What ethnicity is most obese

Overweight adults

2. Things you need to know

What the data measures

The data shows the percentage of adults (people aged 18 and older) in England who were overweight or living with obesity, by ethnicity.

People were asked for their height and weight. These figures were used to determine their body mass index (BMI). This indicates if a person is overweight or living with obesity.

A BMI of:

• 25 or more is classed as overweight
• 30 or more is classed as living with obesity

Percentages are rounded to 1 decimal point.

Not included in the data

16 and 17 year olds have been removed from this analysis to be consistent with the NHS healthy weight calculator.

How BMI is calculated

The metric formula for working out someones BMI is their weight in kilograms divided by their height in metres squared.

The ethnic groups used in the data

Estimates are shown for 7 ethnic groups:

• Asian
• black
• Chinese
• mixed
• white British
• white other
• 'other'

This means estimates are shown for these groups as a whole. This is because the number of people surveyed was too small to make any reliable conclusions about any of the 19 ethnic groups used in the 2011 Census.

Methodology

Read the detailed methodology document for this data.

The figures are based on survey data. Find out more about:

People often underestimate their weight and overestimate their height. This means their self-reported body mass index (BMI) is known to be lower than it actually is. The data adjusts for this bias by using a formula based on observations from the Health Survey for England over several years, which included both self-reported and clinically-measured BMI figures.

In the data file

Download the data for unrounded data.

Differences in child obesity by ethnic group

Introduction

Childhood obesity is one of the most serious global public health challenges of the 21st century.^{[footnote 1]} Obese children and adolescents are at an increased risk of developing various health problems and are more likely to become obese adults;^{[footnote 2]} there is a clear need for understanding and early prevention.

Analyses of the National Child Measurement Programme (NCMP) data have demonstrated that the obesity prevalence of 4 to 5 year olds and 10 to 11 year olds in England varies by ethnic group.^{[footnote 3] [footnote 4]} It is often assumed that these differences are due to deprivation and other known determinants of child obesity which are sometimes associated with ethnicity. The tables and charts presented here show how obesity prevalence varies by ethnicity after adjustment for other explanatory variables.

Data

The NCMP is an annual programme that measures the height and weight of over 1 million children in Reception (aged 4 to 5 years) and Year 6 (aged 10 to 11 years) in state maintained schools in England. Although the NCMP only covers certain age groups, it includes the majority of children in those year groups. This study uses NCMP data collected in the 2016 to 2017 school year when there was an overall 95% participation rate.^{[footnote 5]}

Assessing the body mass index (BMI, weight/height²) of children is more complicated than for adults because a childs BMI changes as they mature. In England the British 1990 growth reference (UK90) for BMI is commonly used to determine weight status according to a childs age and sex.^{[footnote 6]} A UK90 BMI centile of greater than or equal to 95 is considered obese using population monitoring thresholds and is the definition of obesity used here.

Socioeconomic status was classified using the Index of Multiple Deprivation (IMD) (2015).^{[footnote 7]} Ethnic group was allocated using the NHS ethnicity coding scheme.^{[footnote 8]} Numbers of children in many of these detailed ethnic groups are small so they were aggregated to the summary ethnic groups shown in Table 1 for analysis purposes.

Table 1. Summary ethnic groups used in the analysis

Method

Separate multivariable logistic regression models were built for each sex and year group measured by NCMP: Year 6 boys, Year 6 girls, Reception boys, and Reception girls. A binary obesity variable (yes/no) was used as the dependent variable in all models. Two sets of explanatory variables were considered from those available in NCMP data which are known to influence child weight status.^{[footnote 9] [footnote 10]} The first set of explanatory variables includes pupil age in months, quarter of measurement (in the academic year), national deprivation quintile, summary ethnicity, government office region, and the urban or rural status of the lower super output area (LSOA) of the child residence. The second set of explanatory variables includes, in addition, child height. Deprivation and ethnicity were modelled with an interaction term to accommodate possible differences in the socioeconomic gradient of obesity between ethnic groups.

Height was included in the second set of models in order to remove the residual correlation between height and obesity that tends to be associated with measures of BMI in children. Although height can itself be correlated with body fat^{[footnote 11]} it may also reflect pubertal status or other differences between ethnic groups. For example Black African and Black Caribbean children have been found to be on average taller, and South Asians shorter, than White children.^{[footnote 12]}

All models were fitted in R.^{[footnote 13]} The R emmeans package^{[footnote 14]} was used to obtain the marginal predicted probability of obesity by deprivation and ethnicity from each of the fitted models, averaging over the other model variables. The ethnic disparity in obesity was calculated as the contrast (difference) between the predicted probability of obesity in each deprivation quintile and each summary ethnic group compared to White British, with 95% confidence intervals. An adjustment was made to the width of the confidence intervals using the Dunnett method to reflect multiple significance testing.

Results and interpretation

Table 2 gives the number of children in each of the variable categories.

Table 2. Number of children by category in study data

Source: National Child Measurement Programme 2016 to 2017, children with valid height, weight, BMI and home postcode

Predicted ethnic disparities in obesity calculated using each of the eight regression models are given in tables and charts below. As noted the ethnic disparity in obesity compares to White British children in the same deprivation quintile and is highlighted in the tables. A positive value for the disparity indicates that children in the given ethnic group and deprivation quintile have, on average, a higher prevalence of obesity than corresponding White British children. A negative value for the disparity indicates that children in the given ethnic group and deprivation quintile have, on average, a lower prevalence of obesity than corresponding White British children.

Main findings

Ethnicity has an independent effect on obesity prevalence in both Year 6 and Reception boys and girls after pupil age in months, quarter of measurement, national deprivation quintile, height, government office region, and the urban/rural status of the LSOA of the child residence are taken into account.

Ethnic disparities in obesity prevalence are in general greater in Year 6 than in Reception, and there are smaller disparities between the sexes in Reception than in Year 6.

Including height in the models reduces the disparity in predicted obesity prevalence for Black children compared to White children, but has little effect for Asian children.

Controlling for height suggests that previous findings showing that children from Black backgrounds are more likely to be obese^{[footnote 3]} are to some extent due to physical characteristics related to ethnicity, in particular height.

Year 6 Black boys in the most deprived quintile have a predicted obesity prevalence that is on average 2.6 percentage points lower than equivalent White boys after adjustment for other factors including height. In models excluding height Year 6 Black boys in the most deprived quintile have a predicted obesity prevalence that is on average 3.3 percentage points higher than equivalent White boys after adjustment for other factors.

Year 6 Asian boys in the most deprived quintile have a predicted obesity prevalence that is on average 5.2 percentage points higher than equivalent White boys after adjustment for other factors including height. In models excluding height Year 6 Asian boys in the most deprived quintile have a predicted obesity prevalence that is on average 5.5 percentage points higher than equivalent White boys after adjustment for other factors

Year 6 Asian girls in the most deprived quintile have a predicted obesity prevalence that is on average 0.5 percentage points lower than equivalent White girls after adjustment for other factors including height. In models excluding height Year 6 Asian girls in the most deprived quintile have a predicted obesity prevalence that is on average 0.2 percentage points lower than equivalent White girls after adjustment for other factors.

The disparity in predicted obesity prevalence for Year 6 Asian girls compared to equivalent White girls is on average less than 1% across deprivation quintiles and models. This is considerably smaller than the predicted disparity for Year 6 Asian boys compared to equivalent White boys.

The disparity in predicted obesity prevalence for Year 6 Black girls compared to equivalent White girls is greater than the predicted disparity in obesity prevalence for Year 6 Black boys compared to equivalent White boys. This is the opposite of the pattern seen in Asian children.

Year 6 Black girls in the most deprived quintile have a predicted obesity prevalence that is on average 0.2 percentage points higher than equivalent White girls after adjustment for other factors including height. In models excluding height Year 6 Black girls in the most deprived quintile have a predicted obesity prevalence that is on average 7.0 percentage points higher than equivalent White girls after adjustment for other factors.

For Black children, the adjusted disparities in obesity prevalence tend to increase with decreasing deprivation, despite overall obesity prevalence declining as deprivation declines, it does not decline as rapidly in Black children as White children. There is no clear pattern in disparity by deprivation decile in Asian children.

Limitations and recommendation for future research

As the data used comprised only one year of NCMP, the number of children from Asian and Black ethnic groups was not large enough to allow analysis by detailed ethnic group. The summary ethnic groups considered here, therefore, each includes a number of constituent groups which may have different sociodemographic and physical characteristics. This is also the case for the other groups included, for example, children from the Chinese ethnic group were included in Other.

BMI acts as a proxy for lean and fat mass but does not reflect body composition or fat distribution. Inter-ethnic differences in body composition are a confounding factor in comparing obesity levels between ethnic groups: BMI has been found to underestimate body fat in South Asian children and over-estimate body fat in Black African children.^{[footnote 15]} Nevertheless, BMI is non-invasive, quick, easy, cost-effective and very useful as a population measure, such as in the NCMP. It provides consistency in child weight status measurement over time, enabling reliable monitoring of trends and remains the most commonly used, widely accepted, and practical measure of obesity in both children and adults.^{[footnote 16]}

Further work is required to identify ethnic disparities in obesity using more specific ethnic groups than was possible here, potentially combining more years of NCMP data. It would also be informative to examine ethnic disparities in other weight categories such as severe obesity or overweight including obese.

Results tables and charts for models not including height

Year 6 boys

Table 3. Adjusted difference in obesity prevalence by ethnicity and deprivation quintile: Year 6 boys

Year 6 girls

Table 4. Adjusted difference in obesity prevalence by ethnicity and deprivation quintile: Year 6 girls

Reception boys

Table 5. Adjusted difference in obesity prevalence by ethnicity and deprivation quintile: Reception boys

Reception girls

Table 6. Adjusted difference in obesity prevalence by ethnicity and deprivation quintile: Reception girls

Results tables and charts for models including height

Year 6 boys

Table 7. Adjusted difference in obesity prevalence by ethnicity and deprivation quintile: Year 6 boys

Year 6 girls

Table 8. Adjusted difference in obesity prevalence by ethnicity and deprivation quintile: Year 6 girls

Reception boys

Table 9. Adjusted difference in obesity prevalence by ethnicity and deprivation quintile: Reception boys

Reception girls

Table 10. Adjusted difference in obesity prevalence by ethnicity and deprivation quintile: Reception girls

References