Lower BMI cut points warranted in people of non-European descent

Shelley Wood

April 11, 2007

Hamilton, ON - Different ethnic groups develop clinically significant increases in lipid, glucose, or blood-pressure levels at much lower body-mass indexes (BMIs) than those predicted in established BMI cut points, new research shows[1]. The study is the latest to question the validity of BMI—developed and validated in people of European descent—as a tool for risk stratification in people from other ethnic backgrounds.

Fahad Razak (McMaster University, Hamilton, ON) and colleagues publish their findings online April 9, 2007 in Circulation.

According to study coauthor Dr Sonia S Anand (McMaster University), the analysis was prompted by the observation that certain ethnic groups, including South Asians and North American aboriginal people, have an increased prevalence of diabetes at much lower BMIs than do people of European descent.

BMI cut points may not be SHAREd

Razak, Anand, and colleagues analyzed data for 1247 patients, collected in the Study of Health Assessment and Risk in Ethnic Groups (SHARE) and in aboriginal peoples (SHARE-AP), verifying ethnicity from study participants. Subjects underwent tests for glucose, lipids, waist and hip ratio, and BMI; 169 patients were excluded for having established diabetes.

In the remainder, researchers report a "marked ethnic variation" in BMI, with aboriginals having mean BMIs that were more than 7 kg/m2 greater than the Chinese participants, with South Asians and Europeans distributed in between. Glucose levels were significantly higher in aboriginals and South Asians, with South Asians also having the worst lipid profiles, while Chinese subjects had higher blood-pressure levels than Europeans. In a factor analysis, using the parameters of glucose, lipid, and blood pressure, the authors determined the cut point at which nonwhite ethnic groups developed abnormalities in these parameters. The cut point for all three parameters in Europeans was a BMI of 30, the BMI that is also the accepted cut point for obesity. Strikingly, however, cut points in the three other groups were much lower.

"Compared with a BMI in the 30s for European whites, the South Asian, aboriginal, and Chinese individuals started to show abnormalities in their glucose parameters above a BMI of 21, which is far lower than 30," Anand said, adding that each point on the standard BMI scale is equivalent to about six pounds.

"So if we say that's about a 54-pound difference, we're starting to see changes at very low BMIs in the nonwhite groups compared with the white group. We need to reevaluate our normal BMI ranges, because the normal BMI is defined by a BMI of 25, but if we use that as 'normal' in someone of South Asian origin, they in fact might be about 25 pounds overweight, and that's why we see a large number of nonwhite individuals already having diabetes despite having 'normal' BMIs. . . . This calls for second look at cut points for BMI.'

Other differences seen

Other surprising differences also emerged in the data, Anand pointed out. While a low BMI was associated with clinically significant glucose and lipid increases in South Asians, in the Chinese and aboriginal groups, lipids didn't reach abnormal levels until BMIs were in the 26-point range. For blood-pressure increases, only the Chinese seemed to be at risk at lower BMIs. By contrast, aboriginal people did not develop abnormal blood-pressure levels until BMIs hit 33. "We would have expected that aboriginal people would have had some relationship between a lower BMI cut point and blood pressure, but in fact they seem to be able to preserve a low blood pressure despite having very high BMIs," Anand said.

BMI cut points for abnormal blood pressure, lipids, or glucose*

Parameter Aboriginal BMI (kg/m 2 ) South Asian BMI (kg/m 2 ) Chinese BMI (kg/m 2 )
Hypertension >30 28.8 25.3
Dyslipidemia 26.1 22.5 25.9
Dysglycemia 21.8 21.0 20.6
*Cut point in Europeans for all three parameters was 30.0 kg/m 2

Another recent analysis of the INTERHEART study participants, also carried out by McMaster investigators, suggested that BMI be dropped altogether as a risk-stratification tool, precisely because risk stratification based on BMI values can be inaccurate and misleading in different ethnic groups, as reported by heart wire . According to Anand, she and her coinvestigators were unable to include waist circumference or waist-to-hip ratios in the current analysis, but it is something the group intends to look at within the SHARE/SHARE-AP cohort.

"I would concur that data from the INTERHEART study clearly showed it was the distribution of fat that's more important than body weight per se," Anand said. "This raises the point that when different groups around the world, like NHANES or public-health agencies, are trying to put out recommendations for target BMI, they probably should have different BMI targets for different ethnic groups."

While there may be more quibbling about the precise cut points needed for different groups, Anand believes the findings from the current study are immediately relevant for clinicians.

"They need to consider screening people of South Asian, aboriginal, or Chinese origin for problems like glucose and lipids at lower BMIs," she said. "So if a South Asian patient presents with a BMI of 25, physicians shouldn't assume that their blood sugar will be normal. They should be aware that that's actually overweight for a South Asian and screen him or her for glucose and lipid problems."


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.