UK 1990 reference data ) among 15-year-old Scottish children in 1987, 1999, and 2006. Among females, there were significant differences in obesity prevalence by post-code level deprivation in 1987 and 2006, with lower prevalence among those in less deprived settings. There were no differences among males.
One study used a school-based metric to assess SEP disparities in childhood obesity and showed a positive association between deprivation and obesity prevalence , and another used a neighborhood condition metric and found no association . Cecil and colleagues  studied 1,240 boys and 1,214 girls between 4 and 10 years old from 47 primary schools in eastern Scotland. Using a school-based metric based on the number of students entitled to free school meals by school, they found that there was an inverse association with material wellbeing and overweight and obesity (both using IOTF BMI cutoffs ) risk among boys and girls. Hawkins and colleagues  studied 13,188 singletons from the Millennium Cohort Study. Using IOTF BMI cutoffs , they found no association between neighborhood condition, an index of objective measures of neighborhood deprivation, and obesity risk.
There were 16 studies that assessed the relations between household and/or individual-level metrics of SEP and obesity prevalence among children in the UK. Among them, there were 15 different household-level and individual-level SEP metrics assessed as determinants of obesity. Eleven studies assessed the relation between head-of-household occupational social class and obesity prevalence [29,42,49,50,51,60,61,62,63,64,65]. Seven assessed the relation between maternal education and obesity prevalence [50,51,55,61,63,66,67]. Four studies assessed the relation between parental employment and obesity risk [51,55,68]. Three each assessed relationships between household income [55,65,66] or receiving free school meals [50,51,68] and obesity prevalence. Two studies considered the number of people in the household as a determinant of obesity [29,50]. Other household and individual metrics of SEP included material deprivation score , paternal education , number of parents in the household , household benefits receipt , access to a vehicle , and household overcrowding .
The literature about the relation between head-of-household occupational social class and obesity risk generally suggests that low social class is associated with higher risk for obesity. Among studies that utilized this SEP metric, 7 found that social class was inversely associated with obesity risk [29,49,50,60,61,62,63]. For example, among a nationally representative sample of children between 4 and 18 years of age from throughout the UK, Jebb and colleagues  found an inverse relationship between social class and obesity (according to IOTF BMI cutoffs ), with a significantly higher prevalence of obesity in social classes IV and V than in classes I–III (6.5 vs. 2.7%, p = 0.003).
One study found that head-of-household occupational social class was positively associated with obesity risk , Chinn and Rona  assessed trends in triceps skin folds among 20,703 English and 4,094 Scottish 4.5- to 12-year-old children between 1972 and 1990. They found that among boys in England, non-manual paternal social class was associated with higher increases in triceps skin folds than manual social class, but they found the opposite among girls as well as among both boys and girls in Scotland.
However, 3 studies found no association between head-of-household social class and obesity risk [42,64,65]. For example, Saxena and colleagues  found no association between the exposure and obesity (using IOTF BMI cutoffs ) among 5,689 children aged 2–20 years sampled in the 1999 Health Survey for England.
The literature about the relation between maternal education and childhood obesity suggests that low maternal education is an important predictor of childhood obesity. Among 7 studies that assessed this relation, four found a significant inverse association between maternal education and the outcome of interest [50,51,61,66,67]. For example, one study by Semmler and colleagues  found that, although there was no significant difference in increase in overweight (BMI > 91st percentile according to UK 1990 reference data ) or obesity (BMI > 98th percentile of UK 1990 reference data ) between 1998/1999 and 2005/2006 by maternal education among 333 twin children in a 7-year longitudinal study in England and Wales, at last follow-up (age 11 years) 29% of low SEP children were obese compared to 17% of high SEP children (p < 0.05). 86% of children who were obese in low SEP families remained obese at follow-up compared to 41% of children who were obese in high SEP families (p < 0.05). Both findings were statistically significant.
One study found a positive association between maternal education and childhood obesity risk. Duran-Tauleria and colleagues  found that in adjusted models, increasing maternal education was predictive of higher weight-for-height and higher risk for overweight (weight-for-height > 75% percentile, according to UK 1990 reference data ).
There were two studies that found no association between maternal education and childhood obesity [55,63]. Hawkins and colleagues  found no association between maternal education and overweight or obesity (according to IOTF BMI cutoffs ) among 13,188 3-year-old singleton children. Rona and Chinn  found similar results assessing the relation between maternal age at completion of education and triceps skin folds and weight-for-height among 9,815 children aged 5–11 years in Scotland and England.
Four studies assessed relationships between parental employment and obesity risk among children [50,51,55,68]. Among 13,188 3-year-old singleton children, Hawkins and colleagues  found that, while maternal employment itself was not associated with overweight (IOTF BMI cutoffs ), children of mothers working 21 h/week or greater had 1.23 higher odds (95% CI 1.10–1.37) of overweight. Similarly, Duran-Tauleria and colleagues  found that higher working hours among mothers were associated with higher weight-for-height among English, Scottish, and inner-city samples aged 5–11 years. Another study among 2,482 children in East London found that parental unemployment was associated with lower risk for overweight (IOTF BMI cutoffs ) compared to parental employment . Similarly, in a 1982–1983 study of 13,073 5- to 11-year-old children , paternal employment was predictive of higher weight-for-height than paternal unemployment .
Three studies also assessed the relation between household income and obesity risk [55,65,66]. Only one study found an association: Stamatakis and colleagues  found that household income adjusted for family size was associated with lower odds of obesity (using IOTF BMI cutoffs ) in fully adjusted models among data from the National Studies of Health and Growth in 1974, 1984 and 1994, and the 1996–2003 Health Surveys for England. Among 13,188 3-year-old singleton children, Hawkins and colleagues  found no association between household income and overweight (IOTF BMI cutoffs ) in bivariate models, although there was a tendency toward overweight among lower income households (p = 0.11). Matijasevich and colleagues , in a study of just under 7,000 children in Avon, reported a non-significant (p < 0.06) association between household income and overweight among boys, and no association among girls.
Three studies also assessed the relation between receiving free school meals and risk for childhood obesity. They found no association between free school meals and obesity risk [50,51,68]. For example, Taylor and colleagues  showed no association between receiving free school meals and overweight or obesity (using both IOTF BMI cutoffs  and 1990 UK reference data ) among 2,482 11- to 14-year-old children in East London. However, it is important to note that two of the extant studies are derived from data older than 15 years, and one is limited to a highly diverse sample in East London, and therefore may not accurately reflect the relation between receiving free school meals and childhood obesity risk in the UK. Moreover, because free school meals in the UK are allocated based on household income, adjusting for other socioeconomic metrics in multivariable models of this relation may be methodologically questionable, although this was done in each of the studies reviewed here.
Two studies considered the number of people in the household as a determinant of obesity [29,50], with conflicting findings. One study  found in fully adjusted models that larger family size was protective against overweight (weight-for-height above the 75th percentile according the UK 1990 reference data ). Alternatively, Chinn and Rona  found that a higher number of children in the family was associated with higher mean triceps skin fold increase between 1972 and 1990 among English boys, but found the opposite among Scottish boys, and English and Scottish girls.
Other household and individual metrics of SEP that were studied as predictors of childhood obesity included material deprivation , paternal education , number of parents in the household , household benefits reception , access to a vehicle , and household persons per room . Of these, material deprivation  number of parents in the household , and access to a vehicle  were associated with any outcome of interest.
Emerson  studied the relation between an individual metric of material deprivation (using the number of consumer durable goods deemed ‘essentials’ not present in the household) and obesity risk (using IOTF BMI cutoffs ), and found that greater exposure to material deprivation was associated with obesity risk. Hawkins and colleagues , among 13,188 singletons from the Millennium Cohort Study, found that lone-parent status was associated with overweight (using IOTF BMI cutoffs ) risk in fully adjusted models. Taylor and colleagues , in a study of almost 2,500 East London children aged 11–14 years, found that family access to a vehicle was associated higher risk for obesity (according to IOTF BMI cutoffs ) among girls, and higher risk for overweight (according to UK 1990 reference data ) among boys compared to having no access to a vehicle.
A comprehensive review of the peer-reviewed literature regarding socioeconomic inequalities in childhood obesity in the UK between 1980 and 2010 found that both area-level and household/individual-level metrics of SEP are associated with childhood obesity in the UK. In particular, three of the most common area-level indices of SEP in the epidemiologic literature, the Index of Multiple Deprivation, the Townsend Deprivation Index , and the Carstairs Deprivation Index , were reliable predictors of childhood obesity in the UK (although they did not predict obesity in all studies). We also found that common household and individual-level metrics of childhood SEP, including head-of-household social class and maternal education were reliable determinants of childhood obesity in the UK (although they did not predict obesity in all studies). Some other associated household and individual-level metrics, including household income, free school meals, and paternal employment, were not independently associated with childhood obesity after adjustment for income levels. This indicates the importance of supporting analyses with appropriate conceptual casual frameworks that specify the production of inequalities in obesity across levels and therefore guide the construction of analytic models when exploring the roles of socioeconomic predictors of childhood obesity at the area level, household level, and individual level.
While there are relatively few prospective studies of the relation between SEP and obesity in other Western European countries, their findings are comparable to the small number of which we are aware in the UK. For example, in a prospective study of 341 children aged between 6 and 8 years in the south of Italy, Valerio and colleagues  demonstrated that children with less educated mothers had accelerated weight gain over a 3-year period relative to those with more educated mothers. A prospective study of a random sample of 9- and 10-year-old children in Copenhagen schools in Denmark  found that, while parental education and occupation did not predict obesity at 10 years follow-up after adjusting for childhood adiposity and gender, a metric of neighborhood-level deprivation predicted higher risk for overweight after adjustment for parental education and occupation. Similarly, in a prospective cohort of 10-year-old children in Sweden , low neighborhood-level social class was associated with higher obesity risk. By contrast, in a cohort of children in Southeast Sweden , neither maternal nor paternal education was associated with risk of childhood obesity at age 5 (using IOTF BMI cutoffs ).
Although this review demonstrated important gradients in childhood obesity risk by metrics of SEP in the UK, there are several methodological limitations to the current literature about socioeconomic inequalities in childhood obesity that limit our understanding: i) there have been few studies that have simultaneously studied both area-level and household/individual-level determinants of childhood obesity, ii) there are no known studies that have utilized multilevel modeling or complex systems approaches to understand the relative influence of contextual and individual SEP on childhood obesity risk in the UK, iii) there is limited availability of longitudinal studies that have assessed life course and/or intergenerational SEP gradients in childhood obesity, iv) the extant literature has paid little attention to mechanisms underlying the relation between metrics of SEP and childhood obesity, and v) the multiplicity of SEP metrics used in analysis and the lack of adherence to comparable metrics limits comparisons of the literature across metrics.
The first two limitations pose foundational challenges to our understanding of socioeconomic inequalities in childhood obesity in the UK. First, the concept that individuals may interact with, and therefore be influenced by, their ecological contexts is fundamental in social scientific inquiry [73,74,75,76]. Studies about socioeconomic disparities in childhood obesity which only account for variation in the outcome of interest via individual or household metrics of SEP, of which there were 14 of 23 studies reviewed, may not account for ecological manifestations of poverty and their contributions to childhood obesity, yielding an imperfect understanding of the relation between SEP and childhood obesity.
Moreover, 7 of 23 studies reviewed assessed area-level metrics of SEP as determinants of obesity without including individual-level or household-level metrics of SEP. In the absence of individual-level SEP data, investigators often use area-level SEP metrics as proxies for individual SEP. However, area-level SEP metrics are encumbered by substantial measurement error, as not all individuals living in lower SEP areas will have low individual SEP metrics, and vice versa. Moreover, it is difficult to ascertain the mechanisms that relate area-level metrics of SEP to health outcomes of interest. On one hand, area-level SEP metrics can serve as proxies for individual SEP, thereby representing populations with concentrated individual-level poverty, and explaining relations between area-level metrics and outcomes of interest. Alternatively, it is possible that context itself, independent of individual-level SEP, can predict outcomes of interest .
Studies that simultaneously consider both individual and area-level factors as determinants of outcomes are most appropriate, given the following three barriers discussed above: i) Individuals may interact with, and therefore be influenced by, their ecological contexts [73,74,75,76]. ii) Area-level SEP variables may be poor proxies for individual-level SEP. iii) It is difficult to quantify the direct and indirect contributions of area-level metrics to outcomes of interest in epidemiologic analyses that do not include individual-level data. Epidemiologists, therefore, have begun, over the past 10 years, to conceptualize and analyze models of disease with such a multilevel understanding in mind  – a departure from traditional models of disease that focused on features of the individual or proxies thereof exclusively . Multilevel models, which account for clustering within multilevel data, allow investigators to estimate the relations between exposures and outcomes of interest while adjusting for other exposures across levels of influence . Multilevel thinking has allowed investigators to conceptualize and examine how factors operating at multiple levels of influence – characteristics of individuals, their families, contacts, neighborhoods, and societies – can shape, both individually and in cooperation, their health and disease risks . Growing out of the multilevel conceptual paradigm as well as an understanding of the limitations of traditional predictive regression modeling, complex systems approaches feature stochastic modeling techniques that allow investigators to capture bi-directional, dynamic, and relational interactions between traditional ‘exposures’ and ‘outcomes’ at any level of analysis and influence . These approaches, therefore, may be ideally suited to understand the causes, mechanisms, and consequences of socioeconomic disparities in childhood obesity in high-income contexts. Without simultaneous study of SEP metrics on multiple levels using multilevel or complex systems tools, it remains impossible to quantify the contributions of metrics of SEP at multiple levels to childhood obesity risk in this context.
The third methodological limitation is the limited availability of longitudinal studies that have assessed life course and/or intergenerational SEP gradients in childhood obesity in the UK. Increasingly, the life course approach has gained traction in social epidemiology. Defined by Ben-Shlomo and Kuh , the life course framework in epidemiology implies the assessment of long-term effects on disease risk from physical and social exposures throughout the life course. Life course approaches to the question of childhood obesity in the UK have been fruitful: for example, a study about the predictors of childhood obesity among a cohort of 7-year-old children in Avon  found that several exposures occurring before the age of 3, including parental obesity, early adiposity rebound, greater than 8 h/week spent watching television, and short sleep duration at age 3 were all significant predictors of obesity (>95th percentile of UK 1990 reference data ). Moreover, within this framework, evidence has suggested that maternal health and wellbeing, both before and during pregnancy, may influence the health and wellbeing of children. For example, studies have demonstrated that maternal smoking  and maternal obesity  during pregnancy may influence risk for childhood obesity. The paucity of longitudinal studies that have included and/or analyzed socioeconomic data from gestation and/or early childhood as determinants of obesity in later children in meaningful ways limits our understanding of the mechanisms by which SEP may influence childhood obesity risk in the UK.
Growing out of the paucity of longitudinal studies, the fourth methodological limitation to the extant literature about socioeconomic inequalities in childhood obesity in the UK is that the current literature has largely ignored mechanisms by which SEP may influence childhood obesity. For example, it has been shown that food insecurity may mediate the relation between poor SEP and risk for childhood obesity in other countries, as it may lead to binge eating cycles and energy-dense food consumption [82,83]. To our knowledge, this feature of the relation between SEP and obesity among children in the UK remains unstudied. Studies that empirically assess mechanisms by which SEP may influence obesity risk in the UK are crucial for interventions designed to curb the childhood obesity epidemic in this context.
The fifth methodological limitation is that the multiplicity of SEP metrics used in analysis and the lack of adherence to comparable metrics limits comparisons of the literature across metrics. For example, the most regularly studied metric of SEP in the literature reviewed above was head of household social class, about which only 11 out of 23 studies collected and analyzed data. The next most considered metric was maternal education at 7 out of 23 studies. Because the nature of SEP metrics, their relation to childhood obesity, and the mechanism by which they may influence obesity among children may differ substantially, the lack of adherence to comparable metrics among studies assessing SEP differences in childhood obesity risk in the UK disrupts overall comparisons across studies and limits our ability to draw meaningful inference from the extant literature.
The reader should keep in mind several limitations when interpreting the findings of this review. First, because we restricted the studies reviewed above to those published in the peer-reviewed literature, the articles we reviewed and the inferences we have drawn may be subject to a publication bias. Although our inclusion criteria were expansive and included data about many of the largest health surveys in the UK, our findings may not accurately reflect current knowledge about SEP and obesity in the UK. Second, our findings were organized by level of influence and then by metric of analysis. Our organization scheme could possibly have, in part, shaped the inferences we drew from the extant literature. Third, our findings were limited to studies about socioeconomic disparities in obesity risk among children in the UK. Therefore, it would be inappropriate to generalize our findings to other age groups in other national contexts.
Five directions for future research emerge from this review. First, studies that assess SEP metrics at multiple levels of influence – area, household, and individual – are needed to understand the complex relation between SEP and childhood obesity risk.
Second, these studies should utilize multilevel thinking during conception and multilevel and complex systems approaches during analysis so as to quantify SEP influences at multiple levels on childhood obesity risk. These studies could clarify the importance of contextual factors in shaping childhood obesity in the UK for researchers and policymakers alike.
Third, investigators interested in this area might consider studies that include data about maternal SEP during pregnancy as well as early life metrics of SEP among children so as to understand SEP determinants of childhood obesity in the UK throughout the life course.
Fourth, research about the mechanisms relating SEP and childhood obesity in the UK is needed so as to educate potential interventions against childhood obesity.
Fifth, consensus regarding metrics of SEP that are most fruitful in research about socioeconomic disparities in childhood obesity is needed, and researchers in this area should utilize these metrics primarily.
This study was funded in part by the British Heart Foundation and the Rhodes Trust.
The authors claim no conflicts of interest.
Abdulrahman M. El-Sayed
Department of Epidemiology, Mailman School of Public Health
722 W. 168th Street, Room 521, New York, NY 10032 (USA)
Copyright: All rights reserved. No part of this publication may be translated into other languages, reproduced or utilized in any form or by any means, electronic or mechanical, including photocopying, recording, microcopying, or by any information storage and retrieval system, without permission in writing from the publisher or, in the case of photocopying, direct payment of a specified fee to the Copyright Clearance Center.
Drug Dosage: The authors and the publisher have exerted every effort to ensure that drug selection and dosage set forth in this text are in accord with current recommendations and practice at the time of publication. However, in view of ongoing research, changes in government regulations, and the constant flow of information relating to drug therapy and drug reactions, the reader is urged to check the package insert for each drug for any changes in indications and dosage and for added warnings and precautions. This is particularly important when the recommended agent is a new and/or infrequently employed drug.
Disclaimer: The statements, opinions and data contained in this publication are solely those of the individual authors and contributors and not of the publishers and the editor(s). The appearance of advertisements or/and product references in the publication is not a warranty, endorsement, or approval of the products or services advertised or of their effectiveness, quality or safety. The publisher and the editor(s) disclaim responsibility for any injury to persons or property resulting from any ideas, methods, instructions or products referred to in the content or advertisements.