“The use of BMI has long been criticized as fundamentally flawed because it does not distinguish fat mass from lean mass. Despite this limitation, one nationally representative analysis found that BMI is strongly correlated (Pearson correlation coefficient of approximately 0.9) with fat mass adjusted for height as measured by dual x-ray absorptiometry (DXA), which is considered a gold standard. Moreover, BMI is strongly associated with indicators of cardiovascular risk, such as blood pressure and blood lipid levels, and is similar to DXA as a predictor of these risk factors and metabolic syndrome. The correlation between BMI and fat mass measured by DXA is nearly identical among non-Hispanic Black, White, and Hispanic groups. Moreover, correlations between BMI and cardiovascular risk indicators were similar among groups stratified by self-identified race and ethnicity. Among healthy persons who do not smoke, the relation between BMI and mortality is nearly identical among Black persons, White persons, and Asian Americans, based on pooled data from prospective cohort studies. In all groups, the lowest mortality was observed below a BMI of 25 kg/m2.
Although associations expressed as correlations or relative risks are similar across groups defined by race and ethnicity, the absolute risks for diabetes and cardiometabolic disease at a specific BMI are higher for some Asian Americans and other populations, which should be considered in screening or risk prediction. Some concerns about BMI are due to the cut points used to define overweight and obesity. This is inevitable because the underlying relationships, such as with risk for diabetes or coronary heart disease, are continuous and approximately linear and do not represent thresholds of risk. Similar limitations are found with other adiposity measures, as outlined in a previous review, and would exist even with a perfect measure of adiposity; cut points for clinical practice should differ depending on the decision being made (for example, simple counseling versus bariatric surgery).
In clinical practice, BMI can serve as a valuable method to screen for adiposity owing to its robust correlation with DXA, ease of measurement, and low cost. Waist circumference can provide additional information about visceral fat and disease risk, especially in the midrange of BMIs and at older ages, when loss of lean mass makes BMI less informative. However, waist circumference is more difficult to standardize and suffers from the same limitations as BMI when cut points are used. For example, the cut points of 88 cm (35 in) and 102 cm (40 in) for U.S. and Canadian women and men, which are based on clinical guideline recommendations, would allow substantial increases in risk for disease and mortality before these suggested thresholds are reached. The most simple and sensitive variable by which to monitor adiposity is change in weight since early adulthood (which can easily be obtained from the patient’s history) and over time; this takes into account a person’s own lean mass, and even small increases are associated with risks for chronic disease. Importantly, this can identify the need to make adjustments in diet or activity patterns well before arbitrary thresholds are exceeded or complications of adiposity develop. As also suggested by the AMA [American Medical Association] for a more accurate evaluation of obesity-related risks, BMI should be used in conjunction with other assessments; waist circumference and weight change since early adulthood can be readily used for this purpose. Waiting for complications such as hypertension, dyslipidemia, or diabetes to manifest is undesirable because some damage may be irreversible, and weight loss is challenging. Although DXA is considered a gold standard for assessment of body fat, its high cost and intrusive nature render it impractical for routine clinical use, especially in underprivileged clinics or communities with pervasive medical skepticism. [..]
In addition to being a strong predictor of health outcomes, BMI provides a metric by which to comprehend and quantify the ramifications of structural racism and discrimination for population health over the lifespan because data are available across populations and time. Thus, BMI emerges as a modifiable risk factor for existing disparities, enabling researchers to gain deeper insights into the influence of systemic factors—including inequities in resource allocation, educational access, housing security, dietary options, and health care quality—on health outcomes and to develop efficacious interventions to mitigate health disparities. By monitoring interventions and tackling the fundamental origins of structural racism through the lens of BMI data, researchers can identify practices and policies that have a positive and enduring effect.”
Full editorial, AG Cuevas and WC Willett, Annals of Internal Medicine, 2024.7.23