Body Roundness Index (BRI)

Clinical background

BMI is the dominant anthropometric marker of obesity but misses fat distribution. Two individuals with the same BMI may have entirely different amounts of visceral adipose tissue, and it is precisely visceral fat that correlates best with cardiovascular and metabolic risk. Waist circumference captures central obesity better than BMI but does not correct for the fact that taller people have larger circumferences for purely geometric reasons. The waist-to-height ratio is an attempt to address this but rests on a simple linear relationship. The Body Roundness Index (BRI) was developed to model the body as an ellipse and thereby to integrate waist circumference and height into a measure of body shape that reflects visceral fat better than traditional indices [1]. The tool is not intended as a diagnostic test for an individual but as a continuous risk marker, particularly useful for population-based stratification and screening.

Calculating the Body Roundness Index

BRI is derived from a geometric model in which the body is assumed to have the shape of an ellipse, with height forming half the major axis and waist circumference determining the minor axis through the diameter $\text{waist circumference}/(2\pi)$. The eccentricity $e$ is calculated as:

$$e = \sqrt{1 - \left(\frac{\text{waist circumference}/(2\pi)}{0{.}5 \times \text{height}}\right)^2}$$

BRI is obtained by a linear transformation of the eccentricity onto a more readily interpreted scale:

$$\text{BRI} = 364{.}2 - 365{.}5 \times e$$

where waist circumference and height are given in the same unit. A value close to 1 corresponds to a slim, cylindrical body, and higher values indicate a rounder body shape. In the NHANES database the lowest observed BRI was approximately 1 and the highest approximately 16 [1].

The derivation study by Thomas et al. (2013) pooled three databases: NHANES III (6,439 adults with DXA-measured body fat), St. Luke's/Roosevelt Hospital in New York (MRI-measured visceral adipose tissue) and Christian-Albrechts University in Kiel (MRI-measured visceral adipose tissue, used for validation). The aim was to relate BRI to both percentage body fat and percentage visceral adipose tissue, and to compare its predictive performance with BMI, waist circumference and hip circumference [1].

Interpretation in practice

BRI is a continuous variable without an established diagnostic cut-off for an individual patient. In the large NHANES-based mortality study by Zhang et al. (2024), BRI was categorised into quintiles and the association with all-cause mortality followed a U-shaped pattern [2]:

These bands are derived from an American population-based cohort and should not be interpreted as diagnostic thresholds for the individual. They can, however, guide the use of BRI as a screening marker: a value below approximately 3.4 or above approximately 6.9 should prompt further assessment of nutritional status and cardiometabolic risk profile respectively.

In a cross-sectional study by Qiu et al. (2024) of 11,980 adults from NHANES 2007 to 2018, the upper quartile of BRI (≥ 6.40) was associated with an 83 per cent increase in the odds of diabetes or prediabetes compared with the lowest quartile (< 3.81) after adjustment for age, sex, ethnicity and lifestyle factors (OR 1.83, 95% CI 1.29 to 2.58) [3]. Each unit increase in BRI corresponded to a 17 per cent increase in the odds of diabetes or prediabetes (OR 1.17, 95% CI 1.07 to 1.27) [3].

Validation and performance

In the derivation study, combined waist and hip eccentricity explained slightly more of the variance in percentage body fat than BMI, waist circumference or hip circumference alone, but the improvement was marginal [1]. For percentage visceral adipose tissue, waist circumference was the single strongest traditional predictor, and eccentricity-based models were comparable but not superior once covariates such as age and height were included [1].

In the external validation by Zhang et al. (2024), based on 32,995 adults from NHANES 1999 to 2018 with a median follow-up of 9.98 years, a U-shaped association between BRI and all-cause mortality was confirmed [2]. Mean BRI rose from 4.80 to 5.62 over the period, reflecting the increasing prevalence of obesity in the USA.

The comparative study by Zhang et al. (2025) evaluated BRI against BMI and other obesity indices in three prospective ageing cohorts: CHARLS (China, 5,768 participants), HRS (USA, 3,151) and ELSA (England, 3,016), all with participants aged ≥ 45 years [4]. A multivariable Cox model with age, hypertension, systolic blood pressure, BMI and BRI as predictors of incident cardiovascular disease gave a C-index of 0.63 in the training cohort (CHARLS), 0.663 in the test cohort (HRS) and 0.621 in the external validation cohort (ELSA) [4]. BRI showed a stronger association with cardiovascular disease than BMI in both the training and the validation cohorts, but the discrimination of the model was moderate overall [4].

In the diabetes study by Qiu et al. (2024), BRI had the highest AUC (0.68) for predicting diabetes and prediabetes, compared with BMI, waist circumference and ABSI [3]. Here too the discrimination was moderate and insufficient for BRI to replace diagnostic testing.

Limitations

BRI rests on the assumption that the cross-section of the body at the waist is circular, which is rarely true in reality. The model does not take account of differences in muscle mass, age or ethnicity that affect the relationship between body shape and visceral fat. The derivation cohort was predominantly American and German with limited representation of Asian populations, which limits generalisability.

The tool lacks established diagnostic cut-offs for the individual. The bands reported in the literature are population-based quintile or quartile boundaries, not clinically validated thresholds. Using BRI as the sole basis for classifying a patient as healthy or diseased is therefore not justified.

BRI correlates strongly with waist circumference, and the marginal improvement in predictive performance compared with simpler measures such as the waist-to-height ratio is small. In acute clinical settings, or in patients with ascites, pregnancy or abdominal surgery, waist circumference is unreliable and BRI should not be calculated.

References

1. Thomas DM, Bredlau C, Bosy-Westphal A, et al. Relationships between body roundness with body fat and visceral adipose tissue emerging from a new geometrical model. Obesity (Silver Spring). 2013. PMID: 23519954
2. Zhang X, Ma N, Lin Q, et al. Body Roundness Index and All-Cause Mortality Among US Adults. JAMA Network Open. 2024. PMID: 38837158
3. Qiu L, Xiao Z, Fan B, et al. Association of body roundness index with diabetes and prediabetes in US adults from NHANES 2007–2018: a cross-sectional study. Lipids in Health and Disease. 2024. PMID: 39154165
4. Zhang Y, Wang Y, Qiao S, et al. Comparative performance of body roundness index and traditional obesity indices in predicting cardiovascular risk: machine learning insights from three prospective aging cohorts. Frontiers in Endocrinology. 2025. PMID: 41079193