‘Children from middle-class families are generally fatter than their poorer counterparts’, The Independent reports.
The news, covered by much of the media, is based on research reconfirming the fact that obesity is highly prevalent among children across all backgrounds. However, this large and well- designed study produced results that seem to contradict most previous evidence, by finding that childhood obesity was not most prevalent among more ‘deprived’ children.
The three-year study, based in Leeds, found that children aged 11-12 years were more likely to be obese if they came from small areas classed as being around the average on a deprivation rating. Levels of obesity were high in all groups, but the children from the most income deprived and least income deprived areas were less likely to be obese than those in the middle.
On its own, this research is not strong enough to prove that the current way of thinking about the link between deprivation and obesity is wrong. However, it does give pause for thought, and highlights the need for further, more focussed research.
Where did the story come from?
The study was carried out by researchers from Leeds Metropolitan University in the UK. No explicit funding source was described, but the authors declared no conflicts of interest.
The study was published in the peer-reviewed International Journal of Obesity.
The media generally reported the research accurately, with most highlighting that the results challenge the currently held view (based on previous evidence) that obesity levels rise in line with higher levels of deprivation in a relatively linear fashion.
What kind of research was this?
This was a cross-sectional study aiming to investigate associations between a measure of area-level deprivation and three measures of fatness in children:
- body mass index (BMI)
- waist circumference
- waist-to-height ratio
The researchers acknowledge that past findings indicate the prevalence of obesity is highest in more deprived groups. However, they point out a limitation in this evidence base, as it relies on individual levels of deprivation (such as household income), rather than area-level deprivation (the proportion of households in a local area or community that are above or below a threshold for household income).
Researchers wanted to know if the way deprivation was measured influenced the true link between childhood obesity and deprivation. Furthermore, they indicated most previous research has used BMI to measure fatness in children. Other measures of ‘fatness’, such as waist circumference or waist-to-height ratio, may provide a better understanding of the relationship between deprivation and obesity.
A cross-sectional study is a good way of establishing obesity levels at a particular point in time. However, as information is collected at just one point in time, it cannot categorically prove that deprivation causes obesity, only that the two are linked in some way.
Understanding the link between childhood obesity and deprivation is important from a public health perspective. For example, if obesity is found to be linked to different levels of deprivation, local health promotion resources can be targeted proportionately at the groups most in need.
What did the research involve?
Data for this study was obtained from the Rugby League and Athletics Development Scheme (RADS) over a three-year period from 2005 to 2007.
RADS was described by the researchers as a collaboration between Leeds City Council, Leeds Metropolitan University and the Education Authority (Education Leeds), and was set up to identify talented children who were then offered a place on a talent development programme. It was also set up to monitor obesity levels in the city.
The programme involved a series of basic ﬁtness assessments and physical measurements of all year-seven (age 11 years) children from 37 secondary schools in Leeds who agreed to participate. Response rates for the programme were consistently above 80% of pupils. Researchers also measured children’s height, weight and waist circumference while they were at school.
The researchers converted the children’s weight and height measurements to BMI scores. Using standardised reference charts, children were classified as obese if their BMI fell outside the normal range, that is, outside the range you would expect 95 out of 100 children to be in if they had a normal BMI. A waist-to-height ratio of 0.5 was used to “define increased concern”.
Childhood deprivation for each area was estimated by assigning a standard childhood measure of deprivation (the Income Deprivation Affecting Children Index or IDACI) to the geographical areas where the child lived. The geographical areas used are called lower super output areas (LSOAs). These are areas generated to establish similar sized groups (around 1,000 to 1,500 people in each) that share similar house types and tenures (for example, whether the properties are rented, owned, or are occupied by council tenants).
The IDACI score is the proportion of children (aged 0–16 years) in each LSOA who live in income deprived households. ‘Income deprived’ was defined as a household receiving at least one of the following benefits:
- Income Support
- Jobseekers Allowance
- Working Families Tax Credit
- Disabled Persons Tax Credit
The statistical analysis used by the researchers was appropriate and compared whether children were obese or not to levels of deprivation in their area.
What were the basic results?
Over the three years, 15,841 children took part in the study programme. Researchers excluded from their analysis information on any children they did not have complete data for. The final analysis used data from 13,333 children with measures of BMI, and 13,133 children with measures of waist circumference or waist-to-height ratio from 37 schools in 542 LSOAs. The researchers found that:
- There were considerable differences in the prevalence of obesity estimated in the same children using the three different measures of fatness. Combining the three years of data, the obesity levels were 18.6% (using BMI), 26.8% (using waist circumference) and 18.5% (using waist-to-height ratio).
- No statistically significant linear relationship (a straight line on a graph) between area-level deprivation and obesity was found. That is, no direct link was found between higher deprivation levels and higher obesity levels. Even so, a small (non-significant) linear pattern was observed and appeared stronger in girls than boys.
- There was a non-linear pattern (that is, not a straight line but more of a curve on a graph) between area-level deprivation and obesity across all three measures of fatness.
- For all measures of obesity, the highest probability of being obese was around the middle of the income deprivation range (termed ‘middle class’ in the papers). Those in the highest and lowest areas of deprivation were less likely to be obese. The obesity-deprivation relationship varied significantly between boys and girls. The risk of being obese for boys did not peak in the middle deprivation range as much as it did in girls.
- ‘Non-white’ children were more likely to be obese than ‘White-British’ children.
How did the researchers interpret the results?
The researchers concluded there “are inconsistencies between the different measures of obesity” and that “the relationship between obesity and deprivation does not seem to be linear”.
This research describes a non-linear relationship between area-level deprivation (at a local level) and childhood obesity. It suggested those in the middle range of deprivation were the most likely to be obese, more so than children living in the most deprived or least deprived areas.
As the news headlines suggest, this appears to go against other evidence and the commonly held assumption that childhood obesity increases as deprivation increases in a relatively linear (one-to-one) fashion.
The study has important strengths, including its large sample size and use of multiple measures of childhood fatness to estimate obesity, but there are also limitations to consider.
Using lower super output areas
LSOAs are artificial catchment areas and may not always reflect areas important to the individuals living within them. For example, they might divide a housing estate or other significant community boundary that influences deprivation. So using different geographical areas to define deprivation may influence the results.
Measure of deprivation used
The measure of deprivation (IDACI) is based on a number of things, including income of the households where the child lives. It is possible different results would be found using different measures of deprivation. Ideally, multiple measures of deprivation would be used to assess the influence. The researchers used different measures of fatness and noted how this influenced the estimates of obesity massively. Similar differences may be observed using different measures of deprivation. This could influence the results.
Recruitment to the study
Data for this study came from the Rugby League and Athletics Development Scheme that invovled a series of basic ﬁtness assessments and physical measurements of all year-seven (age 11 years) children from 37 secondary schools in Leeds who agreed to participate. It is possible there was selection bias using data obtained through this scheme. That is, the children in the 37 schools that agreed to participate may be systematically different from the schools that chose not to take part in the study. For instance, the schools electing not to take part may have been in more deprived areas with less facilities or interest in sporting activities, or other class related differences affecting the schools' appetite for signing their pupils up for a rugby league development scheme. A different relationship may have been found if all schools in Leeds participated.
Applicability to rest of UK
The study sample was restricted to children aged 11-12 years living in Leeds. This limits how applicable it is to children of other ages living in other areas of the UK. If the research had recruited children from more age groups, a larger geographical area, or from more ethnically diverse groups, the results may have been different.
Furthermore, if other parts of the country were included, it would have been possible to assess the link across a broader range of income deprivation levels. For instance, the middle income deprivation level (or middle class children as the papers put it) in Leeds might not be the same as in other cities like Oxford or Cambridge and this might influence the link between deprivation and childhood obesity levels found in these places, or across the UK as a whole.
Further research along these lines is needed to confirm whether the relationship observed in this group is also true for the rest of the children in the UK. At present this is unclear.
This research is not strong enough on its own to say that the current way of thinking is wrong, but it does given reason for a pause for thought. As with all good science, evidence challenging established ways of thinking is considered on merit and discussed by experts in the field. More research is needed to confirm or refute this current piece of research before any practical changes can be hinged on its unique findings.
This research questions the orthodoxy that childhood obesity increases in line with increasing deprivation. However, it stops short of proving that this is not the case.
At the same time, it serves to reinforce the fact that obesity is highly prevalent among the population: for rich and poor alike.