"It would suggest that any surveys that rely on self-reported height and weight will underestimate true obesity prevalence unless adjusted."
The researchers sought to establish if self-reported height bias or weight bias, or both, was responsible for the declining sensitivity in the obese category between self-reported and clinically measured BMI.
They found self-reported height bias had remained stable over time, but self-reported weight bias had increased for both genders and in all age groups, particularly the obese.
"This latest research supports the notion that as the population continues to gain weight, under-reporting becomes more of an issue," said Tim Crowe, Associate Professor in nutrition at Deakin University.
"The same finding is also seen in under-reporting of types of foods eaten, especially foods high in fat and sugar, and the serving sizes of meals," Professor Crowe said.
The research authors said the results must be cautiously interpreted, as the study numbers used in two of the sample groups included less than 200 people.
Professor Crowe added that BMI itself was far from perfect, with waist circumference a more useful measure of overall health risks.
"Waist circumference specifically targets the most dangerous type of fat - visceral fat," he said.
"Measures above 102cm for men and 88cm for women carry a very high risk of developing type 2 diabetes, high blood pressure, cardiovascular disease and even some forms of cancer."
Professor Proietto said underwater weighing, DEXA scanning, and Bod Pod assessment were more accurate than BMI, however none were feasible in large population studies.
"BMI is simple and cheap and useful, provided one is aware of its limitations."
Read the full study here.
http://theconversation.edu.au/