ABSI Calculator — A Body Shape Index

Background.

Two 45-year-old men are both 178 cm tall and weigh 80 kg, so both have a BMI of 25.25. One measures 90 cm round the waist, the other 100 cm. BMI cannot tell them apart. A Body Shape Index can: it asks whether a waist is large for the body it belongs to, once the part of waist size that simply follows from height and weight has been divided out.

Nir and Jesse Krakauer introduced the index in PLoS ONE in 2012, using 14,105 adults from the US National Health and Nutrition Examination Survey (NHANES) of 1999–2004. In data like that, waist circumference and BMI are so closely correlated, around 0.9, that it is hard to say what a waist measurement adds once BMI is known. Their answer was to find the powers of weight and height that best predict waist, and to define ABSI as the ratio of a person's actual waist to that prediction. The result is almost uncorrelated with height, weight and BMI, yet it tracked mortality: over a mean follow-up of 4.8 years, during which 828 participants died, death rates rose by about 33% for each standard deviation of ABSI above average.

The raw index is an awkward number, about 0.08 in units of m^(11/6)·kg^(−2/3), so the figure to read is the z-score, the number of standard deviations you sit from the published mean for your exact age and sex. That mean climbs with age, by about 14% in men between 18 and 85, so comparing everyone with one all-ages average would flatter older people and penalise younger ones.

ABSI describes shape. It does not measure body fat or diagnose anything, and judgements about your own health belong with a clinician who can see blood pressure, blood tests and history alongside it.

What is absi calculator (a body shape index)?

A Body Shape Index is waist circumference divided by the waist expected for a person's BMI and height: ABSI = WC ÷ (BMI^(2/3) × height^(1/2)), with waist and height in metres and weight in kilograms. The exponents come from the Krakauers' regression, in which waist scaled roughly as weight to the two-thirds times height to the minus five-sixths. Because that predictable part is divided out, ABSI had absolute correlations below 0.1 with height, weight and BMI in the derivation sample. The NHANES adult mean was 0.0808, with a standard deviation of 0.0053.

How to use this calculator.

1. 01Choose your sex and enter your age; together they pick the reference row. Everyone aged 85 or over is compared with the 85 row.
2. 02Pick metric or imperial and fill in the matching weight, height and waist fields; the other set is ignored.
3. 03Measure the waist where the reference data were measured: find the top of the hip bone at your side, run the tape horizontally round at that level, and read it at the end of a normal breath out. The navel and the narrowest point give different numbers.
4. 04Measure twice and use the average. For the man in the worked example, 1 cm of waist moves the z-score by about 0.24, nearly a quarter of a standard deviation.
5. 05Read the z-score first, then the hazard with its interval, then the BMI shown beside them.

The formula.

ABSI = WC ÷ (BMI^(2/3) × H^(1/2)), which rearranges to WC × W^(−2/3) × H^(5/6), with waist WC and height H in metres and weight W in kilograms. Imperial entries are converted first at exactly 0.45359237 kg per pound and 2.54 cm per inch. The index has units, so it is only comparable with the reference values when computed in metres and kilograms.

The z-score is (ABSI − mean) ÷ SD, where the mean and SD are the smoothed age- and sex-specific values in Table S1 of the paper's supplement, one row per year of age from 18 to 85. Age is taken in completed years.

The relative hazard is 1.33^z, the paper's estimate of 33% higher mortality per standard deviation, and the two interval figures are 1.20^z and 1.48^z from its 95% confidence interval. When z is negative the 1.20 figure is the larger of the two, so the page shows the smaller as the low end and the larger as the high end. Each compares you with someone of the same age, sex, height and weight whose ABSI sits exactly on the mean.

Two directions are worth checking against intuition. A bigger waist at the same height and weight raises ABSI. More weight at the same waist and height lowers it, because a heavier body is expected to carry a bigger waist.

The band shown is descriptive only: more than one SD below the mean, below it, above it, or more than one SD above. Five-band risk scales that circulate online could not be traced to the source paper, so none is applied.

A worked example.

Frequently asked questions.

References& sources.

1. [1]Krakauer NY, Krakauer JC. "A New Body Shape Index Predicts Mortality Hazard Independently of Body Mass Index." PLoS ONE. 2012;7(7):e39504. PMID 22815707, PMC3399847. Source of the ABSI definition (Eq. 3), its units m^(11/6) kg^(−2/3), the NHANES sample mean of 0.0808 ± 0.0053, the z-score definition (Eq. 4), the hazard ratio of 1.33 per SD (95 % CI 1.20–1.48, Table 2), the correlations with height, weight and BMI below 0.1 (Table 1), the waist measurement protocol, and the null result for Mexican ethnicity (Table 3). Open access — the PDF was downloaded and every figure above read verbatim.
2. [2]Krakauer NY, Krakauer JC — Table S1, supporting information to the above (doi:10.1371/journal.pone.0039504.s001; figshare item 122583, file Table_S1.txt, version 1, CC BY 4.0). The complete age- and sex-specific smoothed ABSI means and standard deviations from NHANES 1999–2004. All 68 adult rows (ages 18–85) are embedded verbatim in this calculator. Downloaded and MD5-verified against the figshare record (8e3ec4d9abfbd9dfaaf4377a120830a2).
3. [3]Centers for Disease Control and Prevention / National Center for Health Statistics — NHANES Anthropometry Procedures Manual, §3.4.8 Abdominal (Waist) Circumference, reproduced as PhenX Toolkit protocol 21604. Source of the waist measurement instruction: the uppermost lateral border of the right ilium at the midaxillary line, tape horizontal, read at the end of normal expiration.
4. [4]National Institute of Standards and Technology — "Guide for the Use of the International System of Units (SI)," NIST Special Publication 811, 2008 edition, Appendix B.9. Source of the exact conversions 1 lb = 0.45359237 kg and 1 in = 2.54 cm used for imperial input.