Anthropometric low-muscle-mass classification in home-bound older adults is strongly equation-dependent, with prevalence ranging from 0.0% to 50.3% depending on which equation is applied, and SARC-F showed poor discrimination while calf-circumference discrimination was influenced by structural coupling.
Key Findings
Results
Low-muscle-mass prevalence varied dramatically across the five anthropometric equations applied to the same patient population.
Prevalence ranged from 0.0% (Sun equation) to 50.3% (Santos equation) in the same 392 HHC patients.
Kawakami, Xu Wen, and Gomes equations yielded intermediate prevalences of 26.5%, 14.0%, and 13.0%, respectively.
The study included 392 home health care patients followed at a tertiary referral centre between September 2024 and September 2025.
All five equations (Gomes, Kawakami, Santos, Sun, and Xu Wen) are DXA-referenced anthropometric ALST equations.
Results
Agreement between equations was moderate at best, with Bland-Altman bias and kappa statistics indicating substantial inter-equation discordance.
Bland-Altman bias ranged from -1.18 to +4.33 kg/m² across equation pairs.
Cohen kappa values ranged from 0.435 to 0.492, except for the Sun equation which produced a kappa of 0.000.
A kappa of 0.000 for the Sun equation indicates essentially no agreement with other equations in classifying individuals as having low muscle mass.
These findings were derived from a retrospective cross-sectional study design.
Results
SARC-F showed poor discrimination for Kawakami-defined low muscle mass in both women and men.
SARC-F AUC was 0.563 (95% CI 0.480–0.646) in women for Kawakami-defined low muscle mass.
SARC-F AUC was 0.467 (95% CI 0.362–0.571) in men for Kawakami-defined low muscle mass.
An AUC of 0.467 in men is below chance level (0.5), indicating essentially no discriminative ability.
The Strength, Assistance with walking, Rise from a chair, Climb stairs and Falls (SARC-F) questionnaire was used as the functional screening tool.
Results
Calf-circumference discrimination for low muscle mass was higher when calf circumference was a structural component of the outcome-defining equation.
Calf-circumference AUCs were compared according to whether calf circumference was present in the outcome-defining equation.
Equations containing calf circumference as an input variable yielded higher calf-circumference discrimination AUCs than circumference-free equations.
This pattern indicates structural coupling — the predictor performs better when it is also embedded in the definition of the outcome being predicted.
This finding raises concerns about circularity in evaluating calf circumference as a screening tool when using equations that already incorporate it.
Results
In obese patients, height-squared ALST indices markedly under-classified low muscle mass compared to BMI-adjusted circumference markers.
The Kawakami height-squared ALST index classified only 1.5% of obese patients as having low muscle mass.
BMI-adjusted calf circumference classified 86.3% of obese patients as having low muscle mass.
BMI-adjusted mid-upper arm circumference (MUAC) classified 71.0% of obese patients as having low muscle mass.
These sensitivity analyses used BMI-adjusted calf circumference and MUAC as alternative classification approaches.
The authors note that without direct body-composition and muscle-strength measurements, these findings do not establish sarcopenic obesity.
Methods
The study population consisted of older, multimorbid, functionally dependent home health care patients in whom direct body-composition assessment is typically impractical.
The retrospective cross-sectional study included 392 HHC patients followed at a tertiary referral centre.
Data collection spanned September 2024 to September 2025.
HHC patients are described as 'typically older, multimorbid and functionally dependent, limiting direct body-composition assessment.'
Five DXA-referenced anthropometric equations were applied as practical alternatives to direct DXA measurement in this setting.
What This Means
This research suggests that the choice of mathematical formula used to estimate muscle mass in elderly home-bound patients has an enormous impact on whether those patients are classified as having low muscle mass. When five different widely-used equations were applied to the exact same group of 392 older adults receiving home health care, the proportion classified as having low muscle mass ranged from essentially nobody (0%) to more than half (50.3%), depending solely on which equation was used. The level of agreement between equations was poor to moderate, meaning that a patient flagged as having dangerously low muscle mass by one method might be completely missed by another.
The study also found that a commonly used questionnaire called SARC-F — which asks patients about strength, walking ability, and falls — did a poor job of identifying low muscle mass as defined by at least one of these equations, performing no better than chance in men. Additionally, when researchers looked at how well calf measurement predicted low muscle mass, they found the results were artificially inflated when the equation being used to define low muscle mass already included calf size as an input — a form of circular reasoning. Finally, in obese patients, simpler formulas that divide muscle by height dramatically underestimated the proportion with low muscle mass compared to formulas that account for body weight.
This research matters because home health care patients are among the most vulnerable older adults, and accurate identification of low muscle mass (sarcopenia) is important for guiding their care. However, the findings reveal that without direct measurements of body composition and muscle strength — which are difficult to perform in the home setting — clinicians and researchers cannot reliably determine which patients truly have low muscle mass. The study highlights an urgent need for standardized, validated approaches for assessing muscle health in this population rather than relying on any single anthropometric equation or screening questionnaire.
Check Your Own Numbers
Upload your bloodwork. We'll cross-reference your results against this study and 4,700 others.