Problems with the Fraser report Chapter 1: Pitfalls in BMI time trend analysis

The first chapter of the Fraser report "Obesity in Canada: Overstated Problems, Misguided Policy Solutions" presents a flawed and misleading analysis of BMI time trends. The objective of this commentary is to provide a tutorial on BMI time trend analysis through the examination of these flaws. Three issues are discussed: 1. Spotting regions of confidence interval overlap is a statistically flawed method of assessing trend; regression methods which measure the behaviour of the data as a whole are preferred. 2. Temporal stability in overweight (25≤BMI30) prevalence must be interpreted in the context of the underlying population BMI distribution. 3. BMI is considered reliable for tracking population-level weight trends due to its high correlation with body fat percentage. BMI-defined obesity prevalence represents a conservative underestimate of the population at risk. The findings of the Fraser report Chapter 1 are either refuted or substantially mitigated once the above issues are accounted for, and we do not find that the 'Canadian situation largely lacks a disconcerting or negative trend', as claimed. It is hoped that this commentary will help guide public health professionals who need to interpret, or wish to perform their own, time trend analyses of BMI.
Authors (Zotero)
Lo, Ernest
Date (Zotero)
Noviembre, 2014