Obesity is a significant public health problem in Canada. Increasingly, factors that are beyond the individual are being identified as important drivers of the current obesity epidemic. Drivers like prices, taxes, and government policy are jointly identified as population-level determinants of obesity that have resulted in its rapid growth. While these population-level determinants have been identified as important, their impact has not been explicitly quantified in Canada. The geographic distribution of obesity prevalence in Canada is one issue that could potentially highlight the important role that population-level determinants of obesity play.
This dissertation’s objective is to assess the importance of population-level determinants of obesity in Canada by quantifying their impact on individuals living in different regions of the country. Canada has, roughly, an east to west gradient of obesity, with the Atlantic provinces exhibiting the highest prevalence of obesity. I characterize the difference between the Atlantic provinces and other regions of Canada (Quebec, Ontario, the Prairies, and British Columbia) in two ways: the difference in average body mass index (BMI) and the difference in BMI distributions. To estimate the contribution of the population-level determinants to these differences I apply Blinder-Oaxaca decompositions and quantile regression to national level data from the Canadian Community Health Survey. I show that the population-level determinants are important in describing cross-regional differences in obesity in Canada and their importance becomes larger at high percentiles of the BMI distribution, especially in females. I explain how this is consistent with the ecological model of obesity’s portrayal of the population-level determinants of obesity.
Parallel to meeting the overall objective of this dissertation, I assess the added value of corrected BMI values in obesity research. Correction equations are generally used to adjust self-
reported BMI values so they resemble measured BMI values on aggregate. I assess their usefulness by establishing a new correction equation and comparing that correction to established Canadian correction equations, measured BMI, and self-reported BMI. I determine that corrected BMI is not always superior to self-reported BMI and discuss the settings where corrected BMI is useful.