What Adding Race to BMI Can Do

In recent years, the perils of body mass index, or BMI, have become a hobbyhorse for professionals in several fields of medicine and research. For decades, doctors have used BMI to help diagnose and treat obesity, diabetes, and other chronic conditions, even as evidence has accumulated that the metric is a poor proxy for excess fat. BMI factors in height and weight but not actual body composition; many people with high BMIs are the picture of health, and many with “healthy” BMIs are at serious risk of metabolic disease. The case against BMI is strong enough that many in medicine would like to be free of it.

Gripes have been raised, too, about medical guidance that relies on race. Although race can track with some factors that influence health, such as lifestyle and socioeconomic status, its relationship to genetic differences is tenuous: Designations such as “Black” and “Asian” cover so many people, with such varied backgrounds, that they’re essentially meaningless as biological categories. When doctors have used race to assess well-being, they’ve missed diagnoses and discriminated against patients. Experts now widely consider many race-based tools in medicine to be harmful and outdated, and are eager to leave them behind.

But researchers and clinicians still rely deeply on both BMI and race, in some cases at the same time. When screening for type 2 diabetes, for instance, race-sensitive BMI cutoffs identify more at-risk people than either factor alone. And however conflicted experts are over how to use that tool and others like it, finding alternatives comes with its own baggage.

When weighing the risk factors for type 2 diabetes, doctors generally flag a BMI of 25 or higher—what’s usually considered “overweight”—as a factor for further testing. But experts have known for a long time that this universal cutoff makes little sense. The original calculation of BMI arose nearly 200 years ago, was never intended for medical use, and was based on data from primarily white, European populations. And so researchers, clinicians, and policy makers around the world have pushed for people of Asian descent to get that same screening at a lower BMI threshold, of 23. The American Diabetes Association and the U.S. Preventive Services Task Force have supported that guidance for years; the CDC’s online prediabetes test has lower BMI cutoffs for Asian Americans than for people from other backgrounds. In Asian countries such as South Korea and Singapore, the lower threshold has been adopted as the national standard. At this point, the reality for people of Asian descent seems quite clear: “We do know that certain groups would benefit from more aggressive therapy at lower BMI cutoffs,” Fatima Cody Stanford, an obesity-medicine specialist at Massachusetts General Hospital, told me.

In this case, applying a race-and-ethnicity filter may help address some of BMI’s shortcomings. Studies suggest that many people of Asian descent—especially of South Asian descent—might have more trouble regulating their blood sugar than other racial and ethnic groups do, and seem more likely to store fat “in places that it shouldn’t be,” such as around visceral organs, in the abdomen, in the liver, and in muscles, Alka Kanaya, a diabetes researcher at UC San Francisco, told me. That so-called visceral fat seems to drive inflammation and insulin resistance, and has been linked to serious medical issues. But BMI can’t account for the location of fat in the body and so can mask diabetes risk for populations in which bodies might appear thin but have more centralized fat. Using a BMI of 25 to screen for diabetes could mean missing one-third to one-half of Asian Americans with type 2 diabetes; a threshold of 23, meanwhile, could cut that missed proportion in half.

At the same time, racialized cutoffs reveal the drawbacks of relying on race at all. “Asians” is a big group—billions of people—that itself contains immense diversity. And when researchers parse out people of, say, Vietnamese descent from those of Indian, Filipino, Chinese, Korean, or Pacific Islander heritage, they find different risks (without much insight into whether those differences are driven by lifestyle, socioeconomic factors, genetics, or a combination). Not everyone knows their full racial or ethnic makeup; people of mixed backgrounds are one of the fastest-growing demographic groups in the United States. “How do you classify them?” Maria Rosario Araneta, an epidemiologist and a diabetes researcher at UC San Diego, asked me. Ideal screening tools excel both at identifying risky cases and at excluding healthy ones. But lowering the BMI cutoff for people of Asian descent starkly increases the number of patients who are unnecessarily flagged for further testing.

Experts also disagree on what could be used instead of BMI to screen people. Body-composition scans can measure fat directly, but they’re expensive and impractical to use on everyone. Another option could be to screen everyone above a certain age for diabetes, using a fasting glucose test or another test that measures a blood sugar called A1C. But fasting glucose tests—which require, well, fasting—may not come with ideal compliance. And Araneta and her colleagues have found that A1C cutoffs for diagnosing diabetes may need to be reevaluated, especially for certain Asian populations that may develop diabetes at lower levels than people of European descent.

Alternative strategies for estimating excess fat have their challenges too. Goutham Rao, a family physician at the University Hospitals Health System, told me that he favors using waist circumference or waist-height ratio. But other researchers find any tool that relies on measuring waists to be impossibly messy. Even well-trained professionals will sometimes take measurements from different parts of a patient’s midsection; the person being measured, too, can skew the results: “You take a small breath in and you change your waist circumference by two centimeters,” Kanaya said. And research suggests that cutoffs that rely on waist circumference may, yes, also need to take into account a person’s ethnicity or race. “Of course, BMI is not perfect,” George King, the chief scientific officer at the Joslin Diabetes Center, in Boston, told me. “But we don’t really have much else to guide us.”

For now, several researchers told me, race-sensitive BMI risk cutoffs could stand to be used more widely, not less. In the United Kingdom, says Rishi Caleyachetty, a general practitioner and an epidemiologist who has studied BMI, although the National Health Service uses the 23 cutoff for some ethnic populations, including those of Asian descent, those thresholds haven’t been consistently adopted across the country. In the U.S., Stanford said, the Mass General Weight Center still uses a universal set of BMI cutoffs to admit patients, and she has had to overrule those standards in several cases to ensure that certain patients are seen. And many insurance companies have relied on BMI to determine whether they’ll pay for GLP-1 medications, without carving out exceptions for particular racial or ethnic groups that might have distinct risk profiles.

Scientists haven’t been able to rigorously study how much of an impact calls to “screen at 23” have had—in part because Asian Americans weren’t well represented in the U.S.’s National Health and Nutrition Examination Survey, which includes estimates of diabetes prevalence, until 2011. King said he thinks that the available evidence hints at a drop in the prevalence of undiagnosed diabetes in Asian American communities. But one small study from 2022, based on self-reported data on diabetes screening, found no change in diabetes-screening rates among Asian Americans after the change in guidance.

BMI cutoffs that take into account race and ethnicity may be short-lived, as researchers develop better tools and protocols to help people identify and manage chronic metabolic conditions. But BMI is still everywhere for a reason: “No single measure will compete with BMI in simplicity,” Samar El Khoudary, a women’s-health researcher and an epidemiologist at Virginia Commonwealth University’s School of Public Health, told me. Across the board, the researchers I spoke with told me that they understand the serious limitations—and major risks—of overusing or misusing BMI and race, separately or together. But many of them also worry that too hastily casting these categorizations aside could do more harm than good. “To be able to remove it, you need to be able to replace it,” El Khoudary said. And she doesn’t yet see a clear plan for what metric can accomplish that—certainly not one that can also avoid all of BMI’s pitfalls.