Body Roundness Index (BRI)

BRI is a body-shape index introduced by Thomas et al. (2013) that captures body roundness as the eccentricity of an ellipse formed by waist circumference and height. The formula is BRI = 364.2 − 365.5 × √(1 − (WC/(2π))² / (0.5×Ht)²), with WC and Ht in metres. A value closer to 0 indicates a more cylindrical body shape; a higher value indicates more central adiposity. BRI uses the same two measurements as waist-to-height ratio but maps them through a nonlinear shape function instead of a simple ratio.

BMI is mass divided by height squared and does not use waist measurement, so it captures total mass rather than fat distribution. Waist-to-height ratio (WHtR) uses the same two inputs as BRI (waist and height) but as a simple ratio. BRI maps those two inputs through an eccentricity formula that amplifies the effect of waist relative to height, which is why BRI distributions stretch wider in the upper tails than WHtR distributions for the same population.

The Thomas 2013 eccentricity formula amplifies waist relative to height. US women have a lower average height than US men, so the same waist circumference produces a higher BRI in a shorter person. In this NHANES dataset, female p50 sits above male p50 in 10 of the 13 age brackets; the 25-29 bracket is tied, and men edge out women at 35-39 and 75-79. The direction is the opposite of waist-to-hip ratio, which has male medians higher at every bracket. Men still typically have larger absolute waist measurements; the height term reverses the rank order for most age groups.

The site publishes the 2021-2023 single cycle because a sensitivity check against a three-cycle 2015-2023 pool found 64 of 130 (sex × bracket × quantile) cells differ by more than 0.15 BRI units; the largest cell is male 45-49 P95 (11.52 in 2021-2023 versus 9.74 in the pool, a difference of 1.78). The drift is directionally consistent, with recent values systematically higher in upper tails, reflecting the rise in US adult adiposity over the period. Pooling would understate today's distribution. Site metrics with less cross-cycle drift (WHtR, WHR) do pool. The pooled BRI output is kept as an internal sensitivity check.

There is no single widely accepted clinical BRI threshold. Recent studies use study-specific cut-offs derived from outcome-specific ROC analyses: around 4.2 for diastolic dysfunction in Lin et al. (2024), quartile splits in cancer studies, other values for diabetes, NAFLD, and cardiovascular disease. Because the literature does not converge on a single threshold, this page reports percentile distributions rather than risk categories.

These percentiles are derived from NHANES, the CDC's nationally representative health and nutrition survey of US adults. The 2021-2023 cycle was used (n=5,754 adults aged 20 and over). No published study provides US adult BRI percentile tables, so they were calculated directly from CDC public microdata. The method is described on the methodology page.