So while it is true that the prevalence of obesity in two to five year olds declined by an estimated 39.6 percent between 2003-04 and 2011-12, from 13.9 percent to 8.4 percent, it is also true that, according to the same NHANES data, it declined by 23 percent between 2003-04 and 2005-06, and then rose by 19.8 percent between 2007-08 and 2009-10.
The most plausible explanation for these large oscillations can be found in the definition of “childhood obesity” itself. “Childhood obesity” is a brand-new concept, invented a few years ago for essentially political reasons. The official definition is that children who are at or above what was the ninety-fifth percentile of BMI for age in growth charts from the 1960s and 1970s are now classified as “obese.”
It is important to understand two things about this definition. First, it is quite literally arbitrary: it isn’t based on any epidemiological observation that this or that bad outcome is seen among children who are above this particular definitional cut point. Public health authorities needed a definition for their crusade against a newly invented menace, so they made one up.