Appendix 2: Obesity in Canada – Updated economic burden of obesity analysis – Summary of methodology

Appendix 2. Updated economic burden of obesity analysis: summary of methodology

I. Janssen, for PHAC

The economic costs of obesity by year (2000 to 2008) were estimated using a prevalence-based approach that included the following:

• risks of chronic conditions in obese individuals;
• population prevalence of obesity; and
• direct and indirect costs associated with these (specific) chronic conditions.

Calculation of risk of chronic conditions in obese individuals

Risk estimates for the main chronic diseases associated with obesity in men and women were obtained from a 2004 meta-analysis by Katzmarzyk and Janssen,^{Footnote 202} updated to include studies published within the past five years. Eight chronic diseases for which obesity has been consistently shown to be a risk factor were included in the meta-analysis: coronary artery disease, stroke, hypertension, colon cancer, post-menopausal breast cancer, type 2 diabetes, gall bladder disease and osteoarthritis. For each disease, summary relative risk estimates were calculated separately for men and women using a general variance-based method. These summary relative risk estimates represent a weighted average of the relative risk provided in the various studies.

Estimation of the population prevalence of obesity

The methodology used for estimating the prevalence of obesity among Canadian adults depended on the survey year. For 2004, 2005 and 2008, the prevalence was based directly on results from cycles of the CCHS, in which height and weight values used to calculate BMI were directly measured. For the remaining years (2000, 2001, 2002, 2003, 2006 and 2007), the prevalence of obesity based on self-reported height and weight was obtained from nationally representative surveys (cycles of the NPHS and the CCHS conducted in 1994/95, 1996/97, 2000/01, 2003, 2005 and 2007). These prevalence values and corresponding survey years were used to create simple linear regression equations by sex to predict the prevalence of self-reported obesity in 2000, 2001, 2002 and 2006. Next, for those years in which the prevalence of measured obesity was not available, the true prevalence was estimated on the basis of the relative difference in self-reported and measured obesity in the 2005 CCHS (34.2% relative underestimation for men and 35.0% for women).

Determination of population attributable risk (PAR)

The second step was to determine what proportion (or fraction) of each of the eight chronic conditions can be causally attributed to obesity within the adult population in Canada. The PAR combines the summary relative risk (RR) with the population prevalence (P) of obesity:

PAR% - [P(RR-1)]/[1+P(RR-1)]

Determination of direct and indirect costs associated with these chronic conditions

PAR% values were then applied to the total direct and indirect costs for each of the eight target conditions. Costs were based on information in the Economic Burden of Illness in Canada (EBIC) 2000 study and are estimated for the population aged 15 years or older. Direct costs are defined as the value of goods and services for treatment, care and rehabilitation related to the condition, such as hospital care, drug and physician care expenditures, expenditures for care in other institutions and additional direct health expenditures. Indirect costs are defined as the value of economic output lost because of illness, disability or premature death. The indirect costs in the EBIC 2000 were measured in terms of the value of years of life lost due to premature death and the value of activity days lost due to short-term and long-term disability. At the time that the report was prepared, EBIC 2000 had yet to be publicly released. Instead, data were made available from the Population Health Economic Section, Knowledge Information and Data Systems, Office of Public Health Practice, Public Health Agency of Canada.

For the years 2001-2008, the direct health care costs calculated for each chronic condition were inflated to current dollars by using the percentage increase in health care costs in the Consumer Price Index in Canada from the year 2000 to the year of interest. In inflating these values, it was assumed that each disease made up a similar percentage of total health care expenditures throughout that period. The indirect health care costs were inflated to 2001-2008 values using the percentage increase in average earnings in Canada from 2000.