点击显示 收起
1 From the Department of Human Nutrition and Division of Epidemiology and Biostatistics, University of Illinois at Chicago (YW); the Departments of Nutrition and Epidemiology, Harvard School of Public Health, Boston (EBR, FBH, MJS, and WCW); the Channing Laboratory, Department of Medicine, Harvard Medical School and Brigham and Women's Hospital, Boston (MJS, WCW, and EBR).
2 Supported by the National Institutes of Health (CA55075, HL35464, HL65582, and P30DK064408). YW was supported in part by NIH grant R01DK63383. 3 Address reprint requests to Y Wang, Center for Human Nutrition, Department of International Health, Bloomberg School of Public Health, Johns Hopkins University, 615 North Wolfe Street, Baltimore, MD 21205. E-mail: youfwang{at}uic.edu.
ABSTRACT
Background: Obesity is a strong risk factor for type 2 diabetes. However, few studies have compared the predictive power of overall obesity with that of central obesity. The cutoffs for waist circumference (WC) and waist-to-hip ratio (WHR) as measures of abdominal adiposity remain controversial.
Objective: The objective was to compare body mass index (BMI), WC, and WHR in predicting type 2 diabetes.
Design: A prospective cohort study (Health Professionals Follow-Up Study) of 27 270 men was conducted. WC, WHR, and BMI were assessed at baseline. Covariates and potential confounders were assessed repeatedly during the follow-up.
Results: During 13 y of follow-up, we documented 884 incident type 2 diabetes cases. Age-adjusted relative risks (RRs) across quintiles of WC were 1.0, 2.0, 2.7, 5.0, and 12.0; those of WHR were 1.0, 2.1, 2.7, 3.6, and 6.9; and those of BMI were 1.0, 1.1, 1.8, 2.9, and 7.9 (P for trend < 0.0001 for all). Multivariate adjustment for diabetes risk factors only slightly attenuated these RRs. Adjustment for BMI substantially attenuated RRs for both WC and WHR. The receiver operator characteristic curve analysis indicated that WC and BMI were similar and were better than WHR in predicting type 2 diabetes. The cumulative proportions of type 2 diabetes cases identified according to medians of BMI (24.8), WC (94 cm), and WHR (0.94) were 82.5%, 83.6%, and 74.1%, respectively. The corresponding proportions were 78.9%, 50.5%, and 65.7% according to the recommended cutoffs.
Conclusions: Both overall and abdominal adiposity strongly and independently predict risk of type 2 diabetes. WC is a better predictor than is WHR. The currently recommended cutoff for WC of 102 cm for men may need to be reevaluated; a lower cutoff may be more appropriate.
Key Words: Obesity • abdominal adiposity • waist circumference • waist-to-hip ratio • body mass index • type 2 diabetes
INTRODUCTION
Obesity is a powerful risk factor for cardiovascular disease (CVD) (1–4) and type 2 diabetes (5–9). However, the independent effects of abdominal adiposity compared with those of overall obesity are not well studied, and the use of waist circumference (WC) and waist-to-hip ratio (WHR) remains controversial (10–13).
A growing consensus suggests that WC should be used rather than WHR to assess fat distribution (11–15). In a recent statement by the National Institutes of Health and the North American Association for the Study of Obesity, WHR was said to provide no advantage over WC alone (11); a WC cutoff of 40 inches (102 cm) for men and of 35 inches (88 cm) for women were recommended. However, this issue remains inconclusive. For example, the Iowa Women's Health Study, a 12-y prospective cohort study of 31 702 Iowa women aged 55–69 y who were free of cancer, heart disease, and diabetes at baseline, indicated that WHR is a better anthropometric predictor of total mortality than are BMI and WC (16). In a 9-y study that followed a large cohort (n = 12 814) of African-American and white men and women, Stevens et al (17) reported that WC, WHR, and BMI were equivalent in their ability to predict type 2 diabetes. Another important issue concerns the appropriate cutoffs for elevated WC. The cutoffs for WC (102 cm for men and 88 cm for women) and WHR (0.95 for men and 0.88 for women) were recommended by the American Heart Association (18) and the US Department of Agriculture (19). However, lower WC cutoffs of 94 cm for men and of 80 cm for women have also been suggested (20–22).
To look into this question further, we compared the predictive power of WC, WHR, and BMI in diagnosing type 2 diabetes in a large cohort of men (The Health Professionals Follow-Up Study) and evaluated the effect of different cutoffs for WC, WHR, and BMI. We also studied the independent and additive associations of overall obesity (reflected by BMI) and abdominal adiposity (WC and WHR) with risk of type 2 diabetes.
SUBJECTS AND METHODS
Subjects
The Health Professionals Follow-Up Study began in 1986 when 51 529 US male health professionals (dentists, optometrists, pharmacists, podiatrists, osteopaths, and veterinarians), aged 40–75, answered a detailed questionnaire that included a comprehensive diet survey as well as questions about lifestyle practices and medical history. Measures of WC and hip circumference (HC) were not collected in the 1986 baseline survey but were collected in 1987 (see below). Follow-up questionnaires were sent in 1988, 1990, 1992, 1994, 1996, 1998, and 2000 to update information on potential risk factors and to identify newly diagnosed cases of diabetes and other diseases. We excluded from the present analysis men with prior diagnosis of CVD or cancer in 1986 or 1987 (n = 4983) and men with diabetes diagnosed before the 1987 survey (n = 710). Participants who had missing data on WC and HC were also excluded. We followed the remaining 27 270 men for the incidence of type 2 diabetes during the subsequent 13 y.
Diagnosis of type 2 diabetes
A supplementary questionnaire regarding symptoms, diagnostic tests, and hypoglycemic therapy was mailed to men who indicated on any biennial questionnaire that they had been diagnosed with diabetes. A case of diabetes was considered confirmed if at least one of the following was reported on the supplementary questionnaire: 1) one or more classic symptoms (excessive thirst, polyuria, weight loss, or hunger) plus a fasting plasma glucose concentration of 140 mg/dL (7.8 mmol/L) or a random plasma glucose concentration of 200 mg/dL (11.1 mmol/L), 2) 2 elevated plasma glucose concentrations on different occasions [fasting concentration of 140 mg/dL (7.8 mmol/L), random concentration of 200 mg/dL (11.1 mmol/L), or concentration 200 mg/dL after 2 h of oral-glucose-tolerance testing] in the absence of symptoms, or 3) treatment with hypoglycemic medication (insulin or oral hypoglycemic agent).
Our criteria for diabetes classification are consistent with those proposed by the National Diabetes Data Group (23) for 1986–1996. The validity of self-report of type 2 diabetes has been verified in a subsample of 71 men from the Health Professionals Follow-Up Study cohort (24). A physician blinded to the information reported on the supplementary questionnaire reviewed the medical records according to the diagnostic criteria. Of the 59 cases with complete medical records, the diagnosis of type 2 diabetes was confirmed in 57 (97%). One patient denied the diagnosis and another lacked evidence of diabetes in his submitted records. Similarly, 98% of diabetic cases reported by the supplementary questionnaire were confirmed by medical record review in a subsample of participants (n = 62) in the Nurses' Health Study (25).
Anthropometric measures
Self-reported body weights and heights were recorded at baseline in 1986, and self-reported body weights alone were reported in later biennial questionnaire surveys. Information on WC and HC was collected in a brief follow-up questionnaire sent out in 1987 and was collected again in 1996. We included paper tape measures, detailed instructions, and illustrations to assist the men in self-reporting their WC and HC. The participants were instructed to take measurements while standing and to avoid measuring over bulky clothing. The subjects were asked to measure their WCs at the umbilicus and to take their hip measures as the largest circumference between the waist and thighs. Along with the directions, illustrations were included (5). Self-reported waist, hip, and weight measures were found to be reasonably valid when compared with standardized measurements by a technician in a subset of this cohort (26). The crude Pearson's correlation coefficient between self-reported and measured WC was 0.95; the analogous correlation for HC was 0.88. The differences in mean circumference (measured – reported measure) were 0.91 cm (0.36 inches) for the waist and –1.98 cm (–0.78 inches) for the hip. The correlation between measured and self-reported WHR was 0.69 with a mean difference of 0.027. Self reported and measured weights were strongly correlated (r = 0.97), and the difference (–1.06 kg, or –2.34 lb) was relatively small (26). We calculated BMI [weight (kg)/height squared (m)]) and WHR [waist(cm)/hip(cm)] for each subject.
Assessment of other main covariates
The measurements of other main covariates and potential confounders (such as age, dietary intake, smoking, physical activity, family history of type 2 diabetes, and alcohol consumption) that were collected during the questionnaire-based follow-up were described elsewhere (24). Physical activity, smoking, and alcohol consumption were assessed every 2 y, and information on dietary intake was collected every 4 y.
Statistical analysis
We calculated incidence rates and age-adjusted RRs according to the WC and WHR quintiles. Person-time for each participant was calculated from the date of return of the 1987 questionnaires to the date of type 2 diabetes diagnosis, death from any cause, or 31 January 2000, whichever came first. Incidence rates of type 2 diabetes were obtained by dividing the number of cases by person-years in each category of WC, WHR, or BMI. Age-adjusted RRs were computed as the incidence rate in a specific category of WC and WHR divided by that in the reference category, with adjustment for 5-y age categories. Tests for linear trend across increasing categories of WC and WHR were conducted by treating the categories as a continuous variable with the use of the midpoint value for each category.
Next, using Cox proportional hazards models, we estimated multivariate-adjusted RRs for the whole sample and stratified the results by age (<65 or 65), BMI (<25 or 25), and family history of type 2 diabetes (yes or no). In the Cox models, we adjusted simultaneously for potential confounders, including age (40–44, 45–49, 50–54, 55–59, 60–64, 65–69, and 70 y), smoking [never, past, or current (1–14, 15–24, or 25 cigarettes/d)], alcohol consumption (0–4, 5–9, 10–14, 15–29, or 30 g/d), parental history of diabetes, physical activity (quintiles), cereal fiber intake (quintiles), and trans fat intake (quintiles). In further analyses, we included updated BMI (<23, 23–24.9, 25–26.9, 27–29.9, or 30) in the models. Note that in our analysis we also tested glycemic load and the ratio of polyunsaturated to saturated fat as potential confounders, but they were not confounders and thus were not controlled for.
To assess the joint and independent effects of overall obesity (assessed with BMI) and central obesity (WC and WHR) on the risk of type 2 diabetes, we fit Cox models stratified by BMI categories (< 25, 25–29, or 30). Considering the cutoffs recommended for WHR and WC and the distribution of the study sample, the categories of >0.90, 0.90–0.94, and 0.95 were used for WHR, and <90, 90–99, and 100 were used for WC (cm). Thus, on the basis of these BMI and WC categories, the subjects were classified into 9 groups, which were coded with 8 dummy variables. We included these variables in the Cox models along with other potential confounders to estimate the adjusted RRs. Then, similar analyses were conducted to test the joint and independent effects of BMI and WHR.
Finally, we compared the predictive power of baseline WC, WHR, and BMI deciles on risk of type 2 diabetes through receiver operator characteristic (ROC) curve analysis (27). ROC curve is a graphic representation of the relation between sensitivity and specificity for a diagnostic test. It provides a simple tool for comparing the predictive power of different tests or measures, which can assist with the choice of one test over the others. ROC curves are drawn by plotting the sensitivity (true-positive rate) against the false-positive rate (1 – specificity) for several measures or choices. The measure with a ROC curve that is closest to the upper left corner has the highest sensitivity and specificity and, thus, is the best measure. All data analyses were performed with SAS (version 8.2; SAS Inc, Cary, NC).
RESULTS
The characteristics of the study sample by their WC, WHR, and BMI measures at baseline are presented in Table 1 , Table 2 , and Table 3, respectively. Participants with higher WC, WHR, or BMI values tended to be older, were less active, and were more likely to have hypertension and high cholesterol. As expected, BMI and WC were positively correlated, as were BMI and WHR.
View this table:
TABLE 1. Age-adjusted baseline characteristics according to waist circumference quintiles: the Health Professionals Follow-Up Study1
View this table:
TABLE 2. Age-adjusted baseline characteristics according to waist-to-hip ratio quintiles: the Health Professionals Follow-Up Study1
View this table:
TABLE 3. Age-adjusted baseline characteristics according to BMI quintiles: the Health Professionals Follow-Up Study1
Over 13 y of follow-up (354 180 person-years), we documented 884 incident cases of type 2 diabetes. The age-adjusted RRs for the deciles of WC, WHR, and BMI are presented in Figure 1. The risk of diabetes was already significant at the second decile for both WC and WHR and increased more dramatically from the 9th to the 10th decile. A high WC (upper decile) was a strong risk factor for type 2 diabetes (RR: 20.4; 95% CI: 12.3, 33.8), which was much higher than that for the WHR decile (RR: 8.7; 95% CI: 5.8, 13.0) or BMI (RR: 16.5; 95% CI: 10.4, 26.3).
FIGURE 1.. Age-adjusted relative risk (RR) of type 2 diabetes by baseline waist circumference (WC), waist-to-hip ratio (WHR), and BMI deciles. n = 27 270.
As shown in Table 4, the age-adjusted RRs across quintiles of WC at baseline were 1.0, 2.0, 2.7, 5.0, and 12.0, respectively (P for trend < 0.0001). Further adjustment for physical activity, smoking, parental history of diabetes, dietary intakes of trans fat and cereal fiber, and other covariates only slightly attenuated these RRs. This positive association between WC and the risk of type 2 diabetes remained strong even after adjustment for current BMI (updated in each survey year). RRs across quintiles of WC were 1.0, 1.7, 2.0, 3.0, and 4.5, respectively (P for trend < 0.0001).
View this table:
TABLE 4. Age-adjusted and multivariate-adjusted relative risks (RRs) and 95% CIs for type 2 diabetes according to waist circumference quintiles: the Health Professionals Follow-Up Study
Increased WHR at baseline predicted increased risk of type 2 diabetes, but the association was weaker than that for WC (Table 5). The age-adjusted RRs across quintiles of WHR were 1.0, 2.1, 2.7, 3.6, and 6.9 (P for trend < 0.0001). Adjustment for potential confounders did not change these RRs materially. Further adjustment for current BMI attenuated the association considerably, but the RRs (1.6, 2.0, 2.1, and 2.8) remained significant (P < 0.001).
View this table:
TABLE 5. Age-adjusted and multivariate-adjusted relative risks (RRs) and 95% CIs for type 2 diabetes according to waist-to-hip ratio quintiles: the Health Professionals Follow-Up Study
Elevated BMI at baseline increased the risk of type 2 diabetes (Table 6). The association was weaker than that for WC but was stronger than that for WHR. The age-adjusted RRs across BMI quintiles were 1.0, 1.1, 1.8, 2.9, and 7.9 (P for trend < 0.001). Adjustment for potential confounders only changed the RRs slightly. Further adjustment for baseline WC attenuated the association considerably, whereas adjustment for baseline WHR resulted in much smaller changes in the RRs. For example, with these adjustments, the RR for the top BMI quintile dropped from 7.9 to 6.5, 2.7, and 4.8, respectively.
View this table:
TABLE 6. Age-adjusted and multivariate-adjusted relative risks (RRs) and 95% CIs for type 2 diabetes according to BMI quintiles: the Health Professionals Follow-Up Study
Next, we included BMI, WC, and WHR quintiles in the model simultaneously controlling for other covariates and potential confounders. They all remained significant predictors of the risk of type 2 diabetes. Across quintiles, the RRs and 95% CIs were 1, 0.9 (0.6, 1.4), 1.3 (0.9, 1.9), 1.6 (1.1, 2.3), and 3.1 (2.1, 4.4) for BMI; 1, 1.3 (0.9, 2.1), 1.6 (0.9, 2.5), 2.0 (1.3, 3.3), and 3.0 (1.8, 4.9) for WC; and 1, 1.4 (1.0, 2.1), 1.6 (1.1, 2.2), 1.5 (1.1, 2.1), and 1.7 (1.2, 2.4) for WHR, and the trend tests were significant (P < 0.001). Furthermore, when we conducted the analysis excluding current smokers, the results did not change appreciably (data not shown).
The joint relation between overall (BMI) and abdominal (WC and WHR) adiposity and risk of type 2 diabetes is shown in Figure 2. Those with a BMI < 25 and a WHR < 0.90 were the reference group. Within each BMI category (<25, 25–29, and 30), increased WC (<90, 90–99, and 100) predicted a greater risk of type 2 diabetes. The RR increased from 1.0 to 2.4 in those with a BMI < 25, from 2.2 to 5.9 in those with a BMI of 25-29, and from 6.1 to 14.9 in those with a BMI 30. Similar patterns were observed for WHR.
FIGURE 2.. Relative risk (RR) of type 2 diabetes by waist circumference (WC) and waist-to-hip ratio (WHR) relative to baseline BMI (in kg/m2). n = 27 270. Age, physical activity, smoking status, alcohol consumption, and cereal fiber intake measured during the follow-up were adjusted for in Cox models. 95% CIs are provided in parentheses. **There were too few people in the category to allow for the estimate of RR.
To examine the joint effects of overall and abdominal obesity and to test the usefulness of applying BMI, WC, and WHR simultaneously, we conducted similar analyses using the currently recommended cutoffs for BMI (25 or 30), WC (102 cm), and WHR (0.95). The subjects were separated into 8 exclusive groups in which each group had 0, 1, 2, or 3 of the 3 features. Seven dummy variables were created to represent these groups and were tested in a Cox model with adjustment for potential confounders. As shown in Figure 3, men who had both a high BMI (eg, 25) and a high WC (102 cm) had the highest risk (RR: 8.7), compared with those who had low BMI and WC, but they had an increased risk almost identical to that in those who also had a WHR 0.95 (RR: 8.6); ie, for those with a high BMI and WC, WHR added no predictive power. In general, participants who had a high BMI or a high WC had similar increased risks of type 2 diabetes.
FIGURE 3.. Joint and independent effects of overall [BMI (in kg/m2) 25 or 30] and abdominal [waist circumference (WC) 102 cm or waist-to-hip ratio (WHR) 0.95] obesity on the relative risk (RR) of type 2 diabetes. n = 27 270. Reference group 1 consisted of those persons with a BMI < 25, WC < 102, and WHR < 0.95; reference group 2 consisted of those with a BMI < 30, WC < 102, and WHR < 0.95. On the basis of BMI, WC, and WHR cutoffs, the participants were categorized into 8 groups; 7 dummy variables were then created and included in the Cox models. A similar analysis was conducted for BMIs 25 and 30. Age, physical activity, smoking status, alcohol consumption, and cereal fiber intake measured during the follow-up were adjusted for in the models. 95% CIs are provided in parentheses. There was only one significant interaction: (BMI 25) x (WHR 0.95), P = 0.018. All of the main effects were significant.
The ROC curves (Figure 4) indicate that WC and BMI were similar, but were better than WHR in predicting type 2 diabetes. The cumulative proportion (%) of type 2 diabetes cases that could be identified according to BMI, WHR, and WC deciles is shown in Table 7. If the medians of BMI (24.8), WC (94 cm), and WHR (0.94) were used, 82.5%, 83.6%, and 74.1%, respectively, of the type 2 diabetes cases could be identified. In contrast, if the currently recommended cutoffs (ie, 25 for BMI, 102 cm for WC, and 0.95 for WHR) were used, the proportions would be 78.9%, 50.5%, and 65.7%, respectively. We also calculated the population-attributable fractions for these cutoffs, which were 54.6%, 34.5%, and 35.1%, respectively.
FIGURE 4.. Receiver operating characteristic curves for BMI, waist circumference (WC), and waist-to-hip ratio (WHR) deciles. n = 27 270.
View this table:
TABLE 7. Cumulative proportion of type 2 diabetes identified according to BMI, waist circumference (WC), and waist-to-hip ratio (WHR) deciles
DISCUSSION
In this large cohort of men, we found that both overall obesity (reflected by higher BMI) and central adiposity (reflected by higher WC) predict risk of type 2 diabetes. WC appears to be a better predictor than BMI or WHR. Our findings support the contention that the measurement of WC should be used in clinical practice instead of WHR (11, 14, 28) because 1) there is a stronger association between WC and the risk of developing health conditions such as CVD and type 2 diabetes (9, 29, 30), 2) WC is a better predictor of visceral fat (assessed using more advanced techniques such as dual-energy X-ray absorptiometry and computed tomography) than is WHR (31, 32), 3) the measure of WC is simpler than that of WHR and is associated with fewer measurement errors (14, 33), and 4) the biological mechanisms for the association between WHR and health risks are more difficult to explain than are those for WC (28, 34).
Controversies remain regarding the cutoffs for WC that should be used in clinical practice. The influence of abdominal fatness on a health risk such as type 2 diabetes is a continuous one, and, thus, any cutoffs are arbitrary (17). Our findings and those of others suggest that the current recommended WC cutoffs in the United States need to be reevaluated in light of this new evidence. A lower WC cutoff (eg, 95 rather than 102 cm) for men may be more appropriate. Lower WC cutoffs have been recommended by other researchers. Most recently, on the basis of findings from the third National Health and Nutrition Examination Survey, Zhu et al (22) recommended cutoffs of 90 cm for men and 83 cm for women. They examined the associations between WC, BMI, and obesity-associated risk factors such as low HDL cholesterol, high LDL cholesterol, high blood pressure, and high glucose in white study participants and found that the WC (cm) cutoffs that corresponded to the ORs of BMI of 25 and 30 were 90 and 100 for men and 83 and 93 for women, respectively. Previously, Lean et al (21) proposed the use of the following categories to indicate low risk, increased risk, and substantially increased risk, respectively: <94 cm, 94–101 cm, and 102 cm for men and <80 cm, 80–87 cm, and 88 cm for women. These cutoffs were based on data collected in a random sample of 2183 men and 2698 women aged 20–59 from the Netherlands (20, 21) and were adopted in a World Health Organization (35).
We found that only 66% of the type 2 diabetes cases had a WC 102 cm, the recommended cutoff for elevated WC (Table 7). The WC cutoff that corresponds to a BMI of 25 (a widely accepted classification of overweight) was 95 cm in this cohort, and 80% of the type 2 diabetes cases had a WC 95 cm. The population-attributable fractions for a BMI of 25 and a WC of 102 cm were 54.6% and 34.5%, respectively. A WC of 102 cm corresponds to the lower limit of the upper quintile, ie, only 20% of our subjects had central obesity according to this cutoff. In contrast, about half of the subjects were overweight or obese (BMI 25).
We also tested the independent and joint effects of abdominal adiposity (measured with WC and WHR) and overall body fatness (BMI). Baseline BMI was correlated with WC (r = 0.77) and WHR (r = 0.64) in this cohort (note that these correlations might be weaker than those in other reports because BMI was measured 1 y earlier than WC and WHR). When all 3 baseline anthropometric measures (BMI, WC, and WHR) were included in the model simultaneously, they remained significant predictors of risk of type 2 diabetes. In our multivariate analysis we controlled for updated BMI, but the adjustment only moderately attenuated the RRs for type 2 diabetes across categories of these WC and WHR measures. Our stratified analyses (Figure 2) showed that the risk increased with WC or WHR within each category of baseline BMI (<25, 25–30, and 30). Subjects who had both a high BMI (30) and a high WC (102 cm) had more than twice the risk of those who only had a high BMI or a high WC (RR: 8.4 compared with 3.3; Figure 3).
Our findings that increased WC predicted increased risk within BMI categories are consistent with those of Janssen et al's (36), which are based on the National Health and Nutrition Examination Survey data. Although WC and BMI are highly correlated, they measure different aspects of body fatness. BMI tends to indicate overall fatness but not body fat distribution, whereas WC assesses abdominal adiposity. WC can also contribute information about overall adiposity independent of BMI because the latter cannot separate lean from fat mass. Janssen et al (37) reported that WC and BMI independently contributed to the prediction of abdominal subcutaneous, visceral, and nonabdominal fat. They found that WC was a better predictor of visceral fat than was BMI in white men and women. They recommended the use of both WC and BMI in clinical practice.
Finally, it is worth noting that the associations between overall obesity, central adiposity, and the risk of different health outcomes may vary. For example, WC may be more closely associated with the risk of type 2 diabetes than with CVD, possibly because visceral fat is a greater risk factor for type 2 diabetes (6). Overall adiposity may be a more important indicator of risk of coronary heart disease among men younger than 65 y, although central obesity seemed to be a stronger predictor of risk than BMI among men older than 65 y (4). Moreover, further research is needed to understand the ethnic and racial differences in WC and its association with health risks (38–43). To gain a full understanding of the related issues for developing appropriate guidelines for clinical practice, evidence regarding different health outcomes using studies including different ethnic and racial groups should be thoroughly evaluated.
One limitation of the present study is that we followed a homogeneous cohort of health professionals, who were predominately white. Our data did not allow for testing the possible ethnic differences in the relation between WC and the risk of type 2 diabetes. A second limitation is that these participants were likely to have healthier lifestyles and were thinner than the general US public. Thus, the associations we present may be an underestimate of the true risks associated with central obesity.
In conclusion, the present study provides further evidence that abdominal adiposity measured with the use of WC and WHR is a strong risk factor for type 2 diabetes, independent of overall obesity assessed by BMI. WC is a better measure of central obesity than is WHR for predicting the risk of type 2 diabetes. Both BMI and WC are useful for assessing health risk and should be measured in clinical settings and epidemiologic research whenever possible. Our results suggest that the currently recommended cutoff for WC in men needs to be reevaluated on the basis of this new evidence and that a lower cutoff may be more appropriate for guiding individuals to monitor their body weight, assisting overweight individuals to reduce their body weight, and facilitating public health efforts to fight the obesity epidemic.
ACKNOWLEDGMENTS
We are indebted to the participants of the Harvard Health Professionals Follow-Up Study for their outstanding cooperation. We thank Tricia Li and Juhua Luo for their technical assistance.
YW and FBH were responsible for the study concept and design. YW was responsible for the data analysis and interpretation and the draft of the manuscript. YW, EBR, MJS, WCW, and FBH were responsible for the critical revision of the manuscript for important intellectual content. None of the authors had a conflict of interest.
REFERENCES