Limitations

For various reasons, 42.5% of respondents who participated in the 2004 CCHS did not have their weight and height directly measured. This level of non-response might bias estimates if these people differed systematically from those for whom measurements were obtained. A special technique was employed to reduce the possibility of such bias (see Data sources and analytical techniques).

Reasons for not measuring a respondent’s height and weight were: refusal (13.6%); measuring equipment unavailable (9.0%); too tall for interviewer to measure (7.1%); interview conducted by telephone (4.5%); interview setting a problem (3.5%); respondent’s physical condition (1.8%); and other (3.0%). Overall, men had a lower response rate than women: 54% versus 61%.

Percentage distribution of respondents, by provision of directly measured data and reason for non-response

Men’s response rates differed significantly by age, province, and household income; women’s, by province, fruit and vegetable consumption, marital status, and household income.

Response rates to directly measured height and weight, by selected characteristics

Although body mass index (BMI) classifications using data from the 1978/79 Canada Health Survey (CHS), the 1986 to 1992 Canadian Heart Health Surveys (CHHS), the 1999-2002 National Health and Nutrition Examination Survey (NHANES) and the 2004 Canadian Community Health Survey (CCHS) were based on directly measured height and weight, the surveys did not have the same collection methods. For example, for the 2004 CCHS, field interviewers used portable electronic scales to weigh respondents in their homes; for the 1999-2002 NHANES, health professionals measured respondents in mobile laboratories.

BMI has a number of limitations. It does not measure the distribution of body fat, which is important because excess fat in the abdominal areas is associated with increased health risks.⁵ BMI may misclassify young adults who have not reached full growth, people who are naturally very lean or very muscular, people who are very tall or very short, and certain ethnic or racial groups.¹³

BMI should not calculated for pregnant women.⁵ However, pregnancy status was not asked in the 1978/79 CHS, the 1985 and 1990 Health Promotion Surveys, and the 1986 to 1992 Canada Heart Health Surveys.

In the 2004 CCHS, variables other than height and weight were self-reported. The degree to which these variables accurately reflect a person’s health status/characteristics (fruit and vegetable consumption, presence of a chronic condition) is not known.

Respondents were asked about their leisure-time physical activities over the past three months. The results may have been affected by recall problems. Moreover, because physical activity at school and work were excluded, leisure time may not reflect overall physical activity.

The questions on fruit and vegetable consumption pertain to the number of times a day fruit and vegetables are consumed, not the amounts consumed. Because portion size is not specified, compliance with daily intake recommendations, such as the Canada Food Guide, cannot be assessed.

Rather than weight itself, factors associated with weight such as physical activity, body composition, visceral adiposity, physical fitness or dietary intake might be responsible for some or all of the associations of weight with high blood pressure, diabetes and heart disease.²⁷ As well, some diseases cause weight loss, while others are associated with weight gain. This analysis does not take into account recent weight gain/loss, which may be independently associated with poor health.

Type 1, type 2 and gestational diabetes cannot be differentiated in this analysis. Since the risk factors for the various forms of the disease are not the same, the strength of the relationship between BMI and the prevalence of “adult onset” diabetes (type 2) may have been diluted.

Because the CCHS is cross-sectional, no causality between obesity and a health behaviour or outcome can be inferred.