Article Text


Prevalence of severe childhood obesity in England: 2006–2013
  1. Louisa J Ells1,2,
  2. Caroline Hancock2,
  3. Vicky R Copley2,
  4. Emma Mead1,
  5. Hywell Dinsdale2,
  6. Sanjay Kinra3,
  7. Russell M Viner4,
  8. Harry Rutter5
  1. 1Health and Social Care Institute, Teesside University, Middlesbrough, UK
  2. 2Public Health England, Oxford, UK
  3. 3Department of Non Communicable Disease Epidemiology, London School of Hygiene and Tropical Medicine, London, UK
  4. 4Institute of Child Health, University College London, London, UK
  5. 5ECOHOST, Department of Health Services Research and Policy, London School of Hygiene and Tropical Medicine, London, UK
  1. Correspondence to Dr Louisa J Ells, Health and Social Care Institute, Teesside University, Borough Road, Middlesbrough TS1 3BA, UK; l.ells{at}


Background International evidence shows that severe paediatric obesity results in an increased risk of ill health and may require specialised weight management strategies, yet there remains a lack of data on the extent of the problem.

Objective To examine the prevalence of severe obesity in children aged 4–5 and 10–11 years, attending English schools between 2006/2007 and 2012/2013.

Design A retrospective analysis of National Child Measurement Programme (NCMP) data.

Setting Maintained schools in England.

Participants All children aged 4–5 and 10–11 years included in the NCMP dataset.

Main outcome measures Prevalence of severe childhood obesity, defined using the 99.6th centile of the British 1990 (UK90) growth reference for body mass index (BMI), analysed by sex, geography, ethnic group and deprivation.

Results The key findings show that in 2012/2013, severe obesity (BMI ≥UK90 99.6th centile) was found in 1.9% of girls and 2.3% of boys aged 4–5 years, and 2.9% of girls and 3.9% of boys aged 10–11 years. Severe obesity prevalence varies geographically and is more prevalent in children from deprived areas, and among those from black ethnic groups.

Conclusions The findings from this study should help to raise awareness of the prevalence of severe obesity and support the provision of adequate treatment and prevention services both to support children who are already severely obese and reduce the prevalence of extreme weight in the future.

  • Data Collection
  • Obesity

Statistics from

What is already known on this topic

  • Severe paediatric obesity is associated with a number of serious immediate and long-term health problems.

  • Prevalence of severe paediatric obesity has increased in the USA over time.

  • Data from the USA demonstrate inequalities in the prevalence of severe paediatric obesity, with significant sociodemographic variation.

What this study adds

  • This is the first detailed analysis of severe paediatric obesity prevalence in English school children.

  • 1.9% of girls and 2.3% of boys aged 4–5; and 2.9% of girls and 3.9% of boys aged 10–11 were classified as severely obese (falling on or above the 99.6th centile of the UK90 growth charts) in England in 2012/2013.

  • In England, prevalence of severe obesity varies significantly across the country and is highest in children living in the most deprived areas, and from Black ethnic groups.


In children, the relation between body mass index (BMI) and adiposity varies with age and sex, so BMI thresholds are usually defined in terms of specific centiles on a growth reference. There is, however, currently no universal definition of severe paediatric obesity. In England, the British 1990 (UK90) growth charts for BMI are frequently used to define paediatric weight status according to age and sex, with the highest centile shown on the charts being the 99.6th.1 The International Obesity Task Force (IOTF) has recently published a definition for ‘morbid obesity’ in children, which they define as corresponding to a BMI value of 35 kg/m2 at age 18.2 In the USA, an expert committee proposed a classification of severe childhood obesity in 2007 as a BMI ≥99th centile of the Centers for Disease Control growth reference.3 This classification has subsequently been challenged and a new classification of BMI ≥120% of the 95% centile proposed.4–6 Using the new definition, analysis of the US National Health and Nutrition Examination Survey 1999–2012 identified an upward trend in the rates of severe obesity over time, with the latest 2011–2012 data showing a prevalence of 5.9% in children aged 2–19 years.7 This study also demonstrated that severe obesity prevalence was highest among older children, and those from Hispanic and non-Hispanic black populations. These data complement an earlier study that also reported an association between severe obesity and the development of hypertension.8 Further US studies show that severe obesity is associated both with increased paediatric cardiovascular risk factors and risk of severe obesity in adulthood.9 ,10 These studies complement similar findings from a Dutch paediatric surveillance survey,11 which found that two out of three severely obese children (defined using the Dutch age-specific and sex-specific cut points corresponding to an adult BMI of 35 or more) have cardiovascular risk factors.

A 2013 scientific statement from the American Heart Association6 highlighted the associated immediate and long-term cardiovascular, metabolic and other health consequences (obstructive sleep apnoea, non-alcoholic fatty liver disease, muscular-skeletal and psychological problems) of severe paediatric obesity, suggesting the need for specialised treatment services. As these children may require a range of interventions, accurate prevalence data are essential to inform the appropriate planning, delivery and commissioning of such services. These data are also valuable for international comparisons. This paper builds on analyses conducted for the recent Chief Medical Officer's report12 and provides the first detailed description of the prevalence of severe obesity in children in England.



To examine the prevalence and trends for severe obesity in children aged 4–5 and 10–11 years, attending English schools between 2006/2007 and 2012/2013.

Study design

A retrospective analysis of the National Child Measurement Programme (NCMP) was performed for school years 2006/2007–2012/2013. The NCMP data were analysed in Microsoft Access, Excel and R, with 95% CI for the prevalence estimates calculated using the Newcombe method.13 When analysing severe obesity prevalence within ethnic groups and at local authority (LA) level, the last three years of data were combined for the LA analysis, and the last four years combined for the ethnic group analysis, to avoid any small number disclosure and improve statistical strength.

Binary logistic regression was used to assess the relation between NCMP year of measurement and Index of Multiple Deprivation (IMD)i decile and the probability of being severely obese. Four separate models were produced, one for each of the school year and sex combinations. The generalised linear models function in the statistical software Rii was used to fit the models. The models are based on the assumption that the log of the odds of being severely obese is linearly related to the two explanatory variables: year of measurement and IMD decile. IMD decile was considered as a categorical variable. The variables were assessed for significance using the χ2 test. Significance in other data presented was assessed using 95% CIs, with significant differences assumed when intervals did not overlap.


NCMP was established in 2006 and annually collects measured height and weight, sex, age, ethnicity and postcode data from all children in reception (ages 4–5 years) and year 6 (ages 10–11 years), from every maintained (state funded) school across England. Data collection at privately funded and special needs schools is not mandatory. Because coverage of data collection in these schools is low, data from them have been excluded from this analysis. A full description of the methods for NCMP has been previously published.14 The programme is coordinated by Public Health England and is used to inform local planning and delivery of children's services, gather population-level surveillance data and increase awareness of weight issues in children.


Participants are school children in the reception year (aged 4–5) and year 6 (aged 10–11). The numbers measured in previous years were n=873 584 in 2006/2007 (80% participation); n=972 479 in 2007/2008 (88% participation); n=1 003 866 in 2008/2009 (90% participation); n=1 026 368 in 2009/2010 (91% participation); n=1 036 608 in 2010/2011 (93% participation); n=1 056 780 in 2011/2012 (93% participation); and n=1 076 824 in 2012/2013 (93% participation).iii

Main outcome measures

The primary outcome of this study was the prevalence of severe childhood obesity, defined using the 99.6th centile of the UK90 BMI growth reference charts (this equates to a BMI at age 18 of approximately 31.93 for boys and 32.6 for girlsiv). Prevalence was also analysed by sex, geography, ethnic group and deprivation (calculated IMD 2010). As commissioners often rely on higher thresholds to define the need for highly specialised service such as bariatric surgery, two further categories for very severe obesity are also presented, based on the 99.87th (which equates to an adult BMI of approximately 34.46 for boys and 35.09 for girls) and 99.98th (which equates to an adult BMI of approximately 39.38 for boys and 39.73 for girls) centiles of the UK90 growth charts. To facilitate international comparisons, IOTF thresholds2 were also applied to these data.


Table 1 presents the first (2006/2007) and last available (2012/2013) years of prevalence figures for English school children who fell on or above the 99.6th centile of the UK90 growth charts and met the new IOTF morbid obesity classification.

Table 1

Prevalence of severe obesity by international and UK90 proposed definitions

Severe obesity prevalence classified using the IOTF definition aligns most closely to the prevalence defined using the 99.87th centile of the UK90 reference. For the remainder of this paper, the 99.6th centile of the UK90 reference will be used to define severe obesity. This represents the highest centile line marked on the growth charts and is thus easily distinguishable by both lay and clinical personnel using the charts with children in the UK.

The latest available measurement year (2012/2013) shows a prevalence of severe obesity of 1.9% (95% CI 1.81% to 1.92%) for girls and 2.3% (95% CI 2.26% to 2.37%) for boys aged 4–5, and 2.9% (95% CI 2.82% to 2.95%) for girls and 3.9% (95% CI 3.86% to 4.01%) for boys aged 10–11. This equates to a total of 12 316 of 4-year-old to 5-year-old and 16 775 of 10-year-old to 11-year-old children with severe obesity at the time of measurement in 2012/2013. In addition to the differences in severe obesity between age and sex, there is also significant variation by geography (figure 1). Using combined data from the last three years, the prevalence of severe obesity across English local authorities varies from 0.7% to 4.3% for the younger age group and 0.4% to 7.5% for the older children. Furthermore, there is also significant variation in severe obesity prevalence among ethnic groups (figure 2). Data from the last four years combined show that the black ethnic group has higher prevalence than white, mixed, Asian, Chinese and other ethnic groups. The highest prevalence levels are seen in 10-year-old to 11-year-old black Caribbean children, while the lowest rates are seen in Chinese girls of both age groups (table 2).

Table 2

Prevalence of severe obesity (≥99.6th centile UK90) by broad ethnic group, school year and sex (including 95% CIs)

Figure 1

Prevalence of severe obesity (≥99.6th centile UK90) by English lower tier local authority and school year. Data source: National Child Measurement Programme 2010/2011–2012/2013 combined. Crown copyright and database rights 2014 Ordnance Survey 100 016 969. UK90, British 1990.

Figure 2

Prevalence of severe obesity (≥99.6th centile UK90) by ethnic group, school year and sex (including 95% CIs) Data source: National Child Measurement Programme 2009/2010–2012/2013 combined. UK90, British 1990

Prevalence of severe obesity by sex, age group and IMD 2010 decile over time is shown in figure 3, with the supporting logistic regression analysis shown in table 3. These results indicate that both year of measurement and deprivation decile are significant predictors of severe obesity. Table 3 shows that after adjustment for IMD decile the odds of being severely obese have on average increased over time for the older girls and boys by 1.02 times for each additional year of measurement. In contrast, the odds of being severely obese have decreased over time for younger girls and boys, with ORs of 0.99 (95% CI 0.99 to 0.998) and 0.98 (95% CI 0.97 to 0.98), respectively. The 10-year-old to 11-year-old girls in the most-deprived IMD decile have odds of being severely obese, which are 4.35 times those in the least-deprived reference category (95% CI 4.13 to 4.58) (table 3). The difference in severe obesity prevalence between the least-deprived and the most-deprived IMD deciles is less marked for the other age and sex groups (older boys, younger girls and boys), with odds of severe obesity of between 3.17 and 3.64 times those in the least-deprived decile. It was not possible to examine yearly trends by ethnic group as ethnicity reporting was poor in the first few years of the programme and several years of data need to be aggregated to avoid small number disclosure.

Table 3

ORs for severe obesity from four logistic regression models for each sex and school year

Figure 3

Trend in severe obesity (≥99.6th centile UK90) prevalence by school year, sex and deprivation decile (including 95% CIs). Data source: National Child Measurement Programme 2006/2007–2012/2013. IMD, Index of Multiple Deprivation; UK90, British 1990.


This study provides the first detailed analysis of the measured prevalence of severe obesity in school children across England. Severely obese children are at risk of developing a number of serious acute and chronic health problems.6 These children therefore pose a significant concern in terms of their health and well-being, and may require the provision of specialist services.

Nearly 30 000 severely obese (≥99.6th UK90 centile) children in the two primary school year groups were identified in 2012/2013. This is an order of magnitude greater than the number of children falling within the very thin category (<0.4th UK90 centile) (in 2012/2013, 2860 (children aged 4–5: 1311, 0.2%; children aged 10–11: 1549, 0.3%) were classified as very thin). While the strength of this study is the large population size and high participation rate, it is worth noting that the presented figures may underestimate the prevalence of severe obesity for two main reasons: (1) a small number of children with severe obesity may also have other conditions, such as Prader–Willi syndrome, which may lead to them attending schools for children with special needs, most of which do not submit measurements to NCMP; and (2) analysis of previous years’ data and extensive anecdotal evidence suggest that heavier children are more likely to opt out of the measurement programme, especially in the older age group. This is not possible to quantify accurately, and it is possible that some of the increases noted in severe obesity over time are a consequence of the commensurate rise in response rate.

When compared with data from the USA,7 English school children showed similar severe obesity prevalence patterns, with higher rates in boys, older children and black ethnic groups. However, the use of different thresholds, reference populations and age groups in the two countries makes direct comparisons difficult.15 Wider use of the new IOTF classification could avoid this problem. A higher prevalence in the younger age group is also observed when the very high 99.98th centile cut point is applied. The reasons for this remain unknown, although it might result from elevated opt-out rates in very overweight older children, who may be more sensitive to the risks of obesity-related stigma than their younger counterparts. While it would be very interesting to formally assess the impact of opt out, unfortunately the data that would be required for this analysis are not available.

Although in relative terms the prevalence of severe obesity remains low, in absolute terms this represents a very large number of children across the country, many of whom might benefit from top tier service provision. This is a particularly important service consideration when over 4000 children in just the two school years measured by NCMP in 2012/2013 had a BMI falling on or above the exceptionally high 99.98th centile. While NCMP only provides data on 2-year groups, simple extrapolation of these findings to all school-age children would suggest that tens of thousands of children may potentially be suffering from severe obesity.

When examining prevalence using the 99.6th centile, the observed higher prevalence in boys and older children echoes the raised prevalence of general obesity in these groups: in 2012/2013, 20.4% of boys and 17.4% of girls aged 10–11 were classified as obese (≥95th centile of the UK90 reference) compared with 9.7% of boys and 8.8% of girls aged 4–5.14 Severe obesity prevalence patterns also mirror the prevalence patterns for general obesity in the correlation with health inequalities. As reported for general obesity,14 severe obesity prevalence varies significantly by geography and is highest in the most-deprived children and those from black ethnic groups, therefore, suggesting a need for the development and evaluation of more targeted interventions.

The findings from this study should help to raise awareness of the prevalence of severe obesity and support the provision of adequate treatment and prevention services both to support children who are already severely obese and reduce the prevalence of extreme weight in the future. However, there remain several gaps that would benefit from future research: (1) further investigation of the role of the disproportionate drivers of severe obesity in deprived and ethnic minority groups; (2) linkage to other health outcome data to assess short-term and long-term health impacts; (3) longitudinal analyses tracking individual children over time to further understanding of the natural history of this condition and identify any potential predictive factors; and (4) exploration of other international data sources using the new IOTF definition of morbid obesity to support international data comparisons.


Data access for this analysis was granted by Public Health England.


View Abstract


  • Contributors This paper was conceived by LJE, EM, RMV, SK and HR. The analyses were carried out by CH, VRC and HD.

  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement Local authority severe obesity prevalence data will be made available at:

  • i

  • ii R Core Team (2014). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL

  • iii NB due to additional data cleaning these figures may differ slightly from figures previously published by the Health and Social Care Information Centre.

  • iv Analysis utilised LMSgrowth program V.2.77. 2012. Huiqi Pan and Tim Cole. Available as a download from

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles

  • Atoms
    R Mark Beattie