Association of frailty index with sarcopenia: Mediation analysis of triglyceride-glucose-related indices: a cross-sectional study

Abstract

Frailty and sarcopenia are closely related geriatric syndromes that share overlapping pathophysiological mechanisms. However, the metabolic pathways linking frailty burden to sarcopenia remain unclear. The triglyceride-glucose (TyG) index and its derivatives, which serve as surrogate markers of Insulin resistance (IR), may help explain this association. This study aimed to examine the association between frailty index (FI) and sarcopenia and evaluate the potential mediating role of TyG and its derivatives (TyG-BMI, TyG-WC, TyG-WHtR, and TyG-ABSI). This cross-sectional study included 2152 adults aged 20 to 59 years from the 2011–2014 National Health and Nutrition Examination Survey. FI was calculated using a 49-item deficit accumulation model. Sarcopenia was defined according to the for the National Institutes of Health criteria, and muscle quality index (MQI) was additionally analyzed as a continuous indicator of muscle quality. Weighted logistic and linear regression models were used to assess the associations of FI with sarcopenia and MQI. Mediation analyses were further performed to evaluate the role of TyG-related indices. Higher FI scores were associated with increased prevalence of sarcopenia and lower MQI. These associations varied by age (interaction P = .0191) and sex (interaction P = .0261), with stronger associations observed in individuals aged <40 years and in females. Mediation analyses showed that TyG and its related indices significantly mediated the association between FI and sarcopenia. Among these indicators, TyG-WHtR showed the strongest mediation effect, accounting for 28.26% of the total effect. FI was significantly associated with sarcopenia, and this relationship was partially mediated by TyG-related indices. These findings suggest that IR-related metabolic dysfunction may contribute to the observed association between FI and muscle health and provide new insight into the metabolic pathways underlying sarcopenia. Given the cross-sectional design, causal directionality cannot be inferred.

1. Introduction

Sarcopenia is a common geriatric syndrome strongly linked to aging and characterized by the progressive loss of skeletal muscle mass and a decline in muscle function.^[1] Studies have shown that sarcopenia not only significantly increases the risk of falls and fractures but is also associated with various chronic conditions, including cancer, cardiovascular disease, diabetes, respiratory illness, and neurological disorders.^[2–7] Although sarcopenia is more prevalent among older adults, recent studies have indicated that it may also occur in younger populations.^[8] Sarcopenia not only seriously affects individual health and quality of life but also poses a substantial socioeconomic burden. One study reported that hospitalized older adults with sarcopenia incurred approximately 5 times the medical expenses compared to those without the condition.^[9] A U.S.-based study estimated that hospitalizations attributable to sarcopenia resulted in a total cost of $40.4 billion, with an average medical expense of $260 per person.^[10]

Insulin resistance (IR) is a metabolic dysfunction that plays a critical role in skeletal muscle health. It impairs insulin signaling and reduces both glucose uptake and protein synthesis, ultimately leading to reduced muscle mass and functional decline.^[11,12] However, conventional methods for assessing IR, such as the hyperinsulinemic-euglycemic clamp technique, are complex and costly, limiting their applicability in large population studies.^[13] The TyG index, derived from fasting glucose and triglyceride levels, serves as a simple and practical surrogate marker for IR and has been widely used in epidemiological research.^[14] In recent years, several TyG-derived indices have been developed by combining TyG with anthropometric measures, such as BMI (TyG-BMI), WC (TyG-WC), waist-to-height ratio (TyG-WHtR), and a body shape index (TyG-ABSI), to better capture the combined burden of IR and body-fat distribution.^[15]

Aging is a complex biological process involving the decline of multiple physiological systems and is closely related to both sarcopenia and frailty. Studies have reported that the prevalence of sarcopenia ranges from 5 to 13% among individuals aged 60 to 70 years and increases to up to 50% among those over 80 years old.^[16,17] Frailty index (FI), a widely used tool for assessing the health status of older adults, has been employed in the prediction of various adverse health outcomes. Previous studies have shown that FI is not only significantly associated with sarcopenia but also strongly correlated with the TyG index and its derivative indicators.^[18,19]

However, most existing studies are limited to associational analyses, and systematic evidence on the mediating role of TyG-related indices in the relationship between FI and sarcopenia remains lacking. Although FI has often been used as a global indicator of frailty status in older adults, the deficit-accumulation approach can also be applied in population-based studies as a summary measure of cumulative health deficits across adulthood.^[20–23] In the present study, we treated FI as an indicator of overall frailty burden and examined its association with sarcopenia and muscle quality index (MQI). Given the cross-sectional nature of National Health and Nutrition Examination Survey (NHANES), our objective was not to infer temporal causality, but rather to evaluate the strength and robustness of the observed associations and to explore whether TyG-related indices might statistically mediate these relationships. Therefore, based on data from the 2011 to 2014 NHANES, this study aimed to systematically investigate the mediating effects of TyG-related indices on the association between FI and sarcopenia using weighted analytic methods. This study aimed to provide epidemiological evidence for the early identification of sarcopenia and a better understanding of its underlying metabolic mechanisms.

1.1. Study population and data source

NHANES, conducted by the Centers for Disease Control and Prevention (CDC), is a cross-sectional survey based on a nationally representative sample that aims to assess the health and nutritional status of the U.S. population.^[24] The dataset includes questionnaire responses, physical examinations, and laboratory tests, offering multidimensional information across different ages, sexes, races, and socioeconomic backgrounds.

The NHANES protocol was approved by the National Center for Health Statistics Research Ethics Review Board, and written informed consent was obtained from all participants. This secondary analysis of de-identified public data was exempt from institutional review board approval.

The data from the 2011 to 2014 NHANES cycles were accessed for research purposes on April 27, 2025. All NHANES data used in this study are publicly available and de-identified, such that authors did not have access to information that could identify individual participants during or after data collection.

In this study, data from the 2011 to 2014 NHANES cycles were used, with an initial sample size of 19,931 participants. Participants were excluded sequentially based on the availability of exposure, outcome, and covariate data: those with missing FI values (n = 2490; missing > 80% of 49 items); individuals under the age of 20 years (n = 6114); those lacking data on sarcopenia diagnosis indicators, including handgrip strength (HGS, n = 1757), appendicular skeletal muscle mass (ASM, n = 4189), and body mass index (BMI, n = 11); those missing covariates including education (n = 1), poverty-income ratio (PIR, n = 353), smoking (n = 2), alcohol consumption (n = 265), hypertension (n = 2), and diabetes (n = 2); and those missing variables required for TyG calculation, including waist circumference (waist circumference (WC), n = 35), triglycerides (TG, n = 2555), and fasting plasma glucose (fasting plasma glucose, n = 3). Because the sarcopenia assessment required both dual-energy X-ray absorptiometry-derived ASM and handgrip strength data, and these measurements were only available for participants aged 20 to 59 years in NHANES 2011 to 2014, the final analytic sample was restricted to this age range. A total of 2152 eligible participants were included in the final analysis (Fig. 1).

Figure 1.

Flowchart of participant selection from NHANES 2011–2014. HGS handgrip strength, ASM = appendicular skeletal muscle, BMI = body mass index, FPG = fasting plasma glucose, HGS = handgrip strength, NHANES = National Health and Nutrition Examination Survey, PIR = poverty-to-income ratio, TG = triglyceride, WC = waist circumference.

1.2. Exposure and outcome variables

The FI was constructed based on 49 health deficits, including cognitive function, dependency, depression, chronic diseases, healthcare utilization, general health status, anthropometric measures, and laboratory indicators.^[25] FI scores range from 0 to 1, with higher scores indicating more severe frailty. Participants were categorized into 4 groups based on FI values: ≤0.1 (robust), 0.1 to 0.2 (prefrail), 0.2 to 0.3 (mild frailty), and >0.3 (moderate to severe frailty).

Muscle mass data were obtained using dual-energy X-ray absorptiometry employed by NHANES to quantify ASM (kg). The skeletal muscle index was calculated as ASM divided by BMI (kg/m²). According to the criteria from the Foundation for the National Institutes of Health (FNIH) Sarcopenia Project, sarcopenia was defined as skeletal muscle index <0.789 for males and <0.512 for females; these cutoffs have been widely adopted in previous studies.^[26–29] MQI was calculated as the ratio of maximum handgrip strength (HGS, kg) to ASM (kg/kg).^[30] In this study, sarcopenia was primarily defined using the FNIH criteria, whereas MQI was included as a complementary continuous indicator of muscle quality rather than as an alternative diagnostic criterion for sarcopenia.

1.3. Mediating variables

Five TyG-related indices were included as mediating variables in this study. TyG is a triglyceride-glucose–based surrogate marker of IR, and its derived indices combine TyG with anthropometric measures to better reflect adiposity distribution and metabolic burden. Specifically, the following variables were calculated:

1. TyG = Ln [fasting triglyceride (mg/dL) × fasting glucose (mg/dL)/ 2];

2. TyG-BMI = TyG × body mass index (BMI), reflecting the combined burden of IR and general adiposity;

3. TyG-WC = TyG × waist circumference (WC, cm), reflecting the combined burden of IR and central adiposity;

4. TyG-WHtR = TyG × waist-to-height ratio (WHtR), reflecting the combined burden of IR and abdominal fat distribution relative to height;

5. TyG-ABSI = TyG × a body shape index (ABSI), where ABSI = WC (cm)/ [BMI^(2/3) × height^(1/2) (cm)], reflecting body shape–adjusted central adiposity.

1.4. Covariates

The potential covariates considered in this study included sex, age, race/ethnicity, education level, family income-to-poverty ratio (PIR), smoking, alcohol consumption, hypertension, and diabetes. Age and PIR were treated as continuous variables. Socioeconomic status was assessed using PIR and categorized as low (<1.3), middle (1.3–3.5), and high (≥3.5). Categorical variables included sex (male/female), race/ethnicity (Mexican American, non-Hispanic White, non-Hispanic Black, other races), education level (less than high school, high school, high school graduate, college or above), smoking status (never/former/current), alcohol consumption (no/yes), history of hypertension (no/yes), and history of diabetes (no/yes or prediabetes). Smoking status was determined based on whether the participant had smoked more than 100 cigarettes in their lifetime. Alcohol consumption was defined as drinking at least 12 alcoholic beverages per year. Histories of hypertension and diabetes were based on self-reported physician diagnoses.

1.5. Statistical analysis

Continuous variables were expressed as weighted means ± standard deviations, and categorical variables were described using weighted frequencies and weighted percentages. Group differences were assessed using weighted chi-square tests and weighted one-way analysis of variance (ANOVA). All analyses accounted for the complex survey design of NHANES by incorporating sampling weights, clustering (SDMVPSU), and stratification (SDMVSTRA) using R’s survey package. Following NHANES analytic guidance, we used the sample weight corresponding to the smallest analytic subpopulation that included all variables of interest (i.e., the least common denominator weight).^[24] Because TyG was derived from fasting triglycerides and fasting plasma glucose, the fasting subsample weight was applied in the present analyses. For the combined 2011 to 2014 cycles, the corresponding 4-year weight was constructed from the 2-year fasting subsample weight according to NHANES guidance.^[24] Weighted logistic regression models were used to evaluate the association between FI and sarcopenia, while weighted multivariable linear regression models assessed the relationship between FI and MQI. To improve model convergence and coefficient interpretation, FI values were multiplied by 100; thus, a 1-unit increase corresponded to a 0.01 change in the original FI value. As a sensitivity analysis, we additionally performed reverse-direction weighted regression models, treating FI (×100) as the dependent variable and sarcopenia as the independent variable in age-stratified subgroups (≤40 and >40 years) to assess the robustness of the observed association under an alternative model specification. Model 1 was unadjusted; Model 2 adjusted for age, sex, and race/ethnicity; and Model 3 further adjusted for PIR, education level, smoking, alcohol consumption, hypertension, and diabetes. Subgroup analyses were performed based on age (≤40, >40), sex (male/female), race/ethnicity (Mexican American, non-Hispanic White, non-Hispanic Black, other races), education level (less than high school, high school, high school graduate, college or above), PIR (<1.3, 1.3–3.5, ≥3.5), smoking status (never/former/current), alcohol consumption (no/yes), history of hypertension (no/yes), and history of diabetes (no/yes). Interaction tests were conducted to evaluate the consistency of associations across subgroups. Generalized Additive Models were used to fit smooth curves and assess potential non-linear relationships between FI, MQI, and sarcopenia risk. Mediation analyses were conducted to estimate the total effect, direct effect, and indirect effect of FI on sarcopenia and MQI through each TyG-related indicator. In this study, the indirect effect was interpreted as the mediation effect, and the proportion mediated was calculated as indirect effect/ total effect.^[31] All analyses were weighted using NHANES-recommended sampling weights. Statistical analyses were conducted using R software (version 4.3.2) and EmpowerStats (version 4.2), with a two-sided P-value <.05 considered statistically significant.

2. Results

2.1. Baseline characteristics of participants

A total of 2152 participants met the inclusion criteria, including 1115 males and 1037 females. Weighted analyses showed that the prevalence of sarcopenia was 7.62%. Significant differences were observed between participants with and without sarcopenia in terms of age, race, education level, PIR, alcohol consumption, hypertension, diabetes, MQI, and TyG-related indicators (all P < .05). Participants with sarcopenia were older, more likely to be non-Hispanic White, and had a higher prevalence of diabetes and hypertension. They also had higher BMI, TG, fasting plasma glucose, and TyG-related indices, but lower MQI, ASM, and HGS than those without sarcopenia. Detailed weighted means and distributions are shown in Table 1.

Table 1.

Characteristics of participants in NHANES (2011–2014) by sarcopenia.

Characteristics	Non-Sarcopenia (1988)	Sarcopenia (164)	P-value
Age (yr)
≤40	1136 (57.14%)	56 (34.15%)	<.001
>40	852 (42.86%)	108 (65.85%)
Gender
Male	1032 (51.91%)	83 (50.61%)	.748
Female	956 (48.09%)	81 (49.39%)
Race
Mexican American	222 (11.17%)	37 (22.56%)	<.001
Non-Hispanic White	843 (42.40%)	65 (39.63%)
Non-Hispanic Black	422 (21.23%)	12 (7.32%)
Other Race	501 (25.20%)	50 (30.49%)
Education (%)
Under high school	322 (16.20%)	38 (23.17%)	.013
High school	402 (20.22%)	40 (24.39%)
High school graduate	654 (32.90%)	52 (31.71%)
College or above	610 (30.68%)	34 (20.73%)
PIR (%)
≤1.30	679 (34.15%)	77 (46.95%)	.003
1.31–3.5	661 (33.25%)	49 (29.88%)
>3.5	648 (32.60%)	38 (23.17%)
Smoking (%)
Never	1187 (59.71%)	96 (58.54%)	.116
Former	323 (16.25%)	36 (21.95%)
Current	478 (24.04%)	32 (19.51%)
Drinking (%)
No	413 (20.77%)	44 (26.83%)	.068
Yes	1575 (79.23%)	120 (73.17%)
Hypertension (%)
No	1535 (77.21%)	105 (64.02%)	<.001
Yes	453 (22.79%)	59 (35.98%)
Diabetes (%)
No	1859 (93.51%)	144 (87.80%)	.006
Yes	129 (6.49%)	20 (12.20%)
BMI (kg/m2)	28.21 ± 6.60	33.54 ± 8.42	<.001
TG (mg/dL)	122.24 ± 133.94	156.24 ± 124.17	<.001
FPG (mg/dL)	103.19 ± 31.72	111.37 ± 35.22	<.001
MQI	3.39 ± 0.62	3.27 ± 0.65	.021
HGS (kg)	39.67 ± 11.08	33.48 ± 9.73	<.001
ASM (kg)	23.20 ± 6.40	20.29 ± 6.27	<.001
TyG	8.51 ± 0.70	8.83 ± 0.72	<.001
TyG-BMI	241.22 ± 64.75	296.96 ± 79.55	<.001
TyG-WC	821.75 ± 175.99	947.94 ± 193.36	<.001
TyG-WHtR	4.84 ± 1.01	5.93 ± 1.09	<.001
TyG-ABSI	0.68 ± 0.08	0.73 ± 0.08	<.001
FI	0.12 ± 0.08	0.16 ± 0.10	<.001

Mean (SD) for continuous variables, % for categorical variables. All estimates were weighted to account for the complex survey design of NHANES 2011–2014.

ASM = appendicular skeletal muscle, BMI = body mass index, FBG = fasting blood glucose, FI = frailty index, HGS = handgrip strength, MQI = muscle mass index, NHANES = National Health and Nutrition Examination Survey, PIR = poverty-to-income ratio, TG = triglyceride, TyG = triglyceride-glucose, TyG-ABSI = triglyceride-glucose combined with a body shape index, TyG-BMI = triglyceride-glucose combined with body mass index, TyG-WC = triglyceride-glucose combined with waist circumference, TyG-WHtR = triglyceride-glucose combined with waist-to-height ratio.

2.2. Association between FI and prevalence of sarcopenia and MQI

Because MQI reflects muscle strength relative to muscle mass, we additionally analyzed MQI as a complementary continuous outcome to provide a more detailed assessment of muscle quality beyond the binary sarcopenia definition. Higher FI scores were associated with increased sarcopenia risk and lower MQI levels; each 1-unit increase in FI (×100) corresponded to a 0.01 increase in the original FI value. In the unadjusted model (Model 1), each 1-unit increase in FI (×100) was significantly associated with higher sarcopenia risk (OR = 1.06, 95% CI: 1.04–1.07) and lower MQI levels (β = −0.02, 95% CI: −0.02 to −0.01). After adjustment for all covariates (Model 3), these associations remained significant. Each 1-unit increase in FI (×100) was associated with higher sarcopenia risk (OR = 1.06, 95% CI: 1.04–1.09) and lower MQI (β = −0.01, 95% CI: −0.02 to −0.01). Further comparisons across FI categories showed that sarcopenia prevalence was significantly higher in participants with higher FI levels. Compared with the lowest FI group, the highest FI group had a markedly greater risk of sarcopenia (OR = 4.83, 95% CI: 2.27–10.26). Detailed weighted regression results are shown in Table 2.

Table 2.

Association between FI and sarcopenia in multivariate regression models.

	Model 1		Model 2		Model 3
	OR (95% CI)	P-value	OR (95% CI)	P-value	OR (95% CI)	P-value
Sarcopenia
FI	1.06 (1.04, 1.07)	<.0001	1.06 (1.04, 1.08)	<.0001	1.06 (1.04, 1.09)	.0001
FI categorical
≤0.1	Ref.		Ref.		Ref.
>0.1, ≤0.2	2.36 (1.58, 3.51)	.0002	2.40 (1.63, 3.56)	.0002	2.48 (1.61, 3.82)	.0009
>0.2, ≤0.3	2.74 (1.47, 5.12)	.0036	2.87 (1.48, 5.59)	.0048	2.87 (1.37, 6.02)	.0136
>0.3	5.33 (2.79, 10.15)	<.0001	4.64 (2.19, 9.84)	.0005	4.83 (2.27, 10.26)	.001
P for trend	<.0001		<.0001		.0001
MQI
FI	−0.02 (−0.02, −0.01)	<.0001	−0.01 (−0.02, −0.01)	<.0001	−0.01 (−0.02, −0.01)	.0013
FI categorical
≤0.1	Ref.		Ref.		Ref.
>0.1, ≤0.2	−0.19 (−0.25, −0.14)	<.0001	−0.15 (−0.20, −0.09)	<.0001	−0.11 (−0.17, −0.05)	.0032
>0.2, ≤0.3	−0.41 (−0.53, −0.29)	<.0001	−0.32 (−0.44, −0.20)	<.0001	−0.24 (−0.37, −0.12)	.0020
>0.3	−0.42 (−0.69, −0.14)	.0059	−0.31 (−0.58, −0.04)	.0363	−0.17 (−0.44, 0.11)	.2525
P for trend	<.0001		<.0001		.0024

Model 1: Not adjusted. Model 2: Adjusted by age, gender, and race. Model 3: Adjusted by age, gender, race, education level, PIR, smoking, drinking, diabetes, and hypertension. To improve model convergence and interpretability, FI values were multiplied by 100 prior to analysis.

CI = confidence interval, FI = frailty index, MQI = muscle mass index, OR = odds ratio.

2.3. Reverse-direction sensitivity analysis

To address concerns regarding model directionality, we additionally performed reverse-direction weighted regression analyses with FI (×100) as the dependent variable and sarcopenia as the independent variable. In participants aged > 40 years, sarcopenia remained significantly associated with higher FI levels (β = 3.196, 95% CI: 1.550–4.843, P = .001). In participants aged ≤ 40 years, the association was directionally positive but did not reach statistical significance (β = 0.830, 95% CI: −1.736–3.395, P = .504). These findings suggest that the association between frailty burden and sarcopenia was robust in older adults, whereas the evidence in younger adults was less stable. Detailed results are presented in Table S1, Supplemental Digital Content 1.

2.4. Association between TyG-related indicators and prevalence of sarcopenia and MQI

The results of the weighted regression analysis between TyG-related indicators and the risk of sarcopenia and MQI are presented in Table 3. In weighted logistic regression analyses, each 1-unit increase in the TyG index was associated with a higher sarcopenia risk (OR = 1.62, 95% CI = 1.29–2.04) and a 0.12-unit decrease in MQI (β = −0.12, 95% CI = −0.17 to −0.07). In quartile group comparisons, the highest quartile (Q4) had an approximately 3.13-fold higher risk of developing sarcopenia compared with the lowest TyG quartile (Q1) (OR = 3.13, 95% CI = 1.54–6.38, P = .0067), and a reduction in MQI of 0.21 (β = −0.21, 95% CI = −0.31 to −0.10, P = 0016). Weighted analyses of TyG-BMI, TyG-WC, TyG-WHtR, and TyG-ABSI derivatives showed similar direction and statistical significance.

Table 3.

Associations between TyG-related indices and sarcopenia.

	Continuous	Q1	Q2	Q3	Q4	P for trend
Sarcopenia
TyG	1.62 (1.29, 2.04) 0.0007	Ref.	1.36 (0.72, 2.56) 0.3553	2.28 (1.08, 4.81) 0.0469	3.13 (1.54, 6.38) 0.0067	.0062
TyG-BMI	1.01 (1.01, 1.02) <0.0001	Ref.	2.01 (1.12, 3.63) 0.0346	2.34 (1.08, 5.05) 0.0469	9.53 (6.47, 14.05) <0.0001	<.0001
TyG-WC	1.00 (1.00, 1.01) <0.0001	Ref.	3.52 (1.34, 9.24) 0.0219	3.48 (1.49, 8.13) 0.0114	9.65 (4.83, 19.28) <0.0001	<.0001
TyG-WHtR	2.78 (2.25, 3.42) <0.0001	Ref.	4.28 (1.16, 15.79) 0.0452	8.92 (2.68, 29.66) 0.0028	28.10 (9.86, 80.08) <0.0001	<.0001
TyG-ABSI	197.26 (20.77, 1873.21) 0.0003	Ref.	1.88 (0.92, 3.85) 0.1049	2.04 (1.01, 4.10) 0.0653	3.39 (1.83, 6.26) 0.0015	.0016
MQI
TyG	−0.12 (−0.17, −0.07) 0.0002	Ref.	−0.04 (−0.17, 0.09) 0.5228	−0.13 (−0.23, −0.04) 0.0121	−0.21 (−0.31, −0.10) 0.0016	.0003
TyG-BMI	−0.00 (−0.01, −0.00) <0.0001	Ref.	−0.22 (−0.30, −0.14) 0.0001	−0.43 (−0.51, −0.36) <0.0001	−0.79 (−0.89, −0.70) <0.0001	<.0001
TyG-WC	−0.00 (−0.00, −0.00) <0.0001	Ref.	−0.21 (−0.28, −0.14) <0.0001	−0.47 (−0.54, −0.40) <0.0001	−0.78 (−0.88, −0.67) <0.0001	<.0001
TyG-WHtR	−0.29 (−0.32, −0.25) <0.0001	Ref.	−0.18 (−0.25, −0.12) <0.0001	−0.39 (−0.46, −0.32) <0.0001	−0.74 (−0.86, −0.63) <0.0001	<.0001
TyG-ABSI	−0.90 (−1.55, −0.26) 0.0141	Ref.	0.01 (−0.08, 0.10) 0.8194	−0.08 (−0.18, 0.03) 0.1659	−0.15 (−0.28, −0.02) 0.0404	.0244

Models were adjusted for age, gender, race, education level, PIR, smoking, drinking, diabetes, hypertension.

MQI =muscle mass index, PIR = poverty-income ratio, Q = quartile, TyG = triglyceride-glucose, TyG-ABSI = triglyceride-glucose combined with a body shape index, TyG-BMI = triglyceride-glucose combined with body mass index, TyG-WC = triglyceride-glucose combined with waist circumference, TyG-WHtR = triglyceride-glucose combined with waist-to-height ratio.

2.5. Subgroup analysis

To further assess the consistency of the relationship of FI (×100) with sarcopenia and MQI in different populations, multivariate subgroup analyses were performed in this study, and the results are presented in the form of forest plots (Figs. 2 and 3). The analysis results showed a significant interaction between FI and MQI in terms of age (interaction P = .0191) and sex (interaction P = .0261). Specifically, for each 1-unit increase in FI, MQI decreased by 0.02 units (β = −0.02, 95% CI = −0.02 to −0.01) for those under 40 years of age and by 0.01 units (β = −0.01, 95% CI = −0.01 to −0.00) for those over 40 years of age, while the results of the gender stratification showed that for a 1-unit increase in FI, the MQI decreased by 0.01 units in men (β = −0.01, 95% CI = −0.01 to −0.00) and a similar decrease in women (β = −0.01, 95% CI = −0.02 to −0.00), but the association was more statistically stable in the female group. Among the remaining stratification variables, the associations of FI with prevalence of sarcopenia and MQI were not significantly different (all interaction P > .05).

Figure 2.

Subgroup analysis of the association between FI and sarcopenia prevalence. The analysis adjusted for age, gender, race, education level, PIR, smoking, drinking, diabetes, hypertension. To improve model convergence and interpretability, FI values were multiplied by 100 prior to analysis. FI = frailty index, MQI = muscle quality index, PIR = poverty-to-income ratio.

Figure 3.

Subgroup analysis of the association between FI and MQI. The analysis adjusted for age, gender, race, education level, PIR, smoking, drinking, diabetes, hypertension. To improve model convergence and interpretability, FI values were multiplied by 100 prior to analysis. FI = frailty index, MQI = muscle quality index, PIR = poverty-to-income ratio.

2.6. Smooth curve fitting

To further assess whether there was a nonlinear relationship between FI (×100) and the risk of sarcopenia and the level of MQI, the present study used the generalized additive model to perform smoothed curve fitting analysis. The results showed a linear positive trend of correlation between FI and the risk of sarcopenia (Fig. 4), while a linear negative trend was observed between FI and MQI (Fig. 5), with no obvious nonlinear inflection points, supporting the main effect relationship observed in previous regression models.

Figure 4.

Smoothed curve fitting between FI and prevalence of sarcopenia. The solid red line represents the smooth curve fit between variables, and the blue bands represent the 95% confidence interval from the fit. To improve model convergence and interpretability, FI values were multiplied by 100 prior to analysis. FI = frailty index.

Figure 5.

Smoothed curve fitting between the FI and MQI. The solid red line represents the smooth curve fit between variables, and the blue bands represent the 95% confidence interval from the fit. To improve model convergence and interpretability, FI values were multiplied by 100 prior to analysis. FI = frailty index, MQI = muscle quality index.

2.7. Mediating role of TyG-related indices in the correlation between FI and sarcopenia

In addition, mediation analyses were performed to assess the potential mediating role of TyG-related indicators in the associations of FI (×100) with sarcopenia and MQI, as detailed in Table 4. In this analysis, the total effect represents the overall association of FI with the outcome, the direct effect represents the association not operating through the mediator, and the indirect effect represents the mediation effect operating through the corresponding TyG-related indicator. The proportion mediated was calculated as the indirect effect/ total effect. TyG showed a significant indirect (mediation) effect on the associations of FI with sarcopenia and MQI, with proportions mediated of 4.25% and 7.06%, respectively. Among the various TyG-derived indicators, TyG-WHtR showed the strongest mediation, with proportions mediated of 28.26% for sarcopenia and 47.05% for MQI. TyG-BMI, TyG-WC, and TyG-ABSI also showed significant indirect effects, suggesting that these metabolic indicators may play an important bridging role in the association between FI and muscle health. For clarity, the mediation pathways of the strongest mediator, TyG-WHtR, are graphically summarized in Figure 6, including the effect estimates, 95% confidence intervals, and proportion mediated.

Table 4.

Mediation analysis of TyG-related indices in the associations of FI (×100) with sarcopenia and MQI: total effect, direct effect, indirect effect, and proportion mediated (%).

	Total effect (95% CI), P value	Direct effect (95% CI), P value	Indirect effect/ mediation effect (95% CI), P value	Proportion mediated (%) (95% CI), P value
Sarcopenia
TyG	0.03 (0.01, 0.04), <.0001	0.03 (0.01, 0.04), .002	0.001 (0.0002, 0.003), .016	4.00 (1.00, 11.00) .016
TyG-BMI	0.03 (0.01, 0.04), <.0001	0.02 (0.01, 0.03), .008	0.006 (0.003, 0.01), <.0001	20.00 (10.00,50.00), <.0001
TyG-WC	0.03 (0.02, 0.04), <.0001	0.02 (0.01, 0.04), .004	0.005 (0.002, 0.01), <.0001	16.00 (7.00,40.00), <.0001
TyG-WHtR	0.03 (0.01, 0.04), <.0001	0.02 (0.004, 0.03), .014	0.008 (0.005, 0.01), <.0001	28.26 (15.00,67.00), <.0001
TyG-ABSI	0.03 (0.02, 0.04), <.0001	0.03 (0.01, 0.04), .002	0.002 (0.001, 0.004), .002	7.00 (2.00,18.00), .002
MQI
TyG	−0.11 (−0.15, −0.08), <.0001	−0.11 (−0.14, −0.07), <.0001	−0.008 (−0.01, −0.003), <.0001	7.00 (3.00, 3.00), <.0001
TyG-BMI	−0.11 (−0.15, −0.08), <.0001	−0.06 (−0.09, −0.03), <.0001	−0.06 (−0.08, −0.03), <.0001	50.00 (34.00, 69.00), <.0001
TyG-WC	−0.11 (−0.15, −0.08), <.0001	−0.06 (−0.09, −0.03), <.0001	−0.05 (−0.07, −0.03), <.0001	46.00 (31.00, 65.00), <.0001
TyG-WHtR	−0.11 (−0.15, −0.08), <.0001	−0.06 (−0.09, −0.03), <.0001	−0.05 (−0.07, −0.03), <.0001	47.05 (33.00, 66.00), <.0001
TyG-ABSI	−0.11 (−0.15, −0.08), <.0001	−0.11 (−0.14, −0.07), <.0001	−0.01 (−0.01, −0.003), <.0001	7.00 (3.00, 13.00), <.0001

Notes: Mediation analyses were adjusted for age, sex, race, education level, PIR, smoking, drinking, diabetes, and hypertension. The total effect represents the overall association of FI with the outcome. The direct effect represents the association of FI with the outcome not operating through the mediator. The indirect effect represents the mediation effect operating through the corresponding TyG-related indicator. In this study, the indirect effect was interpreted as the mediation effect. Proportion mediated was calculated as indirect effect/ total effect and is presented as a percentage. To improve model convergence and interpretability, FI values were multiplied by 100 prior to analysis.

CI = confidence interval, FI = frailty index, MQI = muscle quality index, PIR = poverty income ratio, TyG = triglyceride-glucose, TyG-ABSI = triglyceride-glucose combined with a body shape index, TyG-BMI = triglyceride-glucose combined with body mass index, TyG-WC = triglyceride-glucose combined with waist circumference, TyG-WHtR = triglyceride-glucose combined with waist-to-height ratio.

Figure 6.

Graphical summary of the mediation pathways of TyG-WHtR in the associations of the FI with sarcopenia and MQI. Panel A and Panel B illustrate the mediation pathways of TyG-WHtR in the associations of FI with sarcopenia and MQI, respectively. The figure presents the total effect, direct effect, indirect effect, and proportion mediated, together with the corresponding 95% confidence intervals. To improve model convergence and interpretability, FI values were multiplied by 100 prior to analysis. CI = confidence interval, FI = frailty index, MQI = muscle quality index, TyG-WHtR = triglyceride-glucose combined with waist-to-height ratio.

3. Discussion

Using a nationally representative sample from NHANES 2011 to 2014, this study systematically evaluated the mediating role of TyG-related indices in the association between FI and sarcopenia, thereby extending current understanding of the role of metabolic dysfunction in muscle health decline. Among the 2152 adults included, higher FI scores were associated with increased sarcopenia prevalence and lower MQI, and these associations remained significant after adjustment for multiple covariates. Mediation analysis further suggested that the association between FI and muscle health was partly mediated by TyG and its derived indices, with TyG-WHtR showing the largest indirect effect and proportion mediated. These findings support the view that IR-related metabolic abnormalities may contribute to the observed association between FI and poorer muscle health. However, because this study was cross-sectional, causal directionality cannot be determined.

In the present study, sarcopenia was defined using the established FNIH criteria, whereas MQI was analyzed as a supplementary continuous metric of muscle quality rather than an alternative diagnostic definition. We chose this approach because the binary sarcopenia definition identifies case status, whereas MQI reflects muscle strength relative to ASM and may capture gradations in muscle quality that are not fully reflected by case-based classification alone. Therefore, the inclusion of MQI was intended to complement the sarcopenia outcome and to provide a more detailed interpretation of muscle health in relation to FI.^[26,30]

FI is often used as an outcome or global health-status indicator in geriatric and interventional research. However, in observational epidemiology, a deficit-accumulation FI can also be interpreted as a summary measure of cumulative health deficits and physiological vulnerability.^[20–22] Therefore, we retained FI as the primary exposure in the main analyses while additionally conducting reverse-direction sensitivity analyses. In those analyses, sarcopenia remained significantly associated with higher FI levels in participants aged > 40 years, whereas the corresponding association in participants aged ≤ 40 years was positive but not statistically significant. Accordingly, the findings in younger adults should be interpreted cautiously and may reflect lower event burden, reduced statistical power, or greater heterogeneity in this age group.^[22,23] Because of the cross-sectional design, the temporal directionality between FI and sarcopenia cannot be determined.

Frailty can be assessed using different instruments, and these tools are not fully interchangeable because they capture partially different frailty constructs. In particular, the frailty phenotype focuses more on the physical manifestations of frailty, whereas the deficit-accumulation frailty index (FI) reflects the cumulative burden of multidomain health deficits.^[32] This distinction is important in the present study because our aim was to evaluate overall health deficit burden at the population level rather than to operationalize a purely physical frailty phenotype. In NHANES-based research, both the FI and frailty phenotype have shown expected frailty characteristics and meaningful associations with disability, self-rated health, and healthcare utilization, while the FI may better discriminate individuals along the lower-to-middle range of the frailty continuum.^[33] In addition, a systematic review of population studies showed that the FI and frailty phenotype had broadly comparable ability to predict all-cause mortality, supporting the validity of the FI as a population-level frailty measure.^[34] At the same time, direct comparisons across multiple frailty instruments have shown that feasibility, administration time, and agreement vary across settings, indicating that no single frailty tool is universally optimal for all research and clinical purposes.^[35] Therefore, although different frailty instruments may yield different classifications, the use of a deficit-accumulation FI in the present NHANES analysis is methodologically reasonable and consistent with prior epidemiological research.

Biologically, the association between FI and poorer muscle health is plausible because frailty reflects multisystem dysregulation, including inflammation, endocrine imbalance, nutritional vulnerability, and reduced physiologic reserve.^{[2,19,36–39]} Likewise, TyG-related indices capture IR-related metabolic disturbance and have been associated with reduced muscle mass and poorer muscle function.^[12,40–43] Therefore, the finding that TyG-related indicators partially mediated the associations of FI with sarcopenia and MQI is biologically reasonable. Nevertheless, these mechanistic considerations should be regarded as supportive background rather than the primary contribution of the present study, which was mainly designed to evaluate the population-level association between FI and sarcopenia.^[2,19]

The subgroup analysis showed that the negative association between FI and MQI varied across age (P for interaction = .0191) and sex (P for interaction = .0261), with a seemingly stronger association in individuals under 40 years of age. This finding appears counterintuitive because sarcopenia is generally considered an age-related condition and should, therefore, be interpreted cautiously. From a methodological perspective, sarcopenia is uncommon in younger adults, which may lead to less stable subgroup estimates, and the corresponding reverse-direction analysis with FI as the outcome did not reach statistical significance in this age group. In addition, MQI, as a continuous indicator of muscle quality, may be more sensitive than the binary sarcopenia definition in capturing early or subclinical muscle impairment in younger adults. From a biological perspective, it is possible that early metabolic dysregulation or reduced physiological reserve reflected by FI may already be associated with subtle declines in muscle quality before overt age-related sarcopenia becomes clinically apparent. Therefore, the younger-age findings should be regarded as exploratory and hypothesis-generating rather than confirmatory.^[23] Previous research has also suggested that the clinical implications of muscle quality may vary across age groups,^[44] and frailty in younger hospitalized adults has been linked to adverse outcomes,^[45] supporting the need for cautious but continued attention to frailty-related risk in younger populations. Regarding gender differences, the negative correlation between FI and MQI was more stable in women, which may be closely related to the key role of estrogen in maintaining skeletal muscle mass and function. Studies have shown that estrogen is involved in the regulation of skeletal muscle metabolism through its nuclear receptors (ERα, ERβ) as well as estrogen-associated receptors (ERRs), and plays an important role, especially in mitochondrial function, energy metabolism, oxidative stress defense, and muscle regeneration processes.^[46,47] Therefore, estrogen may maintain muscle health through multiple mechanisms in women, which also provides a plausible biological basis for explaining the differences in the relationship between FI and MQI between genders.

This study has several strengths. First, it was based on a nationally representative NHANES sample from 2011 to 2014, which enhances external validity and generalizability. Second, the analyses used multivariable adjustment to control for a range of potential confounders, including age, sex, race, education level, PIR, smoking, alcohol consumption, diabetes, and hypertension, and also explored age- and sex-related interactions through subgroup analyses. Finally, to our knowledge, this is the first study to systematically assess the mediating role of TyG-related indicators in the association between FI and sarcopenia in a national population. Nonetheless, the present study has several limitations. First, the study design was cross-sectional, and we were unable to determine a causal relationship between FI, TyG, and sarcopenia. Second, frailty can be operationalized using different instruments, and although the deficit-accumulation FI used here is well established, alternative frailty tools may yield different prevalence estimates and levels of agreement. Therefore, our findings should be interpreted in the context of the specific FI operationalization available in NHANES.^[32–35] Third, although multiple TyG-derived indices were used, they have not yet encompassed all available IR assessment tools for a more comprehensive comparison. Finally, due to database limitations, we included only Americans aged 20-59 years and failed to cover other age groups and countries.

4. Conclusion

To our knowledge, this study is the first to systematically assess the mediating role of TyG-related indicators in the association between FI and sarcopenia. The results showed that elevated FI levels were significantly associated with lower MQI and an increased risk of sarcopenia. The mediation analysis further suggested that TyG-related indicators, especially TyG-WHtR, may partly account for the observed association between FI and sarcopenia. The above findings deepen the understanding of the complex relationship between frailty, metabolic disorders, and skeletal muscle health, and suggest that TyG, as a readily available metabolic marker, may be useful for early identification of people at high risk of sarcopenia. These findings should be interpreted in light of the cross-sectional design, which precludes causal inference.

Acknowledgments

The authors thank all participants and staff of the National Health and Nutrition Examination Survey for their valuable contributions to data collection and management.

Author contributions

Data curation: Zhuanhong Yang, Zhihao Wei, Anran Cheng, Honghong Yang.

Formal analysis: Zhihao Wei, Anran Cheng, Honghong Yang.

Methodology: Honghong Yang.

Project administration: Honghong Yang.

Supervision: Zhuanhong Yang, Honghong Yang.

Visualization: Zhuanhong Yang, Honghong Yang.

Writing – review & editing: Zhuanhong Yang, Honghong Yang.

Writing – original draft: Honghong Yang.