To the Editor:
Obesity is the most prevalent chronic disease in the United States and throughout the world, but there is considerable variation in the prevalence of obesity among racial/ethnic groups. Body mass index (BMI; calculated as the weight in kilograms divided by the height in meters squared) has been the standard measure for defining obesity because it is quick, inexpensive, and noninvasive, and has some correlation to metabolic disease prevalence.1 The BMI cutoffs differ in Asian Americans who have a higher risk of metabolic disease at lower BMI values. As such, the BMI criteria for Asian Americans have been adjusted to lower values for all weight classes.2 There are some conundrums associated with the use of BMI. For example, previous studies suggest that men with an overweight BMI without central adiposity have elevated cardiovascular risk factors but lower all-cause mortality risk than do men of other weight categories.3 In addition, there is so-called obesity paradox in which there appear to be protective effects of obesity in patients who have a history of previously diagnosed heart failure.4
Current BMI cutoffs originated from historical Metropolitan Life Insurance Company (MLIC) actuarial data. In 1942, they developed standard tables to determine “ideal” weight, in 1959 “desirable” weight, and in 1983 “height to weight” tables became the standard.5 To develop their first tables in 1942, they used data from 4,000,000 MLIC policyholders from 1911 to 1935 to assess “ideal” weight on the basis of longevity according to sex, height, and weight. To create a normal distribution curve, they characterized policyholders into small, medium, and large body frames, with obesity being defined as a weight of over 20% to 25%, and severe obesity 70% to 100% over “ideal.” In 1959, they discerned the association between body weight and mortality to define “desirable” weight. With information from the Fogarty International Center Conference on Obesity in 1973 and the National Institutes of Health National Health and Nutrition Examination Survey (NHANES), the current BMI tables were developed.
We sought to redefine BMI’s threshold by sex and race/ethnicity based on association with metabolic disease, thereby linking the definition of obesity to future prediction of medical morbidity and mortality. To determine adjusted BMI thresholds by race and sex, we took an approach similar to that used by Wildman et al6 to recalibrate adjusted BMI cutoffs for Asians. US NHANES data from 1999 to 2016 were used to estimate the distribution of BMI, as well as 3 metabolic disease risk factors: hypertension, dyslipidemia, and diabetes. The projected changes in population prevalence with the adoption of new thresholds in the prevalence of obesity for each population subgroup were determined. All data analyses were conducted using the R statistical platform.
When obesity is defined by a correlation with the presence of metabolic risk factors, the BMI cutoffs to define obesity would change for specific race/ethnicity and sex subgroups instead of a single BMI threshold based on historical statistical data from more than 60 years ago. We believe that our proposed recalibration of BMI values that defines obesity in a more biologically based approach allows for a more individualized approach rather than the current “one-size fits all.”
The validation of additional precision biomarkers that can accurately predict future disease can further clarify the appropriate BMI thresholds and potentially other appropriate biomarkers to define obesity, which may obviate the use of BMIs completely. In the meantime, we implore race/ethnic-based definitions to prevent underdiagnosis and overdiagnosis, which have considerable clinical, psychological, and financial implications for patients.7
The Table indicates that men would undergo a shift toward lower BMI cutoff. Although the projected new BMI cutoffs vary by race/ ethnicity and disease risk factor, the overall trend is in the same direction. For women, the BMI shift for black women would be to higher cutoffs, whereas it would generally be lower for Hispanic and white women. For all groups studied, there would be a change in the prevalence of obesity.
TABLE.
Cutoffs for BMI Based on ROC Curve Analysis
| Obesity Co-morbidity | BMI (kg/m2) |
|||||
|---|---|---|---|---|---|---|
| Men |
Women |
|||||
| Black | Hispanic | White | Black | Hispanic | White | |
| Hypertension | 28 | 29 | 28 | 31 | 28 | 27 |
| Dyslipidemia | 27 | 26 | 27 | 29 | 27 | 25 |
| Diabetes | 29 | 29 | 30 | 33 | 30 | 29 |
| ≤2 risk factors | 28 | 29 | 29 | 31 | 30 | 28 |
| Average | 28 | 28 | 29 | 31 | 29 | 27 |
BMI = body mass index
ROC = receiver operating characteristic.
Acknowledgments
Grant Support: This study was supported by grants NIDDK R01 DK103946–01A1, P30 DK040561, and L30 DK118710 from the National Institutes of Health.
Footnotes
Potential Competing Interests: Dr Hur received consultancy fees from Novo-Nordisk. The other authors report no competing interests.
Contributor Information
Fatima Cody Stanford, Division of Endocrinology-Neuroendocrine, Department of Medicine, MGH Weight Center, Massachusetts General Hospital, Boston, MA; Division of Endocrinology, Department of Pediatrics, Massachusetts General Hospital, Boston, MA; Harvard Medical School, Boston, MA.
Minyi Lee, MGH Institute for Technology Assessment, Boston, MA.
Chin Hur, Harvard Medical School, Boston, MA; MGH Institute for Technology Assessment, Boston, MA; Division of Gastroenterology, Department of Medicine, Massachusetts General Hospital, Boston, MA.
References
- 1.Pandey A, Patel KV, Lavie CJ. Obesity, central adiposity, and fitness: understanding the obesity paradox in the context of other cardiometabolic parameters. Mayo Clin Proc. 2018;93(6): 676–678. [DOI] [PubMed] [Google Scholar]
- 2.WHO Expert Consultation. Appropriate body-mass index for Asian populations and its implications for policy and intervention strategies. Lancet. 2004;363(9403):157–163. Erratum in Lancet. 2004; 363(9412):902. [DOI] [PubMed] [Google Scholar]
- 3.He X, Liu C, Chen Y, He J, Dong Y. Overweight without central obesity, cardiovascular risk, and all-cause mortality. Mayo Clin Proc. 2018;93(6):709–720. [DOI] [PubMed] [Google Scholar]
- 4.Carbone S, Lavie CJ, Arena R. Obesity and heart failure: focus on the obesity paradox. Mayo Clin Proc. 2017;92(2):266–279. [DOI] [PubMed] [Google Scholar]
- 5.Komaroff M. For researchers on obesity: historical review of extra body weight definitions. J Obes. 2016;2016:2460285. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Wildman RP, Gu D, Reynolds K, Duan X, He J. Appropriate body mass index and waist circumference cutoffs for categorization of over-weight and central adiposity among Chinese adults. Am J Clin Nutr. 2004;80(5):1129–1136. [DOI] [PubMed] [Google Scholar]
- 7.Kahan SI. Practical strategies for engaging individuals with obesity in primary care. Mayo Clin Proc. 2018; 93(3):351–359. [DOI] [PubMed] [Google Scholar]