Body Roundness Index — NHANES Microdata Derivation
No published paper provides sex- and age-stratified BRI percentile tables for US adults. The CDC NHANES body measures examination collects waist circumference (BMXWAIST) and standing height (BMXHT), the two inputs to the BRI formula introduced by Thomas et al. (2013). Percentiles were derived as follows:
- Formula:
BRI = 364.2 − 365.5 × √(1 − (WC/(2π))² / (0.5×Ht)²), with waist circumference (WC) and standing height (Ht) in metres. The expression inside the square root is the eccentricity of an ellipse whose major axis is half the height and whose minor axis is the waist circumference divided by 2π. Values closer to 0 indicate a more cylindrical body shape; higher values indicate more central adiposity. - Data: NHANES 2021-2023 release (
BMX_L / DEMO_L). Adults aged 20 and over with complete waist and height measurements were included (n = 5,754). - Weights: MEC examination weights (
WTMEC2YR) were applied as-is. This is a single-cycle table; no pooling adjustment was needed. - Why single-cycle, not pooled: A three-cycle pool of 2015-2016 + 2017-March 2020 + 2021-2023 was computed as a sensitivity check (the same cycle structure used for waist-to-height ratio). The pool disagreed with the 2021-2023 cycle in 64 of 130 (sex × bracket × quantile) cells by more than 0.15 BRI units, with the largest cell at male 45-49 P95 (11.52 in 2021-2023 versus 9.74 in the pool, an absolute difference of 1.78). The drift is directionally consistent: recent values are systematically higher in upper tails, reflecting the rise in US adult adiposity over 2015-2023 amplified by the eccentricity formula. Pooling would understate today's distribution. The 2021-2023 single cycle is published; the pool is retained internally as a sensitivity artifact.
- Quantiles: Weighted empirical quantiles (P5, P25, P50, P75, P95) were computed for each sex × 5-year age bracket from 20-24 through 80+ using linear interpolation on the weighted cumulative distribution. Survey-design standard errors were not computed.
- Negative-discriminant guard: The formula's discriminant
1 − (WC/(2π))² / (0.5×Ht)²becomes negative for extreme body shapes (requires WC > π×Ht). The derivation script drops any such rows; in this dataset, zero rows were affected.
This is an internally derived dataset, not a peer-reviewed publication. The derivation method and its limitations are summarised on the reference page. See also waist-to-height ratio methodology for the related body-composition derivation.
Clinical threshold context
BRI does not have a single widely accepted clinical cut-off. Recent studies report different thresholds for different outcomes: Lin et al. 2024 report a BRI of around 4.2 for diastolic dysfunction; cancer-incidence studies typically use quartile splits rather than a fixed threshold; diabetes and non-alcoholic fatty liver disease (NAFLD) cohorts report yet other values.
Because no single threshold maps cleanly across outcomes, this site reports percentile distributions rather than risk categories. A reader interested in clinical interpretation should consult the outcome-specific literature for the condition in question rather than treating any single BRI number as a diagnostic boundary.