This study was conducted using NHANES data from 2011 to 2018, which included Dual-energy X-ray Absorptiometry (DXA) measures of body composition. NHANES was designed to represent the non-institutionalized U.S. civilian population using a complex, multistage probability sampling method, including oversampling of certain subgroups to produce reliable statistics. All procedural manuals and survey content are indexed and publicly accessible online, ensuring transparency and accessibility for researchers.¹ For this study, participants aged 20–60 years with complete and reliable DXA measurements were considered eligible for the analysis of LMM. We excluded participants with missing data on key variables, such as sample weight, vitamin D intake, DXA results, BMI, physical activity and other relevant covariates. After applying these criteria, a total of 8,443 participants were included in the final analysis. The participant selection process is illustrated in Figure 1.

Dietary intake interviews were conducted in person at the NHANES mobile examination clinics. Survey participants recalled the specific types and amounts of foods consumed the previous day. Dietary intake was evaluated using two 24-h dietary recalls for each participant. The first 24-h recall was collected in person, and the second was conducted by phone 3 to 10 days later. Vitamin D intake was calculated by averaging the values from the two 24-h recalls. If the second 24-h intake data was missing, the intake from the first 24-h recall was used instead. The dietary intake data were assessed using the automated multiple pass method (10), a structured interview technique consisting of five steps: quick list, forgotten foods, time and occasion, detail cycle, and final probe. After the dietary interview, the caloric and nutrient contents of each reported food and beverage item were coded using the U.S. Department of Agriculture (USDA) Food and Nutrient Database for Dietary Studies (FNDDS).² Information on the consumption of specific food groups was extracted from the Food Patterns Equivalents Database (FPED), with each NHANES data cycle analyzed using its corresponding version of FPED.

LMM was defined using the criteria recommended by the Foundation for the National Institutes of Health (FNIH) Sarcopenia Project (11). Specifically, LMM was determined based on ALM adjusted for BMI. Participants were classified as having LMM if their ALM/BMI ratio was less than 0.789 for men and less than 0.512 for women. This definition has been widely used in epidemiological studies and allows for a standardized assessment of muscle mass across different population groups, accounting for variations in body size. The whole-body DXA scans were performed using a Hologic QDR-4500A fan-beam densitometer (Hologic, Inc., Bedford, MA, United States) to assess body composition. Participants were excluded from the DXA examination for several reasons, including pregnancy, self-reported weight over 450 lb. (204 kg), height over 6 ft., 5 in. (195 cm) due to DXA table limitations, and the use of radiographic contrast material (barium) in the past 7 days. Appendicular lean mass (ALM) was calculated as the sum of lean mass for the four limbs.

The covariates included in our analysis were age, gender, race/ethnicity (Mexican American, Non-Hispanic White, Non-Hispanic Black, Other race), family poverty-to-income ratio (PIR) (<1, 1–3, >3), body mass index (BMI), total energy intake, diabetes, cardiovascular disease (CVD) status, cancer, smoking status, and drinking status. Physical activity was measured using Metabolic Equivalent (MET) values. Total MET-minutes were estimated by summing the MET-minutes for each activity based on its type and intensity. Participants reporting no regular physical activity in the past 30 days were classified as sedentary. After excluding these sedentary individuals, the remaining participants were divided into gender-specific tertiles according to their physical activity levels, resulting in four categories: sedentary, low, moderate, and high (12). Smoking status was categorized as nonsmoker (having smoked fewer than 100 cigarettes in a lifetime), former smoker (smoked more than 100 cigarettes but not currently smoking), and current smoker (smoked more than 100 cigarettes and currently smoking). Diabetes status was defined using self-reported diagnosis, HbA1c levels ≥6.5%, or fasting plasma glucose (FPG) levels ≥126 mg/dL. Drinking status was classified as never drinker, abstainer, or current drinker based on self-reported alcohol consumption patterns. The presence of CVD and cancer was determined through self-reported physician diagnoses obtained during structured interviews. Furthermore, laboratory measures included serum vitamin D, serum total cholesterol, serum albumin, and serum calcium. Serum vitamin D concentration (Serum 25-hydroxyvitamin D) was classified into four categories: severe deficiency (<25 nmol/L), deficiency (25–49.9 nmol/L), insufficiency (50–74.9 nmol/L), and sufficiency (≥75 nmol/L) (13).

Due to the complex sampling design used in NHANES, all analyses were conducted using sample weights. Baseline characteristics were presented unweighted to provide an accurate description of the study cohort. Continuous variables were shown as mean ± standard deviation (SD), while categorical variables were displayed as numbers and percentages (%). The Student’s t-test or Wilcoxon rank-sum test was used to compare continuous variables, and chi-square tests were employed to compare categorical variables. The association between vitamin D intake and sarcopenia was evaluated using multivariable logistic regression (MLR). Vitamin D intake was categorized into quartiles, with the lowest quartile (Q1) serving as the reference group. Model 1 was adjusted for age, gender, race, BMI, and PIR. Model 2 included additional adjustments for total energy intake, serum total cholesterol, serum vitamin D, serum albumin, and serum calcium. Finally, Model 3 was further adjusted for diabetes, CVD, cancer, smoking status, drinking status, and physical activity. To examine the independent effect of vitamin D intake, total energy intake was introduced as a covariate in Models 2 and 3. Before its inclusion, we conducted a Pearson correlation analysis between total energy intake and vitamin D intake, yielding an R value of 0.305 (p < 0.001, Supplementary Figure S1), indicating a moderate positive relationship. Additionally, the variance inflation factor (VIF) was calculated as 1.075, demonstrating minimal multicollinearity. Thus, total energy intake was appropriately included in these models to account for potential confounding effects. Results were reported as odds ratios (OR) with 95% confidence intervals (CI). To evaluate the trend between vitamin D intake and LMM, a p-value for trend was calculated by treating categorical exposure variables as ordinal in the MLR. All analyses were performed using R software (version 4.4.0), with statistical significance set at p < 0.05.

The baseline characteristics of the study population (n = 8,443) are summarized (Table 1), highlighting significant differences between participants with LMM (7.3%) and those without (93%). Participants with LMM were older (43.6 ± 12.0 vs. 39.1 ± 11.8 years, p < 0.001), had a higher BMI (35.4 ± 7.6 vs. 28.2 ± 6.2 kg/m², p < 0.001), and a greater prevalence of diabetes (21.5% vs. 6.8%, p < 0.001) and CVD (9.3% vs. 2.7%, p < 0.001). The LMM group also showed a higher proportion of Mexican Americans (26.9% vs. 9.8%, p < 0.001). Additionally, participants with LMM had higher total energy intake, lower serum vitamin D, calcium, and albumin levels (all p < 0.001), and were more likely to be sedentary (27.5% vs. 14.9%, p < 0.001), while current alcohol consumption was lower in this group (59.4% vs. 74.3%, p < 0.001). These differences underscore the association between LMM and adverse health and lifestyle factors.

In the multivariable logistic regression models, we assessed the association between dietary vitamin D intake and the risk of LMM, adjusting for various covariates in three models (Table 2). In Model 1 (Supplementary Table S1), participants in the higher quartiles of vitamin D intake (Q3 and Q4) had significantly lower odds of LMM compared to the lowest quartile (Q1), with odds ratios (OR) of 0.57 (95% CI: 0.41–0.79) for Q3 and 0.54 (95% CI: 0.37–0.79) for Q4 (p for trend <0.001). In Model 2 (Supplementary Table S2), after adjusting for total energy intake and additional covariates, the association between higher dietary vitamin D intake and lower risk of LMM was attenuated and did not reach statistical significance (Q3: OR = 0.71, 95% CI: 0.50–1.01; Q4: OR = 0.74, 95% CI: 0.50–1.11). The trend across quartiles was borderline significant (p for trend = 0.051), suggesting a potential relationship that may warrant further investigation. In Model 3 (Supplementary Table S3; Figure 2), after further adjusting for additional health-related factors, the association between vitamin D intake and LMM reached statistical significance (p = 0.044). Participants in Q3 had an OR of 0.69 (95% CI: 0.49–0.97), and those in Q4 had an OR of 0.70 (95% CI: 0.47–1.05), with a significant trend across quartiles (p for trend = 0.029).