SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Disclosure
  9. REFERENCES

The objective of this study was to determine whether the bias in self-reported estimates of obesity has changed over time and followed different patterns in Canada and the United States. Using age-standardized data from three waves of the National Health and Nutrition Examination Survey (NHANES) in the United States and the Canadian Community Health Survey (CCHS) and the Canadian Heart Health Survey (CHHS) in Canada, discrepancies were compared between reported and measured estimates of height, weight, and obesity (based on the BMI) from 1976 to 2005. Results indicated that obesity increased in both countries, but rates were higher in the United States. The discrepancy between self-reported and measured obesity was small in the United States with reported data underestimating measured prevalence by about 3%; this stayed relatively constant over time. In Canada, the discrepancy was large and doubled in the past decade (from 4 to 8%). In the United States, self-reported data may be more accurate in monitoring changes in obesity over time, as the estimates have consistently remained about 3% below the measured estimates, whereas in Canada, monitoring obesity based solely on self-reported height and weight may produce inaccurate estimates because of the increasing discrepancy between self-reported and measured data.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Disclosure
  9. REFERENCES

Obesity is an important public health issue, with well-documented links between excess body weight and disease (1,2). In Canada, 59% of adults are overweight or obese (3) and in the United States, 66% (ref. 4). Over a quarter of children and youth in the United States and Canada are considered overweight or obese (5) with increasing numbers being under the age of 5 years. If current trends continue, the number of obese individuals is expected to rise to over 1 billion worldwide by 2030 (ref. 6).

Self-reported obesity, often assessed using the BMI, is commonly used for surveillance purposes because reported height and weight can readily be collected on national surveys at a fraction of the cost of directly measuring respondents. Research has found, however, that self-reports overestimate height and underestimate weight, which leads to an underestimation of the BMI and consequently of obesity prevalence (7,8). The consequences of this underreporting are substantial, not only in terms of underestimating the burden of obesity, but also in distorting our understanding of the relationship between obesity and obesity-related diseases. For instance, recent research has shown that when estimates of obesity are based on self-reported data, the relationship between obesity and obesity-related conditions such as heart disease, diabetes, and hypertension are exaggerated when compared to estimates based on directly measured values (9,10,11). Correction equations that statistically adjust self-reported estimates of obesity to a more close approximation of their measured values have been developed with some success (12,13,14,15), but the generalizability of these equations depends on the stability of the self-reporting bias over time and across populations.

Little is known about how the discrepancy between reported and measured obesity has changed over time, but there is some evidence to suspect that the discrepancy is growing. First, there has been a dramatic increase in the prevalence of obesity in recent years (4,16,17), and studies have consistently shown that heavier individuals are more likely to underreport their weight (14,18,19). Second, in this same period of time, the awareness of obesity and its health consequences has increased (2,16), as has the pressure to be thin (20), and the prevalence of weight discrimination (21), all of which could lead individuals to feel more pressure to present themselves in a socially desirable manner; it has been suggested that a change in awareness and attention to issues such as obesity could affect the way individuals respond to self-reported questions (22).

If the bias is constant or changing systematically over time, then self-reported estimates may still be valuable for monitoring trends and could be statistically adjusted to increase their accuracy (i.e., to a more close approximation of their measured values). However, if there is no systematic trend in the relationship between reported and measured obesity over time, it may limit the usefulness of reported estimates for monitoring public health, informing policy and program development, and understanding the relationship between obesity and disease.

To our knowledge, this study presents the first historical and international comparison of the bias in self-reported obesity in Canada and the United States. Specifically, we examined the discrepancy between reported and measured height, weight and BMI, which were used to calculate reported and measured obesity, over ∼30 years, between 1976 and 2005. Both Canadian and American data were used to determine whether the bias changed and whether the rate of change was consistent in both countries.

Methods and Procedures

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Disclosure
  9. REFERENCES

Data sources

This analysis used national Canadian and American data sets that contained self-reported estimates of height and weight, and also direct measurements of these. For the United States, this included the National Health and Nutrition Examination Surveys (NHANESs) from 1976 to 1980 (NHANES II), 1988 to 1994 (NHANES III), and 2003 to 2004 (NHANES). In Canada, only two data sets were available with both measured and reported data: the Canadian Heart Health Surveys (CHHSs; 1986–1992) and the Canadian Community Health Survey (CCHS; 2005).

NHANES is a “program of studies designed to assess the health and nutritional status of adults and children in the United States”(23, p. 1). The surveys began in 1971 and in 1999, they became continuous. Each contains a nationally representative sample of the US noninstitutionalized civilian population, collected through a stratified multistage probability cluster sampling design. Some cycles include oversampling of certain groups such as blacks and Mexican Americans in NHANES III. NHANES combines a household interview with physical examinations undertaken in mobile clinics. Reported height and weight were collected at the home, with the measured values collected in the clinic a few days to several weeks later (see Table 1).

Table 1.  Survey characteristics
inline image

The CHHSs were a series of 10 provincial surveys conducted between 1986 and 1992 as part of the Canadian Heart Health Initiative to provide information about cardiovascular disease and its risk factors in Canada. Data on measured and reported height and weight were only collected in eight provinces (Saskatchewan and Nova Scotia did not collect this information); individuals living in military institutions or Indian reserves were also excluded. Respondents were sampled through a stratified multistage probability sampling design. The medical insurance registers of the provinces were used as the sampling frame from which the sample was drawn. In each province, all persons on the medical insurance register within the selected areas were further stratified into six age/sex groups, and independent samples of persons were randomly selected from each age/sex stratum to give the required number of responses. Self-reported information was collected at home and respondents were measured at a subsequent clinic visit.

The CCHS is designed to provide timely cross-sectional estimates of health determinants, health status, and health system utilization in Canada. The CCHS samples its population through a multistage cluster sampling technique that is representative of over 98% of the Canadian population (individuals living on Indian reserves or Crown Lands, residents of institutions, Canadian Armed Forces, and certain remote regions are excluded). Three sampling frames were used to select the sample of households for the 2005 CCHS: 49% of households came from an area frame; 50%, from a list frame of telephone numbers; and the remaining 1%, from a random digit dialing sampling frame. Measured height and weight were collected for only a subsample of respondents because of cost considerations, all of whom were from the area frame. Reported and measured height and weight data were collected as part of a household interview. The characteristics of each survey are described in Table 1.

Study population

The study is limited to adults aged 18–74 years. Children were excluded, as it has been suggested that reporting error in children and adolescents may be of a different nature than that in adults (14). Women who were pregnant, individuals who did not have both self-reported and measured data available, and those who had completed a home visit rather than attending the clinic were excluded.

Data were appropriately weighted, and to adjust for the complex survey designs, variances were estimated with a Taylor linearization method (NHANES) and the bootstrap technique (CCHS), whereas for the 1986–1992 CHHS, the formula for simple random sampling was used to estimate standard errors with the incorporation of a design effect (set at 1.5).

All data were age-standardized (using direct standardization) with the 2001 Canadian Census as the standard population to allow for comparison among the surveys and over time. In the Canadian surveys, height could be reported in centimeters or inches (and feet), and weight could be reported in kilograms or pounds. Most respondents reported their height and weight in the imperial system units. BMI was calculated as weight in kilograms divided by height in meters squared (data were converted to metric units as required). World Health Organization guidelines were used to assign BMI cut points to identify overweight (BMI 25.0–29.9 kg/m2) and obese (BMI ≥30.0 kg/m2) respondents (24). All analyses were conducted with SAS (version 9.1; SAS Institute, Cary, NC) and SUDAAN (version 9.0; RTI International, Research Triangle Park, NC).

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Disclosure
  9. REFERENCES

In the United States, mean measured weight increased from 72 kg in 1976–1980 to 82 kg in 2003–2004, with similar patterns observed in men and women (see Table 2). In Canada, mean measured weight was 73 kg in 1986–1992 and rose to 77 kg in 2005, with similar increases experienced for men and women. However, the contrast between the self-reported and measured weights (reported–measured) was pronounced when comparing the two countries. In the United States, the difference between reported and measured weight was −0.4 kg in 1976–1980, rose slightly at the second time point (−0.7 kg) before falling back to −0.6 kg in 2003–2004. This underreporting was driven entirely by women, as men overestimated rather than underestimated their weight. In Canada, the reported–measured differences were greater; they rose from −2 kg in 1986–1992 to −2.3 kg in 2005. Women in Canada underestimated by a greater degree than men, but both Canadian men and women underestimated their weight.

Table 2.  Mean differences between reported and measured values
inline image

Men and women in both countries consistently overestimated their height, with men overestimating to a greater degree than women. For American men, this overestimation appeared to be decreasing slightly over time, but in women, the pattern was not consistent, with the overestimation decreasing from NHANES II to NHANES III and then increasing again in 2003–2004. In Canada, the overestimation was increasing between the two time points in both men and women.

The mean BMI rose in both countries; from 25 to 28 kg/m2 in the United States and from 25 to 27 kg/m2 in Canada. The underestimation in weight and overestimation in height resulted in an underestimation of BMI in both countries and sexes, but this was greater in Canada. The underestimation of BMI is also increasing in Canada over time, rising from an underestimate of 0.8 kg/m2 in the CHHS to 1.1 kg/m2 in the CCHS. The discrepancy was slightly larger in women. In the United States, the discrepancy between reported and measured BMI stayed relatively consistent over time.

Figure 1a illustrates the changes in obesity prevalence over time, based on measured and self-reported data in the two countries. In the United States, the prevalence of measured obesity more than doubled from 1976–1980 to 2003–2004, from 15% (95% CI 14–15%) to 32% (95% CI 29–35%), with the reported prevalence rising from 12% (95% CI 11–13%) to 29% (95% CI 27–32%) in the same time period. The discrepancy between the reported and measured prevalence (about a 3% difference) has stayed relatively constant over time. In Canada, the rise in the prevalence of obesity from the late 1980s and early 1990s until 2005 has been of a similar magnitude as the increase in the United States from the same time period, but the absolute prevalence remains lower in Canada, rising from 14% (95% CI 13–15%) to 24% (95% CI 22–26%). Despite the lower prevalence of obesity, the discrepancy between the reported and measured values appears to have doubled in Canada in this time, going from a difference of 4% in the late 1980s to a difference of 8% in 2005. Similar patterns were seen in men and women in both countries (data not shown).

image

Figure 1. Measured and reported prevalence of (a) obesity and (b) overweight, Canada and United States, 1976–2005, men and women combined. Note: Not all surveys covered the full range of years.

Download figure to PowerPoint

The percentage of the population who were considered overweight (not including obesity) based on measured data has changed little in either country, ranging from 32 to 33% in the United States and 34 to 35% in Canada, whereas the reported prevalence of overweight has increased slightly (see Figure 1b). In the United States though, the prevalence of overweight was underreported in NHANES II by 1.4%, and then in NHANES III and 2003–2004 was overreported (by 2.1 and 0.9%, respectively). In Canada, the prevalence of overweight was underreported in both surveys, but the degree of underreporting decreased between the two surveys, from −3.5 to −1.2%. In Canada, this decrease was driven almost entirely by the men whose reporting bias went from −5% to almost no difference, whereas the women stayed consistent around −2%.

The reporting bias (reported–measured) in the prevalence of obesity by age is displayed in Figure 2. In Canada, the bias was always smaller in the CHHS compared to the more recent CCHS, suggesting that the bias across all age-groups may be increasing over time in Canada. In the NHANES surveys, the bias was consistently greater in all age-groups in NHANES III compared to NHANES II. The NHANES 2003–2004 showed the greatest bias at the younger age ranges that decreased with increasing age, with the lowest bias seen for those aged 45–74 years. This was the only survey in which the bias was not the greatest in the highest age-groups. The discrepancy at the oldest ages (65–74 years) ranged from a reported–measured difference in obesity prevalence of −4% (NHANES 2003–2004) to −17% (CCHS 2005). The reporting bias in the prevalence of overweight showed no clear pattern (data not shown).

image

Figure 2. Bias in obesity prevalence (reported–measured) by age-group, men and women, 1976–2005.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Disclosure
  9. REFERENCES

Consistent with past research (7), this study found that respondents in national health surveys in both Canada and the United States have a tendency to underestimate their weight and exaggerate their height, which results in an underestimation of BMI when it is based on these values. This trend has remained consistent since 1976 in the United States and since 1986 in Canada, with one exception: American men did not underestimate their weight in any of the NHANES cycles that we examined. The discrepancy in reported weight appears to be driven mainly by an underestimation of weight in women, whereas the overestimation of height in men seems to be the main driver of the height discrepancy.

In the past 30 years, average weights and BMIs have increased substantially in both countries, although both the average BMI and the prevalence of obesity was higher in the U.S at all time points, with Canada seeming to lie approximately a decade behind the United States in terms of obesity prevalence. At any given prevalence value, however, the discrepancy between the reported and measured estimates was greater in Canada. For instance, in NHANES III (1988–1994), the prevalence of obesity was 23% corresponding to the Canadian prevalence of 24% from the 2005 CCHS; yet in NHANES III, the bias was only 4% and in the CCHS, it was twice this at 8%. This indicates that the reporting bias may be changing differently in the two countries. This is an interesting observation that warrants further research to investigate the source of this difference and the implications of such differences for interpreting trends based on self-reported height and weight.

Over the 30 years studied in this paper, the bias in reported BMI and obesity has stayed relatively constant in the United States, whereas in Canada, this bias has increased substantially. A Swiss study that examined the reporting bias over time also found that the discrepancy was relatively constant in that population (25). Therefore, it is likely that in the United States, self-reported data may be more accurate in monitoring trends in obesity over time, as the estimates have consistently remained about 3–4% below the measured estimates. More caution is required, however, when examining patterns in self-reported data for specific age-groups, as the bias tends to be greater for those in the older age ranges.

Although a third data point is required in order to establish a trend in Canada, the increasing discrepancy from the late 1980s to 2005 indicates that the reporting bias may not be constant over time. This makes it difficult to interpret self-reported data in the absence of the measured estimates, as it introduces an underestimate, the magnitude of which is difficult to quantify, and that appears to be changing over time. In addition, the age-based discrepancies, especially in the 2005 CCHS are substantial, with the older age-groups underestimating their obesity prevalence by ∼17 percentage points. This appears to be due to both the overreporting of height (mostly in men) and the underreporting of weight (in women). We would recommend caution particularly when researchers are using self-reported obesity estimates to quantify obesity prevalence in the older age-groups.

The bias in reported obesity is not a new phenomenon, and researchers have attempted to correct for this bias in many ways. One of the most common methods has been to use regression models to adjust self-reported data based on measured values (13). Recently, this has been successfully attempted with the 2005 CCHS data (15), but the change in the bias over time in Canada suggests that formulas derived from one data set may not be applicable to Canadian data collected at other time points. Visscher et al.'s work has also concluded that correction equations may only be applicable to the data sets in which they are generated (26). This means that in order to generate appropriate adjustments to self-reported data, a subsample of data with both measured and reported information would be required in all data sets. This may not be feasible in small-scale or telephone surveys.

The smaller bias in the US data is an unexpected finding given the higher prevalence of obesity among the Americans and the fact that the bias in reported obesity has been shown to be greater as BMI increases (14,18,19). Other factors such as age, education, ethnicity, and activity levels (8,14,18,27) have also been associated with increased discrepancies, but to date, there is no research of which we are aware that has compared the discrepancies or their determinants between Canada and the United States. Further research is required to understand the reasons for the discrepancies between the two countries.

It is possible that cultural differences between the two countries are responsible for the differences, but little evidence exists to help quantify what these differences are or how they operate. There is some evidence within both countries that indicates that ethnic differences can influence perceptions of one's weight. For instance, within Canada, foreign-born women are more likely to perceive themselves as overweight than their Canadian-born counterparts (28), whereas in the United States, whites are more likely to perceive themselves as overweight than African or Mexican Americans (29).

It is also conceivable that with rising obesity rates, weight perceptions and individual views of socially acceptable weight may be changing. With approximately two-thirds of Americans being overweight or obese, those with BMIs >25 kg/m2 may be perceived as the new norm, as they now outnumber those in the normal weight range. This “normalizing” of overweight is not without consequences, and research is showing that it is affecting people's perceptions of their weight status, which in turn could be influencing how they respond to national surveys. Although we may have expected an increase in the proportion of Americans who perceive themselves as being overweight corresponding with the rise in obesity rates, research with the NHANES data has actually shown that the number of overweight people who perceive themselves as overweight is declining as obesity rates rise (30). In addition, a 22-country comparison found that women in the United States and the United Kingdom, areas where weight has increased dramatically in recent years, were less concerned about their weight than women from Asian countries such as Japan, where women have actually been getting slimmer over time (31). A Finnish study has further demonstrated that this trend seems to begin in adolescence, with its finding that between 1979 and 1999, the proportion of adolescents perceiving themselves as overweight had decreased (32). Although it is counterintuitive that the perception of being overweight has decreased whereas actual overweight and the idealization of thinness have increased, researchers postulate that it is the shift in our perception of overweight and its normalization that is changing, as we increasingly compare ourselves to our peers, who are themselves heavier, rather than to idealized media images (32).

The implications of these changes both for measuring obesity and addressing the epidemic are not clear, but it does offer one potential explanation for why we are not seeing an increase in the discrepancy between reported and measured estimates in the United States. More work is needed to determine whether this change in overweight perception is related to the way people report data on self-reported questionnaires and whether this is also occurring in the Canadian population. Should Canadian obesity levels continue to climb, it will be interesting over time to see if the reporting bias in Canada decreases as our obesity rates reach the levels in the United States.

There are many limitations inherent in any international or historical research. Surveys have different designs, measurement protocols, and methodologies that could lead to differences in results. The large discrepancy between the CCHS data and the other data sources, for instance, raises the question of the comparability of the results. Rounding off the height data in the CCHS to the nearest inch is a limitation that could introduce bias into the estimates, and we recommend that in future cycles of the CCHS, this practice be eliminated. Nevertheless, we believe that the use of standardized nationally representative data with similar analytic designs improves the accuracy of our findings. In addition, a minimum of three time points of data are preferable for establishing trends and patterns over time. Further research will be required when more Canadian data become available to determine whether the discrepancy between reported and measured values continues to diverge and whether a trend can be established.

Past research has demonstrated that respondents to questions about height and weight often intentionally or unintentionally round their responses to values ending in 0 or 5, a practice known as end-digit preference (33). This practice may increase reporting bias if respondents who are already overestimating their height further the overestimation by rounding up to a unit ending in 0 or 5. We were not able to assess the degree of bias that may have been due to end-digit preference in this study because of differences in how the data were captured among the surveys.

Data for all five surveys were collected through in-person interviews. Telephone interviews have shown a greater bias in self-reported obesity than in-person interviewing (34,35), and the degree to which the current findings apply to data collected by phone is not yet known. Reporting bias can also be increased if respondents are unaware that they will subsequently be measured (33). In all surveys except the 2005 CCHS, we could reasonably suspect that participants had an indication that they would be measured, as they were participating in an in-depth health survey that required a follow-up visit to a clinic for physical testing. In the CCHS, a household survey, a subsample of the population was selected to have their height and weight measured at the end of the survey; this was the only physical measures data that were collected, and there was no prior indication that any measurements would be taken. It is therefore possible that the increase in bias seen in the CCHS may be due, in part, to the fact that participants were unaware that their self-reported responses would be validated. This and the fact that the reported and measured values are both collected on the same occasion are both important strengths of the CCHS methodology that may have contributed to more accurate results.

To our knowledge, this paper is the first to undertake an international comparison of the bias in reported obesity and how it has changed over time. We have found indications of a consistent reporting bias in the United States and of an increasing discrepancy in Canada, indicating that self-reports could be more accurately used in the United States to monitor trends over time. More research is required to provide insight into the determinants of the bias in the two countries and possible data transformation procedures to adjust self-reported data to reflect better objective measures.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Disclosure
  9. REFERENCES

We thank Georgia Roberts for her methodological advice and assistance in variance estimation for the historical surveys. Funding for this work was partially provided by Statistics Canada.

REFERENCES

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Disclosure
  9. REFERENCES
  • 1
    Health Canada. Canadian Guidelines for Body Weight Classification in Adults. Health Canada: Ottawa, Ontario, 2003.
  • 2
    US National Institutes of Health. Clinical Guidelines on the Identification, Evaluation, and Treatment of Overweight and Obesity in Adults: The Evidence Report. National Institutes of Health: Bethesda, MD, 1998.
  • 3
    Tjepkema M. Adult obesity. Health Rep 2006; 17: 925.
  • 4
    Ogden CL, Carroll MD, Curtin LR et al. Prevalence of overweight and obesity in the United States, 1999–2004. JAMA 2006; 295: 15491555.
  • 5
    Shields M. Overweight and obesity among children and youth. Health Rep 2006; 17: 2742.
  • 6
    Kelly T, Yang W, Chen CS, Reynolds K, He J. Global burden of obesity in 2005 and projections to 2030. Int J Obes (Lond) 2008; 32: 14311437.
  • 7
    Gorber SC, Tremblay M, Moher D, Gorber B. A comparison of direct vs. self-report measures for assessing height, weight and body mass index: a systematic review. Obes Rev 2007; 8: 307326.
  • 8
    Engstrom JL, Paterson SA, Doherty A, Trabulsi M, Speer KL. Accuracy of self-reported height and weight in women: an integrative review of the literature. J Midwifery Womens Health 2003; 48: 338345.
  • 9
    Yannakoulia M, Panagiotakos DB, Pitsavos C, Stefanadis C. Correlates of BMI misreporting among apparently healthy individuals: the ATTICA study. Obesity (Silver Spring) 2006; 14: 894901.
  • 10
    Chiolero A, Peytremann-Bridevaux I, Paccaud F. Associations between obesity and health conditions may be overestimated if self-reported body mass index is used. Obes Rev 2007; 8: 373374.
  • 11
    Shields M, Gorber SC, Tremblay MS. Effects of measurement on obesity and morbidity. Health Rep 2008; 19: 7784.
  • 12
    Bolton-Smith C, Woodward M, Tunstall-Pedoe H, Morrison C. Accuracy of the estimated prevalence of obesity from self reported height and weight in an adult Scottish population. J Epidemiol Community Health 2000; 54: 143148.
  • 13
    Kuskowska-Wolk A, Bergström R, Boström G. Relationship between questionnaire data and medical records of height, weight and body mass index. Int J Obes Relat Metab Disord 1992; 16: 19.
  • 14
    Rowland ML. Self-reported weight and height. Am J Clin Nutr 1990; 52: 11251133.
  • 15
    Gorber S Connor, Shields M, Tremblay MS, McDowell I. The feasibility of establishing correction factors to adjust self-reported estimates of obesity. Health Rep 2008; 19: 7182.
  • 16
    World Health Organization. Obesity: Preventing and Managing the Global Epidemic. Technical Report Series No. 894. World Health Organization: Geneva, 2000.
  • 17
    Shields M, Tjepkema M. Trends in adult obesity. Health Rep 2006; 17: 5359.
  • 18
    Shields M, Gorber SC, Tremblay MS. Estimates of obesity based on self-report versus direct measures. Health Rep 2008; 19: 6176.
  • 19
    Elgar FJ, Roberts C, Tudor-Smith C, Moore L. Validity of self-reported height and weight and predictors of bias in adolescents. J Adolesc Health 2005; 37: 371375.
  • 20
    Katzmarzyk PT, Davis C. Thinness and body shape of Playboy centerfolds from 1978 to 1998. Int J Obes Relat Metab Disord 2001; 25: 590592.
  • 21
    Andreyeva T, Puhl RM, Brownell KD. Changes in perceived weight discrimination among Americans, 1995–1996 through 2004–2006. Obesity (Silver Spring) 2008; 16: 11291134.
  • 22
    Tremblay M. The need for directly measured health data in Canada. Can J Public Health 2004; 95: 165168.
  • 23
    Centers for Disease Control and Prevention, National Center for Health Statistics. National Health and Nutrition Examination Survey, 2005–2006 Overview. U.S. Department of Health and Human Services, Centers for Disease Control and Prevention: Hyattsville, MD, 2006.
  • 24
    World Health Organization. Physical Status: The Use of and Interpretation of Anthropometry, Report of the WHO Expert Committee. WHO Technical Report Series No. 854. World Health Organization: Geneva, 1995.
  • 25
    Faeh D, Marques-Vidal P, Chiolero A, Bopp M. Obesity in Switzerland: do estimates depend on how body mass index has been assessed? Swiss Med Wkly 2008; 138: 204210.
  • 26
    Visscher TL, Viet AL, Kroesbergen IH, Seidell JC. Underreporting of BMI in adults and its effect on obesity prevalence estimations in the period 1998 to 2001. Obesity (Silver Spring) 2006; 14: 20542063.
  • 27
    Gillum RF, Sempos CT. Ethnic variation in validity of classification of overweight and obesity using self-reported weight and height in American women and men: the Third National Health and Nutrition Examination Survey. Nutr J 2005; 4: 27.
  • 28
    Gucciardi E, Wang SC, Badiani T, Stewart DE. Beyond adolescence: exploring Canadian women and men's perception of overweight. Women's Health Issues 2007; 17: 374382.
  • 29
    Paeratakul S, White MA, Williamson DA, Ryan DH, Bray GA. Sex, race/ethnicity, socioeconomic status, and BMI in relation to self-perception of overweight. Obes Res 2002; 10: 345350.
  • 30
    Johnson-Taylor WL, Fisher RA, Hubbard VS, Starke-Reed P, Eggers PS. The change in weight perception of weight status among the overweight: comparison of NHANES III (1988-1994) and 1999–2004 NHANES. Int J Behav Nutr Phys Act 2008; 5: 9.
  • 31
    Wardle J, Haase AM, Steptoe A. Body image and weight control in young adults: international comparisons in university students from 22 countries. Int J Obes (Lond) 2006; 30: 644651.
  • 32
    Kaltiala-Heino R, Kautiainen S, Virtanen SM, Rimpelä A, Rimpelä M. Has the adolescents' weight concern increased over 20 years? Eur J Public Health 2003; 13: 410.
  • 33
    Bopp M, Faeh D. Who gives me five? Rounding preference for self reported height depends on language. BMJ 2008; 337: 1463.
  • 34
    Ezzati M, Martin H, Skjold S, Vander Hoorn S, Murray CJ. Trends in national and state-level obesity in the USA after correction for self-report bias: analysis of health surveys. J R Soc Med 2006; 99: 250257.
  • 35
    Béland Y, St-Pierre M. Mode effects in the Canadian Community Health Survey: a comparison of CATI and CAPI. In: Lepkowski JM, Tucker C, Brick JM et al. (eds). Advances in Telephone Survey Methodology. Wiley: New York, 2008, pp 297314.