Information identified as archived is provided for reference, research or recordkeeping purposes. It is not subject to the Government of Canada Web Standards and has not been altered or updated since it was archived. Please "contact us" to request a format other than those available.
For this article...
This study compares the bias in self-reported height, weight and body mass index (BMI) in the 2008 and 2005 Canadian Community Health Surveys and the 2007 to 2009 Canadian Health Measures Survey. The feasibility of using correction equations to adjust self-reported 2008 Canadian Community Health Survey values to more closely approximate measured values is assessed.
Data and methods
Data are from the 2008 and 2005 Canadian Community Health Surveys and the 2007 to 2009 Canadian Health Measures Survey. In these surveys, respondents reported their height and weight, and were subsequently measured. Regression equations based on the 2007 to 2009 Canadian Health Measures Survey and the 2005 Canadian Community Health Survey were applied to self-reported 2008 Canadian Community Health Survey data. These equations predicted measured BMI based on self-reported BMI..
The bias in reporting height was similar across all three surveys, but the bias in reporting weight was larger in the two Canadian Community Health Surveys, and as a result, discrepancies in estimates of obesity between self-reported and measured values were greater. Application of correction equations based on 2005 Canadian Community Health Survey data to self-reported values in the 2008 Canadian Community Health Survey produced more accurate estimates of obesity than did equations based on Canadian Health Measures Survey data.
Survey context may influence the magnitude of the bias in self-reported weight. Respondents who are aware that they will be weighed may report their weight more accurately. Additional data points are required to determine whether the bias in self-reported measures in the Canadian Community Health Survey is changing.
Body mass index, direct measure, measurement error, misclassification, prevalence, sensitivity, specificity
The health consequences of excess body weight have made obesity a public health challenge throughout the world.1 Accurate monitoring of the prevalence of obesity is critical in the assessment of intervention programs. [Full Text]
Margot Shields (1-613-951-4177; email@example.com) is with the Health Analysis Division at Statistics Canada, Ottawa, Ontario, K1A 0T6. Sarah Connor Gorber is with the Public Health Agency of Canada, Ottawa, Ontario. Ian Janssen is with Queens University, Kingston, Ontario. Mark S. Tremblay is with the Children's Hospital of Eastern Ontario Research Institute and the University of Ottawa, Ottawa, Ontario.
What is already known on this subject?
- Body mass index values based on self-reported height and weight underestimate the true prevalence of obesity.
- For fiscal and logistical reasons, the practice of collecting self-reported height and weight will continue in Statistics Canada's Canadian Community Health Survey.
- Correction equations based on half of the 2005 Canadian Community Health Survey subsample, for whom both measured and self-reported values were collected, were successfully applied to the other half of the sample to produce more accurate estimates of obesity.
What does this study add?
- The bias in obesity estimates appears to depend on survey context.
- The bias in weight in the 2007 to 2009 Canadian Health Measures Survey (respondents were aware that direct measures would be taken) was substantially lower than the bias in the Canadian Community Health Survey (respondents were not informed before self-reporting that direct measures would be taken).
- Correction equations based on 2005 Canadian Community Health Survey data were successfully applied to self-reported 2008 Canadian Community Health Survey values to produce more accurate estimates of obesity.
- Differences between corrected estimates of obesity from the Canadian Community Health Survey and measured estimates from the Canadian Health Measures Survey should be monitored over time to determine if the bias in self-reported values is changing and if new correction equations need to be developed.
- Date modified: