Having two different reference charts with different

Having two different reference charts with different indications for use, has the potential
to cause confusion.
Some authors have found the Cole charts to be inferior to both the CDC and locally
derived reference charts in terms of sensitivity for identifying obesity, but comparable
when identifying overweight (Fu et al 2003; Reilly et al 2000; Reilly 2002). They
recommend using population-specific BMI cut-offs for screening, epidemiological and
clinical purposes, and international cut-offs for international comparisons only. Given
that New Zealand has no population-specific reference data or cut-offs, using the
international charts by Cole et al seems the best available option, while recognising
their limitations in application to certain ethnic groups and the reduced sensitivity of
identifying obesity. The sensitivity for identifying overweight using the Cole charts is
high (over 90%) while retaining acceptable specificity.
As a population indicator, BMI does not need to be a precise measure (high sensitivity
and specificity) at an individual level. However, population BMI collection will allow
trends to be followed, both in terms of mean BMI and its distribution and the prevalence
of overweight and obesity for the population and within subgroups such as ethnic
groups or communities. It would be a useful tool to detect differences between groups
concurrently or over time. By identifying trends in the whole population or subgroups of
the population, targeted interventions and forecasting for future service needs will be
possible. The usefulness of collecting population BMI will be further discussed in
Section 5.
22
An Analysis of the Usefulness and Feasibility of a
Population Indicator of Childhood Obesity

2 Timing of BMI Collection
When assessing the best age to collect BMI at a population level the following issues
need to be considered:

accessibility to a health professional

prevalence of obesity in the age group

ability to measure change

minimisation of potential harm

predictability of obesity in adulthood

the presence of modifiable lifestyle factors

potential responsiveness of children or young people and their families to lifestyle
and other interventions.
Of the above factors, the most relevant to deciding the best age for collecting BMI for
the purpose of population monitoring are the first four: accessibility, prevalence of
obesity, the consequent increased ability to measure change, and minimising potential
harm. The final three factors are more relevant to screening, which is not the purpose
of this proposal. They would be highly relevant issues if we were screening to identify
significantly obese children for intervention (see Appendix 2).
To analyse the optimal age to collect BMI, the age brackets of five to six years, seven
to nine years, 10 to 11 years and adolescents (12 years and over) have been used.