Appendix 1: Obesity in Canada – Estimating the prevalence of obesity – methodology and additional 2007/08 CCHS analyses
Appendix 1. Estimating the prevalence of obesity: methodology and additional 2007/08 CCHS analyses
Using the Canadian Community Heath Survey (CCHS), new analyses presented in this report estimate the prevalence of obesity in Canadian populations stratified by age, sex, geographic location, and income and education levels. Data from the 2007 and 2008 CCHS share files, both separately and combined, were the focus of the new analyses. Data from previous cycles, covering the years 2000/2001, 2003, 2004 and 2005, were also used. Detailed descriptions of the variables used, analyses conducted and resulting limitations are outlined in this appendix.
Canadian Community Health Survey
The target population of the CCHS is all Canadians aged 12 and over (for the 2004 CCHS Nutrition Cycle (2.2), the target population was extended to all age groupsFootnote *). People living on Indian Reserves and Crown Lands, institutional residents, full-time members of the Canadian Forces and residents of certain remote regions were not included in the survey. The survey population is representative of approximately 98% of the total population in the provinces, 90% in the Yukon, 97% in the Northwest Territories and 71% in Nunavut.Footnote 193
Prior to 2007, core component data were collected every two years over a one-year period. Since 2007, data collection has occurred on an ongoing basis across 12-month collection periods. For both the annual data and two-year combined data (2007/08), sampling and bootstrap weights are provided by Statistics Canada such that resulting weighted estimates are representative of the population in the specified period.Footnote 194 Information on approximate CCHS sample size specific to each collection period is provided in Table 3.
|2000/2001 CCHSFootnote 195||2003 CCHSFootnote 196||2004 CCHSFootnote 197||2005 CCHSFootnote 198||2007 CCHSFootnote 199||2008 CCHSFootnote 200|
|Available survey sample size||130,827||134,072||35,107||132,947||65,946||66,013|
|Self-report BMI class available||86,000 (ages 20-64)||111,000 (18+)||3,200 (2-17) 7,300 (18+)||11000 (12-17) 114,000 (18+)||4,700 (12-17) 57,300 (18+)||4,900 (12-17) 56,900 (18+)|
|Measured BMI class available||Not available||Not available||8,660 (2-17) 11,800 (18+)||480 (12-17) 4,200 (18+)||Not available||400 (12-17) 4,100 (18+)|
Valid responses for BMI are available for part of the sample, as illustrated in Table 4. All prevalence estimates for obesity are based on the total population for which BMI was available; non-responses and ineligible respondents for whom BMI was not calculated (e.g., pregnant women) were excluded from the analyses. In particular, for the combined 2007/08 obesity estimates, over 8,000 persons from the sample of approximately 132,000 are excluded.
Being female was the characteristic of the sample that showed the highest proportion for whom BMI class was not determined (Table 4), and this was due to the exclusion of pregnant women. Otherwise, characteristics were quite consistent between the sample excluded and the remaining sample.
Source: Canadian Community Health Survey 2007/08, Statistics Canada.
Variables used in analyses
In addition to age, sex, province/territory and health region, the following variables were used in the analyses.
Obesity: The BMI is a derived variable calculated by dividing the respondent’s measured body weight (in kilograms) by the square of the respondent’s height (in metres). This calculation is done similarly for self-reported, parental-reported or measured height and weight. Overweight and obese categories for children and youth were developed on the basis of IOTF cut-offs (Cole method)Footnote 23 with specific ranges for children by age and sex. For adults, BMI classes are based on international standards developed by the WHO.Footnote 7
BMI is not calculated for adult respondents with a height of less than 0.91m (3') or more than 2.13m (7'), or for women who either reported being pregnant or did not respond to the question on pregnancy.
Aboriginal Peoples: The variable used in the analyses was based on the question “Are you an Aboriginal person, that is, North American Indian, Métis, or Inuit?”Footnote 193 Footnote 198 Footnote 199
Income Deciles: This variable was derived by Statistics Canada and is based on self-reported household income before taxes. Ten categories with approximately the same percentage of respondents in each group were generated by province/territory according to the ratio of household income to the low income cut-off corresponding to the respondent's household and community size.
Household Educational Attainment: This variable was derived by Statistics Canada and represents the highest level of education attained by any member of the household.
Methods of analysis
Descriptive analyses were used to estimate the prevalence of obesity across different population subgroups by age, sex, education, income decile, Aboriginal identity, province/territory and health region. Corresponding sample sizes by CCHS year are available in Table 3, which highlights the number of respondents with a valid BMI classification. Non-overlapping 95% confidence intervals were considered as reflecting significantly different point estimates of obesity prevalence. Bootstrapping techniques were used in these analyses to generate confidence intervals, as this technique takes the complex survey design into account. Thus, more accurate estimates of the variability of prevalence values were provided.
New analyses presented in the report have various limitations, depending on the population of subgroups used and the data cycle considered. One caveat when comparing obesity between Aboriginal and non-Aboriginal populations is that the CCHS does not include respondents who reside on reserves or in some remote communities. Therefore, there is no single obesity estimate that includes all Aboriginal peoples.
In addition, the estimates of obesity shown are not age-standardized, and therefore the age differences in obesity among groups or over time are not accounted for. Table 5 illustrates one example of the difference in age-standardized and non-age-standardized obesity prevalence. For the total population, using the four age groups specified to age-standardize the obesity estimate has a very minor impact, of less than half a percent, on the resulting estimate. However, for subgroups with different age and sex distributions, age and sex standardization might be warranted. The focus in this report was to obtain current prevalence estimates.
Analysis was limited to those measures captured by the CCHS. For example, nutrition was captured primarily by fruits and vegetables consumption only, and only in terms of number of times per day and not actual servings.
Source: Analysis of Canadian Community Health Survey 2007/08, Statistics Canada.
As discussed earlier in the report, there are limitations and critiques of the BMI classification system. As noted in Health Canada’s Canadian Guidelines for Body Weight Classification in Adults,Footnote 7 particular caution should be used when classifying people who are very lean or very muscular, some ethnic and racial groups, and seniors.Footnote 201 In addition, research has shown that women are more likely to underestimate their weight and men more likely to overestimate their heights, both of which would result in a more conservative estimate of BMI.Footnote 4
Finally, obesity estimates based on directly measured heights and weights could be calculated for a large representative population for the 2004 CCHS, but only for subsamples in the 2005 and 2008 CCHS. The lack of routine, national, measured obesity estimates has been a noted surveillance information gap. All other obesity estimates are calculated according to self-reported height and weight information which, as discussed earlier in this report, has been shown to underestimate BMI and therefore obesity.
Additional analyses: 2007/08 CCHS
Figure 5 illustrates the prevalence of obesity (self-reported data) in the top and bottom 10 ranked health regions. Obesity prevalence estimates for all health regions in Canada are provided below (Table 6).
Source: Analysis of Canadian Community Health Survey 2007/08, Statistics Canada.
Report a problem or mistake on this page
- Date modified: