Calf Circumference: NHANES Microdata Derivation

These percentiles are a FitnessNorms derivation from public CDC NHANES microdata. The cohort definition is informed by Gonzalez et al. (2021), but the final percentile table is presented separately because one published cleaning step cannot be exactly reconstructed.

Files, variables, and eligibility

• Cycles: 1999-2000 (BMX / DEMO / DXX), 2001-2002 (_B), 2003-2004 (_C), and 2005-2006 (_D).
• Measurement: maximal right-calf circumference (BMXCALF) in centimetres, measured while seated to the nearest 0.1 cm.
• Eligibility: age 18 or older, valid sex, calf measurement, and positive MEC examination weight; known pregnancy and an unknown pregnancy result were excluded. Complete measured weight, height, BMI, appendicular DXA lean fields, total fat, and total lean mass were required.
• DXA imputations: each public DXX release contains five imputed records per participant. _MULT_ = 1 was used to establish one participant record; body-composition values are not used to calculate calf circumference.
• Age groups: 18-19, 20-29, 30-39, 40-49, 50-59, 60-69, and 70+, matching the comparison study's native grouping. Public RIDAGEYR is top-coded at 85, so the oldest bracket combines observed ages 70-84 with all participants aged 85 and older. Its survey-weighted mean recorded age is 76.8; the trend chart plots the bracket at age 77 rather than the generic open-bracket value of 75.

Cohort reconstruction and outliers

The four DXX files contain 29,026 unique participants, matching the supplement's starting population. Restricting to adults with valid calf circumference and MEC weights, excluding pregnancy, excluding unknown pregnancy status, and requiring the primary anthropometric and DXA fields produces 17,856 eligible participants exactly, the count reported before outlier removal by Gonzalez et al.

The paper then reports removing 67 calf-circumference outliers using an "IQR method with a threshold of 3." That sentence does not define whether the fence was global or grouped. A conventional global Tukey fence is Q1 - 3 × IQR to Q3 + 3 × IQR. In the reproduced cohort, Q1 is 35.3 cm, Q3 is 40.8 cm, and the resulting 18.8-57.3 cm fence removes 17 records, leaving 17,839. Applying the same rule within documented sex, age, race/ethnicity, and BMI combinations did not reproduce both the paper's 67 removals and final sex counts. The site therefore uses the conventional global result and labels it as a FitnessNorms derivation; it does not claim to reproduce the paper's final n=17,789.

Weights and percentiles

• Eight-year weights: WTMEC4YR × 0.5 for 1999-2000 and 2001-2002; WTMEC2YR × 0.25 for 2003-2004 and 2005-2006. This follows the CDC rule that each source period contributes its duration divided by the combined eight-year duration.
• Quantiles: observations are ordered by circumference and P5, P25, P50, P75, and P95 are interpolated on midpoint cumulative examination weight, then rounded to 0.1 cm.
• Checks: all 14 age-by-sex cells exceed 900 unweighted observations, percentile order is monotonic, and weighted overall means are 39.3 cm for men and 38.2 cm for women, the same rounded means reported by Gonzalez et al.
• Uncertainty: examination weights are used for nationally representative point estimates, but strata, clusters, standard errors, and confidence intervals are not estimated.

Interpretation and population context

Calf circumference combines muscle and adipose tissue, and that mixture changes with age. These neutral distribution percentiles describe position within the 1999-2006 US population; they are not the BMI-adjusted screening cutoffs proposed by Gonzalez et al. and cannot diagnose sarcopenia.

Population and protocol matter. A multinational study pooling community cohorts in Japan, Malaysia, and Taiwan reported lower calf-circumference values in many adult age groups and a decline after about age 60 (Chen et al. 2025). That comparison is context, not a conversion: the cohorts, body size distributions, and measurement protocols differ.

The complete derivation, fixed gates, and generated percentile artifact are retained in scripts/nhanes_calf_circumference_percentiles.py and scripts/nhanes_calf_circumference_percentiles.json. See the reference page for a compact source summary.

View Calf Circumference norms →