---

Analysis Code for Diagnostic Performance of Rectus Femoris Ultrasound for the Diagnosis of Sarcopenia in Older Adults: A Multicenter Study

Data Availability & Requirements

Due to data privacy regulations (GDPR), the original dataset cannot be shared.

To run these scripts, your input dataset (data/analysis_data.csv) should contain the following variables:

Data dictionary

Datasets

Object	Dimensions	Description
Ana_all	130 × 32	Full analysis dataset: both centers, all ages, one row per participant
all_old	78 × 32	Older cohort only (Age >= 65), used for all diagnostic accuracy analyses

Both objects share the same 32 columns. all_old is a row subset of Ana_all.

Each row is one participant. ID is unique within each center and was used as the join key between the bioimpedance (BIA) export and the clinical dataset.

Variables

Variable Name	Type	Description
ID	chr	Participant identifier, unique within center
Age	num	Age at assessment, years
Sex	Factor: male, female	Sex, harmonised from the site-specific codes
Height	num	Body height, cm
Weight	num	Body weight, kg
BMI	num	Body mass index, kg/m². Provided by the Nuremberg dataset; computed as Weight / (Height/100)^2 for Antwerp
SARCF	num	SARC-F screening questionnaire, total score (0–10)
SARCF_pos	Factor: Negative, Positive	SARC-F screen positive, defined as SARCF >= 4
HGSr	num	Handgrip strength, right hand, kg
HGSl	num	Handgrip strength, left hand, kg
HGS	num	Handgrip strength used in the analysis: the higher of HGSr and HGSl, kg
CST	num	Five-times chair stand test, seconds. Recorded as "not possible" or 999 in the source data and set to NA here
GS	num	Gait speed, m/s. Nuremberg recorded gait time, which was inverted (1/GS) to give speed
PA	num	Whole-body bioimpedance phase angle, degrees
ASM	num	Appendicular skeletal muscle mass from BIA, kg (sum of the four limb segments)
corASM	num	Hydration-corrected ASM, kg. Antwerp: each limb segment × 0.39 / segmental water mass, then summed. Nuremberg: ASM × 74.3 / Hydration
ASMh	num	ASM indexed to height: 10000 × ASM / Height^2 (Height in cm), kg/m²
ASMb	num	ASM indexed to BMI: ASM / BMI
MTmean	num	Rectus femoris muscle thickness, mm. Mean of three repeated ultrasound measurements at 50% femur length
CSAmean	num	Rectus femoris cross-sectional area, cm². Mean of three repeated measurements
PAmean	num	Rectus femoris pennation angle, degrees. Mean of three repeated measurements
FL	num	Rectus femoris fascicle length, mm. Derived as MTmean / sin(PAmean × π / 180)
asm_problem	logi	Low muscle mass: ASM < 20 kg (men) or ASM < 15 kg (women)
grip_problem	logi	Low muscle strength: HGS < 27 kg (men) or HGS < 16 kg (women)
cst_problem	logi	Impaired physical performance: CST > 15 s, or CST missing
MusLos	Factor: No-Loss, Loss	Reduced muscle strength / performance: cst_problem OR grip_problem
sarcopenia	Factor: Non-sarcopenic, Sarcopenic	EWGSOP2 sarcopenia: asm_problem AND (grip_problem OR cst_problem)
group3	Factor: Group I, Group II, Group III	Group I = age < 65; Group II = age ≥ 65 without sarcopenia; Group III = age ≥ 65 with sarcopenia
AgeG	chr	Age group: "young" (< 65) or "old" (≥ 65)
datas	Factor: antwerpen, nuremberg	Recruiting center

Classification cut-offs (EWGSOP2)

Criterion	Men	Women
Handgrip strength	< 27 kg	< 16 kg
Appendicular skeletal muscle mass	< 20 kg	< 15 kg
Chair stand test	> 15 s	> 15 s

Notes on the data

Missing-value coding. The source datasets use 999 as a missing-value sentinel. All numeric 999 values were converted to NA during cleaning. For handgrip strength this conversion happens before the cut-offs are applied, because 999 would otherwise be read as an extremely high grip strength.

Chair stand test. A participant unable to complete the test is treated as having impaired performance: cst_problem is TRUE when CST is missing as well as when CST > 15 s.

Factor level order. sarcopenia is coded with Non-sarcopenic as the first level and Sarcopenic as the second, and Sex with male first. This order matters: the ROC analyses pass levels = c("Non-sarcopenic", "Sarcopenic") so that the first level is treated as the control group, and the sex-stratified outputs follow the Sex level order.

group3 in all_old. Because all_old is restricted to age ≥ 65, only Group II and Group III occur. The three-group comparison (Figure 1) is therefore run on Ana_all, not on all_old.

Ultrasound measurements. All rectus femoris parameters are the mean of three repeated measurements taken at 50% of femur length. The three individual measurements are dropped during cleaning and are not present in the analysis datasets.

Exclusions. Participants with a pacemaker were excluded from the Antwerp sample, since bioimpedance analysis is contraindicated. Nuremberg participants were restricted to those flagged Include == 1, and participants missing all three ultrasound measurements were removed.

Script Execution Order

1. 01_Data_clean_and_tidy_Sarcopenia.R - Cleans raw variables and handles missing values.
2. 02_Data_Analysis_Sarcopenia.R - Fits primary logistic regression models.