This study used data from the Korean Genome and Epidemiology Study (KoGES), conducted by the Korea National Institute of Health. The KoGES is a large-scale consortium project consisting of six prospective cohort studies to investigate and assess the genetic and environmental etiologies of non-communicable diseases in Korea, including obesity, hypertension, diabetes mellitus, and cardiovascular diseases [15]. For this study, we used 2003–2013 data from the KoGES_Health Examinee (HEXA) study, including 173,202 urban residents aged 40–79 years, as well as data from the fourth wave of the KoGES_Ansan and Ansung study (2007–2008), including 6,688 participants, aged 44–76 years, who lived in Ansan (an urban area) or Ansung (a rural area). As specific information on RT levels could be retrieved from the fourth wave of the KoGES_Ansan and Ansung study, we included this and not the baseline data. All participants underwent physical examinations and face-to-face surveys conducted by trained medical staff. A detailed description of the KoGES cohort studies has been provided previously [15].

Among the 179,890 participants from the cohorts, 53,551 were excluded from the present study based on the following exclusion criteria: lack of data on fat-free mass (n = 44,195), lack of data on leisure-time PA levels (n = 4,360), and no data available for the covariates (n = 4,996). Overall, 126,339 participants (81,263 women) were included in the final analysis (Additional file 1). This study was approved by the Institutional Review Board Committee of the Korea National Institute of Health, Korea Disease Control and Prevention Agency (Approval No. 2021-04-02-P-A).

All participants completed questionnaires containing details on RT regularity and leisure-time PA levels. RT was defined as any training program involving muscle contraction against external resistance using body weight, weight machines, barbells, or dumbbells. The frequency (per week), training time (min/week), and training period (months) of RT were assessed. Regular RT was defined as participation in an RT program for more than 1 day per week. Participants were classified into two groups based on the regularity of RT: “Non-RT (not performing RT),” and “RT (performing RT).” To investigate the presence of an inverse dose–response relationship between RT levels and the risk of low muscle mass, the training period (months) and frequency (per week) of RT were used. Based on the frequency of RT, participants were categorized into one of five subgroups: “Non-RT (not performing RT),” “1 day/week,” “2 days/week,” “3–4 days/week,” and “≥5 days/week.” Similarly, participants were classified into one of four subgroups based on training period of RT: “Non-RT (not performing RT),” “<12 months,” “12–23 months,” and “≥24 months.”

Regarding leisure-time PA levels, we assessed the intensity, frequency (per week), and duration (min/week) during a typical week. Moderate-intensity leisure-time PA was defined as participating in sports or engaging in exercise that results in sweating. Based on the PA guideline (moderate-intensity leisure-time PA for at least 150 min per week) [16] and RT regularity, participants were categorized into one of four subgroups: “Low-PA (not meeting the guideline),” “Low-PA+RT (not meeting the guideline but performing RT),” “High-PA (meeting the guideline),” and “High-PA+RT (meeting the guideline and performing RT).”

Low muscle mass was defined based on the fat-free mass index (FFMI), which was determined using fat-free mass measured by bioelectrical impedance analysis (BIA) (InBody 3.0, Biospace, Seoul, Korea). The FFMI was calculated by dividing the fat-free mass (kg) by the square of the height (m) (kg/m2). According to a recent study on the screening of low muscle mass, the cutoff points of FFMI were 17.5 kg/m2 for men and 14.6 kg/m2 for women [17].

Our analyses encompassed various sociodemographic and health-related factors, including age, sex, educational level, drinking and smoking habits, PA-time, body mass index (BMI), waist circumference (WC), fat-free mass, blood pressure (BP), hypertension, diabetes mellitus, and laboratory parameters. Educational level was classified as elementary school graduate or lower, middle or high school graduate, and college graduate or higher. Drinking and smoking habits were classified as “never,” “former,” and “current.” PA-time was defined as the total time (min/week) spent engaging in moderate-intensity leisure-time PA.

Anthropometric data, including height, body weight, and WC, were measured by trained healthcare providers using standardized methods. BMI was calculated as body weight (kg) divided by height (m) squared (kg/m2). Trained healthcare providers also measured BP using standard protocols. Systolic BP (SBP) and diastolic BP (DBP) were obtained by averaging two readings from the arm with the highest SBP after the participant had rested for 5 min in a seated position. Blood samples were collected after an overnight fasting period of 8 h. Biochemical assays were performed to determine levels of total cholesterol (T-Chol), high-density lipoprotein cholesterol (HDL-C), triglyceride (TG), and fasting blood glucose (FBG). Hypertension was defined based on a previous diagnosis by a physician, current use of antihypertensive drugs, SBP ≥140 mmHg, or DBP ≥90 mmHg. Diabetes mellitus was defined based on a previous diagnosis by a physician, current use of antidiabetic medications, including insulin and oral hypoglycemic agents, FBG ≥126 mg/dL, or glycated hemoglobin ≥6.5%. Detailed information on the biochemical analyses is available elsewhere [15].

All statistical analyses were conducted using SAS software (version 9.4; SAS Institute, Cary, North Carolina, United States). Participant characteristics are presented as descriptive statistics. Continuous variables are presented as mean ± standard deviation, whereas categorical variables are expressed as absolute frequencies and percentages (%). The chi-square test was used to compare educational levels, drinking and smoking habits, RT regularity, and the prevalence of low muscle mass and non-communicable diseases (e.g., hypertension and diabetes mellitus) between the groups. Independent t-tests were used to compare age, PA-time, BMI, WC, fat-free mass, FFMI, SBP, DBP, T-Chol, HDL-C, TG, and FBG levels between groups.

A multiple logistic regression model was used to evaluate odds ratios (ORs) and 95% confidence intervals (CIs) for the prevalence of low muscle mass. The models were adjusted for age, sex, drinking, smoking, educational level, BMI, hypertension, and diabetes mellitus. Subgroup analyses were conducted to examine the relationship between RT levels and the risk of low muscle mass, taking into account age (<65 and ≥65 years), sex (male and female), educational level (≤middle school and ≥high school), current drinking habits (no and yes), smoking status (never and ever), BMI (<25 and ≥25 kg/m2), hypertension (no and yes), and diabetes mellitus (no and yes). The p-value for the interaction was estimated to assess the consistency of the associations across the subgroups. All tests were two-tailed, and statistical significance was set at a p-value < 0.05.