Association between obesity and high blood pressure - Nature

International Journal of Obesity (1998) 22, 771±777 ß 1998 Stockton Press All rights reserved 0307±0565/98 $12.00 http://www.stockton-press.co.uk/ijo

Association between obesity and high blood pressure: Reporting bias related to gender and age Y Chen1, DC Rennie2, LA Lockinger2 and JA Dosman2 1

Department of Epidemiology and Community Health, Faculty of Medicine, University of Ottawa, Ottawa, Ontario and 2 Centre for Agricultural Medicine and Department of Medicine, University of Saskatchewan, Saskatoon, Saskatchewan, Canada

OBJECTIVE: To examine the validity of self-reported information on obesity and high blood pressure (HBP) in relation to gender and age, and to explore the impacts of their misclassi®cation on the association between obesity and HBP. DESIGN: Community based cross-sectional study. SUBJECTS: 1791 adult subjects living in Humboldt, Saskatchewan, Canada. MEASUREMENTS: Objectively measured HBP was positive if systolic blood pressure (BP) was 140 mm Hg, diastolic BP was 90 mm Hg or the subject was currently using antihypertensive medication. Self-reported HBP was positive if the subjects gave an af®rmative response to the question: `Has a doctor ever said you had high blood pressure?' Body mass index (BMI) was calculated as weight (kg)=height (m)2. Obesity was de®ned as a BMI > 27 kg=m2. Measured obesity and reported obesity were based on measured and self-reported information on height and weight, respectively. RESULTS: The sensitivity of self-reported HBP was low, and was lower for men than for women, and for younger subjects than for older subjects. The speci®city was similar for both genders. Obese individuals had higher sensitivity and lower speci®city than non-obese individuals. The differential misclassi®cation of self-reported HBP caused a bias away from the null when the relative risk for HBP in relation to obesity was estimated. CONCLUSIONS: As a result of the gender- and age-related misclassi®cation of self-reported HBP, the modi®cation role of gender and age on the association between obesity and HBP could be altered. The bias caused by self-reported obesity was relatively small and was either toward or away from the null. Keywords: bias; information bias; obesity; blood pressure; gender; age; epidemiology

Introduction Reporting bias, as one type of information bias, happens frequently, and is a major concern in questionnaire studies. As a result of information bias, a study subject may be misclassi®ed in terms of disease and=or exposure status.1 Dependent on whether or not the classi®cation error of disease or exposure status was the same for each comparison group, the misclassi®cation can be classi®ed into two types: nondifferential and differential.1 Ignoring either type of misclassi®cation may lead to biased results. Previous studies have documented that overweight or obesity is associated with elevated blood pressure (BP) in men and women.2 ± 17 Overweight or obesity in epidemiological studies is often de®ned by using weight and height measures.18 ± 20 A number of studies have examined the validity of self-reported as compared to measured weight and=or height.21 ± 34 As continuous variables, the correlations between the self-reported and measured sources are high; however, Correspondence: Dr Yue Chen, Department of Epidemiology and Community Health, Faculty of Medicine, University of Ottawa, 451 Smyth Road, Ottawa, Ontario K1H 8M5, Canada. Received 3 November 1997; revised 9 March 1998; accepted 19 March 1998

there is a tendency to underestimate weight and overestimate height. Some investigators observed substantial misclassi®cation when they de®ned a categorical variable of obesity by using self-reported information on weight and height.26,30,31,34 Researchers have commonly used self-reported high blood pressure (HBP) in epidemiological studies, for example, the US Nurses' Health Study.35 A study showed that registered nurses underreported their HBP.36 Nurses are more likely to know and reliably report their health conditions than the general population.35 The validity of self-reported information on HBP could be worse in the general population, which has not been well documented as yet. The purpose of the present analysis is to examine the validity of self-reported information on obesity and HBP, and to explore the impacts of gender- and age-related misclassi®cation on the association between obesity and HBP.

Methods A community study was conducted in the town of Humboldt, Saskatchewan in 1993.16,37 The study

Reporting bias of obesity and high blood pressure Y Chen et al

772

population included all town residents aged 6±74 y. This analysis was based on the data from adult subjects aged 18±74 y. In the study of adults, approximately 150 volunteers canvassed the town and asked each eligible subject to participate in the project and to provide a written consent for involvement. If the subjects were not willing to participate in the study, the reasons for nonparticipation were identi®ed. Of the 2327 potential subjects aged 18±74 y (1051 men and 1276 women), 1998 participated, resulting in a response rate of 85.9%. The response rate was higher in women (87.8%) than in men (83.5%). The subjects completed a questionnaire in the home by themselves. The questionnaire asked for information on socio-demographic factors, smoking, alcohol consumption, exercise, the home environment, individual and family history of pulmonary and cardiovascular diseases and diabetes. Each subject reported height and weight as well. Each participant had a clinic visit which included measurements of lung function, blood pressure, height and weight. Guidelines for blood pressure measurements were those recommended by the Canadian Coalition for High Blood Pressure.38 We used standard mercury sphygmomanometers and 15 inch stethoscopes as well as appropriately sized cuffs, based on subjects' arm circumference: a regular adult cuff (22±33 cm), a large adult cuff (33±41 cm) and a thigh cuff (> 41 cm). Prior to BP reading, each subject rested quietly for a minimum of 5 min. We obtained two BP readings and used the mean of the two measurements for this analysis. We measured weight to the nearest 0.1 kg using a calibrated hospital spring scale with subjects dressed in normal indoor clothing but without shoes. We measured height in centimetres against a wall, using a ®xed tape measure with subjects standing in stockinged feet on a hard surface. Objectively measured HBP was positive if systolic BP was 140 mm Hg, diastolic blood pressure was 90 mm Hg or the subject was currently using antihypertensive medication. Self-reported HBP was positive if the subjects gave an af®rmative response to the question: `Has a doctor ever said you had high blood pressure?'

We calculated BMI as weight (kg)=height (m)2. We de®ned obesity as a BMI > 27 kg=m2. Objectively measured obesity and reported obesity were based on objectively measured and self-reported information on height and weight, respectively. Assuming that objectively measured HBP and obesity were valid, we examined construct validity by calculating sensitivity, speci®city and positive and negative predictive values of self-reported HBP and obesity. Sensitivity was the proportion of self-reported HBP or obesity among all objectively measured HBPs or obesity. Speci®city was the proportion classi®ed as without self-reported HBP or obesity among all of those not classi®ed as HBP or obesity by objective measuring. Positive predictive value was the proportion of those who self-reported HBP or obesity, and actually had objectively measured HBP or obesity. Negative predictive value was the proportion of those who self-reported no HBP or obesity and did not have measured HBP and obesity. Furthermore, assuming that the calculation of relative risk based on objective measures of obesity and HBP produces a valid estimate, bias is calculated by: RR* ÿ RR=RR RR* is a relative risk estimate based on self-reported information either on obesity or HBP or both. RR is a relative risk estimate based on both objectively measured obesity and HBP.

Results Overall, 1998 adult subjects participated in the study. We excluded 207 who did not answer the questions about HBP, height and=or weight. Table 1 shows the comparison of objectively measured and self-reported HBP and obesity. In men, the prevalence of reported HBP was markedly lower than that of objectively measured HBP. The difference decreased with increasing age, namely, the largest difference was observed with the 18±34 y age group, the lowest with the 55±74 y group and an intermediate difference with the 35±54 y group. For women, the prevalence of self-reported HBP was also lower than that of

Table 1 Number and prevalence (%) of self-reported and objectively measured high blood pressure (HBP) and obesity by gender and age HBP

Obesity

Gender

Age (y)

n

Measured

Reported

Measured

Reported

Men

18 ± 34 35 ± 54 55 ± 74 Total 18 ± 34 35 ± 54 55 ± 74 Total

225 282 296 803 284 341 363 988

25.8 30.9 53.7 37.9 6.7 19.9 58.4 30.3

8.0 14.2 33.1 19.4 9.5 13.5 41.0 22.5

38.2 56.0 64.5 54.2 26.4 36.1 54.5 40.1

35.1 49.6 58.4 48.8 25.0 34.0 49.6 37.1

Women


measured HBP; however, the difference was less striking when compared to men. The prevalence of reported obesity was lower than that of measured obesity for both genders (Table 1). The difference seemed larger for older subjects than for younger subjects. Table 2 presents the gender- and age-speci®c sensitivity and speci®city of obesity based on reported height and weight, when compared to obesity based on objectively measured height and weight. The sensitivity was 86.0% for men and 87.9% for women, and the speci®city was 95.1% for men and 96.8% for women. The positive and negative predictive values were 95.4% and 85.2% for men, respectively. For women, they were 97.6% and 85.9%, respectively. The data did not show clear patterns of the validity across gender, age and HBP strata. Table 3 shows the gender- and age-speci®c sensitivity and speci®city of reported HBP as compared to measured HBP for persons with and without measured obesity. Overall, the sensitivity for reported HBP was low, and lower for men (41.1%) than for women (61.2%). The speci®city was relatively high and showed little difference between genders (93.8% for men and 94.3% for women). Men had a lower negative predictive value (72.3%) than did women (84.9%). The positive predictive value was similar for men (80.1%) and women (82.4%). The sensitivity was higher in obese subjects than in non-obese subjects (men: 46.9% vs 30.6%; women: 68.2% vs 50.8%), while the speci®city was lower in obese vs non-obese

subjects (men: 90.0% vs 97.3%; women: 90.3% vs 96.2%). The sensitivity increased with increasing age in both obese and non-obese men. There was no discernible trend for sensitivity in obese women. The positive predictive value was improved in older subjects while the negative predictive value was relatively high in younger subjects. We estimated relative risk and 95% con®dence intervals for HBP in relation to obesity based on the objectively measured and self-reported information (Table 4). For men, the relative risk was similar whether obesity was classi®ed by using objectively measured or self-reported height and weight. However, the relative risk increased if HBP was based on self-reporting, particularly in the younger age group. In women as a whole, the relative risk for selfreported HBP was greater than for measured HBP, but the difference was small when compared to that found in men and was not consistent across the age groups. Table 5 shows the reporting bias in the association between obesity and HBP. If the classi®cation of obesity was based exclusively on reported information, the bias for the association between obesity and HBP was small for both men and women with an absolute value less than 10%, and showed no consistent direction. When HBP was based on self-reporting, the bias in men was very large in the 18±34 y age group at 367% and 434% depending on whether obesity was determined based on measured or selfreported height and weight. The bias in men decreased

Table 2 Sensitivity, speci®city, and positive and negative predictive values (PVs) of self-reported obesity, as compared to objectively measured obesity by gender, age and high blood pressure (HBP) Men Measured HBP No

Age (y) 18 ± 34

35 ± 54

55 ± 74

Total

Reported obesity No Yes

No Yes

No Yes

No Yes

Accuracy measure

Sensitivity Speci®city Positive PV Negative PV Sensitivity Speci®city Positive PV Negative PV Sensitivity Speci®city Positive PV Negative PV Sensitivity Speci®city Positive PV Negative PV

Women Measured HBP Yes

No

Measured obesity

Yes Measured obesity

No

Yes

No

Yes

No

Yes

No

Yes

107 4

8 48

26 2

5 25

198 4

8 55

7 0

0 12

19 83

29 2

6 50

183 5

10 75

29 1

12 69

47 2

11 99

76 6

11 58

80 3

39 200

102 6

22 174

457 15

29 188

116 4

88 5

53 3

248 12

85.7% 94.4% 92.3% 93.0% 81.4% 94.6% 94.3% 82.2% 85.2% 94.6% 95.8% 81.5% 83.7% 95.4% 94.3% 86.4%

83.3% 92.9% 92.6% 83.9% 89.3% 93.5% 96.2% 82.9% 90.0% 95.9% 98.0% 81.0% 88.8% 94.4% 96.7% 82.3%

87.3% 98.0% 93.2% 96.1% 88.2% 97.3% 90.4% 94.8% 84.1% 92.7% 90.6% 87.4% 86.6% 96.8% 92.6% 94.0%

100.0% 100.0% 100.0% 100.0% 92.1% 96.7% 92.2% 90.6% 87.6% 96.4% 97.4% 83.3% 89.4% 96.7% 97.6% 85.9%

3 35

16 113

19 160

773


774

Table 3 Sensitivity, speci®city and positive and negative predictive values (PVs) of self-reported high blood pressure (HBP) as compared to objectively measured HBP by gender, age and obesity Men Measured obesity No

Age (y)

Reported HBP

18 ± 34

No Yes

35 ± 54

Sensitivity Speci®city Positive PV Negative PV

No Yes

55 ± 74


No Yes

Total

Accuracy measure


No Yes


Women Measured obesity Yes

No

Measured HBP

Yes Measured HBP

No

Yes

No

Yes

No

Yes

No

Yes

109 2

27 1

49 7

22 8

192 10

7 0

55 8

3 9

23 8

93 9

36 20

183 5

17 13

77 8

25 24

73 8

46 64

79 3

35 48

64 5

75 33

215 24

104 92

454 18

59 61

196 21

90 3

54 2

253 7

3.6% 98.2% 33.3% 80.1% 25.8% 96.8% 72.7% 79.6% 49.0% 96.4% 92.3% 68.4% 30.6% 97.3% 82.5% 77.1%

26.7% 87.5% 53.3% 69.0% 35.7% 91.2% 69.0% 72.1% 58.2% 90.1% 88.9% 61.3% 46.9% 90.0% 79.3% 67.4%

0.0% 95.0% 0.0% 96.5% 43.3% 97.3% 72.2% 91.5% 57.8% 96.3% 94.1% 69.3% 50.8% 96.2% 77.2% 88.5%

75.0% 87.3% 52.9% 94.8% 52.6% 90.6% 71.4% 81.1% 72.1% 92.8% 94.9% 64.0% 68.2% 90.3% 85.3% 77.5%

18 20

36 93

57 122

Table 4 Relative risk (95% con®dence intervals) for high blood pressure (HBP) in relation to obesity based on objectively measured and self-reported information Information Information source HBP Men Measured

Age (y) Obesity

18^34

35^54

55^74

Total

Measured

1.73 (1.12 ± 2.69) 8.08 (2.41 ± 27.10) 1.61 (1.04 ± 2.49) 9.24 (2.76 ± 30.96)

1.42 (0.98 ± 2.05) 2.07 (1.08 ± 3.98) 1.51 (1.05 ± 2.16) 2.67 (1.39 ± 5.14)

1.23 (0.97 ± 1.57) 1.52 (1.04 ± 2.23) 1.24 (0.99 ± 1.55) 1.54 (1.07 ± 2.20)

1.54 (1.27 ± 1.86) 2.45 (1.76 ± 3.42) 1.52 (1.27 ± 1.83) 2.59 (1.88 ± 3.55)

4.78 (1.95 ± 11.68) 4.74 (2.27 ± 9.88) 5.14 (2.11 ± 12.56) 3.75 (1.84 ± 7.63)

2.24 (1.47 ± 3.43) 2.76 (1.59 ± 4.78) 2.18 (1.43 ± 3.32) 2.76 (1.60 ± 4.74)

1.30 (1.08 ± 1.55) 1.60 (1.22 ± 2.09) 1.23 (1.03 ± 1.46) 1.55 (1.20 ± 2.00)

2.23 (1.84 ± 2.70) 2.71 (2.12 ± 3.45) 2.06 (1.70 ± 2.48) 2.48 (1.96 ± 3.14)

Reported

Measured

Measured

Reported

Reported

Reported

Women Measured

Measured

Reported

Measured

Measured

Reported

Reported

Reported

with increasing age and approximated 25% in the 55±74 y age group. The direction of the bias due to reported HBP was always away from the null in men. In women, the bias for the association between obesity and reported HBP was less striking and showed no clear trend related to age. The direction of bias was either toward or away from the null.

Discussion The overall sensitivity and speci®city for reported obesity was slightly higher in women than in men. Compared to men, women are more likely to underreport their weight, but are more accurate in reporting


775

Table 5 Reporting bias* in the relationship between obesity and high blood pressure (HBP) Information source HBP Men Measured Reported Measured Reported Women Measured Reported Measured Reported

Age (y) Obesity

18^34

35^54

Measured Measured Reported Reported

3.67 70.07 4.34

0.46 0.06 0.88

Measured Measured Reported Reported

70.01 0.08 70.22

0.23 70.03 0.23

55^74 (Reference)

(Reference)

Total

0.24 0.01 0.25

0.59 70.01 0.68

0.23 70.05 0.19

0.22 70.08 0.11

*(RR* 7 RR)=RR. RR* is a relative risk estimate based on self-reported information either on obesity or HBP or both. RR is a relative risk estimate based on both objectively measured obesity and HBP.

their height, which results in greater accuracy of relative weight categories.31 Our data showed a decreased validity of reported obesity in younger men as compared to their female counterparts. The validity of reported obesity in terms of sensitivity and speci®city was similar for men and women in the 55± 74 y age groups. The overestimation of height and underestimation of weight results in an underestimation of relative weight.26,31 Previous studies have indicated that using the self-reported relative weight measures as a continuous variable may have little impact on results; however, substantial misclassi®cation can occur when using relative weight as a categorical variable.26,31 In our case, the bias of the relative risk estimation for HBP due to misclassi®cation of obesity was less than 10% in all gender and age groups, if HBP was objectively measured. The bias was either positive or negative. However, if HBP was self-reported, the bias was as high as 29% in men aged 35±54 y ((2067±2.07)=2.07). Our data showed that the subjects remarkably underreported HBP, and that this trend was more marked in men than in women and more marked in younger subjects than in older subjects. There is some evidence for the validity of self-reported HBP in the existing literature. Based on a small sample of the US Nurses' Health Study cohort, Colditz et al36 found that 7% of the female nurses who had not reported HBP, actually had BP elevations. Those registered nurses were more knowledgeable about their medical conditions than are the general population. In the present study, we found that the overall speci®city and negative predictive value of reported HBP were similar in men and women, while the sensitivity and positive predictive value were much worse in men than in women. The sensitivity and positive predictive value of self-reported HBP in men were the worst in the youngest age group and improved with age, but the trend was less clear in obese women. Obese men and women were more likely to know their HBP condition than were their non-obese counterparts, indicating that the misclassi®cation was differential. The speci®city of reported HBP was relatively stable across the gender and age strata. The data suggested that self-

reported HBP is biased, which is related to gender, age and obese status. It seems a reasonable thought that if individuals are older and obese, they may seek medical attention to a greater degree than younger and non-obese persons do. We could not identify factors which might affect the measurement of blood pressure and the de®nition of HBP and which were responsible for the genderand age-related differences in the validity of reported HBP. It has been indicated that incorrect cuff size may lead to an underestimation of blood pressure in lean individuals and overestimation in obese individuals.35 However, the sensitivity of self-reported HBP was higher in obese subjects than in non-obese subjects in the present study. The differential misclassi®cation of self-reported HBP appeared to cause biased estimates of the association between obesity and HBP. The association was stronger based on self-reported HBP vs measured HBP and the bias was always away from the null. The bias of a relative risk estimate was larger in men (59%) than in women (22%), which was tallied with the difference in the validity of self-reported HBP between men and women. Since obese individuals were more likely to know their status with regard to HBP, the estimate based on reporting data suggested an arti®cially stronger association than that based on measuring data. The relationship between obesity and HBP was stronger in the younger subjects than in the older subjects, which is consistent with previous studies.10,12,14 ± 16 However, the trend in men became more prominent when the HBP was based on selfreporting. The younger the men, the lower the validity of self-reported HBP. As mentioned above, a higher degree of the differential misclassi®cation of HBP caused a stronger association between obesity and HBP. The validity of self-reported HBP in women shows no clear age-related trend. It seems that the estimation of age-related association between obesity and HBP based on self-reporting in women was more reliable compared to men. The modi®cation of gender on the association between obesity and HBP is also of great interest.


776

Based on the measured data, the association was stronger in women than in men, particularly in younger age groups. For those aged 18±24 y, the relative risk was 4.78 in women compared to 1.73 in men. In the same age group, the relative risk estimate was similar in women (RR 4.74) if HBP was based on self-reporting but increased dramatically in men (RR 8.08), which suggests that the conclusion of gender-related association can be altered because of gender-related misclassi®cation of self-reported HBP.

Conclusion In summary, the misclassi®cation of self-reported HBP was extensive in the general population and was gender- and age-related. When examining the association between obesity and HBP, the misclassi®cation was differential, which caused the bias away from the null. The misclassi®cation of HBP varied across gender and age strata. As a result, the modi®cation role of gender and age on the association between obesity and HBP could be altered. The bias caused by misclassi®cation of reported obesity was relatively small, and was either toward or away from the null. Acknowledgements

The authors thank D. Still (Mayor), W. Herman (Town Administrator), W. German, N. Trach, J. Hergott (Co-Chairs of the local organizing committee), M. Gillis-Cipywnyk and all of the volunteers who provided canvassing and other ®eld work. This work was supported by a grant from the Saskatchewan Health Services Utilization and Research Commission. Yue Chen is a National Health Research Scholar of the National Health Research and Development Program (NHRDP), Health Canada. References

1 Kleinbaum DG, Kupper LL, Morganstern H. Epidemiological Research. Lifetime Learning Publication: Belmont, 1982, pp 220 ± 241. 2 Johnson AL, Corononi JC, Cassels JC, Tyroler HA, Heyden S, Hames CG. In¯uence of race, sex and weight on blood pressure behavior in young adults. Am J Cardiol 1975; 35: 523 ± 530. 3 Hypertension Detection and Follow-up Program Cooperative Group. Race, education and prevalence of hypertension. Am J Epidemiol 1977; 106: 351 ± 361. 4 Hsu PH, Mathewson FAL, Rabkin SW. Blood pressure and body mass index patterns ± a longitudinal study. J Chronic Dis 1977; 30: 93 ± 113. 5 Stamler R, Stamler J, Riedlinger WF, Algera G, Roberts RH. Weight and blood pressure: ®ndings in hypertension screening of 1 million Americans. JAMA 1978; 240: 1607 ± 1610. 6 MacMahon SW, Blacket RB, Macdonald GJ, Hall W. Obesity, alcohol consumption and blood pressure in Australian men and women: the National Heart Foundation of Australia Risk Factor Prevalence Study. J Hypertens 1984; 2: 85 ± 91.

7 Neser WB, Thomas J, Semenya K, Thomas DJ, Gillum RF. Obesity and hypertension in a longitudinal study of black physicians: the Meharry Cohort Study. J Chronic Dis 1986; 39: 105 ± 113. 8 Dyer AR, Elliott P, Shipley M for the INTERSALT Cooperative Research Group. Body mass index versus height and weight in relation to blood pressure: ®ndings for the 10 079 persons in the INTERSALT Study. Am J Epidemiol 1990; 131: 589 ± 596. 9 Stamler J. Epidemiologic ®ndings on body mass and blood pressure in adults. Ann Epidemiol 1991; 132: 347 ± 362. 10 Ford ES, Cooper RS. Risk for hypertension in a national cohort study. Hypertension 1991; 18: 598 ± 606. 11 Bunker CH, Ukoli FA, Nwankwo MU, Omene JA, Currier GW, Holi®eld-Kennedy L, Freeman DT, Vergis EN, Yeh LL, Kuller LH. Factors associated with hypertension in Nigerian civil servants. Prev Med 1992; 21: 710 ± 722. 12 He J, Klag MJ, Whelton PK, Chen JY, Qian MC, He GQ. Body mass and blood pressure in a lean population in southwestern China. Am J Epidemiol 1994; 139: 380 ± 389. 13 Jones DW, Kim JS, Andrew ME, Kim SJ, Hong YP. Body mass index and blood pressure in Korean men and women: the Korean National Blood Pressure Survey. J Hypertens 1994; 12: 1433 ± 1437. 14 Epstein FH, Francis T Jr, Hayner NS, Johnson BC, Kjelsberg MO, Napier JA, Ostrander LD Jr, Payne MW, Dodge HJ. Prevalence of chronic diseases and distribution of selected physiologic variables in a total community, Tecumseh, Michigan. Am J Epidemiol 1965; 81: 307 ± 322. 15 Havlik RJ, Hubert HB, Fabsitz RR, Feinleib M. Weight and hypertension. Ann Intern Med 1983; 98: 855 ± 859. 16 Chen Y, Rennie DC, Reeder BA. Age-related association between body mass index and blood pressure: The Humboldt Study. Int J Obes 1995; 19: 825 ± 831. 17 Kaufman JS, Durazo-Arvizu RA, Rotimi CN, McGee DL, Cooper RS. Obesity and hypertension prevalence in populations of African origin. Epidemiology 1996; 7: 398 ± 405. 18 Bray GA. De®nition, measurement and classi®cation of the syndrome of obesity. Int J Obes 1978; 2: 99 ± 112. 19 Harrison GG. Height-weight tables. Ann Intern Med 1985; 103: 989 ± 994. 20 James WPT. Epidemiology of obesity. Int J Obes 1992; 16 (suppl. 2): S23 ± S26. 21 Stunkard AJ, Albaum JM. The accuracy of self-reported weights. Am J Clin Nutr 1981; 34: 1593 ± 1599. 22 Pirie P, Jacobs D, Jefferey R, Hannan P. Distortion in selfreported height and weight data. J Am Diet Assoc 1981; 78: 601 ± 606. 23 Schlichting P, Hùilund-Carlsen PF, Quaade F. Comparison of self-reported height and weight with controlled height and weight in women and men. Int J Obes 1982; 5: 67 ± 76. 24 Palta M, Prineas RJ, Berman R, Hannan P. Comparison of self-reported and measured height and weight. Am J Epidemiol 1982; 115: 223 ± 230. 25 Stewart AL. The reliability and validity of self-reported weight and height. J Chronic Dis 1982; 35: 295 ± 309. 26 Stewart AW, Jackson RT, Ford MA, Beaglehole R. Underestimation of relative weight by use of self-reported height and weight. Am J Epidemiol 1987; 125: 122 ± 126. 27 Jalkenen L, Tuomilehto J, Tanskanen A, Puska P. Accuracy of self-reported body weight compared to measured body weight. A population survey. Scand J Soc Med 1987; 15: 191 ± 198. 28 Le Marchand L, Yoshizawa CN, Nomura AMY. Validation of body size information on driver's licenses. Am J Epidemiol 1988; 128: 874 ± 877. 29 Kuskowska-Wolk A, Karlsson P, Stolt M, RoÈssner S. The predictive validity of body mass index based on self-reported weight and height. Int J Obes 1989; 13: 441 ± 453. 30 Rowland ML. Self-reported weight and height. Am J Clin Nutr 1990; 52: 1125 ± 1133.


31 Neito-Garcia FJ, Bush TL, Keyl PM. Body mass de®nitions of obesity: sensitivity and speci®city using self-reported weight and height. Epidemiology 1990; 1: 146 ± 152. 32 Kaskowska-Wolk A, Bergstrom R, Bostrom G. Relationship between questionnaire data and medical records of height, weight and body mass index. Int J Obes 1992; 16: 1 ± 9. 33 Zhang J, Feldblum PJ, Fortney JA. The validity of selfreported height and weight in perimenopausal women. Am J Public Health 1993; 83: 1052 ± 1053. 34 Roberts RJ. Can self-reported data accurately describe the prevalence of overweight? Public Health 1995; 109: 275 ± 284. 35 Fiebach NH, Hebert PR, Stampfer MJ, Colditz GA, Willett WC, Rosner B, Speizer FE, Hennekens CH. A prospective study of high blood pressure and cardiovascular disease in women. Am J Epidemiol 1989; 130: 646 ± 654.

36 Colditz GA, Martin P, Stampfer MJ, Willett WC, Sampson L, Rosner B, Hennekens CH, Speizer FE. Validation of questionnaire information on risk factors and disease outcomes in a prospective cohort study of women. Am J Epidemiol 1986; 123: 894 ± 900. 37 Chen Y, Rennie DC, Dosman JA. In¯uence on environmental tobacco smoke on asthma in nonallergic and allergic children. Epidemiology 1996; 7: 536 ± 539. 38 Canadian Coalition for High Blood Pressure. Prevention and control. Guidelines for the measurement of blood pressure. 1988.

777