International Journal of (2002) 26, 797–804 ß 2002 Nature Publishing Group All rights reserved 0307–0565/02 $25.00 www.nature.com/ijo PAPER Trends in and obesity among adults in (1970 – 1992): evidence from national surveys using measured height and weight

GM Torrance1*, MD Hooper2 and BA Reeder3

1Physical Activity Unit, , Ottawa, , Canada; 2Office of Nutrition Policy and Promotion, Health Canada, Ottawa, Ontario, Canada; and 3Department of Community Health and Epidemiology, University of , Saskatoon, Saskatchewan, Canada

PURPOSE: To examine secular trends in obesity and overweight among Canadian adults between 1970 and 1992. The impact of education level and smoking on weight trends is explored. DATA: Adults aged 20 – 69 participating in three national health surveys which obtained measured height and weight: the Nutrition Canada Survey conducted between 1970 and 1972 (analysis sample n ¼ 5963); the Canada Health Survey of 1978 – 1979 (analysis sample n ¼ 3622); and the Canadian Heart Health Surveys conducted between 1986 and 1992 (analysis sample n ¼ 17 699). METHODS: Comparison of percentage overweight (age-standardized body mass index (BMI) 25.0 – 29.9) and obese (age- standardized BMI  30.0) by sex, education level and smoking status across the three surveys. RESULTS: Among men, the proportion overweight and obese increased steadily from 1970 – 1972 to 1986 – 1992. Among women, there was a substantial increase in the proportion overweight and obese between 1970 – 1972 and 1978 – 1979, then an increase in proportion obese, but not overweight, between 1978 – 1979 and 1986 – 1992. Although the prevalence of obesity increased in all education levels, the sub-groups with the greatest relative increase are men in the primary education category, and women in the secondary and post-secondary between 1970 – 1972 and 1986 – 1992. An increase in the prevalence of obesity was greatest among current smokers and, to a lesser extent, among former smokers. CONCLUSION: While excess weight has become an increasing public health problem among Canadian adults, the rate of increase in prevalence of obesity since 1970 varied with sex, education level and smoking status. There is a need for new data on measured heights and weights of Canadian adults and children and youth to update trends. International Journal of Obesity (2002) 26, 797 – 804. doi:10.1038=sj.ijo.0801991

Keywords: obesity; trends; Canada; smoking; education

Introduction In 1997, total direct costs of obesity in Canada were esti- Excess weight is a significant health issue in Canada and mated at $1.8 billion, or 2.4% of the total health care around the world. Recent estimates show the prevalence of expenditures for all diseases.3 overweight and obesity increasing at alarming rates, in both The body mass index (BMI; kg=m2) is the most commonly developed and developing countries.1,2 Overweight and obe- used indicator of obesity as it provides useful estimates of sity are related to a number of health consequences, includ- weight-for-height at the population level. Using the criteria ing type 2 diabetes, coronary heart disease, stroke, for classification of obesity recommended by the World hypertension, gallbladder disease, certain types of cancer, Health Organization (WHO) Expert Consultation on Obesity sleep apnea, psychosocial disturbances and osteoarthritis.1 (BMI  30),1 it is estimated that 13% of men and 14% of women in Canada are ‘obese’.4 Estimated prevalence rates of obesity vary worldwide. In European countries, prevalence *Correspondence: GM Torrance, 291 Pleasant Park Road, Ottawa, ranges from 10 to 20% among men and 10 to 25% among Ontario, Canada K1H 5M7. women;1 in Australia, 9% of men and 11% of women are E-mail: [email protected] 5 Received 29 March 2001; revised 22 November 2001; considered obese, and in the United States, obesity affects accepted 18 December 2001 20% of men and 25% of women.6 Obesity trends in Canadian adults GM Torrance et al 798 International studies are consistent in showing secular Various studies have shown that current smokers tend to trends toward increased obesity, but there are variations in be lighter than non-smokers, who in turn are lighter than ex- the timing, the demographic groups at risk, the correlates smokers. International studies suggest that declines in smok- and the proposed explanations. In the United States, the ing explain a relatively minor part of the population increase percentage obese (BMI  30) using measured data increased in overweight and obesity.10 – 12 However, Simmons and slowly from 1960 – 1962 to 1976 – 1981, then more rapidly to colleagues estimated that in New Zealand, the decline in 1988 – 1991.6 A more recent study using self-reported height smoking was responsible for 7% of the increase in BMI for and weight found a further large increase between 1991 and men and 10% for women between 1982 and 1994.10 Flegal 1998.7 and colleagues estimated that the decline in the trends are difficult to interpret, in part due accounted for one-quarter of the increase in overweight in to the different methodologies of various surveys.4,8 For men, and one-sixth in women, between 1978 and 1990 in example, when using the cut-off of BMI  27 to define the United States.11 obesity, Macdonald and colleagues concluded that the pre- The purpose of this study is to examine the historical valence of obesity had changed very little between the late trend in obesity prevalence among Canadian adults using 1970s and early 1990s; however, the same researchers found height and weight data from national surveys. The impact a substantial increase in the prevalence of obesity when of education level and smoking on body weight are using BMI  30 as the cut-off. Investigating obesity patterns explored. over the same time period, Cairney and Wade were unable to conclude with certainty that obesity had increased or decreased.8 The authors, who compared changes over time using a combination of self-reported and measured data, note that the variation in prevalence over time may be due Methods to differences in methodology employed in the surveys and National population-level data sources that have used self- misclassification due to errors associated with self-reported reported height and weight include the 1985 and 1991 data. Most recently, the Statistical Report on the Health of General Social Surveys, the 1985 and 1990 Health Promotion reported a sharp increase in BMI  27 among Surveys, and the 1994 – 1995, 1996 – 1997 and 1998 – 1999 men between 1985 and 1996 – 1997 based on self-reported National Population Health Surveys. Sources using measured height and weight.9 Overweight among women increased in height and weight include the Nutrition Canada Survey of parallel with men until 1994 – 1995, then declined slightly 1970 – 1972, the Canada Health Survey of 1978 – 1979, the between 1994 – 1995 and 1996 – 1997. Canada Fitness Survey of 1981, the Campbell Survey on the The majority of studies in industrialized countries have Well-Being of Canadians of 1988, and the Canadian Heart found that lower education and income groups are most at Health Surveys of 1986 – 1992. risk of overweight and obesity.1,4 However, the most recent An exploratory analysis of the various national Canadian report from the United States concluded that ‘obesity sources indicated they would best be grouped into sets of increased in men and women and across all sociodemo- surveys with similar methodology, with each set treated graphic groups ... with the highest increase among the separately, rather than as time points on a single continuum. youngest ages and higher education levels.’7 Whether this On this principle, Figure 1 shows the proportion of Cana- represents a new trend or an anomaly remains to be seen. dians between 20 and 64 y of age with BMI  30 for:

Figure 1 Obesity trends among men and women (20 – 64 y) across five sets of surveys.

International Journal of Obesity Obesity trends in Canadian adults GM Torrance et al 799 (1) the 1985 and 1991 General Social Surveys (GSS; self- survey clinic. It also had a volunteer sample of 3295 indivi- reported data); duals. Three trained teams comprising physicians, dentists, (2) the 1985 and 1990 Health Promotion Surveys (HPS; self- nurses, nutritionists, dental hygienists and support staff reported data); conducted the field work for the survey which included (3) the 1994 – 1995 and 1998 – 1999 National Population clinical, dental and anthropometric examinations, dietary Health Surveys (NPHS; self-reported data); interviews and blood and urine analyses. The teams travelled (4) the 1981 Canada Fitness Survey (CFS) and the 1988 to clinics in different locations organized by local public Campbell Survey on the Wellbeing of Canadians health units. Anthropometric measurements including skin- (CSWB, measured data); folds, height and weight were taken by an anthropometrist (5) the 1970 – 1972 Nutrition Canada Survey (NCS), the as part of the clinical interview when subjects reported to 1978 – 1979 Canada Health Survey (CHS), the 1986 – the survey clinics. Weight was measured with a beam bal- 1992 Canadian Heart Health Surveys (CHHS; measured ance, with the subject wearing light underclothing and a data). paper examination gown. Height was measured without shoes to the nearest mm using a stadiometer pole. Height It can be seen that, although the percentage obese (BMI  30) and weight measures were obtained on 7181 people in the varies across the different sets, the pattern within each set is probability sample between the age of 20 and 69 (5963 usually consistent — an increase in obesity over time. excluding pregnant women). Our analysis included 2662 Self-reported data tend to over-report height and under- men and 3301 non-pregnant women in the probability report weight.13,14 It is, therefore, not unexpected that the sample and excluded those in the volunteer sample. For percentage obese in sets based on self-reports is generally lower analysis, the sample data were weighted up to population than in those from measured data series. Comparing the two proportions. sets based on measured data, the percentage obese in the two The Canada Health Survey (CHS) of 1978 – 1979 had an fitness surveys is considerably lower than in the three ‘health’ initial planned sample of over 20 000 households.17 It con- surveys. For instance, comparing the 1978 – 1979 Canada sisted of both a questionnaire and a physical measures Health Survey with the more-or-less contemporaneous 1981 component. A sub-sample of the questionnaire sample was Canada Fitness Survey gives 13% obese vs 9%, and comparing selected for physical measures which included anthropo- the 1988 Campbell Survey on Wellbeing with the 1986 – 1992 metric measurements, a fitness test and blood samples col- Heart Health Surveys gives 10% obese vs 14%. lected by survey nurses on a subsequent visit to the subject’s Our preliminary investigation indicated that the two fitness home. The response rate for the physical measures compo- surveys (CFS, CSWB) tended to have substantially lower pro- nent (excluding blood samples) was 72% of the interview portions in the overweight and obese BMI categories than the sample. Subjects wore light indoor clothing without shoes other surveys using measured height and weight. It appears for the anthropometry. Weight was measured using a beam that heavier individuals may have either been screened out or balance. Height was measured to the nearest cm with sub- had a higher refusal rate because of the fitness test component jects standing on a hard surface next to a wall, using a square of these surveys (Cora Craig, CFLRI, personal communication, and tape measure fixed to the wall. The analysis sample 31 January 2000). As a result, it seemed prudent to treat these consisted of 1623 men and 1999 women for a total of 3622 two surveys as a separate series. individuals between ages 20 and 69. These were weighted While it is probable that any of the sets would be suitable up to population proportions using the for analysis within itself, the set we have chosen to analyse in clinical data case weights. depth is the one containing the three health surveys — the The Canadian Heart Health Surveys (CHHS) were con- Nutrition Canada Survey (NCS), the Canada Health Survey ducted between 1986 and 1992 in all 10 provinces using the (CHS) and the Canadian Heart Health Surveys (CHHS). These health insurance registration files to select a probability three surveys give the longest window into the past and sample and using the same protocol to collect core informa- share similarities in context and procedures, thus were most tion.18 Questionnaires were completed by trained survey likely to yield comparable data. Each survey was done in nurses at the subjects’ homes, followed by a clinic visit a medical, rather than a fitness context, with physical within 2 weeks during which anthropometric measure- measurements taken by health professionals. ments, blood pressure and a blood sample were taken. Anthropometric measures were taken in the morning after an overnight fast with participants wearing light indoor Data clothing without shoes.19 Weight was measured using a The Nutrition Canada Survey (NCS) was conducted between beam balance in all provinces except Ontario, where a October 1970 and October 1972 to determine the nutritional calibrated spring scale was used. Height was measured to status and dietary intakes of the Canadian population.15,16 the nearest cm with subjects standing on a hard surface next The survey was based on a three-stage, stratified national to a wall, using a square and tape measure fixed to the wall. probability sample with 12 795 people (46% of those con- The analysis sample in the 20 – 69 y age group consisted of tacted) responding to an initial invitation and attending a 8711 men and 8988 women for a total of 17 699 individuals.

International Journal of Obesity Obesity trends in Canadian adults GM Torrance et al 800 Measures Table 1 Raw sample size by 10-y age groups and sex for NCS,a CHSb c BMI was calculated as weight (kg)=height (m)2. BMI was and CHHS calculated from original height and weight data available NCS CHS CHHS in the NCS and CHS datasets, whereas derived BMI was (1970 – 1972) (1978 – 1979) (1986 – 1992) available in the CHHS dataset. Education level was originally categorized differently in Age Men Women Men Women Men Women the three surveys. For our purposes, three categories were 20 – 29 525 690 460 540 2625 2826 created: primary (elementary complete plus some second- 30 – 39 509 699 345 448 2570 2677 ary); secondary (secondary completedÆ non-university post- 40 – 49 559 740 289 371 1146 1197 50 – 59 476 585 287 362 901 904 secondary); and post-secondary (any university). 60 – 69 593 587 242 278 1469 1384 Current smoking status was available as a three-part Total 2662 3301 1623 1999 8711 8988 classification in the NCS, current smoker, former smoker aNutrition Canada Survey, 1970 – 1972. and never smoker. For the CHS and CHHS, the categories b Canada Health Survey, 1978 – 1979. for type of smoker were never smoked, former smoker, cCanadian Heart Health Surveys, 1986 – 1992. regular cigarette smoker, occasional cigarette smoker, pipe or cigar smoker. The last three categories were combined as current smokers.

both the overweight (BMI 25.0 – 29.9) and obese (BMI  30.0) Statistical analysis categories. Among women, there was a substantial increase The weight classification recommended by the WHO Expert in the proportion overweight and obese between 1970 – 1972 Consultation on Obesity is used in describing the results of and 1978 – 1979, then an increase in proportion obese, but this study.1 That is BMI between 25.0 and 29.9 denotes not overweight, between 1978 – 1979 and 1986 – 1992. ‘overweight’ and BMI of 30.0 or greater represents ‘obesity’. In all three surveys for the entire 20 – 69 y age group, a The analysis is mainly confined to simple exploratory and higher proportion of men than women were in the over- descriptive methods. Because of the complex survey designs weight category (BMI 25.0 – 29.9), while the opposite was and the lack of detail available on the older surveys, we were true in the obese category (BMI  30.0). An examination of unable to estimate precise confidence intervals for differ- the prevalence of obesity (BMI  30.0) by age group in men ences in proportions. For each sex, 1978 – 1979 and 1986 – demonstrates a greater relative increase over the period in 1992 data were age-sex standardized to the age distribution the younger than in the older group. Among women, the of the 1970 – 1972 survey using the direct method. pattern differed between younger and older women in the two time periods. In the younger group, the greatest change occurred between 1978 – 1979 and 1986 – 1992 while in the Raw sample size older group it occurred in the earlier period with a slight Table 1 gives the raw sample size by 10 y age-groups and sex decline in the latter period. In the overweight category (BMI for the three surveys. 25.0 – 29.9), the situation was stable from 1970 – 1972 to 1978 – 1979 in young men with an increase in the later period. In the older age group, the change was largely in Results the former period, with a stable situation between 1978 – Median height and weight increased progressively over the 1979 and 1986 – 1992. In both younger and older women the three surveys (Table 2). Men were about 6 cm taller and 5.5 kg increase in the overweight category was mainly in the earlier heavier, and women were 4 cm taller and 4 kg heavier, over time period. the period from 1970 – 1972 to 1986 – 1992. The percentile distribution of BMI in the three surveys is Figure 2 demonstrates that, among men, the proportion given in Table 3. In men, an increase is seen at all percentile overweight and obese increased steadily from 1970 – 1972 to values but is greater at the upper end of the distribution. In 1986 – 1992. The greatest change was between 1970 – 1972 women the increase is almost exclusively seen at the upper and 1978 – 1979. The increase occurred rather uniformly in end of the distribution.

Table 2 Median heights and weights for men and women aged 20 – 69, 1970 – 1972, 1978 – 1979 and 1986 – 1992

Men Women

1970 – 1972 1978 – 1979 1986 – 1992 1970 – 1972 1978 – 1979 1986 – 1992

Weight (kg) 73.5 75.4 79.0 59.7 61.3 64.0 Height (cm) 172 173 178 159 160 163

International Journal of Obesity Obesity trends in Canadian adults GM Torrance et al 801

Figure 2 Distribution of overweight and obesity by sex and age group. *1978 – 79 and 1986 – 92 data are age-sex-standardized to 1970 – 72 values.

Impact of education Impact of smoking The increase in the prevalence of obesity was not confined to An increase in the prevalence of obesity over time was seen any one education category, however, the rate of increase most prominently among current and former smokers (Table was greater in those with lower levels of education than in 5). The data illustrate an increase in obesity among both men those with higher levels (Table 4). As a result, one sees a and women never smokers between 1970 – 1972 and 1978 – growing difference in the levels of obesity between educa- 1979, followed by a decline among men and no change tional groups over the time period. When the joint effects of among women between 1978 – 1979 and 1986 – 1992. sex and education are examined, the sub-groups with the greatest relative increase are men in the primary education category (9.6 – 20.0%), and women in the secondary (5.0 – Discussion 13.9%) and post-secondary education category (2.6 – 7.2%) The main finding of this study accords with those of other between 1970 – 1972 and 1986 – 1992. Among men with studies in Canada — that there has been an increase in over- post-secondary education, the prevalence of obesity weight and obesity according to the BMI in the adult popu- decreased from 9.6 to 7.9% between 1978 – 1979 and lation from the early 1970s to the 1990s.9 This study adds to 1986 – 1992. the knowledge gained from other Canadian studies by

Table 3 Percentiles of BMI distribution for men and women aged 20 – 69, 1970 – 1972, 1978 – 1979 and 1986 – 1992

Men Women

1970 – 1972 1978 – 1979 1986 – 1992 1970 – 1972 1978 – 1979 1986 – 1992

1% 17.3 18.2 18.3 17.1 17.0 17.1 5% 19.3 20.0 20.3 18.2 18.5 18.6 10% 20.3 21.1 21.3 19.1 19.4 19.5 25% 22.2 23.0 23.1 21.0 21.3 21.3 50% 24.7 25.2 25.6 23.2 23.9 23.9 75% 27.4 27.6 28.0 26.6 27.5 27.6 90% 29.4 30.6 30.8 30.8 31.5 32.0 95% 31.3 32.7 33.0 33.5 34.1 34.7 99% 34.0 36.7 37.5 39.3 40.6 41.0

International Journal of Obesity Obesity trends in Canadian adults GM Torrance et al 802 grouping existing surveys into separate sets according to not the major factor in the rising rates of overweight and procedures and design and comparing within similar sets, obesity since BMI increased in smokers as well as in never- by extending the time window back to the early 1970s, by smokers and former smokers. using measured anthropometric data rather than self- Another possible causal factor is a decline in physical reported data for the detailed analysis, by refining the activity — either leisure-time activity or that associated analysis of age, gender and educational subgroups and with work, commuting and household activity. Although by examining the impact of the decline of smoking on the surveys examined did not have comparable data on these population weight trends. factors to allow a direct examination of their effects, there is The increase in weight-for-height has occurred in the some evidence from other sources on their secular trends in context of a quite dramatic increase in average height and Canada. Between 1981 and 1997, the proportion of the weight among Canadian adults. The rise in the prevalence of population that participated in leisure-time physical activ- obesity (BMI  30) is due, in part, to a shift of individuals ities appears to have increased from 21 to 37%.20 Most of the upward from the healthy weight category, and — especially increase has been in less-vigorous activities such as walking among women — to a shift toward greater degrees of adipos- and gardening. For those in the paid labour force, it would ity among those who are overweight or obese. The preva- appear that physical activity has declined as a result of lence of obesity has increased across all levels of education, continuing mechanization and computerization of tasks. with the greatest relative increases for men in lower educa- The Canadian work force is increasingly engaged in service tion categories, and for women, in the higher education industries and less in primary industries such as farming, levels. In absolute terms, however, the disparity in the mining and manufacturing. Several other countries have prevalence of obesity between those with lower and higher noted the paradox of increasing overweight and obesity at levels of education has increased during the time period. the same time as the population’s leisure-time physical The international literature has attributed the rise of activity has either remained stable or increased.10,21,22 It obesity to several, chiefly environmental factors. The sort would appear that, at a population level, the increase in of changes implicated are broad social or technological energy expenditure in leisure-time physical activity does not changes that are widely diffused across the population such offset the substantial reductions in occupational and house- as changes in eating patterns, physical activity levels, and hold physical activity energy expenditure. smoking habits. Changes in the racial and ethnic composi- Information on secular trends in Canadian eating patterns tion of the Canadian population due to immigration and is scarce; however, a recent national survey and two provin- different rates of natural increase may also have played some cial surveys undertaken in the early 1990s demonstrate an part. overall reduction in energy consumption since the early In Canada, the decline in smoking rates probably made 1970s. The 1997 – 1998 ‘Food Habits of Canadians’ survey some contribution to the increase. In 1970 – 1972, about indicates a general reduction in energy intake in Canadian 46% of Canadian population aged 20 – 69 smoked; by adults since the early 1970s.23 While the decrease in energy 1978 – 1979 it was down to 44% and by 1986 – 1992 to intake is most pronounced among men aged 20 – 39 y (14% 34%. The proportion of former smokers rose from 16 to 28 decrease), intakes also decreased among women of the same to 33% over the period. Yet smoking cessation was clearly age (7% decrease) and men 40 to 64 y (8% decrease). No

Table 4 Proportion of men and women aged 20 – 69 with BMI  30.0 by education, 1970 – 1972, 1978 – 1979 and 1986 – 1992

Men Women Both

1970 – 1972 1978 – 1979 1986 – 1992 1970 – 1972 1978 – 1979 1986 – 1992 1970 – 1972 1978 – 1979 1986 – 1992

Primary 9.6 14.5 (14.8) 20.0 (20.0) 16.8 20.1 (19.3) 23.1 (19.8) 13.4 17.6 (17.4) 21.5 (20.1) Secondary 6.8 10.5 (9.6) 11.8 (11.2) 5.0 9.3 (10.7) 13.8 (13.9) 5.8 9.9 (10.4) 12.8 (12.8) Post-secondary 5.1 6.4 (9.6) 8.3 (7.9) 2.6 5.0 (6.5) 7.9 (7.2) 4.1 5.9 (8.1) 8.2 (7.7)

Note: numbers in parentheses are age – sex-standardized to 1970 – 1972 values.

Table 5 Proportion of men and women aged 20 – 69 with BMI  30.0 by smoking status, 1970 – 1972, 1978 – 1979 and 1986 – 1992

Men Women Both

1970 – 1972 1978 – 1979 1986 – 1992 1970 – 1972 1978 – 1979 1986 – 1992 1970 – 1972 1978 – 1979 1986 – 1992

Current smoker 5.9 12.8 (13.1) 13.2 (13.4) 10.3 12.1 (12.2) 13.0 (13.9) 7.9 12.5 (14.4) 13.1 (15.8) Former smoker 14.3 10.1 (11.2) 17.3 (17.3) 10.9 10.3 (10.3) 17.0 (16.7) 13.2 10.2 (11.3) 17.2 (17.3) Never smoker 7.4 11.4 (11.4) 8.2 (8.2) 14.6 16.9 (16.0) 15.7 (16.0) 12.2 15.0 (15.3) 12.6 (13.9)

Note: numbers in parentheses are age – sex-standardized to 1970 – 1972 values.

International Journal of Obesity Obesity trends in Canadian adults GM Torrance et al 803 change in energy intake was seen among older women. The sive modifications in lifestyle and creation of supportive report of the Que´bec Nutrition Survey in 1990 shows a drop environments. of 13% in the energy intake of the Que´bec population since the Nutrition Canada Survey (1970 – 1972).24 Similarly, the Nutrition Survey reports a drop of about 10% Acknowledgements 25 over the same time period in the province. Despite these This work was supported in part by Health Canada. apparent reductions in energy intake, actual energy intakes appear to exceed energy expenditure, resulting in excess weight. References 1 World Health Organization. Obesity: preventing and managing the global epidemic. Report of a WHO Consultation on Obesity. WHO: Geneva; 1998. Limitations 2 Popkin BM, Doak CM. The obesity epidemic is a worldwide The main limitation of this study is the uncertainty asso- phenomenon. Nutr Rev 1998; 56:106– 114. ciated with the extent of selection bias and degree to which 3 Birmingham CL, Muller JL, Palepu A, Spinelli JJ, Anis AH. The cost of obesity in Canada. Can Med Assoc J 1999; 160: 483 – 488. it was statistically adjusted by weighting in the data sources. 4 Macdonald SM, Reeder BA, Chen Y, Depre´s J-P, Canadian Heart The 46% response rate of the Nutrition Canada Survey raises Health Surveys Research Group. Obesity in Canada: a descriptive concerns. Men aged 20 – 29 were most under-represented; analysis. Can Med Assoc J 1997; 157(1 Suppl): s3 – s9. however, it is not known whether there was a systematic 5 National Health and Medical Research Council. Acting on Austra- lia’s weight: a strategic plan for the prevention of overweight and non-response bias by the more obese or those with more risk obesity. Commonwealth of Australia: Canberra; 1997. factors. The analysis by education suggests that those with 6 Flegal KM, Carroll MD, Kuczmarski RJ, Johnson CL. Overweight less education were under-represented in the physical mea- and obesity in the United States: prevalence and trends, 1960 – sures sample relative to the interview sample. To the extent 1994. Int J Obes Relat Metab Disord 1998; 22:39– 47. 7 Mokdad AH, Serdula MK, Dietz WH, Bowman BA, Marks JS, that low education is associated with higher BMI, this could Koplan JP. The spread of the obesity epidemic in the United be a source of bias resulting in under-representation of the States, 1991 – 1998. JAMA 1999; 282: 1519 – 1522. overweight and obese. 8 Cairney J, Wade TJ. Correlates of body weight in the 1994 There are also concerns about selective drop-out between National Population Health Survey. Int J Obes Relat Metab Disord 1998; 22: 584 – 591. interview and clinical examination components of the 9 Health Canada and Statistics Canada. Statistical report on the health 1978 – 1979 Canada Health Survey, and uncertainty about of Canadians.Prepared for the Federal, Provincial and Territorial how this influenced post-stratification weighting. The report Advisory Committee on Population Health. Minister of Public of the CHS notes that healthy men aged 20 – 29 were the Works and Government Services: Ottawa, 1999. 10 Simmons G, Jackson R, Swinburn B, Yee RL. The increasing most likely to drop out, a factor which should not inflate prevalence of obesity in New Zealand: is it related to recent BMI. Nevertheless, the percentage obese seems high in some trends in smoking and physical activity? NZ Med J 1996; sub-groups relative to the previous and succeeding surveys. 109(1018): 90 – 92. The pattern of a decline in the prevalence of obesity among 11 Flegal KM, Troiano RP, Pamuk ER, Kuczmarski RJ, Campbell SM. The influence of smoking cessation on the prevalence of over- older women from 1978 – 1979 to 1986 – 1992, for example, weight in the United States. New Engl J Med 1995; 333: 1165 – seems unprecedented in the international literature and 1170. runs contrary to results based on self-report for 1985 to 12 Boyle CA, Dobson AJ, Egger G, Magnus P. Can the increasing 1990 – 1991 from the General Social Survey and the Health weight of Australians be explained by the decreasing prevalence of cigarette smoking? Int J Obes Relat Metab Disord 1994; 18:55– Promotion Survey. 60. 13 Tienboon P, Wahlqvist ML, Rutishauser IH. Self-reported weight and height in adolescents and their parents. J Adolesc Health 1992; Conclusion 13: 528 – 532. 14 Rowland ML. Self-reported weight and height. Am J Clin Nutr Excess weight remains a significant public health problem 1990; 52: 1125 – 1133. among Canadian adults. The prevalence of overweight and 15 Nutrition Canada. Nutrition: a national priority. A Report by obesity is increasing and is an issue for men and women in Nutrition Canada to the Department of National Health and all age groups. While limited, available data indicate increas- Welfare. Information Canada: Ottawa; 1973. 16 Demirjian A. Anthropometry Report: height, weight and body dimen- ing prevalence of obesity despite increased leisure-time phy- sions. A Report from Nutrition Canada. Minister of National sical activity and decreased caloric intake. There is a clear Health and Welfare: Ottawa; 1980. need for up-to-date measured height and weight data on 17 Health and Welfare Canada, Statistics Canada. The health of Canadian adults, as well as children. To better understand Canadians. Report of the Canada Health Survey. Minister of Supply and Services: Ottawa; 1981. and address the factors contributing to the increase in pre- 18 MacLean DR, Petrasovits A, Nargundkar M, Connelly PW, valence of overweight and obesity in Canada, there is an MacLeod E, Edwards A, Hessel P. Canadian heart health surveys: urgent need for updated, ongoing national data on dietary aprofile of cardiovascular risk. Survey methods and data analysis. Can Med Assoc J intake and non-leisure-time physical activity. In the mean- 1992; 146: 1969 – 1974. 19 Reeder BA, Angel A, Ledoux M, Rabkin SW, Young TK, Sweet LE. time, however, efforts are needed to effectively promote Obesity and its relation to cardiovascular disease risk factors in daily physical activity and healthy eating through progres- Canadian adults. Can Med Assoc J 1992; 146: 2009 – 2019.

International Journal of Obesity Obesity trends in Canadian adults GM Torrance et al 804 20 Craig CL, Russell SJ, Cameron C, Beaulieu A. Foundation for joint 23 Gray-Donald K, Jacobs-Starkey L, Johnson-Down L. Food habits action: reducing physical inactivity. Canadian Fitness and Lifestyle of Canadians: reduction in fat intake over a generation. Can J Publ Research Institute: Ottawa; 1999. Health 2000; 91: 381 – 385. 21 Bennett SA, Magnus P. Trends in cardiovascular risk factors in 24 Sante . Les Quebecoises et les Quebecois mangent-ils mieux? Australia. Results from the National Heart Foundation’s Risk Rapport de l’enquete quebecoise sur la nutrition 1990. Sante Quebec: Factor Prevalence Study, 1980 – 1989. Med J Aust 1994; 161: Quebec, 1995. 519 – 527. 25 Nova Scotia Heart Health Program, Nova Scotia Department of 22 Posner BM, Franz MM, Quatromoni PA, Gagnon DR, Sytkowski Health, Health and Welfare Canada. Report of the Nova Scotia PA, D’Agostino RB, Cupples LA. Secular trends in diet and risk Nutrition Survey, 1993. factors for cardiovascular disease: the Framingham Study. JAm Diet Assoc 1995; 95: 171 – 179.

International Journal of Obesity