Ethics statement

The study was approved by the Multicentre Research Ethics Committee (Wales). Informed written consent was obtained from each pupil's parent or guardian.

Study design

The Child Heart and Health Study in England (CHASE) is an investigation of the cardiovascular health of British school children aged 9–10years of white European, South Asian and black African-Caribbean origin. Full details of the study design have been reported elsewhere [15], [19]. In brief, the study took place in 200 primary schools in London, Leicester and Birmingham, sampled to include 100 schools with a high proportion (20–80%) of South Asian pupils and 100 schools with a high proportion of black African-Caribbean pupils. All Head Teachers were approached by the Principal Investigator and invited to participate; 140 (70%) agreed. Non-participating schools were replaced by a school from the sampling frame with a similar ethnic mix and in the same or a neighbouring borough. The final sample included 183 schools from London, 14 from Birmingham and 3 from Leicester.

Measurements

Assessments were carried out during school terms by a single research team visiting schools in different areas in rotation. Participating children had physical measurements (anthropometry, blood pressure and spirometry) and provided a fasting blood sample. Height was measured to the last complete millimetre with a portable stadiometer (Chasmors Ltd, London, UK) and weight with an electronic digital scale (Tanita Inc, Tokyo, Japan). Ponderal index was calculated as kg/m³. Waist circumference was measured at the midpoint between the lower margin of the ribs and the iliac crest in the mid-axillary line. Right-sided skinfolds (biceps, triceps, subscapular, suprailiac) were measured and summed for analysis. Body composition was measured using leg to arm bioimpedance (Bodystat 1500 bioimpedance monitor, Bodystat Ltd, Isle of Man, UK). Fat mass was calculated using equations derived for children using dual-emission X-ray absorptiometry [20], and presented as fat mass index independent of height (fat mass/m⁵).

Blood samples were transferred for analysis of HbA1c, glucose, and blood lipids within 48 hours of collection. Glucose was measured in plasma using the hexokinase method. HbA1c was measured in whole blood by ion exchange high performance liquid chromatography and adjusted for abnormal haemoglobin variants or for increased amounts of normal variant fetal haemoglobin where present. Serum triglyceride and HDL-cholesterol were measured using an Olympus auto-analyser. Serum, separated and frozen on dry ice after collection, was measured for insulin using an ELISA method and for C-reactive protein (CRP) using ultra-sensitive nephelometry (Dade Behring, Milton Keynes, UK). The homeostasis model assessment (HOMA) equations were used to provide an estimate of insulin resistance [21].

Parents completed a questionnaire on their self-defined ethnicity, ethnicity of the participating child and their occupations. Ethnicity of the children was defined using the ethnicity of both parents or (if not available) the ethnicity of the child or (if not available) place of birth of parents and grandparents provided by the child (1%). In the present analyses ‘white European’ includes children whose ethnic origin was defined as ‘white British’ ‘white Irish’ and ‘white European’ (or a combination of these). ‘South Asian’ includes ‘Indian’ ‘Pakistani’ ‘Bangladeshi’ and ‘Sri Lankan’ (or a combination of these). ‘Other Asian’ includes ‘Asian other’ and ‘other’ with a specified Asian place of origin (mainly Afghanistan, China and Turkey). ‘Black African-Caribbean’ includes ‘black African’ ‘black Caribbean’ ‘black British’ and ‘black other’ (or a combination of these). The ‘other’ ethnic group includes all other categories of individual and mixed ethnic origins. The ethnic subcategories ‘Indian’ ‘Pakistani’ ‘Bangladeshi’ includes children whose parents both originated in the same country; ‘black African’ and ‘black Caribbean’ groups those who originated in the same region. For primary analyses, all children of South Asian origin and all black children originally of African origin have been considered together; in a second level of analysis SEP patterns have been examined separately in children of Indian, Pakistani and Bangladeshi origin and separately in black children of African and Caribbean origin.

Current socio-economic position (SEP) was measured according to the UK National Statistics Socioeconomic Classification (NS-SEC), coded from parental occupation (Standard Occupational Classification 2000) using an Office for National Statistics matrix [22]. The SEP of the child was defined on the basis of the parent with the highest NS-SEC grade (or that of the sole parent in single parent households, 25%). NS-SEC was operationalized using both the three class version (NS-SEC3 categorised as professional and managerial, intermediate, routine and manual) and the five class version (NS-SEC5 categorised as managerial and professional, intermediate, small employers and own account workers, lower supervisory and technical, semi-routine and routine). An additional category of ‘economically inactive’ was added at the lower end of both NS-SEC versions, in accordance with published guidance on NS-SEC use [22]. Individuals who could not be classified into an NS-SEC group are shown in Tables as ‘unclassified’ but have not been included in statistical analyses.

Statistical analysis

All analyses were carried out in STATA (version 11.1; StataCorp LP, College Station, TX, USA). Multilevel linear regression models fitting school as a random effect in order to take account of the natural clustering of children within school were used to provide adjusted means and adjusted differences in risk factors and their 95% confidence intervals. Analyses were adjusted for sex, age, month of assessment and observer (physical measurements only) as fixed effects. All outcome variables were log transformed and ethnic and socio-economic differences in these variables were expressed as percentage change (exp(ß)-1)*100 in order to achieve normality and to enable comparisons across outcomes.

Associations between NS-SEC3 and T2DM risk factors have treated NS-SEC3 both as an ordinal variable and as a nominal variable [22]. Associations between NS-SEC5 and T2DM risk factors have treated NS-SEC5 entirely as a nominal variable. The principal analyses have included ‘economically inactive’ individuals, treating them as the most deprived group in analyses of trends across ordinal categories; sensitivity analyses have been carried out excluding the economically inactive groups and therefore restricted to currently employed groups alone.

We examined the influence of NS-SEC on ethnic differences on T2DM risk factors by (i) adjusting ethnic differences for NS-SEC5, and (ii) where there was evidence of interaction (for markers of adiposity and insulin resistance in particular and more weakly for triglyceride levels) by examining the ethnic differences stratified by NS-SEC3 groups. We also investigated the effect of adjustment for parental education and an index of household amenities (data not presented) in order to address the potential for residual confounding by other dimensions of SEP.

Tests of interaction between ethnic group and NS-SEC and between gender and NS-SEC were carried out using the likelihood ratio (LR) test; these tests were carried out with NS-SEC fitted both as ordinal and nominal variables. The associations between NS-SEC and T2DM risk factors were similar in boys and girls, both overall and within specific ethnic groups, and data are therefore presented for both groups combined, adjusted for sex. Analyses were undertaken for the main ethnic groups and repeated using ethnic sub-groups.

Descriptive characteristics

Of 8,641 children invited, 5,887 (68%) participated and 4843 (82%) children without type 1 diabetes provided fasting blood samples. Analyses were based on 4804 children (1158 white European, 1201 African-Caribbean, 1314 South Asian, 295 other Asian children and 836 of other ethnic origins) with complete data on parental employment status. The mean age of participants was 10.0 years (s.d. = 0.4 years) and 51% were female. Participation rates were unrelated to age but were slightly lower among males (65%) than females (71%). Participation rates were similar among white Europeans, South Asians, other Asians and other ethnic groups (69%, 73%, 70% and 71% respectively) with slightly lower participation among black African-Caribbeans (65%). The socioeconomic status of children who did and did not provide blood samples did not differ appreciably (both groups included 27% managerial/professional and 17% economically inactive).

There were marked differences in the distribution of NS-SEC categories between the main ethnic groups (Table 1). White European and black African-Caribbean children had higher proportions of parents in managerial/professional occupations and lower proportions in routine/manual occupations and economically inactive than South Asian and other Asian children. Within the main ethnic groupings, Pakistani and Bangladeshi children were slightly more disadvantaged than Indian children and black African children more so than black Caribbean children.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Socio-economic position (NS-SEC) by ethnic group in CHASE, n (%).

https://doi.org/10.1371/journal.pone.0032619.t001

NS-SEC and type 2 diabetes risk factors in main ethnic groups

Geometric mean T2DM risk factor levels for each NS-SEC3 category and the percentage change in risk factor levels per one NS-SEC group decrement are presented for the whole sample and separately by main ethnic group in Tables 2 to 5.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Adjusted means and percentage differences in height, weight and ponderal index by NS-SEC and ethnic group.

https://doi.org/10.1371/journal.pone.0032619.t002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Adjusted means and percentage differences in skinfolds, fat mass index and waist circumference by NSSEC and ethnic group.

https://doi.org/10.1371/journal.pone.0032619.t003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Adjusted means and mean differences in HbA1c, glucose and insulin resistance by NS-SEC and ethnic group.

https://doi.org/10.1371/journal.pone.0032619.t004

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Adjusted means and mean differences in triglyceride, HDL-cholesterol and C-reactive protein by NS-SEC and ethnic group.

https://doi.org/10.1371/journal.pone.0032619.t005

NS-SEC and type 2 diabetes risk factors in ethnic sub-groups

The associations between NS-SEC3 and T2DM risk factors were examined separately in black African and black Caribbean children, and separately in Indian, Pakistani and Bangladeshi children (Tables S3 and S4). There was little evidence that the (generally weak) associations between NS-SEC3 and T2DM risk factors in these ethnic sub-groups differed between black Africans and black Caribbeans, or between Indians, Pakistanis or Bangladeshis, either when NS-SEC3 was treated as an ordinal or nominal variable. Among the associations referred to above, there was some evidence that the associations between higher NS-SEC3 and greater adiposity (particularly ponderal index and fat mass index) and insulin resistance were stronger among black Africans than black Caribbeans. However, there was no strong evidence that any associations between NS-SEC and T2DM risk factors differed between Indian, Pakistani or Bangladeshi children (Tables S3 and S4).

Ethnic differences in type 2 diabetes risk factors: influence of NS-SEC

As previously reported [15], compared to white European children, South Asian children had higher sum of skinfolds, fat mass index, HbA1c, glucose, insulin resistance, triglycerides and C-reactive protein and lower HDL-cholesterol (Table S5). Black African-Caribbean children had less marked increases in HbA1c, insulin resistance and C-reactive protein, but conversely, had lower triglycerides and higher HDL-cholesterol; adiposity levels were not consistently increased (Table S5).

Adjustment for NS-SEC5 made no material difference to the size and direction of these ethnic differences (Table S5). Further analyses including parental education and household amenities had no further effect on these differences (data not presented). The stratified analysis showed that the higher levels of skinfolds, fat mass index, insulin resistance and triglyceride levels observed in South Asians compared with white Europeans were particularly marked in the higher NS-SEC3 groups (Table S6). Similarly, the higher fat mass index and insulin resistance observed in black African-Caribbeans compared with white Europeans were particularly marked in the higher NS-SEC3 groups; in contrast, the lower triglyceride levels were particularly marked in the lower NS-SEC3 groups.

Main findings

In this school-based study of 9–10 year old children living in London, Birmingham and Leicester, SEP (measured by NS-SEC) was associated with adiposity and insulin resistance, although these associations appeared to vary between different ethnic groups. White European children from lower NS-SEC groups had higher levels of adiposity and insulin resistance, while black African-Caribbean children from lower NS-SEC groups (particularly the economically inactive group) had lower adiposity levels, findings that were particularly marked in African children. Little association was apparent for South Asian children and there was little evidence of heterogeneity between South Asian groups. Marked ethnic differences in T2DM risk factors were unaffected by adjustment for NS-SEC5; in stratified analyses, ethnic differences in fat mass index and insulin levels in particular were largest in the higher socio-economic groups.

Comparison with other studies

In adults, low SEP is consistently related to higher T2DM prevalence and associated morbidity and mortality, particularly in white Europeans in high income countries including the UK [6], [7]. However, data on socio-economic gradients in T2DM in UK South Asians and black African-Caribbeans are limited; current evidence to date suggests that the inverse association between SEP and T2DM and its risk factors are weaker in South Asians than white Europeans [2], [3], [8]. Previous studies of socio-economic gradients in T2DM in black African-Caribbeans have been less consistent and depend on the SEP measure used. Using the Registrar General's occupational social class, the National Survey of Ethnic Minorities failed to show a clear association with diabetes prevalence in Caribbean adults, but found that tenants were more likely to report diabetes than homeowners [3]. The Health Survey for England reported an inverse association between income and T2DM among Caribbean adults [2].

In UK children, there is little direct evidence on associations between SEP and T2DM risk factors. In the present study, most variation in T2DM risk factors (particularly adiposity) occurred at individual rather than school level, a finding consistent with observations in previous school-based trials of obesity prevention in children [23]. Associations between low SEP and greater adiposity have been increasingly apparent in recent studies, particularly in white Europeans [18] and have recently been reported in 10 year-olds in the large ALSPAC Study (a predominantly white European population) based on DXA total body fat measurements [24]. This pattern, though not consistently observed in studies before 2000 [17], [18], is consistent with our own findings and suggests that associations between low SEP and greater adiposity are present in the first decade of life, particularly in UK white European children; similar patterns have also been reported in studies of predominantly white children in the US [25]–[27]. No UK studies have to our knowledge reported patterns of associations between SEP and insulin resistance in children, but our observations of an inverse association in white Europeans are consistent with the results of the European Youth Heart Study in Denmark [28], studies in the USA [29], and with the socio-economic gradient in T2DM seen for UK adults [5], [8]. Earlier UK reports have shown little consistent evidence of associations between SEP and blood lipids in white Europeans [17]; our finding of an association between low SEP and high triglyceride is however consistent with the increasing evidence of associations between low SEP and greater adiposity in other studies [18]. The lack of an association between SEP and T2DM risk factors in South Asian children is consistent with adult data [2]–[4]. The associations between low SEP and lower adiposity and insulin resistance among black African-Caribbean children, though less consistent with adult studies, are consistent with the patterns observed for adiposity and insulin resistance patterns in children in less affluent countries undergoing socio-economic transition [28], [30] and with adiposity patterns in African American children in the US [25].

Strengths and limitations

Particular strengths of CHASE were its large size (designed to detect modest differences in T2DM risk factors between main ethnic groups), strong representation of ethnic minority groups and detailed measurements, particularly including adiposity indices that are more suitable than body mass index for studies of ethnically diverse children [31]–[33]. Insulin resistance was estimated from fasting glucose and insulin concentrations using the homeostatic model assessment (HOMA) method, a validated measure of insulin resistance shown to be associated with an increased risk of developing T2DM in longitudinal studies and appropriate for use in large, epidemiological studies where only a single fasting plasma sample is required [34]. Results using the HOMA-IR method, which has also been validated in children [35], were consistent with findings using fasting insulin (data not presented).

The study sampled schools from three cities in which most UK South Asians and black African-Caribbeans reside [36] and included similar numbers of South Asian children of Indian, Pakistani and Bangladeshi origin and of black African-Caribbean children of black African and black Caribbean origin. Thus, the study is likely to have stronger general representation of ethnic minority groups than white Europeans, higher proportions of whom live in other parts of the UK. The effect of non-participation at school level (30% of invited schools, with replacement of each non-participant school with a school of similar characteristics) is difficult to assess; decisions were often made for reasons which appeared to be unrelated to pupil characteristics (e.g. impending school inspections or senior staff changes). The restriction of the study to state schools and the moderate individual response rates may have limited representation of high SEP and low SEP participants respectively. However, response rates were very similar in most ethnic groups and comparisons of the characteristics of responders and non-responders provided little evidence of selection bias. The proportions of white European children in CHASE whose highest NS-SEC parent was in managerial/professional (31%) and routine/manual (24%) occupations were reassuringly similar to the 2001 Census (27% and 28% respectively), though intermediate occupations were somewhat overrepresented (28% vs 17% in the Census) [37]. In black African-Caribbeans, professional/managerial occupations were over-represented (33% vs 25% in Census), while among South Asians, routine/manual occupations were over-represented (35% vs 24% in Census). Nevertheless, the proportion of children living in a “workless household” (17%) was comparable to the overall Labour Force Survey (LFS) estimate of 16%, while the prevalences in specific ethnic groups closely matched national data for all groups except black Caribbeans [38]. Although the white European population in CHASE may not be fully representative of the wider UK white European population, the associations between low NS-SEC and higher T2DM risk markers observed in white European children in CHASE (particularly affecting adiposity, insulin and triglyceride) were reassuringly consistent with those in previous recent reports [18], [23].

Our main analyses focused on the three main ethnic groups living in the UK – white Europeans, South Asians and black African-Caribbeans. These groups were used because of their common origins and similar within-group T2DM risks (very high among South Asian groups, moderately high among black Africans and black Caribbeans). However, supplementary analyses examining differences in the relations between SEP and T2DM risk factors among Indian, Pakistani and Bangladeshi children and among black Africans and black Caribbeans were also carried out. We did not find strong evidence for heterogeneity in the associations between NS-SEC and T2DM risk factors amongst South Asian children, although some heterogeneity was apparent for black African-Caribbean children, particularly for adiposity. However, even with a study population of this size, statistical power for the detection of differences is limited.

The use of the NS-SEC (a recently developed national occupational classification taking account of current employment patterns and including non-employed categories) for defining SEP is a further strength of the study. The NS-SEC places people into classes according to their occupational title and employment relations and conditions such as employers or employees, whether they have a wage or salary, levels of autonomy and prospects for promotion [39]. As such, the NS-SEC is considered a better theoretical indicator of SEP than the Registrar General's Social Class (based primarily on occupation) and also may be better suited to studying health inequalities in ethnic groups [39]. The NS-SEC3 can be treated as an ordinal or nominal variable, while the NS-SEC5 is used primarily as a nominal variable. The validity of the NS-SEC in different ethnic groups in the present study is supported by the association between low NS-SEC and shorter stature (a key marker of childhood nutrition) in the whole study population, with no strong evidence of heterogeneity in this association between ethnic groups (Table 2). Its identification of small employers and own account workers (self-employed) may be particularly relevant for South Asians [39]. In accordance with official NS-SEC guidance [22], the substantial group of economically inactive subjects were classified into a separate low SEP group placed below the lowest employed group; similar results were obtained using the alternative recommended strategy of merging the economically inactive group with the lowest employed group [22]. This approach ensured the widest possible representation of the range of SEP in this study population and maximized the number of participants included in analysis.

Implications

An important finding of the present study is that while among white Europeans, low NS-SEC groups have greater adiposity and insulin resistance, the patterns of association between SEP and T2DM risk factors in other ethnic groups within the UK population may differ. These patterns were apparent whether NS-SEC was treated as an ordinal or nominal variable, and whether fitted as a three class or five class variable. They were also similar when the outcome variables were treated as dichotomous (focussing on the top fifths of the distributions of adiposity, insulin and blood lipids). However, the results (particularly the evidence that the associations between NS-SEC and T2DM risk factors differed between ethnic groups) were dependent on the inclusion of the economically inactive groups in the analysis; exclusion of this group markedly reduced the evidence of heterogeneity in associations between NS-SEC and T2DM risk factors between ethnic groups. This could simply reflect the reduced number of groups and study participants and the more limited range of social circumstances and the consequent reduction in statistical power, or it could reflect the exclusion of particular economically inactive groups, perhaps especially the black African-Caribbean group, with low levels of adiposity and insulin resistance. However, the observed patterns are consistent with earlier reports of the differing associations between SEP and adiposity in black and white populations in the USA [25] and with evidence from other sources which suggest that associations between SEP and chronic disease risks differ between times, places and population groups [40].

The associations between low SEP and higher levels of T2DM risk factors in white European children are consistent with the socio-economic patterning of childhood obesity reported in other studies both in the UK [24] and more widely [18], [30] and with the socio-economic patterning of chronic disease (particularly T2DM and cardiovascular disease) observed in current and previous adult generations, reflecting the later stage of the socioeconomic transition [41]. The weaker associations observed in South Asian and black African-Caribbean children, (with some evidence of an opposite association among black African-Caribbean children) would be consistent with earlier stages of the socio-economic transition where adverse exposures are less concentrated among lower SEP groups than in later stages [41]–[44] a pattern which has also been reported for adiposity and insulin resistance in children in less affluent countries [28], [30] and for adiposity in African American children [25].

The findings in our study could reflect the evolution of socially determined exposures such as dietary characteristics (particularly total calorie and fat intakes) and physical inactivity (influenced by patterns of family car use). However, overall associations between SEP and physical activity are weak and patterns in individual ethnic groups do not correspond to those described here for adiposity and insulin resistance (C Thomas, unpublished data). Dietary factors could therefore be important and need further exploration. As diets have changed in migrant populations to reflect those of the host population, particularly among younger age groups [45], [46] investigation into dietary patterns and practices, including food purchasing and cooking patterns and the extent of dietary acculturation (perhaps especially in black African-Caribbean children from low NS-SEC families) could shed further light on the role of health behaviours in explaining ethnic differences in T2DM risk factors.

The analyses examining the impact of NS-SEC adjustment on ethnic differences in T2DM risk factors suggested that ethnic differences in SEP did not directly account for ethnic differences in T2DM risk factors. Despite its strengths, the NS-SEC classification cannot represent all dimensions of SEP and it is therefore possible that these analyses are affected by residual confounding [3]. However, this possibility is made less likely by the very limited effect of NS-SEC adjustment on the size of ethnic differences, by the minimal effect of additional adjustments for parental education and household amenity score, as well as by the evidence that the association between NS-SEC and T2DM risk factors differs by ethnic groups, which render simple approaches to NS-SEC adjustment potentially inappropriate. This conclusion was further strengthened by the results of stratified analyses, which showed that ethnic differences were apparent at all SEP levels, although they were particularly marked in the highest SEP groups, which argues against the “underclass” hypothesis [47].

These results have important implications for strategies for early prevention of T2DM risk. Ethnic differences in NS-SEC do not appear to explain why UK black African-Caribbean and South Asian children are more adipose and insulin resistant than white European children. Moreover, ethnic differences in T2DM risk markers appear to be largest in the most affluent socio-economic groups. These findings highlight important implications for strategies in early T2DM prevention. First, they suggest that strategies aimed at reducing socio-economic inequalities in emerging T2DM risk in childhood will not be effective in reducing ethnic differences in T2DM risk, a conclusion also reached by other investigators [25]. However, efforts to reduce social inequalities in T2DM risk could be particularly important in white European children, in whom low SEP is strongly associated with T2DM risk factors. Efforts to identify and control the determinants of adiposity and insulin resistance among white European children from low SEP groups are a key priority for early T2DM prevention.