## Services on Demand

## Article

## Indicators

## Related links

## Share

## Revista de Saúde Pública

*Print version* ISSN 0034-8910

### Rev. Saúde Pública vol.41 n.5 São Paulo Oct. 2007

#### http://dx.doi.org/10.1590/S0034-89102007000500013

**ORIGINAL
ARTICLES**

**Relative
importance of body mass index and waist circumference for hypertension in adults**

**Flávio
Sarno ^{I}; Carlos Augusto Monteiro^{II}**

^{I}Programa
de Pós-Graduação. Faculdade de Saúde Pública
(FSP). Universidade de São Paulo (USP). São Paulo, SP, Brasil

^{II}Departamento
de Nutrição. FSP-USP. São Paulo, SP, Brasil

**ABSTRACT**

**OBJECTIVE:** To assess the relative importance of Body
Mass Index (BMI) and waist circumference for the determination of
hypertension in adults.

**METHODS:** Cross sectional analysis of a sample of
employees (N=1,584), aged 18 to 64 years, from a private general
hospital in the city of São Paulo, Brazil. Data collection
included the application of a structured questionnaire and blood
pressure, weight, high, and waist circumference measurements.
Hypertension was defined as blood pressure levels __>__ 140/90 mmHg
or reported use of anti-hypertensive medication. The relative
importance of BMI and waist circumference was evaluated by
calculating the attributable fraction of hypertension
corresponding to each anthropometric indicator, employing both
the usual cut-off points as well as cut-off points based on the
observed distribution of the indicator in the population. In
addition, an indicator combining simultaneously BMI and abdominal
circumference values was also developed.

**RESULTS:** Prevalence of hypertension was 18.9% (26.9% in
men and 12.5% in women). In men, the fraction of hypertension
attributable to BMI exceeded the fraction attributable to waist
circumference based on the usual cut-off points for the
indicators (56% vs. 48%, respectively) and also considering the
quartiles of the observed distribution for these indicators (73%
vs. 69%, respectively). In women, the fraction of hypertension
attributable to waist circumference was slightly higher than the
fraction attributable to BMI based on the usual cut off points
for both indicators (44% vs. 41%), but the reverse was true when
quartiles of the observed distribution were used (41% vs. 57%,
respectively). In women only, the fraction of hypertension
attributable to the indicator combining BMI and waist
circumference (64%) was higher that observed using each indicator
alone.

**CONCLUSIONS:** Both BMI and abdominal circumference were
positively and independently associated with the occurrence of
arterial hypertension, the influence of BMI being higher among
men.

**Key words:**
Hypertension. Body mass index. Abdominal circumference. Obesity. Attributable
fraction.

**INTRODUCTION**

Several studies provide evidence for an association between
arterial hypertension and anthropometric indicators that reflect
excess adipose tissue. Major indicators in this context are
abdominal circumference^{7,22} – which would provide a
measure especially of visceral fat – and the body mass index
(BMI), obtained by dividing weight in kilograms by the square of
height in meters, which would reflect the proportion of adipose
tissue in the total body mass, regardless of
localization.^{2,4,11}

Studies attempting to compare the associations with BMI and
abdominal circumference have arrived at conflicting conclusions.
While certain studies show better associations for BMI, others
find that abdominal circumference is the better indicator, with
results that often vary according to
sex.^{1,6,8,20,21,25}

The objective of the present study was to evaluate, in a population of Brazilian adults, the relative importance of BMI and abdominal circumference in the occurrence of arterial hypertension.

**METHODS**

The study population comprised a sample from the operational, administrative, and health care staff of a private general hospital in the municipality of São Paulo, Brazil. In November 2001, a campaign was conducted to diagnose the level of cardiovascular risk among the hospital staff. At the time, the total number of staff members was 3,623 (1,403 men and 2,220 women). All workers were invited to participate by means of posted signs and through the hospital's internal computer network. Of these, 1,584 workers agreed to take part in the campaign, of which 707 were men and 877 were women. Our study population was composed of these subjects. Studied and non-studied staff members did not differ significantly in terms of sex, age, and schooling.

Data collection was carried out during the five days of the
campaign – which lasted from Monday to Friday, from 7 a.m. to 9
p.m. – by teams composed of nursing students, who were trained
and supervised. Information was obtained through a structured
questionnaire, which included data on age, sex, skin color,
schooling, physical activity, drinking, work shift, and smoking.
Standardized techniques were used to obtain
anthropometric^{10} and blood pressure^{19}
measurements. Weight was measured with subjects wearing light
clothing and no shoes, using previously calibrated
microelectronic scales (Tanita) with 100 g precision. Height was
measured using a stadiometer (Seca) mounted on the wall, with 0.1
cm precision. Arterial pressure was measured once with the
subject standing, after a resting period of approximately five
minutes, using a previously calibrated, certified, BP 3BTO-A
instrument (Microlife) with 1 mmHg resolution.^{13}
Abdominal circumference was measured at the midpoint between the
last rib and the iliac crest, using an inextensible measuring
tape.

We considered as hypertensive all subjects with systolic
arterial pressure __>__140 mmHg and/or systolic arterial pressure
__>__90 mmHg, as well as all those under antihypertensive
medication.

The classification of subjects according to anthropometric
indicators initially considered the standard cutoff points for
BMI and abdominal circumference, which define values considered
as normal, moderately high, and high. In the case of BMI,
regardless of sex, these ranges correspond to <25, 25-29.9,
and __>__30 kg/m^{2}, respectively.^{24} In the case
of abdominal circumference, these same classes correspond to,
respectively, <94, 94-101.9, and __>__102 cm for men, and <80,
80-87.9, and __>__88 cm for women.^{12} As a next step, we
alternatively classified BMI and abdominal circumference based on
quartiles of distribution in the studied population, separated by
sex.

In order to produce a classification that took into account both indicators simultaneously, BMI and abdominal circumference values were transformed into z-scores in order to create a new indicator. This indicator was defined as the sum of the z-scores for each indicator, and was classified into quartiles, as previously done for BMI and abdominal circumference.

The study of the importance of BMI, abdominal circumference, and the combined indicator in determining arterial hypertension included initially an evaluation of bivariate associations between anthropometric indicator classes and presence of arterial hypertension. For this we used test based on the chi-squared distribution. We then evaluated the association of potential confounders in the association between anthropometric indicators and arterial hypertension. These included age, schooling, skin color, frequency of physical exercise, smoking, frequency of consumption of alcoholic beverages, and work shift. All potential confounders whose association with arterial hypertension showed p-values below 0.2 in the chi-squared test were introduced, one-by-one, into logistic regression models for arterial hypertension as a function of, alternatively, BMI, abdominal circumference, and the combined indicator. Adjusted odds ratios for arterial hypertension were given by final regression models that included all confounders determining variations of at least 10% in the odds ratios associated with the anthropometric indicators. We also checked for the potential occurrence of significant interactions between each anthropometric indicator and the variables included in the final regression models.

The importance of each anthropometric indicator in the
determination of arterial hypertension was quantified based on
the calculation of the corresponding population attributable
fraction. The attributable fraction was calculated based on the
formula [H – 1/H * 100, with H = f1 x 1 + f2 x OR2 + f3 x OR3 +
f4 x OR4], where f1 is the frequency of subjects in the baseline
category of the anthropometric indicator ("unexposed to risk"),
f2, f3, and f4, are the frequencies in the risk categories of the
indicator, and OR2, OR3, and OR4 are the adjusted, odds ratios
for arterial hypertension in each risk category for that
indicator.^{9}

Given the systematic differences found between sexes regarding the association between anthropometric indicators and arterial hypertension, all analyses were performed separately for both sexes. All analyses were carried out using SPSS version 10.0 software.

All staff members included in the study signed a term of free and informed consent. The study protocol was approved by the Departamento de Medicina do Trabalho and the Instituto de Ensino e Pesquisa do Hospital Israelita Albert Einstein, as well as by the Research Ethics Committee of the Faculdade de Saúde Pública da USP.

**RESULTS**

Prevalence of arterial hypertension was 26.9% among males, 12.5% among females, and 18.9% in the general population. Among men, there was a significant increase in prevalence associated with age and with working the night shift. Among women, prevalence of hypertension varied significantly with age (direct relationship), schooling (inverse relationship), and ethnicity (higher prevalence among nonwhites) (Table 1).

A little over one-half
of men and a little over one-third of women had BMIs above the upper limit of
the normal range (__>__25 kg/m^{2}). Abdominal circumference values above
the upper limit (94 cm for men and 80cm for women) were found in a little over
one-third of men and a little over one-half of women (Table
2).

For both sexes, there was a uniform and significant increase (p<0.01 for linear trend) in prevalence of arterial hypertension with increases in BMI and abdominal circumference. This occurred both when we used the standard cutoffs for BMI and/or abdominal circumference and when we used a classification based on quartiles for the analysis (Table 3).

Table
4, restricted to males, shows the adjusted odds ratios for hypertension
and the corresponding fractions of disease attributable to the anthropometric
indicators. Age and schooling emerged as adjustment variables in all final regression
models. There were no significant interactions between the anthropometric indicators
and these variables. The fraction of arterial hypertension attributable to BMI
exceeded that attributable to abdominal circumference both when the standard
cutoff points for the indicator (56% vs. 48%, respectively) and the quartiles
of the observed distribution (73% vs. 69%, respectively) were employed. Classification
according to quartiles showed an increase in occurrence of arterial hypertension
for intervals of BMI (between 22.9 and 25.2 kg/m^{2}) and abdominal
circumference (between 84 and 91 cm) considered as normal based on the traditional
cutoff points for these indicators. The attributable fraction for the indicator
combining BMI and abdominal circumference among men was 67%, and therefore lower
than the attributable fraction for each indicator alone.

Table 5 presents the adjusted odds ratios for hypertension and
the corresponding fractions of disease attributable to the
anthropometric indicators among females. Again, age and schooling
were kept in the final regression models as confounding
variables, and again there was no significant interaction between
anthropometric indicators and these variables. The fraction of
female arterial hypertension attributable to abdominal
circumference was slightly higher than that attributable to BMI
when standard cutoff points were used for both indicators (44%
vs. 41%, respectively). However, when quartiles were employed,
the explanatory power of BMI exceeded that of abdominal
circumference (attributable fractions of 57% and 41%,
respectively). Again, classification according to quartiles
showed increased occurrence of arterial hypertension in intervals
of BMI (between 21.7 and 23.6 kg/m^{2}) considered as
normal according to the standard classification. However, the
same was not true for abdominal circumference. The explanatory
power of the combined indicator for women (attributable fraction
of 64%) was higher than that of BMI or abdominal circumference
alone.

**DISCUSSION**

Our results indicate that both BMI and abdominal circumference have an important association with arterial hypertension in both sexes, even after control for relevant confounders. Depending on the classification employed, the fraction of arterial hypertension attributable to these indicators ranged from 48% to 73% among men and 41% to 64% among women. Among men, explanatory power for the occurrence of arterial hypertension, as measured by the population attributable fraction, was greater for BMI than for abdominal circumference, regardless of the classification used. Among women, the greater power of BMI over abdominal circumference was only apparent when the classification based on quartiles of the observed distribution in that population was used. The explanatory power of the indicator combining BMI and abdominal circumference was greater than that of each indicator alone only among women. The classification of indicators based on the distribution observed in the population rather than on the usual cutoff points increased the explanatory power of BMI among women and of both BMI and abdominal circumference among men. These results indicate the occurrence of arterial hypertension at levels of BMI and abdominal circumference lower than those established by traditional classifications.

Certain limitations should be taken into consideration when
interpreting the results of the present study. The first concerns
the particularities of the study sample – the staff of a private
hospital in the city of São Paulo – which limits the
extrapolation of results to other settings. The second limitation
is that diagnosis of arterial hypertension was based on a single
measurement, when ideally it would be based on two measurements
obtained at different times.^{5} Finally, the
cross-sectional design of the study does not ensure the temporal
precedence of anthropometric measures over the occurrence of
arterial hypertension.

On the other hand, strengths of the present study include a wide demographic and socioeconomic diversity within the sample, the obtainment of anthropometric and blood pressure measures by direct measurement rather than by self-report, control for relevant confounders in the estimates of association between anthropometric indicators and hypertension, and the analytic procedures employed, which ensure comparability of evaluation and in terms of explanatory power of anthropometric indexes in the determination of arterial hypertension.

Studies comparing the relative importance of BMI and abdominal
circumference among adults frequently employ regression analyses
or analyses based on ROC curves. Thus, a comparison of these
results with those of the present study are not direct or
immediate. While certain studies show stronger associations for
BMI, others find that abdominal circumference is the better
indicator, with results that often vary according to
sex.^{1,6,8,20,21,25} However, most of these studies
employ a priori classifications for these two indices, which do
not necessarily maximize the explanatory power of the indicators
in an equal manner. In some of these studies, it was shown that
the simultaneous consideration of both BMI and abdominal
circumference could increase explanatory power above that of
individual indices alone, both among women only,^{1,8} as
found in the present study, or for both
sexes.^{23,26}

The choice of using the attributable fraction as a means to measure the explanatory power of each index was based mainly on the ease of interpretation provided by this method. The attributable fraction indicates the proportion of occurrence of the disease that would be eliminated if individuals remained unexposed to the risk condition under study ("high values" of BMI, abdominal circumference, or the combined index). In addition, sensitivity and specificity calculations and ROC curves require dichotomous variables, which is not the case for BMI and abdominal circumference. Finally, the attributable fraction allows for confounder control in the association between studied variable and outcome.

We found only six studies in the literature using the
population attributable fraction to investigate the importance of
abdominal circumference and/or BMI in determining arterial
hypertension. Five of these evaluate only abdominal
circumference, identifying attributable fractions for arterial
hypertension ranging from 5.8% to 30% among men and 11.1% to
66.5% among women.^{14-18} One of these studies
calculated the attributable fraction of hypertension associated
with circumference with and without controlling for BMI, and
identified a substantial reduction in the explanatory power of
circumference after controlling for BMI in both
sexes.^{16} The only study comparing the attributable
fractions of arterial hypertension associated to BMI and
abdominal circumference found similar explanatory power for both
indicators (population attributable fraction of about 40% for
both sexes).^{3} There is no record in the literature of
studies calculating the attributable fraction of arterial
hypertension in relation to an indicator combining BMI and
abdominal circumference.

In conclusion, the results of the present study confirm data from the literature that indicate a high explanatory power for both BMI and abdominal circumference in determining arterial hypertension. This suggests that increase in fat deposits may increase risk of disease, be it in the abdominal region, or in other parts of the body. Our results indicate that not only fat deposits in the abdominal region should be considered as hazardous to health, since the population attributable fraction for hypertension associated with BMI was greater than that found for abdominal circumference, especially among men. Furthermore, among women, the combination of BMI and abdominal circumference increased the explanatory power of each index alone. We also draw attention to the observation that the classification of BMI and abdominal circumference based on the distribution of these indexes in the studied population showed that values usually regarded as normal are already associated with increased occurrence of arterial hypertension, which indicates a need for revision of traditional classifications.

**REFERENCES**

1. Ardern CI, Katzmarzyk PT, Janssen I, Ross R. Discrimination of health risk by combined body mass index and waist circumference. *Obes Res.* 2003;11 (1):135-42. [ Links ]

2. Arroyo P, Fernandez V, Avila-Rosas H. Overweight and Hypertension: data from the 1992-1993 Mexican survey. Hypertension. 1997;30(3 Pt 2): 646-9. [ Links ]

3. Bei-Fan Z, Cooperative Meta-Analysis Group of Working Group on Obesity in China. Predictive values of body mass index and waist circumference for risk factors of certain related diseases in Chinese adults: study on optimal cut-off points of body mass index and waist circumference in Chinese adults. *Asia Pac J Clin Nutr*. 2002;11(Supl 8):685-93. [ Links ]

4. Brown CD, Higgins M, Donato KA, Rohde FC, Garrison R, Obarzanek E, et al. Body mass index and the prevalence of hypertension and dyslipidemia. *Obes Res.* 2000;8(9):605-19. [ Links ]

5. Chobanian AV, Bakris GJ, Black HR, Cushman WC, Lee A, Green LA, et al. The Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure: the JNC 7 report. JAMA. 2003;289(19):2560-72. [ Links ]

6. Cox BD, Whichelow MJ, Ashwell M, Prevost AT, Lejeune SR. Association of anthropometric indices with elevated blood pressure in British adults. *Int J Obes Relat Metab Disord.*1997;21(8):674-80. [ Links ]

7. Grievink L, Alberts JF, O'Niel J, Gerstenbluth I. Waist circumference as a measurement of obesity in the Netherlands Antilles; associations with hypertension and diabetes mellitus. *Eur J Clin Nutr.* 2004;58(8):1159-65. [ Links ]

8. Janssen I, Katzmarzyk PT, Ross R. Waist circumference and not body mass index explains obesity-related health risk. *Am J Clin Nutr.* 2004;79(3):379-84. [ Links ]

9. Kleinbaum DG, Kupper LL, Morgenstern H. Epidemiologic Research: principles and quantitative methods. New York: Lifetime Learning Publications; 1982. p. 160-164. [ Links ]

10. Lohman TG, Roche AF, Martorell R. Anthropometric standardization reference manual. Champaign: Human Kinetics Books; 1988. [ Links ]

11. Must A, Spadano J, Coakley EH, Field AE, Colditz G, Dietz WH. The disease burden associated with overweight and obesity. *JAMA.* 1999;282(16):1523-9. [ Links ]

12. National Institutes of Health. Clinical Guidelines on the Identification, Evaluation, and Treatment of Overweight and Obesity in Adults––The Evidence Report. National Institutes of Health. *Obes Res.* 1998;6 (Supl 2):51-209. [ Links ]

13. O'Brien E, Waeber B, Parati G, Staessen J, Myers MG. Blood pressure devices: recomendations of the European Society of Hypertension. *BMJ.* 2001;322(7285): 531-6. [ Links ]

14. Okosun IS, Cooper RS, Rotimi CN, Osotimehin B, Forrester T. Association of waist circumference with risk of hypertension and type 2 diabetes in Nigerians, Jamaicans, and African Americans. *Diabetes Care.* 1998;21(11):1836-42. [ Links ]

15. Okosun IS, Prewitt TE, Cooper RS. Abdominal obesity in the United States: prevalence and attributable risk of hypertension. *J Hum Hypertens.*1999;13(7): 425-30. [ Links ]

16. Okosun IS, Forrester TE, Rotimi CN, Osotimehin BO, Muna WF, Cooper RS. Abdominal adiposity in six populations of West African descent: prevalence and population attributable fraction of hypertension. *Obes Res.* 1999;7(5):453-62. [ Links ]

17. Okosun IS, Choi S, Dent MM, Jobin T, Dever GE. Abdominal obesity defined as a larger than expected waist girth is associated with racial/ethnic differences in risk of hypertension. *J Hum Hypertens.* 2001;15 (5):307-12. [ Links ]

18. Okosun IS, Boltri JM, Anochie LK, Chandra KM. Racial/ethnic differences in prehypertension in American adults: population and relative attributable risks of abdominal obesity. *J Hum Hypertens.* 2004;18(12):849-55. [ Links ]

19. Pickering TG, Hall JE, Appel LJ, Falkner BE, Graves J, Hill MN, et al. Recommendations for blood pressure measurement in humans and experimental animals: part 1: blood pressure measurement in humans: a statement for professionals from the Subcommittee of Professional and Public Education of the American Heart Association Council on High Blood Pressure Research. *Circulation*. 2005;111(5):697-716. [ Links ]

20. Sakurai M, Miura K, Takamura T, Ota T, Ishizaki M, Morikawa Y, et al. Gender differences in the association between anthropometric indices of obesity and blood pressure in Japanese. *Hypertens Res.* 2006;29(2):75-80. [ Links ]

21. Sanchez-Castillo CP, Velazquez-Monroy O, Berber A, Lara-Esqueda A, Tapia-Conyer R, James WP. Encuesta Nacional de Salud (ENSA) 2000 Working Group. Anthropometric cutoff points for predicting chronic diseases in the Mexican National Health Survey 2000. *Obes Res.* 2003;11(3): 442-51. [ Links ]

22. Snijder MB, Zimmet PZ, Visser M, Dekker JM, Seidell JC, Shaw JE. Independent and opposite associations of waist and hip circumferences with diabetes, hypertension and dyslipidemia: the AusDiab Study. *Int J Obes Relat Metab Disord.* 2004;28(3):402-9. [ Links ]

23. Wildman RP, Gu D, Reynolds K, Duan X, Wu X, He J. Are waist circumference and body mass index independently associated with cardiovascular disease risk in Chinese adults? *Am J Clin Nutr.* 2005;82(6):1195-202. [ Links ]

24. World Health Organization. Obesity: preventing and managing the global epidemic. Report of a WHO consultation. *World Health Organ Tech Rep Ser.* 2000;894:i-xii,1-253. [ Links ]

25. Zhu S, Wang Z, Heshka S, Heo M, Faith MS, Heymsfield SB. Waist circumference and obesity-associated risk factors among whites in the third National Health and Nutrition Examination Survey: clinical action thresholds. *Am J Clin Nutr.* 2002;76 (4): 743-9. [ Links ]

26. Zhu S, Heshka S, Wang Z, Shen W, Allison DB, Ross R, Heymsfield SB. Combination of BMI and Waist Circumference for Identifying Cardiovascular Risk Factors in Whites. *Obes Res.* 2004;12 (4):633-45. [ Links ]

**Correspondence:**

Carlos Augusto Monteiro

Departamento de Nutrição

Faculdade de Saúde Pública da USP

Av. Dr. Arnaldo 715

01246-907 São Paulo, SP, Brasil

E-mail: carlosam@usp.br

Received: 8/22/2006

Reviewed:5/2/2007

Accepted: 5/28//2007

Article based on the master's dissertation by F Sarno, presented at Faculdade de Saúde Pública da Universidade de São Paulo, in 2005.