Association of anthropometric indices with elevated blood ... - Nature

International Journal of Obesity (1997) 21, 674±680 ß 1997 Stockton Press All rights reserved 0307±0565/97 $12.00

Association of anthropometric indices with elevated blood pressure in British adults BD Cox1, MJ Whichelow1, M Ashwell2, AT Prevost3 and SRE Lejeune2 1

Department of Community Medicine, Institute of Public Health, Robinson Way, Cambridge, CB2 2SR, UK; 2 Ashwell Associates, Ashwell Street, Ashwell, North Herts, SG7 5PZ, UK and 3 Centre for Applied Medical Statistics, Department of Community Medicine, Institute of Public Health, Robinson Way, Cambridge, CB2 2SR, UK

OBJECTIVE: To investigate which of six anthropometric indices was most strongly associated with elevated blood pressure and frank hypertension in a representative population sample of young and middle-aged British adults. SUBJECTS: A representative random sample of British adults (2712 men and 3279 women) aged between 18 and 64 y, who were resident in England, Scotland and Wales, were studied in the 1984±85 Health and Lifestyle Survey. MEASUREMENTS: Following an interview where demographic, health and lifestyle details were recorded, measurements of height, weight and waist and hip circumference were made by a nurse at a home visit, where blood pressure and medication to control blood pressure were also recorded. BMI and the ratios of waist circumference to height (WHTR) and waist circumference to hip (WHR) were calculated. Respondents with a blood pressure above 140 mm Hg (systolic) and/or 90 mm Hg (diastolic) or who were being treated for hypertension were classi®ed as having elevated blood pressure. RESULTS: For men and women aged 18±39 and 40±64 y the prevalence of elevated blood pressure increased across the quintiles of BMI, weight, waist, WHTR and WHR was P < 0.001 for each, with waist and WHTR having the highest odds ratios. Waist and height adjusted for each other were independently related to the prevalence of elevated blood pressure in 40±64 y old men and women. Height, on its own, was inversely related (P < 0.05) only in the 40±64 y old men. The age adjusted partial correlations between systolic and diastolic blood pressure measurements and the measurements of BMI, weight, waist, WHTR and WHR were close and signi®cant, P < 0.001 for each. The ranking and signi®cance of the correlations were hardly affected by excluding the treated hypertensives. CONCLUSIONS: The prevalence of elevated blood pressure was associated with quintiles of BMI, waist, WHTR, WHR and weight, with WHTR and WHR having the highest odds ratios. Waist and height were independently related to the prevalence of elevated blood pressure. Keywords: blood pressure; hypertension; waist height ratio; waist circumference; BMI; waist hip ratio

Introduction Adiposity has long been associated with the risk of developing cardiovascular disease, but which measure of obesity is the best predictor of cardiovascular disease has recently become the subject of much investigation. BMI has been shown to be associated with cardiovascular risk factors including elevated blood pressure1±7 and is routinely used in the clinic situation as an indicator of health risk. However, BMI does not necessarily re¯ect regional adiposity and as Segal et al8 showed, men with a high BMI but a normal proportion of body fat have a low cardiovascular risk pro®le, similar to non-obese men, whereas obese men with a high body fat (> 25%) have higher risk values. Many studies have found that anthropometric measures which include an assessment of fatness, particularly the regional distribution of fat, show closer Correspondence: Dr BD Cox. Received 2 Ocober 1996; revised 7 April 1997; accepted 17 April 1997

associations with a range of risk factors and are better predictors of mortality. Some studies have found that the waist to hip ratio (WHR) is a better indicator than BMI of risk factors1,9 and a better predictor of myocardial infarction and mortality in women10 and of CHD in elderly men.4 In younger men the regional distribution of fat seems less important as BMI and WHR are both well correlated with risk factors.3,11 Other studies have found close associations with the waist to thigh ratio,12,13 the waist circumference12,14± 16 and the waist to height ratio (WHTR).5,17 Short stature has been shown to be related to the occurrence of coronary heart disease in men4 and women.18 Comparison of skinfold thicknesses at a number of sites, as an indicator of regional fat distribution, has been employed to show associations with cardiovascular disease.12,13,19,20 The amount of intra-abdominal adipose tissue (IAT) as determined by computed tomography has been found to be closely associated with cardiovascular risk factors, including elevated blood pressure, in women,21 strengthening the hypothesis that it is this adipose tissue which is associated in some way with cardiovascular disease. IAT correlates closely with waist measurements and changes in waist

Anthropometry and blood pressure BD Cox et al

circumference parallel those in IAT,22,23 but the waist to height ratio has been shown to have the strongest association with IAT.24 In Britain, the Government's targets are for reductions in both obesity and blood pressure by the year 2005,25 in the face of a rising prevalence of obesity.26 From the public health point of view, it is desirable to identify a simple non-invasive, and non-intrusive anthropometric measurement which gives the most reliable indication of risk. Data from the nationwide random sample of adult British subjects who took part in the Health and Lifestyle Survey27 have been analysed to compare the associations of elevated blood pressure and frank hypertension with six anthropometric indices: BMI, height, weight, the waist to hip ratio, waist circumference and the waist to height ratio. A brief report of these ®ndings has been presented elsewhere.28

Methods The 1984±85 Health and Lifestyle Survey was carried out on a representative sample of 9003 adults, aged 18±97 y, resident in England, Scotland and Wales. They were randomly selected from two wards in each of 198 constituencies. The response rate at viable addresses was 77.6%. The respondents were interviewed in their own homes by a trained interviewer using a structured questionnaire. The topics covered were socio-demographic details, past and present illness, current health status, attitudes to health and dietary, smoking, drinking, exercise and leisure activities. Following this a nurse visited the consenting 7414 respondents and carried out measurements of height, weight and waist girth, as well as blood pressure and lung function. Current medication was also recorded by the nurse. This report covers the 2712 men and 3279 women aged 18±64 y on whom valid anthropometric and blood pressure measurements were made. Hip circumference was also measured in 1659 and 2186 of these men and women respectively. A full description of the survey is described elsewhere.27 Height was measured to the nearest 1 mm in stockinged feet, using a portable stadiometer, with the subject's head positioned so that the eye was level with the external auditory meatus (the Frankfurt line). Weight was measured with electronic scales to the nearest 100 g, in indoor clothing, with a correction made for what the nurse judged to be heavy or light clothing. The scales were regularly calibrated. Waist circumference was measured halfway between the lowest rib and the top of the iliac crest using a plastic measure. Hip circumference was measured at the widest part. From the measurements Body Mass Index (BMI), the waist/hip ratio (WHR) and the waist/height ratio (WHTR) were calculated. Blood pressure measurements were made with the

respondent relaxed and seated upright with the right forearm supported in a resting position. An automatic blood pressure measuring device, the Datascope `Accutorr', was employed to record four measurements of systolic, diastolic and mean blood pressure and pulse rate. The lowest of each of the readings were those used for analysis. The use of the `Accutorr' eliminated observer error and the dif®culties of hearing systolic and diastolic sounds in a noisy domestic situation. The Accutorr machines were evaluated by comparison with intra-arterial catheterization, and checked regularly against a mercury manometer. Details of any medication for hypertension or which could affect the blood pressure were recorded. Respondents who were normotensive at measurement, and with no history of hypertension, but who were on therapy which could affect (lower) blood pressure were excluded, 21 men and 28 women. Those respondents in whom valid anthropometric measurements could not be made because of pregnancy, missing limbs or an inability to stand upright, were also excluded, 22 men and 127 women. The survey was carried out in three waves, and hip measurements were only made on respondents in waves two and three. Those with a history of hypertension and currently on medication to lower blood pressure, or a systolic blood pressure over 140 mm Hg and a diastolic blood pressure over 90 mm Hg, were classi®ed as having elevated blood pressure. Respondents with a history of treated hypertension or a systolic blood pressure above 160 mm Hg and/or diastolic blood pressure above 95 mm Hg were classi®ed as hypertensive. The smoking categories used were non-smokers: never having smoked as much as one cigarette, cigar or pipe a day for six months, smokers: those currently smoking at least one cigarette, pipe or cigar a day, and ex-smokers: those who used to smoke regularly, for more than six months. For each anthropometric index, quintiles were derived separately for men and women from data from the whole survey population, aged 18±97 y.

Statistical analysis

Logistic regression analysis (SPSS) was used to describe trends of the prevalence of elevated blood pressure across the quintiles of each anthropometric index adjusting for age, entered as a continuous variable, in each age group. The odds ratios give the incremental odds of the occurrence of elevated blood pressure from one quintile to the next. The relationship of measured systolic and diastolic blood pressure to the measured anthropometric indices was calculated by partial linear correlation, adjusting for age in each group. The co-adjusted effects of height and waist with the prevalence of elevated blood pressure was also examined using logistic regression analysis, again adjusting for age.

675


676

Results Table 1 shows the ranges of values in each quintile for the six anthropometric indices studied. Figure 1(a) and (b) shows (unadjusted) proportions of those aged 18±39 y with elevated blood pressure in each quintile for the six anthropometric indices. For each index, except height, the prevalence of elevated blood pressure rose from the lowest to the highest anthropometric quintile, and in each quintile the prevalence was higher in the men than in the women. In the middle aged, 40±64 y group, the proportions of men and women with elevated blood pressure were higher in each quintile (Figure 2(a) and (b)) than for younger subjects. The prevalence of elevated blood pressure increased from the ®rst to the ®fth quintile for each index except height. The proportion of hypertensives in each quintile is also shown, and reveals similar trends. The prevalence of elevated blood pressure and hypertension increased with decreasing height in men and women. The trends across the quintiles for elevated blood pressure in two age groups are shown in Table 2. For each index, except height, the increments of the prevalence of elevated blood pressure from quintile 1, through to quintile 5 were highly signi®cant. The odds ratios were higher for all indices for women than for men, re¯ecting a steeper gradient. In all groups the odds ratios were higher for waist and WHTR than for the other indices, including WHR. Analyses of BMI, WHTR, waist, weight and height on the reduced sample for whom WHR measurements were available revealed that the odds ratios and con®dence intervals were very close to those using the full sample. There was a signi®cant inverse association with height only in middle aged men. Correlations between measured systolic and diastolic blood pressure levels and the anthropometric indices within each age group, are shown in Table 3. Except for height, systolic and diastolic blood pressure levels were closely associated with all the indices. In men the correlations with systolic blood pressure were closer in the younger age group than in

Figure 1 Prevalence of elevated blood pressure (SBP > 140 mm Hg and/or DBP > 90 mm Hg or treatment for hypertension) by quintile of waist/height (WHTR), waist (W), waist/hip (WHR), BMI, weight (WT) and height (HT) in (a) men aged 18±39, and (b) women aged 18±39.

the middle aged group. Overall, for men, the correlations with WHTR were marginally closer than for waist and BMI and considerably closer than for WHR or weight. In the women, BMI showed the closest

Table 1 Anthropometric indicesÐrange of values in each quintile (Q) (WHTR waist/height, WHR waist/hip) Men Q

BMI

Weight-kg

Height-cm

Waist-cm

WHTR

WHR

1 2 3 4 5

10.00±21.68 21.69±23.65 23.66±25.37 25.38±27.53 27.54±60.00

30.0±64.5 64.6±70.6 70.7±76.1 76.2±84.1 84.2±160.0

110.0±167.5 167.6±171.6 171.7±174.5 174.6±179.5 179.6±220.0

40.0±79.9 80.0±86.0 86.1±91.5 91.6±98.0 98.1±145.0

0.15±0.46 0.47±0.49 0.50±0.52 0.53±0.56 0.57±1.00

0.60±0.86 0.87±0.89 0.90±0.92 0.93±0.95 0.96±1.30

20.0±68.0 68.1±73.0 73.1±78.2 78.3±86.1 86.2±145.0

0.15±0.42 0.43±0.45 0.46±0.49 0.50±0.54 0.55±1.00

0.60±0.74 0.75±0.77 0.78±0.80 0.81±0.84 0.85±1.30

Women 1 2 3 4 5

10.00±20.89 20.90±22.63 22.64±24.57 24.58±27.41 27.42±60.00

25.0±53.4 53.5±58.4 58.5±63.4 63.5±70.9 71.0±160.0

110.0±155.0 155.1±158.8 158.9±162.0 162.1±166.0 166.1±210.0


treated hypertensives, whose blood pressure would have been modi®ed by treatment, slightly reduced the closeness of correlations but did not substantially alter the ranking of the indices. Although the distribution of smokers, ex-smokers and non-smokers was not uniform across the quintiles of the anthropometric indices (except for height and WHR) adjusting for smoking did not alter the associations with hypertension or elevated blood pressure. Thus no independent association of smoking with blood pressure could be detected. Table 4 reveals that there are independent effects of waist and height in relation to the prevalence of elevated blood pressure in the middle aged group, with the effect stronger for men than women. There was a greater prevalence of elevated blood pressure in the shorter subjects. There were no signi®cant trends in the younger age group.

(a)

Discussion

(b) Figure 2 Prevalence of elevated blood pressure (SBP > 140 mm Hg and/or DBP > 90 mm Hg or treated for hypertension) and hypertension (SBP > 160 mm Hg and/or DBP > 95 mm Hg or treatment for hypertension) by quintile of waist/height (WHTR), waist (W), waist/hip (WHR), BMI, weight (WT) and height (HT) in (a) men aged 40±64 and (b) women aged 40±64.

correlations, followed by WHTR and waist, and weight for the 18±39 y old group. Whereas the correlations for WHR for men were fairly similar to those of the other indices (except height) the WHR correlations for women were much lower. Exclusion of the

The present study shows that all the anthropometric indices, except height, were closely associated with the prevalence of elevated blood pressure with WHTR and waist having higher OR values than other anthropometric indices. As short stature is associated with an increased risk of elevated blood pressure, in the 40±64 y old group, when waist is taken into account WHTR has been shown to be a good index of risk, at least in middle aged subjects. The closer association of WHTR than waist with intra-abdominal fat measurements24 also indicates that WHTR is an index of value. The weaker, although still highly signi®cant, association of waist/hip (WHR) with elevated blood pressure, hypertension and intra-abdominal fat suggest that this index is of less value than WHTR or waist. The association between the risk of cardiovascular disease, or the presence of an elevated blood pressure and increased intra-abdominal adipose tissue now seems well established. However, it is not practical

Table 2 Age adjusted incremental odds ratios for elevated blood pressure (> 140/90 mm Hg or treated for hypertension) by quintiles, from the lowest to the highest, of anthropometric indices Men Age

BMI WHTR Waist Weight Height

18^39 OR 95% CI

40^64 OR 95% CI

18^39 OR 95% CI

40^64 OR 95% CI

n 1409

n 1303

n 1690

n 1589

2.01 (1.60±2.52) 2.17 (1.74±2.71) 2.21 (1.75±2.78) 1.76 (1.38±2.17) 0.89b (0.73±1.09)

1.52 (1.38±1.68) 1.70 (1.52±1.88) 1.68 (1.51±1.87) 1.41 (1.28±1.54) 0.93b (0.85±1.01)

n 816

n 1106

n 1026

1.33 (1.19±1.49)

1.63 (1.26±2.10)

1.37 (1.22±1.52)

1.53 (1.34±1.76) 1.58 (1.38±1.81) 1.60 (1.39±1.83) 1.49 (1.30±1.71) 1.02b (0.89±1.16) n 843

WHR

1.35 (1.14±1.61) a

Women

b

P < 0.001 for all except P < 0.05 and NS.

1.33 1.36 1.34 1.23 0.91a

(1.21±1.46) (1.23±1.50) (1.22±1.48) (1.12±1.34) (0.83±1.00)

677


678

Table 3 Partial correlations by age group (adjusting for age) of blood pressure measurements with anthropometric measurements Men All subjects Age

18^39

Women Excl. treated hypertensives

40^64

40^64

All subjects 18^39

Excl. treated hypertensives 40^64

40^64

Systolic blood pressure BMI WHTR Waist Weight Height

n 1409

n 1303

n 1196

n 1690

n 1589

n 1451

0.298 0.298 0.287 0.255 70.029b

0.212 0.220 0.206 0.168 70.038b

0.205 0.215 0.200 0.159 70.044b

0.278 0.254 0.252 0.264 70.009b

0.285 0.258 0.274 0.267 70.020b

0.282 0.258 0.262 0.267 70.006b

n 843

n 816

n 751

n 1106

n 1026

n 940

0.248

0.179

0.181

0.130

0.175

WHR

0.163

Diastolic blood pressure BMI WHTR Waist Weight Height

n 1409

n 1303

n 1196

n 1690

n 1589

n 1451

0.250 0.263 0.251 0.212 70.028b

0.241 0.241 0.228 0.194 70.035b

0.236 0.237 0.220 0.184 70.047b

0.235 0.229 0.219 0.209 70.021b

0.286 0.265 0.268 0.269 70.021b

0.285 0.254 0.259 0.269 70.013b

n 843

n 816

n 751

n 1106

n 1026

n 940

0.213

0.186

0.189

0.102

0.155

0.147

WHR

b

P < 0.001 for all except NS. Table 4 Logistic regression of elevated blood pressure (> 140/ 90 mm Hg or treated for hypertension) by waist and height, adjusting for age in subjects aged 40±64 Men (n 1303) Quintile

OR

Smallest 1 2 3 4 Largest 5

1.00 1.83 2.18 2.79 4.47

Waist, adjusting for Height Ð 1.00 Ð (1.04±3.23) 1.94 (1.04±3.62) (1.27±3.74) 2.62 (1.46±4.69) (1.65±4.72) 3.86 (2.18±6.86) (2.64±7.57) 9.11 (5.17±16.03)

1.00 1.21 1.52 1.71 1.88

< 0.001 < 0.001 Height, adjusting for Waist Ð 1.00 Ð (0.81±1.82) 1.18 (0.79±1.76) (1.02±2.29) 1.22 (0.80±1.84) (1.15±2.54) 1.62 (1.10±2.40) (1.25±2.83) 1.53 (1.02±2.29)

P value for trend Tallest

5 4 3 2 Shortest 1

P value for trend

95% CI

Women (n 1589)

< 0.001

OR

95% CI

< 0.001

in mass screening or clinic situations to conduct computed tomography, so the established association between waist circumference and IAT22,23 and waist/ height and IAT24 enables measurements involving the waist to be used as a surrogate for IAT. The ®ndings in most studies that WHR is not only less closely correlated with IAT, cardiovascular disease and its risk factors than is waist circumference alone, waist/ thigh or waist/height22,23,27 indicates that lower abdominal fat may be incorporated in the hip girth measurement or that WHR is also measuring muscle mass. Thigh measurements which have been employed in association with the waist circumference, are more

intrusive than measurements of weight, height or waist, a point to be taken into consideration when selecting a simple but reliable index for routine risk assessment. The relationship between WHTR, which is a very simple anthropometric index to obtain, and elevated blood pressure suggests that WHTR can be used routinely to screen for those who need to reduce their intra-abdominal adipose tissue, particularly middle aged subjects, who are at greatest risk of premature cardiovascular disease. Indeed, prospective studies from HALS using mortality data on those who died in the ten years following the 1984±85 survey, and on respondents seen again in the follow-up survey of 1991±92 showed that WHTR was more closely associated with all causes of mortality and the development of, or death from, cardiovascular disease than BMI.29,30 Hsieh and Yoshinaga5 showed, in Japanese men, that the WHTR ratio was a good predictor of a number of CHD risk factors including raised systolic blood pressure. In the Health Professionals Follow-up Study in the USA, waist circumference adjusted for height was also shown to be a better discriminator for multivariate coronary risk than waist circumference, hip girth or hip/height ratio, particularly in older men.4 In this study it has also been shown that WHTR had a close association with the prevalence of elevated blood pressure, particularly in those aged 40±64 y. WHR occupied a relatively poor position compared to WHTR, waist and BMI and even weight. Bonara et al7 have shown that WHR was not such a good indicator of cardiovascular risk as other mea-


sures as it did not distinguish between subcutaneous and visceral fat, whereas subscapular fat thickness, used as an indicator of central adiposity, was closely related to blood pressure in middle-aged American subjects in the NHANES I study.19 The association was closer in women than men. Similarly, the present study has shown a steeper gradient for the prevalence of elevated blood pressure across the anthropometric quintiles (except for height) in women than men. This indicates that increased weight, or adipose tissue in women is more hazardous, in terms of blood pressure, than for men. The measured values of blood pressure measurements in treated hypertensives may range from normal to above 160/95 mm Hg, depending on the effectiveness of treatment and compliance. By grouping all treated hypertensives with the newly recognised hypertensives and examining them by quintiles of anthropometric indices any spurious effect of antihypertensive medication is eliminated. Studies reporting correlations between BMI or indices of fat distribution and blood pressure have not always indicated how the measured blood pressure values of those on anti-hypertensive treatment were dealt with in the analyses.2,3,5,6,10,11,13,15,19 Bonara et al 7 did exclude those taking any drugs and Lundgren et al 1 reported adjusting for anti-hypertensive drugs in their study of adipose tissue distribution in relation to diabetes. The linear correlations between measured blood pressure in this study and the various anthropometric indices are similar with and without the inclusion of the treated hypertensives and this might suggest that the relationships are with measured rather than intrinsic blood pressure. This would mean that drug treatment of blood pressure would also considerably alter anthropometric indices, which is clearly not the case. However, since in the present study only 9% of the women and 8% of the men were treated hypertensives, their contributions to the overall correlations were small (Table 3). Furthermore nearly half of these hypertensive subjects had blood pressure measurements above the normal range. In studies where the proportions of treated hypertensives, especially if they are being successfully treated, are greater this factor may need to be taken into account. Although smoking is a very potent independent risk factor for cardiovascular disease, the associations between elevated blood pressure and the anthropometric indices appear to be similar for non-smokers, ex-smokers and smokers as has been found for BMI.6 However, since there was an excess of male exsmokers in the higher quintiles of WHTR, BMI, waist and weight, with the increased prevalence in these quintiles of elevated blood pressure, the public health message to ex-smokers to control their weight and waist circumference needs to be emphasised. It is apparent that, although waist reduction is concurrent with weight reduction, intra-abdominal adipose tissue is more responsive than adipose tissue in other

regions, to weight reduction regimes,14 and waist changes correlate with those of IAT, at least in women.23,31 The observation that height was related (inversely) to the prevalence of elevated blood pressure in the middle aged, but not the younger, subjects raises the question of the mechanism of action. Barker and coworkers32 have postulated that nutritional deprivation in foetal life and infancy not only predispose to various diseases, including cardiovascular disease, diabetes and hypertension, but also results in failure to obtain full growth potential. One possible explanation for the current ®ndings is that the middle aged subjects were mostly born before World War II when ante-natal and post-natal services were not so well developed. There may therefore have been more subjects in that age group who were inadequately nourished in early life, so that this is a cohort effect. Alternatively, those in the younger group, who were not adequately nourished in utero or infancy, had not yet reached the age where the risks to health become manifest.

References

1 Lundgren H, Bengtsson C, Blohme G, Lapidus L, SjoÈstrom S. Adiposity and adipose tissue distribution in relation to incidence of diabetes in women: results from a prospective population study in Gothenburg, Sweden. Int J Obes 1989; 13: 413±423. 2 Troisi RJ, Weiss ST, Segal MR, Cassano PA, Vokonas PS, Landsberg L. The relationship of body fat distribution to blood pressure in normotensive men: the normative aging study. Int J Obes 1990; 14: 515±525. 3 Seidell JC, Cigolini M, Deslyper J-P, Charzewska J, Ellsinger B-M, Cruz A. Body fat distribution in relation to serum lipids and blood pressure in 38 year old European men: the European fat distribution study. Atherosclerosis 1991; 86: 251±260. 4 Rimm EB, Stampfer MJ, Giovannucci E, Ascherio A, Spiegelman D, Colditz GA, Willett WC. Body size and fat distribution as predictors of coronary heart disease among middle-aged and older US men. Am J Epidemiol 1995; 141: 1117±1127. 5 Hsieh SD, Yoshinaga H. Abdominal fat distribution and coronary heat disease risk factors in menÐwaist/height ratio as a simple and useful predictor. Int J Obes 1995; 19: 585±589. 6 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. 7 Bonara E, Targher G, Branzi P, Zenere M, Saggiani F, Zenti MG, Travia D, Tonoli M, Muggeo M, Cigolini M. Cardiovascular risk pro®le in 38-year and 18-year old men. Contribution of body fat content and regional fat distribution. Int J Obes 1996; 20: 28±36. 8 Segal KR, Dunaif A, Gutin B, Albu J, Nyman A, Pi-Sunyer FA. Body composition, not body weight, is related to cardiovascular disease risk factors and sex hormone levels in men. J Clin Invest 1987; 80: 1050±1055. 9 Hartz AJ, Rupley DC, Rimm AA. The association of girth measurements with disease in 32,856 women. Am J Epidemiol 1984; 119: 71±80. 10 Lapidus L, Bengtsson C, Larsson B, Pennert K, Rybo E, Sjostrom L. Distribution of adipose tissue and risk of cardiovascular disease and death: a 12 year follow-up of participants in the population study of women in Gothenburg, Sweden. BMJ 1984; 289: 1257±1261.

679


680

11 Bonara E, Zenere M, Branzi P, Bagnani M, Maggiulli L, Tosi F, Travia D, Cacciatori V, Querena M, Moghetti P, Micciolo R, Ferrari A, Muggeo M. In¯uence of body fat and its regional localization on risk factors for atherosclerosis in young men. Am J Epidemiol 1992; 135: 1271±1278. 12 Seidell JC, Bakx JC, de Boer E, Deurenberg P, Hautvast JGAJ. Fat distribution of overweight persons in relation to morbidity and subjective health. Int J Obes 1985; 9: 363±374. 13 Mueller WH, Wear ML, Hanis CL, Emerson JB, Barton SA, Hewett-Emmett D, Schull WJ. Which measurement of body fat distribution is best for epidemiologic research? Am J Epidemiol 1991; 133: 858±869. 14 Egger G, Bolton A, O'Neill M, Freeman D. Effectiveness of an abdominal obesity reduction programme in men: the GutBuster `waist loss' programme. Int J Obes 1996; 20: 227±231. 15 Han TS, van Leer EM, Seidell JC, Lean MEJ. Waist circumference action levels in the identi®cation of cardiovascular risk factors: prevalence study in a random sample. BMJ 1995; 311: 1401±1405. 16 Lean MEJ, Han TS, Morrison CE. Waist circumference as a measure for indicating need for weight management. BMJ 1995; 311: 158±161. 17 Ashwell M, Lejeune S. Ratio of waist circumference to height may be better indicator of need for weight management. BMJ 1996; 312: 337. 18 Rich-Edwards JW, Manson JE, Stampfer MJ, Colditz GA, Willett WC, Rosner B, Speizer FE, Hennekens CH. Height and the risk of cardiovascular disease in women. Am J Epidemiol 1995; 142: 909±917. 19 Blair D, Habicht J-P, Sims EAH, Sylwester D, Abraham S. Evidence for an increased risk for hypertension with centrally located body fat and the effect of race and sex on this risk. Am J Epidemiol 1984; 119: 526±540. 20 Duciemetiere P, Richard J, Cambien F. The pattern of subcutaneous fat distribution in middle-aged men and the risk of coronary heart disease: the Paris prospective study. Int J Obes 1986; 10: 229±240. 21 Williams MJ, Hunter GR, Kekes-Szabo T, Trueth MS, Snyder S, Berland L, Blaudeau T. Intra-abdominal adipose tissue cut-points related to elevated cardiovascular risk in women. Int J Obes 1996; 20: 613±617.

22 Ross R, Rissanen J, Hudson R. Sensitivity associated with the identi®cation of visceral adipose tissue levels using waist circumference in men and women: effects of weight loss. Int J Obes 1996; 20: 533±538. 23 Lemieux S, Prud'homme D, Tremblay A, Bouchard C, Depres J-P. Anthropometric correlates to changes in visceral adipose tissue over 7 years in women. Int J Obes 1996; 20: 618±624. 24 Ashwell M, Cole TJ, Dixon AK. Ratio of waist circumference to height is a strong predictor of intra-abdominal fat. BMJ 1996; 313: 559±560. 25 Secretary of State for Health. The Health of the Nation: A Strategy for Health in England. HMSO: London, 1993. 26 Cox BD. Changes in body measurements. In: Cox BD, Huppert FA, Whichelow MJ (eds). The Health and Lifestyle Survey: Seven Years on. Dartmouth Publishing Co Ltd: Aldershot, 1993, pp 103±117. 27 Cox BD, Blaxter M, Buckle ALJ, Fenner NP, Golding JF, Gore M, Huppert FA, Nickson J, Roth M, Stark J, Wadsworth MEJ, Whichelow MJ. The Health and Lifestyle Survey: Preliminary report of a nationwide survey of the physical and mental health, attitudes and lifestyle of a random sample of 9,003 British adults. Health Promotion Research Trust: Cambridge, 1987. 28 Lejeune SRE, Ashwell MA, Cox BD, Whichelow MJ. Waist:height ratio is a simple anthropometric index which is closely associated with blood pressure in middle-aged British adults. Proc Nutr Soc 1996; 55: 230A. 29 Cox BD, Whichelow MJ, Ashwell M, Prevost AT. Comparison of anthropometric indices as predictors of mortality in British adults. Int J Obes Suppl 1996; 20/4: 141. 30 Cox BD, Whichelow MJ. Ratio of waist circumference to height is better predictor of death than body mass index. BMJ 1996; 313: 1487. 31 Zamboni M, Armellini F, Turcato E, Todesco T, Bissoli L, Bergamo-Andreis IA, Bosello O. Effect of weight loss on regional body fat distribution in premenopausal women. Am J Clin Nutr 1993; 58: 29±34. 32 Barker DJP. The foetal and infant origins of inequalities in health in Britain. J Public Health Med 1991; 13: 64±68.