To provide meaningful advice for the prevention of childhood obesity, studies are needed to assess obesity prevalence and to investigate risk factors for obesity. Such research requires international consensus on the definition of childhood obesity, in order to form a global perspective of the epidemic. Much progress has been made since researchers first started to investigate childhood obesity. Originally, the definition of obesity was derived from the corresponding definition of underweight, and since then, a variety of growth indices, cut-offs and weight-for-age or weight-for-height references have been proposed. A step towards harmonization occurred when the body mass index (BMI) was first applied to children  and the first BMI reference curves were published in the early 1980s . The earliest BMI reference curves were constructed using national samples with statistically based centile cut-offs. BMI has since emerged as the most practical, widely available and inexpensive indicator for classifying overweight and obesity in children.
In 2000, the International Obesity Task Force (IOTF) proposed an international definition  for childhood obesity based on representative data from six countries, and cut-offs were constructed in reference to World Health Organization (WHO) health-related adult obesity cut-offs  rather than based on centile distributions.
The WHO growth standards, released in 2006 for assessing the growth of children from birth to 5 years of age, were created from samples of healthy breastfed children from six countries around the world , and were intended to represent a ‘standard’ of physiological growth rather than a descriptive ‘reference’. To extend these growth curves to school-age children and adolescence, references for 5- to 19-year-olds were developed using data from US surveys . In addition, various other countries have constructed their own references, primarily for clinical use, as they are more representative of the country's children. The adoption of BMI to define overweight and obesity and the availability of BMI reference curves both represent progress, but the multiplicity of references, and hence definitions, makes it difficult to choose between them .
Thus, introducing yet another definition might be viewed as a further shift away from harmonization. However, the paper by Cole and Lobstein in this issue  does not propose a new definition of obesity – an important point to be stressed. On the contrary, the authors propose several simplifications and improvements to the existing IOTF  and thinness references . Reference values for overweight, obesity and thinness are now presented together. In addition, the article provides L (lambda), M (mu) and S (sigma) values allowing standard deviation (SD) scores to be calculated. A different approach as compared to the previous papers [3, 9] was used to build the new cut-offs, which involved averaging the LMS curves for the six datasets rather than averaging their cut-offs. Differences between the original and revised cut-offs appear, but they are so small that they cannot really impact on estimates of overweight or obesity prevalence. For example, the prevalence of obesity in 2- to 17.9-year-old boys from the NHANES 2005/6 is 10.2% using both previous IOTF and new definitions, while it is respectively 12.1% and 12.3% in girls. Differences are slightly larger for thinness in the youngest age groups. Thus, prevalence data based on the new references can be compared with previous data based on IOTF  and Cole et al.  definitions.
There are some limitations of the original IOTF cut-offs : they provided only limited centile ranges, they did not allow calculation of SD scores and there was no high cut-off defining morbid obesity. These limitations are resolved in this new article , but other limitations remain. The references start at 2 years of age and refer only to BMI, with no references for weight, height or skin-fold measurements.
All moves towards harmonization are good news, but there is still a long way to go to reach an internationally accepted definition. Even the most optimistic observers do not expect the adoption of a common international reference in the near future. As a compromise given the availability of numerous definitions, the European Childhood Obesity Group recently proposed  that prevalence data should be published using several of the main references (IOTF, WHO and additionally Centers for Disease Control and Prevention [CDC] and national references), thus allowing more opportunities for comparison between studies.
Other IOTF cut-off tables
A new and extended set of cut-offs is now available to download free at the website of the International Association for the Study of Obesity.
The tables provide international child cut-offs corresponding to the following body mass index (BMI) values at 18 years:
- 16 – thinness grade 3
- 17 – thinness grade 2
- 18.5 – thinness grade 1
- 23 – overweight (unofficial Asian cut-off)
- 25 – overweight
- 27 – obesity (unofficial Asian cut-off)
- 30 – obesity
- 35 – morbid obesity
The tables provide cut-offs at 1-month intervals from age 2 years to age 18 years, for boys and girls separately.
The values are derived from the methods described in the paper ‘Extended international (IOTF) body mass index cut-offs for thinness, overweight and obesity’ by Tim Cole and Tim Lobstein, in this issue of Pediatric Obesity.
Download the tables in Excel or PDF format from the following site:
Another important problem persists, and this could be solved more easily by consensus. BMI cut-offs to assess nutritional status have three main levels and four BMI categories, globally termed thin, normal, overweight (not obese) and obese. However, the use of the term overweight is often ambiguous as it refers to different BMI ranges [7, 10]. In adults, according to WHO , ‘overweight’ includes grade 1 (25 ≤ BMI<30) and grade 2 (BMI≥30) overweight. In children , grade 1 corresponds to ‘overweight ‘ for CDC, to ‘overweight excluding obesity’ for IOTF and WHO (5–19 years) and to ‘at risk of overweight’ for WHO (0–5 years). Grade 2 overweight corresponds to ‘overweight’ for WHO (0–5 years) and to ‘obesity’ for the other definitions. So why should not we agree to simplify the language and use the common terminology ‘grade 1’ and ‘grade 2’ overweight in children? In a second stage, terms like ‘at risk of overweight’ etc could be used in particular cases. This would be particularly useful in the clinician's office where the terminology could be adapted to the age of the child and possibly other health or familial parameters, avoiding judgment or stigmatization . The universal use of grades 1 and 2 overweight would improve communication, provide clarity to the different definitions and be particularly useful for international comparisons.
Simplification of definitions is important, but the main problem is to find out the definition that best identifies good health and later risk of pathology. WHO responds partially to this issue by selecting healthy breastfed infants in different parts of the world, representing optimal growth, but longer follow-up through all periods of growth should be needed. More evidence on the association between child BMI level and later health would help improve the choice of cut-offs to define childhood obesity, but efforts to reach international consensus are still needed.