Reanalysis suggests that the new definition for child obesity has questionable validity in puberty
EDITOR- Cole at al (1) suggest a new definition of child obesity
which they argue is less arbitrary and potentially more international. It
is unclear how pooling data from six non-representative countries and
linking the cut-off to a certain adult cut-off value, which was
arbitrarily defined in itself, serves either of the two purposes greatly.
If anything, it muddies the waters a bit as one of the advantages of using
cut-offs based on one population, however inappropriate, is that it allows
the user to predict the direction of the bias, which becomes harder with a
The authors argue that 'the sensitivity of the curve to the timing of
puberty may affect performance of the cut-off points where puberty is
appreciably delayed, although differences of less than two years are
unlikely to make much difference'. To test this claim on the behalf of the
authors, I have compared the standard definition used in the UK, which was
also provided by Cole, (BMI above the 98th centile of UK reference) (2) to
the new definition using a local data set of 20,802 children, aged 5.5 -
14 years, from Plymouth, UK (3). Overall, there were 5% (n=1035) children
considered obese by the old definition as against 3.0% (n=614) by the new
definition; of these 611 children were classified obese by both the
definitions (kappa statistic=0.73; p<_0.001. most="most" of="of" this="this" difference="difference" in="in" classification="classification" however="however" was="was" manifest="manifest" the="the" pubertal="pubertal" age="age" group="group" see="see" table="table" _1.="_1." p="p"/>
|Age quartiles (years)||BOYS||GIRLS||BOTH SEXES|
|Old def.||New def.||Old def.||New def.||Old def.||New def.|
|5.3-7.1||3.8% (104)||2.2% (60)||3.2% (80)||3.1% (78)||3.5% (184)||2.7% (138)|
|7.1-8.9||5.0% (131)||2.8% (74)||3.8% (97)||3.3% (84)||4.4% (228)||3.0% (158)|
|8.9-11.7||5.7% (154)||2.7% (74)||5.0% (125)||3.2% (83)||5.4% (279)||3.0% (157)|
|11.7-14.1||6.2% (161)||2.1% (53)||7.0% (183)||4.1% (108)||6.6% (344)||3.1% (161)|
|ALL AGES||5.2% (550)||2.5% (261)||4.8% (485)||3.5% (353)||5.0% (1035)||3.0% (614)|
|* All differences significant at 5% level (chi2 test p|
Significant differences in the timing of the pubertal onset between
populations are well known (4). A key determinant of variation in pubertal
timing between populations is socio-economic status and five out of the
six countries (Brazil excluded) have Gross Domestic Products (GDP's) way
above most of the African and Asian countries. Most childhood obesity is
manifest in puberty and it is exactly at this stage that the new
definition is likely to misclassify. Definitions such as these, produced
by respected authorities, often get accepted very quickly without
sufficient debate. Over time, the word 'international' sticks and gives
people the impression of universality; long after the original baseline
population gets forgotten. The authors make the time immemorial
misassumption of regarding children as little adults and trying to extend
the relatively greater homogeneity of adult populations to children.
Inconvenient as it may be, we will just have to learn to live with the
quirks of childhood growth, instead of trying to fudge them in one all
encompassing international definition.
Sanjay Kinra specialist registrar in public health medicine
South & West Devon Health Authority, Dartington TQ9 6JE
1. Cole TJ, Bellizzi MC, Flegal K, Dietz WH. Establishing a standard
definition for child overweight and obesity worldwide: international
survey. BMJ 2000;320:1240-3.
2. Cole TJ, Freeman JV, Preece MA. Body mass index reference curves for
the UK, 1990. Arch Dis Child 1995;73:25-9.
3. Kinra S, Nelder R, Lewendon G. Deprivation and childhood obesity: a
cross-sectional study of 20,973 children in Plymouth, UK. The Journal of
Epidemiology & Community Health (in press).
4. Proos LA. Anthropometry in adolescence - secular trends, adoption,
ethnic and environmental differences. Hormone Res 1993;39 Suppl 3:18-24.
Competing interests: Table 1. Numbers (in brackets) and proportions of children consideredobese by old and new definitions (def.)*