ences, Faculty of Medicine, University of Manitoba,. Winnipeg (Drs Young and Gelskey), and Cadham Pro- vincial Laboratory, Winnipeg, Manitoba (Dr Gelskey).
Obesity Metabolically Benign? Is Noncentral
Implications for Prevention T. Kue
From
a
Population Survey
Young, MD, DPhil, Dale E. Gelskey, DrPH
Objective.\p=m-\Todetermine if individuals who are overall obese but have low waist-to-hip ratios have unfavorable lipid profiles, blood pressures, and glucose statuses.
Design.\p=m-\Cross-sectionalstudy. Setting.\p=m-\TheManitoba Heart Health Survey surveyed a representative sample of residents of the Canadian province of Manitoba. Participants.\p=m-\Atotal of 2792 adults aged 18 to 74 years were interviewed, 2339
of whom underwent clinical examinations. Main Outcome Measures.\p=m-\Bloodpressure, fasting plasma glucose, total cholesterol, triglycerides, high-density lipoprotein, and low-density lipoprotein measurements were compared across categories of body mass index (BMI) and waist\x=req-\ to-hip ratio (WHR). Main Results.\p=m-\Manitobanswith noncentral obesity tend to occupy positions between those of the nonobese and the centrally obese in terms of the effect on blood pressure, plasma lipids, and glucose. In multiple linear regression models involving age, BMI, and WHR as independent variables and one of the metabolic variables as dependent variables, both BMI and WHR are significant independent predictors of most metabolic variables. Where both are significant, BMI tends to be the stronger predictor, with a larger standardized regression coefficient. Conclusions.\p=m-\Noncentralobesity is not metabolically benign; BMI as an overall measure of obesity is as important as, and sometimes more important than, WHR in predicting metabolic effects. The recognition of the epidemiological significance of the WHR as a centrality measure of obesity should not divert attention from the metabolic risk status of noncentrally obese individuals who require continued health education to reduce weight. (JAMA. 1995;274:1939-1941)
THE ADVERSE health effects of obe¬ sity are well recognized. In the past two decades, there has been renewed inter¬ est in Vague's original observation of From the
Department of Community Health SciFaculty of Medicine, University of Manitoba, Winnipeg (Drs Young and Gelskey), and Cadham Provincial Laboratory, Winnipeg, Manitoba (Dr Gelskey). Reprint requests to Department of Community Health Sciences, University of Manitoba, 750 Bannatyne Ave, Winnipeg, Manitoba, Canada R3E 0W3 (Dr Young). ences,
the importance of central obesity,1 also referred to as abdominal, android, fran¬
cai,
or
upper-body obesity, although
these are not identical entities. Clinical studies and cross-sectional surveys have shown an association between indexes of central obesity (such as the waist-tohip ratio [WHR]) with a variety of meta¬ bolic variables, cardiovascular risk fac¬ tors, and glucose-intolerant states.25 Furthermore, cohort studies have also shown that central obesity is predictive
of future development of diabetes, is¬ chemie heart disease, and stroke.67 The recognition of central obesity has been communicated to the general public. Many health-conscious consumers are conversant with the body mass index (BMI) and also increasingly knowledge¬ able about the WHR. The existing literature, however, is in¬ consistent and inconclusive with regard to whether individuals who are overall obese (ie, high BMI) without being cen¬ trally obese (ie, high WHR) are also at an increased risk for various adverse health consequences. This is an important issue in the planning of health promotion strat¬ egies, since health professionals and in¬ dividuals may be less motivated to con¬ trol overweight if they find their clients or themselves in the noncentrally obese category. We used the results from a large population-based survey of Cana¬ dians to determine if noncentral obesity is significantly associated with such meta¬ bolic effects as elevated blood pressure, an unfavorable plasma lipid profile, and indicators of glucose intolerance. Methods The Manitoba Heart Health Survey (MHHS) was part of a national effort to estimate the prevalence and distribution patterns of cardiovascular risk factors and assess the level of knowledge re¬ garding cardiovascular health among Canadians. The survey included an in¬ terviewer-administered risk factor ques¬ tionnaire, a self-administered nutrition questionnaire, and clinical and laboratory investigations. Detailed descriptions of the design and methods of the MHHS are
available.8
The MHHS sampled the adult (aged 18 to 74 years), noninstitutionalized popu¬ lation of the province of Manitoba. A to¬ tal of 2792 individuals from 46 sampling units or sites were interviewed, with a response rate of 77%. Of these, 2339 (84%) returned for the clinical and laboratory investigations. Because of the higher prevalence of obesity and various car¬ diovascular risk factors among North American Indians,9 the 169 residents of Indian reserves in the sample were ex¬ cluded from analysis. Interviews, anthro¬ pometry, and blood pressure measure¬ ments were performed by nurses who had undergone a standardized training program to ensure national comparabil¬ ity of the various provincial surveys. Blood specimens were collected after at least 14 hours of fasting and tested for
total cholesterol, high-density lipoprotein
(HDL), low-density lipoprotein (LDL), triglycérides, and glucose. Lipid analy¬ ses were performed at the national lipid reference laboratory at the University of Toronto (Ontario). Details on the tech¬ nical procedures, standardization, and quality control used by this laboratory have been published.10 Glucose analyses
Table 1.—Division of Survey Waist-to-Hip Ratio (WHR)*
Sample
Into
performed at the Cadham Provin¬ Laboratory in Winnipeg, Manitoba, by an Abbott Spectrum Analyzer (Ab¬ bott Laboratories, Mississauga, Ontario) using an enzymatic method. were
cial
Four blood pressure measurements were obtained on two separate occasions, twice during the interview and twice dur¬ ing the clinic visit. Because of intraperson variability, the means of the four values were used in the analysis. Anthropom¬ etry consisted of height and weight using a balance beam scale with participants in indoor clothing and without shoes. Waist and hip measurements were taken with the subjects in underclothing. Waist was taken as the narrowest circumference of the abdomen and hips at the level of the
greatest gluteal protuberance. In the analysis, BMI (measured
as
weight in kilograms divided by the square of height in meters [kg/m2]) and WHR for and women were divided into fer¬ tiles. Three categories were defined when BMI and WHR were cross-tabulated (Table 1): (1) the nonobese (low BMI, low WHR): 651 men and 601 women; (2) the noncentrally obese (high BMI, low WHR): 161 men and 145 women; and (3) the cenmen
Obesity Categories According
to
Body
Mass Index
(BMI) and
BMI Tertlle*
WHR tertilet 1
Nonobese
Nonobese
2
Nonobese
3
Nonobese
Nonobese Centrally obese
Noncentrally obese Noncentrally obese Centrally obese
*BMI tertlle 1: 15.0 to 24.6 (M), 16.2 to 23.1 (F); BMI fertile 2: 24.7 to 28.0 (M), 23.2 to 27.4 (F); BMI tertlle 3: 28.1 (M), 27.5 to 54.2 (F). tWHR tertlle 1: 0.71 to 0.90 (M), 0.61 to 0.76 (F); WHR fertile 2: 0.91 to 0.96 (M), 0.77 to 0.82 (F); Tertile 3: 0.97 to 1.24 (M), 0.83 to 1.34(F). to 48.7
Table
2.—Comparison
of Mean Values of Selected Metabolic Variables
trally obese (high BMI, high WHR): 305
and 325 women. Individuals who reported being on antihypertensive medications (n=338), be¬ men
ing on cholesterol-lowering drugs (n=54), and having previous diagnoses of diabe¬ tes mellitus
Results The three categories are compared in terms of mean systolic blood pressure, mean diastolic blood pressure, fasting glu¬ cose, fasting total cholesterol, HDL, LDL, and triglycérides, adjusting for age. To avoid confusion, in the following discus¬ sion the terms/aiwaoZe and unfavorable are used. The former term refers to low systolic blood pressure, diastolic blood pressure, total cholesterol, LDL, and tri¬ glycérides and high HDL, and the latter term refers to high systolic blood pres¬ sure, diastolic blood pressure, total cho¬ lesterol, LDL, and triglycérides and low HDL. Table 2 shows that men with noncentral obesity have mean values in all metabolic indicators that are more un¬ favorable (P
