Study: Estimates and effects of obesity based on self-reported data versus direct measures

View the most recent version.

Since the mid-1990s, Statistics Canada's major health surveys have generally relied on respondents to report their weight and height in order to estimate body mass index (BMI) and produce rates of obesity.

Two articles released today in Health Reports quantify the bias resulting from the use of self-reported data, as well as its impact.

The analysis was based on 4,567 respondents aged 12 or older to the 2005 Canadian Community Health Survey (CCHS), who, during a face-to-face interview, provided self-reported values for height and weight. Their height and weight were then measured by trained interviewers.

Respondents tended to slightly over-report their height. On average, males over-reported their height by 1 centimetre, and females, by half a centimetre.

The tendency to under-report weight was more pronounced. Females under-reported their weight by an average of 2.5 kilograms, while males under-reported it by 1.8 kilograms.

Moreover, the heavier they were, the more they under-reported. Females who were in the highest category of obesity under-reported their weight by an average of 8.6 kilograms, and males in this category, by 5.0 kilograms.

Not surprisingly, many studies have found that self-reported data yield lower estimates of the prevalence of obesity than do estimates based on measured data. However, few studies have examined the effect of this misclassification on the relationship between BMI categories and obesity-related health conditions.

Based on self-reported height and weight data from the CCHS, a substantial proportion of individuals with excess body weight are erroneously placed in lower BMI categories. Consequently, associations between weight and obesity-related diseases such as diabetes, high blood pressure and heart disease, may be exaggerated because people in the overweight and obese categories based on self-reports are actually heavier than those in the corresponding categories based on measured values.

Because the use of self-reported data underestimates the prevalence of obesity, the total burden of obesity-related disease (as measured by the number of cases) is underestimated.

For example, among those classified as obese based on self-reported data, 360,000 people aged 40 years or older had diabetes. But among those classified as obese based on measured values, 530,000 people or nearly 50% more, had diabetes. These differences simply reflect the greater number of people who are classified as obese when measured data are used.

Statistics Canada is working on methods of addressing these issues.

Definitions, data sources and methods: survey number 3226.

The articles, "Estimates of obesity based on self-report versus direct measures" and "Effects of measurement on obesity and morbidity," which are part of today's Health Reports, Vol. 19, no. 2 (82-003-XWE, free) online release, are now available from the Publications module of our website.

For more information, or to enquire about the concepts, methods or data quality of this release, contact Margot Shields (613-951-4177; margot.shields@statcan.gc.ca) or Sarah Connor Gorber (613-951-1193; sarah.connorgorber@statcan.gc.ca), Health Information and Research Division, or Mark Tremblay (613-951-4285; mark.tremblay@statcan.gc.ca), Physical Health Measures Division.

For more information about Health Reports, contact Christine Wright (613-951-1765; christine.wright@statcan.gc.ca), Health Information and Research Division.