Abstract
Background
Sarcopenia and diabetes are major public health burdens. Body mass index (BMI) represents a complex factor in diabetes-sarcopenia relationship, implicated in both low muscle mass and the high-risk phenotype of sarcopenic obesity, but its role as a statistical mediator in this association remains unclear in population-based studies.
Methods
This cross-sectional study included 7,037 U.S. adults aged 20–59 years from National Health and Nutrition Examination Survey (NHANES) 2011–2018. Sarcopenia was primarily defined by low appendicular skeletal muscle mass index (ASM/height2; sex-specific cut-offs), with BMI-adjusted low muscle mass (ASM/BMI) as a secondary definition. Survey-weighted multivariable logistic regression and non-parametric bootstrap formal mediation analyses were performed, adjusting for demographic, socioeconomic, lifestyle and comorbidity covariates.
Results
Using the ASM/height2 definition, diabetes was inversely associated with sarcopenia in models without BMI (Model 2 OR = 0.60, 95% CI 0.47–0.77), whereas the association reversed after additional adjustment for BMI (Model 3 OR = 1.84, 95% CI 1.42–2.39). Higher BMI was strongly associated with lower odds of sarcopenia (OR = 0.49 per kg/m2, 95% CI 0.46–0.53). Mediation analysis suggested inconsistent (competitive) mediation: diabetes was associated with higher BMI (OR = 1.13, 95% CI 1.09–1.17), producing a negative indirect effect via BMI that offset the positive direct effect and yielded an inverse total effect. In contrast, using the ASM/BMI definition, higher BMI was positively associated with sarcopenia (OR = 1.13, 95% CI 1.11–1.15) and explained approximately two-thirds of the diabetes-sarcopenia association (proportion mediated = 66.47%).
Conclusion
The association between diabetes and sarcopenia differed by the operational definition of low muscle mass. BMI showed definition-dependent mediation: under ASM/height2, BMI acted as a suppressor (competitive) pathway that offset the positive direct association between diabetes and sarcopenia, whereas under ASM/BMI, BMI was a positive mediator accounting for about two-thirds of the association. These findings underscore the importance of specifying diagnostic criteria and body composition when interpreting relationships between diabetes and sarcopenia. Longitudinal studies with functional measures are needed to support causal inference.
Graphical Abstract
Supplementary Information
The online version contains supplementary material available at 10.1186/s12889-026-26303-w.
Keywords: National Health and Nutrition Examination Survey (NHANES), Diabetes, Body mass index, Sarcopenia, Mediation
Introduction
Sarcopenia, the age-related loss of muscle mass and function, is a prevalent condition with substantial health and economic costs, including increased disability, mortality, and healthcare expenditures [1–3]. Among its risk factors, diabetes, a kind of chronic metabolic disease with status of persistent hyperglycemia caused by unusual glucose metabolism, has been consistently implicated [4, 5]. This association is biologically plausible, as the metabolic dysregulation in diabetes such as insulin resistance and hyperglycemia can disrupt the balance of protein metabolism, adversely affecting muscle mass and function [6] and is supported by epidemiological evidence showing higher sarcopenia prevalence and risk in individuals with diabetes [7–9]. The established associations between these diseases, combined with the ongoing escalation of the global diabetes pandemic and its complications such as increased risk of cancer, cognitive impairment and affective disorders [10, 11], constitute a dual public health challenge, highlighting their upward trajectory.
Given the established associations of BMI with both diabetes and sarcopenia, BMI is likely to be a key intermediate factor linking these two conditions. Previous work using network and ROC analyses identified BMI as strong predictors of diabetes [12, 13]. While for sarcopenia, low BMI is a known risk factor and high BMI is associated controversially, partly due to the different diagnostic criteria for sarcopenia and the phenomenon of sarcopenic obesity, where obesity (high BMI) coexists with low muscle mass and function, creating a phenotype with high risk [14, 15]. Given that higher BMI is a well-established risk factor for diabetes and has also been linked to muscle loss and physical impairment, it is important to understand whether adiposity statistically mediates the association between diabetes and sarcopenia.
Despite extensive research on diabetes and sarcopenia, and growing interest in sarcopenic obesity, critical research gaps remain. No study has employed a formal mediation analysis framework in a nationally representative sample to quantitatively assess the degree of mediation by BMI in the diabetes-sarcopenia association. Most existing work remains at the level of establishing simple associations, lacking a quantitative exploration of the potential pathway of BMI. Therefore, this study aims to investigate the association between diabetes and sarcopenia using nationally representative data from the NHANES 2011–2018, with a specific focus on evaluating the mediating role of BMI. We hypothesize that diabetes is associated with sarcopenia in U.S. adults and that BMI may statistically mediate this association (with the direction potentially varying by the operational definition of low muscle mass). By elucidating the mediating pathway through BMI, this research may provide novel insights into the mechanistic underpinnings of sarcopenia and inform the development of targeted preventive interventions.
Materials and methods
Study design and participants
NHANES is an ongoing survey conducted by the National Center for Health Statistics (NCHS), which operates under the Centers for Disease Control and Prevention (CDC). It aims to collect nationally representative data from the US civilian, noninstitutionalized population. This program was conducted with approval from the NCHS Ethics Review Committee and all participants provided written informed consent.
The data in our study were derived from NHANES 2011–2018, which included 39,156 participants. Whole-body DXA examinations were administered only to participants aged 8–59 years in these cycles, therefore, we restricted our analyses to adults aged 20–59 years to align with adult sarcopenia cut-offs. We then excluded participants with missing data on diabetes status, BMI, DXA-derived appendicular lean mass (required to define sarcopenia) and prespecified covariates. Pregnant women are ineligible for DXA, so no pregnant participants remained after applying the above exclusions. Ultimately, 7,037 participants comprised the final analytic sample (Fig. 1).
Fig. 1.
Study population inclusion process
Exposure, mediator and outcome
The exposure, mediator, and outcome in this mediation analysis were diabetes status, BMI, and sarcopenia, respectively. NHANES does not consistently distinguish type 1 from type 2 diabetes, therefore our definition of diabetes included all types of diabetes based on self-reported physician diagnosis and biomarker criteria. Diabetes was defined as meeting at least one of the following criteria: (1) a prior physician diagnosis of diabetes; (2) current use of insulin or oral hypoglycemic agents; or (3) fulfillment of any one of the following laboratory parameters: hemoglobin A1c ≥ 6.5%, fasting plasma glucose ≥ 126 mg/dL (7.0 mmol/L), or 2-h postprandial plasma glucose ≥ 200 mg/dL (11.1 mmol/L). BMI was calculated using weight in kilograms divided by height in meters squared. In NHANES, the appendicular skeletal muscle mass (ASM), representing the aggregate lean mass in the arms and legs, was measured by DXA. Sarcopenia was operationalized as low appendicular skeletal muscle mass indexed to height squared (ASM/height2; < 7.26 kg/m2 in men and < 5.5 kg/m2 in women) [16]. Additionally, as a secondary outcome for supplementary analyses, BMI-adjusted low muscle mass (ASM/BMI; < 0.789 in men and < 0.512 in women) was used [17].
Assessment of covariates
Covariates were preselected based on established risk factors for diabetes and sarcopenia [4, 18–20]. Demographic characteristics and health-related parameters were systematically collected according to previous studies. The analysis included potential confounders such as age, gender, race (including Mexican American, non-Hispanic black, non-Hispanic white, other Hispanic, and others race), education (including Less than high school, High school or equivalent and College or above) and marital status (including Married, Widowed, Divorced, Separated, Never married and Living with partner). Poverty income ratio (PIR) was calculated as the ratio of family income to the federal poverty threshold for the survey year, as defined by the U.S. Census Bureau, and categorized as < 1.30, 1.30–3.49, and ≥ 3.50.
Chronic disease status was defined based on a combination of self-reported diagnoses and objective measurements. Hypertension was defined as meeting at least one of the following criteria: (1) self-reported prior diagnosis by a physician; or (2) systolic blood pressure (SBP) ≥ 140 mmHg or diastolic blood pressure (DBP) ≥ 90 mmHg. Chronic kidney disease (CKD) was defined as either an estimated glomerular filtration rate (eGFR) < 60 mL/min/1.73 m2 or a urinary albumin-to-creatinine ratio (UACR) ≥ 30 mg/g. Cardiovascular disease (CVD) was considered if the participant answered “yes” to any of the following questions: “Has a doctor or other health expert ever told you that you have congestive heart failure/coronary heart disease/angina/myocardial infarct/stroke?”.
Health-related behaviors were assessed using standardized questionnaires. Smoking status was self-reported and categorized as never (< 100 lifetime cigarettes), former (≥ 100 cigarettes, currently quit), or now (≥ 100 cigarettes, currently smoking). Drinking status was self-reported and categorized based on the intake of at least 12 drinks during the preceding year. Physical activity level was derived from the physical activity questionnaire. Weekly metabolic equivalent (MET) minutes were calculated based on the type, frequency, duration, and intensity of reported activities. Participants were then categorized as having low (< 500 MET-minutes/week) or high (≥ 500 MET-minutes/week) physical activity according to national guideline thresholds. The complete variable definitions are available in https://wwwn.cdc.gov/nchs/nhanes/Default.aspx.
Statistical analysis
All analyses accounted for the NHANES complex, multistage probability sampling design, including examination sample weights, strata, and primary sampling units (PSUs). For pooled analyses across 2011–2018 (four 2-year cycles), we constructed 8-year weights by dividing the 2-year MEC examination weights (WTMEC2YR) by four, following NHANES analytic guidelines. Continuous variables are presented as survey-weighted means (standard errors, SE), and categorical variables as survey-weighted percentages. Group differences were assessed using survey-weighted linear regression (Wald tests) for continuous variables and Rao–Scott χ2 tests for categorical variables.
We used survey-weighted multivariable logistic regression to examine associations of diabetes and BMI with sarcopenia, reporting odds ratios (ORs) and 95% confidence intervals (CIs). Model 1 was unadjusted. Model 2 adjusted for age, gender, race, marital status, education, PIR, smoking, drinking, physical activity, and comorbidities (hypertension, CKD, and CVD). Model 3 additionally adjusted for BMI when diabetes was the exposure, and for diabetes when BMI was the exposure. Prespecified subgroup analyses were conducted for the diabetes–sarcopenia association across key covariates.
We conducted mediation analyses to evaluate the potential mediating role of BMI in the diabetes–sarcopenia association. Using the R package “mediation”, we fitted the mediator and outcome models incorporating survey weights as probability weights and estimated the average causal mediation effect (ACME), average direct effect (ADE), total effect, and proportion mediated via non-parametric bootstrap with 1,000 replications. Given the cross-sectional design and the strong assumptions required for causal mediation, these results are presented as statistical (associational) mediation patterns rather than definitive causal effects. All tests were two-sided with P < 0.05. Analyses were performed in R (version 4.3.2).
Results
Characteristics of participants
In terms of inclusion criteria, a total of 7037 subjects were selected from 39156 participants for this analysis. The baseline features of those subjects are detailed in Table 1. Participants were categorized into two groups: sarcopenia (769, 10.93%) and non-sarcopenia (6268, 89.07%). Statistically significant differences were observed between the sarcopenia and non-sarcopenia groups with respect to age, BMI, marital status, race, smoking, physical activity, hypertension and diabetes (P < 0.05) where individuals with sarcopenia had lower BMI. But when using ASM/BMI definition, we could find that BMI in sarcopenia group was higher than that in non-sarcopenia. Detailed data are shown in Table S1.
Table 1.
Characteristics of participants from the NHANES (2011–2018)
| Variable | Total (n = 7037) | Non-sarcopenia (n = 6268) | Sarcopenia (n = 769) | t/χ2 | p |
|---|---|---|---|---|---|
| Age, Mean (SE) | 39.44 (0.30) | 39.67 (0.30) | 37.46 (0.70) | −3.51 | 0.001 |
| BMI, Mean (SE) | 28.75 (0.16) | 29.62 (0.14) | 21.32 (0.13) | −49.91 | <.001 |
| Marital status, n(%) | 73.95 | <.001 | |||
| Married | 3421 (52.41) | 3086 (53.69) | 335 (41.52) | ||
| Widowed | 92 (1.21) | 78 (1.06) | 14 (2.48) | ||
| Divorced | 646 (9.38) | 584 (9.53) | 62 (8.15) | ||
| Separated | 233 (2.45) | 213 (2.57) | 20 (1.37) | ||
| Never married | 1869 (24.18) | 1608 (22.95) | 261 (34.58) | ||
| Living with partner | 776 (10.37) | 699 (10.19) | 77 (11.90) | ||
| Gender, n(%) | 1.45 | 0.239 | |||
| Male | 3558 (50.78) | 3184 (51.02) | 374 (48.68) | ||
| Female | 3479 (49.22) | 3084 (48.98) | 395 (51.32) | ||
| Race, n(%) | 104.68 | <.001 | |||
| Mexican American | 1006 (9.70) | 934 (10.05) | 72 (6.71) | ||
| Other Hispanic | 713 (6.58) | 648 (6.67) | 65 (5.77) | ||
| Non-Hispanic White | 2675 (64.44) | 2368 (64.07) | 307 (67.64) | ||
| Non-Hispanic Black | 1465 (10.91) | 1413 (11.74) | 52 (3.89) | ||
| Other Race | 1178 (8.37) | 905 (7.47) | 273 (15.99) | ||
| Smoking, n(%) | 34.75 | <.001 | |||
| Never | 4209 (58.40) | 3743 (59.04) | 466 (52.96) | ||
| Former | 1204 (19.36) | 1098 (19.71) | 106 (16.34) | ||
| Now | 1624 (22.24) | 1427 (21.25) | 197 (30.71) | ||
| Education, n(%) | 3.75 | 0.252 | |||
| Less than high school | 1196 (12.45) | 1083 (12.44) | 113 (12.53) | ||
| High school or equivalent | 1524 (20.71) | 1367 (20.39) | 157 (23.36) | ||
| College or above | 4317 (66.85) | 3818 (67.17) | 499 (64.11) | ||
| Drinking, n(%) | 3.07 | 0.095 | |||
| No | 1699 (18.52) | 1468 (18.24) | 231 (20.89) | ||
| Yes | 5338 (81.48) | 4800 (81.76) | 538 (79.11) | ||
| Physical activity, n(%) | 35.36 | <.001 | |||
| Low | 2149 (28.19) | 1849 (27.10) | 300 (37.50) | ||
| High | 4888 (71.81) | 4419 (72.90) | 469 (62.50) | ||
| PIR, n(%) | 7.58 | 0.212 | |||
| < 1.30 | 2299 (23.21) | 2034 (22.75) | 265 (27.14) | ||
| 1.30–3.49 | 2486 (34.27) | 2242 (34.37) | 244 (33.48) | ||
| ≥ 3.50 | 2252 (42.52) | 1992 (42.89) | 260 (39.38) | ||
| Hypertension, n(%) | 40.20 | <.001 | |||
| No | 5088 (73.27) | 4447 (72.12) | 641 (83.03) | ||
| Yes | 1949 (26.73) | 1821 (27.88) | 128 (16.97) | ||
| Diabetes, n(%) | 15.78 | <.001 | |||
| No | 6228 (90.76) | 5506 (90.29) | 722 (94.76) | ||
| Yes | 809 (9.24) | 762 (9.71) | 47 (5.24) | ||
| CKD, n(%) | 0.00 | 0.969 | |||
| No | 6390 (92.17) | 5681 (92.17) | 709 (92.13) | ||
| Yes | 647 (7.83) | 587 (7.83) | 60 (7.87) | ||
| CVD, n(%) | 1.09 | 0.356 | |||
| No | 6782 (96.94) | 6036 (96.86) | 746 (97.56) | ||
| Yes | 255 (3.06) | 232 (3.14) | 23 (2.44) |
SE Standard Error, t Wald tests, χ2 survey-weighted tests, CKD Chronic kidney disease, CVD Cardiovascular diseases, PIR Poverty income ratio, Sarcopenia is defined by ASM/height2
Logistic regression analysis between diabetes, BMI and sarcopenia
The results of the survey-weighted multivariate logistic regression analyses for the primary and alternative sarcopenia definitions are presented in Table 2 and Supplementary Table S2, respectively.
Table 2.
Logistic regression models of relationship between diabetes, BMI and sarcopenia
| Variables | Model 1 | Model 2 | Model 3 | |||
|---|---|---|---|---|---|---|
| OR (95%CI) | p | OR (95%CI) | p | OR (95%CI) | p | |
| Diabetes | 0.51 (0.35 ~ 0.76) | 0.002 | 0.60 (0.38 ~ 0.93) | 0.033 | 1.84 (1.10 ~ 3.08) | 0.029 |
| BMI | 0.55 (0.52 ~ 0.58) | <.001 | 0.49 (0.45 ~ 0.53) | <.001 | 0.49 (0.45 ~ 0.53) | <.001 |
Model 1: Crude;
Model 2: Adjust for age, gender, race, marital status, education, PIR, smoking, drinking, physical activity, Hypertension, CKD and CVD
Model 3: Additionally adjusted for BMI when diabetes was the exposure, and for diabetes when BMI was the exposure
OR Odds Ratio, CI Confidence Interval, Sarcopenia is defined by ASM/height2, CKD Chronic kidney disease, CVD Cardiovascular diseases, PIR Poverty income ratio
In the primary analysis, prior to adjustment for BMI, diabetes showed a significant protective association with sarcopenia. In the unadjusted Model 1, patients with diabetes had significantly lower odds of sarcopenia compared to non-diabetics (OR = 0.51, 95% CI: 0.35 ~ 0.76, P = 0.002). This protective association, though attenuated, remained significant after adjustment for demographic, socioeconomic, behavioral, and clinical covariates in Model 2 (OR = 0.60, 95% CI: 0.38 ~ 0.93, P = 0.033). Notably, after additional adjustment for BMI in Model 3, the direction of association reversed, with diabetes becoming a significant risk factor for sarcopenia (OR = 1.84, 95% CI: 1.10 ~ 3.08, P = 0.029). Conversely, a higher BMI was consistently and strongly associated with lower odds of sarcopenia across all models. In the crude model, each unit increase in BMI was associated with an OR of 0.55 (95% CI: 0.52 ~ 0.58, P < 0.001). This significant negative association was slightly strengthened and remained stable in both Model 2 (OR = 0.49, 95% CI: 0.45 ~ 0.53, P < 0.001) and Model 3 (OR = 0.49, 95% CI: 0.45 ~ 0.53, P < 0.001), indicating a robust independent protective effect of BMI against sarcopenia.
When sarcopenia was alternatively defined using ASM/BMI, the relationship patterns differed. Diabetes was a strong and significant risk factor for sarcopenia in the unadjusted and partial adjusted models (Model 1: OR = 3.28, P < 0.001; Model 2: OR = 1.91, P = 0.002) and was attenuated to non-significance after additional BMI adjustment (Model 3: OR = 1.19, P = 0.367). Higher BMI was consistently associated with significantly higher odds of sarcopenia in all models (OR = 1.13, P < 0.001).
Subgroup analyses
To explore potential effect modification, we conducted subgroup analyses based on the fully adjusted model. The results are shown in Fig. 2. The test for interaction was not statistically significant across most predefined subgroups, including race (P for interaction = 0.197), education level (P = 0.378), drinking status (P = 0.917), physical activity (P = 0.638), PIR (P = 0.742), and comorbid conditions such as hypertension (P = 0.808), CKD (P = 0.655), and CVD (P = 0.844). This suggests that the association between diabetes and sarcopenia was generally consistent across these strata.
Fig. 2.
Forest plot illustrating the result of subgroup analysis of the association between diabetes and sarcopenia
However, there were suggestive trends of effect modification by gender and smoking status, though the interaction terms did not reach conventional statistical significance (P for interaction = 0.061 and 0.066, respectively). Visually, the point estimates indicated a stronger association in women (OR = 2.55, 95% CI: 1.35 ~ 4.83) than in men (OR = 1.56, 95% CI: 0.79 ~ 3.09), and in never smokers (OR = 2.48, 95% CI: 1.28 ~ 4.83) compared to former smokers. Some strata showed imprecise point estimates with CIs not crossing 1. However, interaction tests were not statistically significant and subgroup findings should be interpreted as exploratory.
Mediation analysis of BMI
The mediation analysis of the primary and secondary outcome are presented in Fig. 3 and Supplementary Table S3, respectively. The exposure, mediator, and outcome in this mediation analysis were diabetes status, BMI, and sarcopenia, respectively. Both mediation analyses were conducted after adjusting for marital status, gender, race, smoking, education, drinking, physical activity, hypertension, CKD, CVD, age and PIR.
Fig. 3.
The mediation effect of BMI on the relationship between diabetes and sarcopenia
The result of primary analysis revealed a competitive mediation effect, which mean that the direct risk effect of diabetes on sarcopenia was offset by a stronger and opposing indirect protective effect mediated by higher BMI. As presented in Fig. 3, the total effect of diabetes on sarcopenia was significant (β = −0.08, P < 0.001). This overall association was composed of a significant direct effect (β = 0.03, P = 0.002) and a significant indirect effect through BMI (β = −0.11, P < 0.001). The detailed path coefficients for this model are shown in Supplementary Table S4. When sarcopenia was defined by ASM/BMI in a secondary analysis, BMI demonstrated major mediation in the association between diabetes and sarcopenia, accounting for approximately 66% of the total effect. Detailed results are provided in Supplementary Table S3.
Discussion
This study investigated the association between diabetes and sarcopenia, and quantitatively assessed the mediating role of BMI using a nationally representative sample of U.S. adults. Our result reveals a complex relationship which is dependent on the operational definition of sarcopenia. When using the ASM/height2 definition, we observed a robust and protective association between higher BMI and sarcopenia. Mediation analysis further showed a competitive mediation pattern, which mean the direct and detrimental effect of diabetes on sarcopenia was offset by a stronger and opposing indirect effect mediated through higher BMI, resulting in a net protective total effect. In contrast, when using the ASM/BMI definition, which inherently adjusts muscle mass for adiposity, BMI was a consistent risk factor and partially mediated the association between diabetes and sarcopenia. These divergent results underscore BMI as a pivotal and definition-dependent factor in the diabetes-sarcopenia pathway.
Epidemiologic studies from Asian and Western populations have reported positive associations between diabetes and sarcopenia, with odds ratios ranging from approximately 1.1 to 2.7 [7, 8, 21, 22]. However, these effect estimates vary considerably across studies. This heterogeneity may stem from differences in study populations, such as community-dwelling individuals versus clinical cohorts, or middle-aged versus elderly populations. It could also arise from varying definitions of sarcopenia and the extent of adjustment for confounding factors like obesity. After full adjustment in our primary analysis, we found a positive association between diabetes and sarcopenia, which is consistent with the main research direction in the existing literature. However, in unadjusted or partially adjusted models, the strength and even the direction of association vary significantly across studies. We observed the significantly protective association before adjusting for BMI (OR = 0.51), while less commonly reported, underscores a critical methodological insight. This difference may not only be due to differences in population characteristics or diagnostic criteria, but also to the degree of body composition adjustment. Our analysis shows that the observed epidemiological association can be completely reversed if BMI is not considered. This occurs because BMI exhibits a significant negative confounding effect where it is positively associated with diabetes, but negatively associated with the sarcopenia according to the ASM/height2 definition. Under this definition, individuals with low BMI are more likely to be classified as having low muscle mass, resulting in a strong negative correlation between BMI and outcome. This is a "definition mechanism" rather than a purely biological protective effect. Consequently, its unadjusted effect masks the underlying metabolic risk of diabetes, creating a spurious protective effect. This reveals the primary source of heterogeneity in previous studies and suggests that rigorous adjustment for obesity severity is not only recommended but also a necessary condition to avoid fundamentally misleading conclusions in sarcopenia etiology research. In our secondary analyses using BMI-adjusted muscle mass (ASM/BMI), participants classified as sarcopenic exhibited substantially higher BMI, consistent with a sarcopenic-obesity phenotype rather than generalized wasting. This definition-dependent pattern helps explain why the diabetes–sarcopenia association can differ across operational definitions and underscores the need to interpret findings in the context of the chosen muscle-mass index.
These conflicting patterns require a dual-pathway physiological hypothesis to link diabetes to muscle health. According to the ASM/height2 definition, this direct positive effect is consistent with the known pathophysiological mechanisms of diabetes, namely that insulin resistance, chronic inflammation, and metabolic dysregulation directly impair muscle protein synthesis through the mTOR pathway and promote its degradation via pathways such as the ubiquitin–proteasome system [6, 23–26]. In this definition, higher BMI is typically associated with an overall increase in mass, which may be related to higher absolute lean body mass, thereby diluting the detection of true sarcopenia. This may also represent a phenotype with greater metabolic reserve, temporarily offsetting the effects of muscle loss. This pathological mechanism highlights the diagnostic limitations of adjusting mass solely based on body size, as it fails to distinguish muscle mass and fat content. Conversely, under the ASM/BMI definition, BMI, primarily referring to its fat component, is the central pathological vehicle. Diabetes promotes central obesity and ectopic fat deposition [27], subsequently leading to adipose tissue dysfunction, lipotoxicity, and chronic inflammation [28]. These factors directly contribute to muscle atrophy and mass loss. The ASM/BMI definition, by adjusting for BMI, essentially captures this core adiposity-mediated pathophysiology, making BMI a major statistical mediator. It identifies individuals with significantly low muscle mass and high fat content, a hallmark of sarcopenic obesity. Together, these pathways are not mutually exclusive but coexist. The innovation of this study not only lies in formally quantifying the extent to which BMI statistically mediates the association between diabetes and sarcopenia but also to quantify how their relative statistical dominance is switched by the researcher’s definitional choice, reflecting different underlying physiological emphases.
Subgroup analyses based on primary outcome demonstrated that the association between diabetes and sarcopenia remained consistent across most demographic and clinical strata, suggesting the robust core associations. For smoking, we did not observe a significant difference in smoking status between participants with and without sarcopenia, which is different from previous studies [29, 30]. This may reflect the relatively young age distribution of our sample (20–59 years), the complex interplay between smoking, BMI, and physical activity, and potential residual confounding. It is also possible that the effects of smoking on muscle health become more apparent at older ages or with longer exposure. Our findings should thus be viewed as inconclusive with respect to smoking and sarcopenia in this age group. Although suggestive, the trends toward enhanced effects observed in women and never-smokers were not statistically significant and warrant further validation in larger studies.
It is noted that methodological details require attention. When using ASM/BMI to define sarcopenia, the mediator (BMI) is mathematically embedded in the outcome, thereby creating inherent endogeneity. This structural correlation may overestimate the estimated mediation effect, as the path from BMI to ASM/BMI reflects a formulaic inverse relationship, while any biological effects coexist. Thus, the observed major mediation primarily indicates a strong statistical association. This caution underscores the greater methodological robustness of our primary analysis using ASM/height2 definition, which avoids circular reasoning and reveals a more biologically plausible competitive mediation mechanism. However, BMI shares the same denominator as ASM/height2, and this mathematical structure may also exaggerate the relationship between BMI and sarcopenia. Future research should employ muscle function indicators independent of BMI, such as grip strength and walking speed, or directly assess muscle mass and adipose infiltration through imaging, to validate these pathways within a more robust methodological framework. Moreover, mediation analyses incorporated sampling weights as probability weights. However, the approach does not fully account for NHANES stratification and clustering, therefore variance estimates may be conservative or biased.
Several limitations must be acknowledged. First, the cross-sectional design cannot establish causal relationships between the temporal sequence of diabetes, BMI changes, and the onset of sarcopenia. Reverse causation and residual confounding may also contribute to the observed associations. Longitudinal studies are essential to confirm the directionality of these associations. Second, whole-body DXA was administered only to participants aged 8–59 years in NHANES 2011–2018, therefore, our analytic sample was restricted to adults aged 20–59 years. As shown in Supplementary Table S5, excluded participants (primarily aged 60 years or older) were older and had a higher burden of chronic conditions, which limits generalizability to older adults. In addition, for remaining missing covariate data within the eligible age range, we used a complete-case approach which may introduce selection bias if missingness was related to exposure, outcome, or covariates. Third, while DXA-derived muscle mass is objective, it fails to fully reflect either muscle mass or function, both of which are critical diagnostic criteria for sarcopenia in clinical guidelines. Although sarcopenia is typically framed as age-related, our operational definition captures low muscle mass in middle-aged adults and may not fully represent clinically confirmed sarcopenia requiring functional criteria. Finally, the final sample lacked data from individuals aged 60 years and above, which is the most susceptible population to sarcopenia, limiting the generalizability of the study findings to older adults. Future research could address these limitations through a coherent agenda based on our framework. First, longitudinal studies containing the same and older cohorts are required to establish temporality and validate the definition-dependent mediation dynamics that vary over time. Second, phenotype-specific studies in multicenter cohorts can apply both definitions simultaneously to compare the incidence, outcomes, and metabolic profiles. Third, mechanistic validation through direct measurements of intramuscular fat, systemic inflammation and adipose function would biologically confirm the proposed dual-pathway model. Finally, precision trials based on individualized phenotypic testing for sarcopenia to customize lifestyle or pharmacological intervention strategies can translate these findings into more effective and personalized preventive measures.
Our findings carry important practical implications. For clinical diagnostic practice, these findings necessitate that clinical guidelines and research explicitly define the operational definition of sarcopenia and recognize that ASM/height2 and ASM/BMI may identify different pathophysiological phenotypes. Moreover, our findings support a more precise and phenotype-based patient management approaches. For individuals identified using the ASM/height2 definition, who may present with pure low muscle mass, the clinical focus should shift toward body composition remodeling. This entails implementing combined interventions, such as resistance training and adequate protein intake, aimed at preserving or increasing lean mass while cautiously managing adiposity, rather than pursuing weight loss simply. In contrast, for individuals identified by the ASM/BMI definition, a profile closely aligned with sarcopenic obesity, the primary treatment goal should be substantial reduction in fat mass. This strategy should not only include conventional glycemic control but also incorporate targeted weight management measures, as reducing fat mass may be pivotal in addressing the relative muscle deficiency reflected by this metric. In general, management strategies should be differentiated according to the diagnostic criteria used, and a gradual shift should be made towards more targeted and mechanism-based interventions.
Conclusion
This study demonstrates that the relationship between diabetes and sarcopenia is fundamentally shaped by the operational definition of sarcopenia, with BMI serving as a pivotal, definition-dependent mediator. When sarcopenia is defined by ASM/height2, higher BMI exerts a protective indirect effect, masking the direct detrimental impact of diabetes and leading to an apparent protective association prior to adjustment. Conversely, when using the ASM/BMI definition, higher BMI acts as a core pathological mediator, significantly explaining the positive association between diabetes and sarcopenia. These differences indicate that the two definitions may capture distinct clinical phenotypes, with low muscle mass and sarcopenic obesity each involving different physiological pathways. Consequently, rigorous adjustment for adiposity is essential in etiological research to avoid biased conclusions. Clinically, these insights advocate for a phenotype-informed approach, which involves guiding targeted management strategies through diagnostic criteria, prioritizing either lean body mass preservation or obesity reduction. Future longitudinal and mechanistic studies are also required to validate these pathways and translate definitional precision into effective and personalized interventions.
Supplementary Information
Acknowledgements
We would like to express our sincere thanks to all participants in the NHANES survey.
Abbreviations
- NHANES
National Health and Nutrition Examination Survey
- BMI
Body mass index
- PIR
Poverty income ratio
- CKD
Chronic kidney disease
- CVD
Cardiovascular diseases
- CI
Confidence Interval
- SE
Standard Error
- OR
Odds Ratio
- χ2
Chi-square test
Authors’ contributions
Conceptualization, Z.L. and L.X.; methodology, Z.L. and Y.G.; software, Z.L. and S.C.; validation, Z.L., J.W. and L.D.; formal analysis, Z.L.; data curation, L.X.; writing—original draft preparation, Z.L. and L.X.; writing—review and editing, Z.L., H.Z and S.C.; visualization, L.X. and H.Z.; supervision, S.C and Z.Z.; project administration, M.M. and Z.Z. All authors have read and agreed to the published version of the manuscript.
Funding
This research was supported by the 2024 Shantou Science and Technology Plan Medical and Health Project (Grant No: 240506186498668, 240922169602002), the 2025 Shantou Science and Technology Plan Medical and Health Project (Grant No: 250928209618515).
Data availability
This study used publicly available data from the National Health and Nutrition Examination Survey (NHANES). The datasets analyzed are available from the NHANES website (https://www.cdc.gov/nchs/nhanes/) following the survey’s data-use guidelines.
Declarations
Ethics approval and consent to participate
The NCHS Research Ethics Review Committee and the NHANES both examined and approved the research that involved human subjects. To take part in this study, the participants gave their written informed consent. This study utilized de-identified, publicly available NHANES secondary data. All personally identifiable information (including names, identification numbers, and other direct identifiers) was removed through standardized de-identification procedures in compliance with ethical requirements for human subject data protection. In accordance with prevailing research ethics guidelines, secondary analysis of such pre-approved, de-identified public datasets qualifies for exemption from additional institutional review, a standard this study rigorously followed.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Zhikai Li and Lei Xie contributed equally to this work.
Contributor Information
Zhigang Zhong, Email: stzzg@163.com.
Shangmin Chen, Email: 18smchen@stu.edu.cn.
References
- 1.Sayer AA, Cooper R, Arai H, et al. Sarcopenia. Nat Rev Dis Primers. 2024;10(1):68. [DOI] [PubMed] [Google Scholar]
- 2.Kirk B, Cawthon P M, Arai H, et al. The conceptual definition of sarcopenia: delphi consensus from the Global Leadership Initiative in Sarcopenia (GLIS). Age Ageing. 2024;53(3). [DOI] [PMC free article] [PubMed]
- 3.Gong H, Duan S, Lin X, et al. The association between dietary index for gut microbiota and sarcopenia: the mediating role of dietary inflammatory index. Front Nutr. 2025;12:1514209. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Yuan S, Larsson SC. Epidemiology of sarcopenia: prevalence, risk factors, and consequences. Metabolism. 2023;144:155533. [DOI] [PubMed] [Google Scholar]
- 5.Liu M, Yu D, Pan Y, et al. Causal roles of lifestyle, psychosocial characteristics, and sleep status in sarcopenia: a Mendelian randomization study. J Gerontol A Biol Sci Med Sci. 2024;79(1). [DOI] [PubMed]
- 6.Lalla E, Papapanou PN. Diabetes mellitus and periodontitis: a tale of two common interrelated diseases. Nat Rev Endocrinol. 2011;7(12):738–48. [DOI] [PubMed] [Google Scholar]
- 7.Qiao YS, Chai YH, Gong HJ, et al. The association between diabetes mellitus and risk of sarcopenia: accumulated evidences from observational studies. Front Endocrinol (Lausanne). 2021;12:782391. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Chung SM, Moon JS, Chang MC. Prevalence of sarcopenia and its association with diabetes: a meta-analysis of community-dwelling Asian population. Front Med Lausanne. 2021;8:681232. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Feng L, Gao Q, Hu K, et al. Prevalence and risk factors of sarcopenia in patients with diabetes: a meta-analysis. J Clin Endocrinol Metab. 2022;107(5):1470–83. [DOI] [PubMed] [Google Scholar]
- 10.Worldwide trends in diabetes prevalence and treatment from 1990 to 2022: a pooled analysis of 1108 population-representative studies with 141 million participants. Lancet, 2024, 404(10467): 2077-93. [DOI] [PMC free article] [PubMed]
- 11.Tomic D, Shaw JE, Magliano DJ. The burden and risks of emerging complications of diabetes mellitus. Nat Rev Endocrinol. 2022;18(9):525–39. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Kowsar R, Mansouri A. Multi-level analysis reveals the association between diabetes, body mass index, and HbA1c in an Iraqi population. Sci Rep. 2022;12(1):21135. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Chandrasekaran P, Weiskirchen R. The role of obesity in type 2 diabetes mellitus-an overview. Int J Mol Sci. 2024;25(3). [DOI] [PMC free article] [PubMed]
- 14.Curtis M, Swan L, Fox R, et al. Associations between body mass index and probable sarcopenia in community-dwelling older adults. Nutrients. 2023;15(6). [DOI] [PMC free article] [PubMed]
- 15.Merchant RA, Seetharaman S, Au L, et al. Relationship of fat mass index and fat free mass index with body mass index and association with function, cognition and sarcopenia in pre-frail older adults. Front Endocrinol (Lausanne). 2021;12:765415. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Cawthon PM. Recent progress in sarcopenia research: a focus on operationalizing a definition of sarcopenia. Curr Osteoporos Rep. 2018;16(6):730–7. [DOI] [PubMed] [Google Scholar]
- 17.Chen LK, Woo J, Assantachai P, et al. Asian Working Group for Sarcopenia: 2019 Consensus Update on Sarcopenia Diagnosis and Treatment. J Am Med Dir Assoc. 2020;21(3):300-7.e2. [DOI] [PubMed] [Google Scholar]
- 18.Weber MB, Oza-Frank R, Staimez LR, et al. Type 2 diabetes in Asians: prevalence, risk factors, and effectiveness of behavioral intervention at individual and population levels. Annu Rev Nutr. 2012;32:417–39. [DOI] [PubMed] [Google Scholar]
- 19.Duarte MP, Almeida LS, Neri SGR, et al. Prevalence of sarcopenia in patients with chronic kidney disease: a global systematic review and meta-analysis. J Cachexia Sarcopenia Muscle. 2024;15(2):501–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Gao K, Cao LF, Ma WZ, et al. Association between sarcopenia and cardiovascular disease among middle-aged and older adults: findings from the China health and retirement longitudinal study. EClinicalMedicine. 2022;44:101264. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Wang T, Feng X, Zhou J, et al. Type 2 diabetes mellitus is associated with increased risks of sarcopenia and pre-sarcopenia in Chinese elderly. Sci Rep. 2016;6:38937. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Shi H. Unraveling the dual threat: sarcopenia and insufficient physical activity in diabetes risk. Front Endocrinol (Lausanne). 2024;15:1507657. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Russell ST, Rajani S, Dhadda RS, et al. Mechanism of induction of muscle protein loss by hyperglycaemia. Exp Cell Res. 2009;315(1):16–25. [DOI] [PubMed] [Google Scholar]
- 24.Lawrence JC Jr. mTOR-dependent control of skeletal muscle protein synthesis. Int J Sport Nutr Exerc Metab. 2001;11 Suppl:S177-85. [DOI] [PubMed] [Google Scholar]
- 25.Goossens GH, Blaak EE, Theunissen R, et al. Expression of NLRP3 inflammasome and T cell population markers in adipose tissue are associated with insulin resistance and impaired glucose metabolism in humans. Mol Immunol. 2012;50(3):142–9. [DOI] [PubMed] [Google Scholar]
- 26.Giha HA, Alamin OAO, Sater MS. Diabetic sarcopenia: metabolic and molecular appraisal. Acta Diabetol. 2022;59(8):989–1000. [DOI] [PubMed] [Google Scholar]
- 27.Buch A, Carmeli E, Boker LK, et al. Muscle function and fat content in relation to sarcopenia, obesity and frailty of old age–an overview. Exp Gerontol. 2016;76:25–32. [DOI] [PubMed] [Google Scholar]
- 28.Meex RCR, Blaak EE, Van Loon LJC. Lipotoxicity plays a key role in the development of both insulin resistance and muscle atrophy in patients with type 2 diabetes. Obes Rev. 2019;20(9):1205–17. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Lee SR, Lee SY, Park EJ, et al. Association between cumulative pack-year smoking exposure and sarcopenia: a KoGES cohort study. Int J Epidemiol. 2025;54(3). [DOI] [PubMed]
- 30.Rom O, Kaisari S, Aizenbud D, et al. Sarcopenia and smoking: a possible cellular model of cigarette smoke effects on muscle protein breakdown. Ann N Y Acad Sci. 2012;1259:47–53. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
This study used publicly available data from the National Health and Nutrition Examination Survey (NHANES). The datasets analyzed are available from the NHANES website (https://www.cdc.gov/nchs/nhanes/) following the survey’s data-use guidelines.