Body Mass Index Thresholds for Asians: A Race Correction in Need of Correction?

Due to a higher tendency toward central adiposity, Asian Americans face a substantial risk for type 2 diabetes, cardiovascular disease, and other comorbidities at lower body mass indices (BMIs) than other populations (1, 2). As such, in 2004, the World Health Organization (WHO) suggested new BMI categories for many Asian populations, based primarily on data measuring body fat percentage: 18.5 to 22.9 kg/m² for increasing but acceptable risk (normal weight), 23 to 27.5 kg/m² for increased risk (overweight), and 27.5 kg/m² or greater for high risk (obesity) (1). BMI is primarily based on data from non-Hispanic White people (3); however, the WHO recognized this metric's utility by lowering Asian cutoffs to reduce bias. Twenty years later, the United States still has not developed more granular BMI categories despite heterogeneity among Asian American ethnic groups (4, 5). With recent pushes toward disaggregated data and personalized medicine, we believe that increasing granularity for Asian Americans can pave the way for similar efforts among all racial and ethnic groups, making BMI more accurate, tailored, and equitable.

Although its expert consultation explained variations in health risks between Asian subgroups, the WHO did not attempt to redefine cutoffs for each population separately; instead, it provided guidance for countries to create their own definitions (1). As such, obesity cutoffs vary widely in Asia, from 25 kg/m² in India to 28 kg/m² in China (6); however, Asian Americans continue to be treated as a monolith without official disaggregated cutoffs. In 2015, the American Diabetes Association (ADA) recommended using lower BMI cutoffs (≥23 kg/m²) when screening Asian Americans for prediabetes and type 2 diabetes (4), and in 2022, the American Society for Metabolic and Bariatric Surgery (ASMBS) published new guidelines defining a BMI of 27.5 kg/m² or greater for Asian patients as qualifying for surgery (5). These efforts are commendable because they use high-quality evidence about adverse outcomes and because standard BMI thresholds can result in missed opportunities for diagnosis, intervention, and treatment, exacerbating health inequities. More work remains to be done: The Centers for Disease Control and Prevention still does not present nuanced thresholds in its public-facing materials. However, on a more structural level, the efforts of the ADA, the ASMBS, and other organizations have arguably not gone far enough. After all, Asians are a heterogeneous group with diverse genetic profiles, unique cultural practices, and varying diets, and they account for more than 60% of the world's population (2).

Indeed, where there is evidence from disaggregated studies, the findings point to significant variation between subgroups. The California Health Interview Survey (CHIS), conducted in Mandarin, Cantonese, Korean, and Vietnamese, is a well-known data source for studying Asian Americans. In the CHIS analysis evaluating BMI cutoffs, the prevalence of type 2 diabetes was consistently higher for Asian subgroups than for non-Hispanic Whites. However, these data also suggested that, in the 23 to 24.9 kg/m² range, Filipino, Korean, South Asian, and Vietnamese populations—but not Chinese or Japanese populations—had higher diabetes prevalence than non-Hispanic Whites, suggesting varying diabetes susceptibility (7). In one of the most extensive analyses examining BMI cutoffs, with 1 472 819 people from the United Kingdom, researchers similarly demonstrated that all ethnic minority groups that were studied developed diabetes at lower BMIs than the White population. Specifically, the age- and sex-adjusted BMI cutoffs were 23.9 kg/m² in South Asian populations, 26.6 kg/m² in Arab populations, 26.9 kg/m² in Chinese populations, and 28.1 kg/m² in Black populations, compared with the standard cutoff of 30.0 kg/m² (8). These findings raise important questions. Does it make sense for people of Chinese descent to use the same BMI threshold as the South Asian group when their “equivalent risk cutoff” is closer to that of the Arab and Black groups who share the standard BMI threshold?

Most data in this area are cross-sectional and not the longitudinal data needed to make such a determination. However, according to Hsu and colleagues, 4 prospective cohort studies between 2009 and 2014 reported type 2 diabetes incidence in disaggregated Asian American populations (6). All found substantial heterogeneity, with one study reporting incidences of 9.3 per 1000 person-years in Chinese Canadians, 9.5 per 1000 person-years in White Canadians, and 20.8 per 1000 person-years in South Asian Canadians and another reporting incidences of 4.6 per 1000 person-years in Vietnamese Americans, 6.3 per 1000 person-years in White Americans, and 20.3 per 1000 person-years in Korean Americans (6). In both studies, BMI cutoffs were lower among all studied Asian ethnic groups than among non-Hispanic Whites. However, other groups, such as Cambodian and Laotian Americans, were excluded due to data availability issues. Pan-Asian cutoffs move in the right direction but obscure vast heterogeneity within the Asian race; more granular thresholds—and research—are needed.

We thus believe that the ADA, the ASMBS, and other organizations should acknowledge BMI thresholds as temporary placeholders and review data to determine which Asian American ethnicities, if any, can have more sensitive and specific thresholds defined. This strategy would initially result in a patchwork of precise thresholds and proxy values. Still, it would recognize that Asian Americans are not a monolith and would spotlight the need for disaggregated data rather than hiding this under an “Asian cutoff.” Professional organizations should then fund research to replace proxy values with evidence-based thresholds over time. The WHO's lowered cutoff was never meant to be an end in itself; instead, it is a steppingstone that should be refined based on 2 decades of research and further longitudinal studies.

Of course, the need for disaggregated data is not unique to Asian Americans, especially given BMI's broad biases and the heterogeneity of all racial and ethnic groups (3). General efforts to create more granular cutoffs should proceed in parallel; however, our recommendations focus on Asian Americans for 2 main reasons. First, success would create a precedent for complete disaggregation and help ensure that other groups do not stall at an intermediary level. Second, substantial research into Asian ethnic groups—and the WHO's precedent 20 years ago—creates a solid foundation to build upon (1, 2).

We recognize several faults of this argument. For one, the use of race-based modifiers in guidelines has come under increased scrutiny because, as seen with estimated glomerular filtration rates, they risk inadvertently masking and perpetuating inequalities (9). In the case of BMI, however, these modifiers can reverse some of the metric's structural deficiencies and thus expand access to care. Another potential limitation is that some data suggest that there is no increased risk for mortality among Asian Americans with BMIs of 20 to 25 kg/m²; however, we are not arguing for lower BMI cutoffs but for disaggregation of data to better define these values (10). Finally, although we focus on Asian Americans, more granular cutoffs are needed for Asians in the United Kingdom, in Canada, and across the Asian diaspora. Given limited funding, geographically diverse evidence may be necessary to formulate any country's thresholds.

The 20th anniversary of the WHO's expert consultation should remind us of the importance of disaggregation and how BMI can better serve patients. Alternative measures for diagnosing obesity and estimating risk, including waist circumference, waist-to-hip ratio, and waist-to-height ratio, are becoming increasingly prominent, with Annals' recent In the Clinic article on obesity recommending that waist circumference, in particular, be measured for persons of South Asian, Southeast Asian, and East Asian descent with a BMI of 23 kg/m² or greater (3). We welcome these efforts to address BMI's inherent limitations, but continued disaggregation is still necessary given this metric's convenience and staying power. Ultimately, advancing equity will require funding research that engages diverse Asian communities and developing tailored interventions for all ethnicities.