Grip strength-to-weight ratio, all-cause and cardiovascular mortality, and cardiovascular disease prevalence: Evidence from NHANES and CHARLS

Abstract

Objective

The study aimed to investigate the relationships between grip strength-to-weight ratio (GSWR) and all-cause mortality, cardiovascular mortality, and prevalence of cardiovascular disease (CVD).

Methods

Data were analyzed from two nationally representative cohorts: the National Health and Nutrition Examination Survey (NHANES) and the China Health and Retirement Longitudinal Study (CHARLS). Weighted Cox proportional hazards models were used to estimate hazard ratios (HRs) for mortality outcomes, while weighted logistic regression evaluated the association between GSWR and CVD prevalence. Multivariable-adjusted restricted cubic splines (RCS) were used to examine potential non-linear relationships.

Results

Over median follow-up of 82.19 months and 100.07 months, individuals in the highest GSWR quartile exhibited a 73 % reduction in all-cause mortality (HR: 0.27, 95 % CI: 0.16–0.43) in American adults, and a 47 % reduction (HR: 0.53, 95 % CI: 0.43–0.65) in Chinese middle-aged and elderly populations, compared to those in the lowest quartile. After a median follow-up of 83.66 months, cardiovascular mortality was reduced by 69 % (HR: 0.31, 95 % CI: 0.22–0.44), 69 % (HR: 0.31, 95 % CI: 0.16–0.59), and 79 % (HR: 0.21, 95 % CI: 0.09–0.48) for each increase in GSWR quartile among American adults. An typical l-shaped non-linear relationship was observed between GSWR and both all-cause and cardiovascular mortality, with a similar non-linear association identified for CVD prevalence.

Conclusion

The GSWR was non-linearly associated with all-cause mortality, cardiovascular mortality, and prevalence of CVD. It can serve as a valuable health index, encouraging strength training among middle-aged and elderly individuals to maintain and enhance functional abilities.

1. Introduction

Muscular strength is a fundamental component of physical function, essential for activities of daily living such as walking, climbing stairs, lifting objects, and maintaining balance. Age-related physiological changes lead to a gradual decline in muscle strength, thereby heightening the risk of frailty, falls, and functional disability. Handgrip strength, measured via hand-held dynamometer, provides a simple, rapid, and cost-effective means of evaluating overall muscular capacity. Moreover, grip strength serves as a broad indicator of health, with demonstrated associations to mental health, nutritional status, cardiovascular diseases (CVDs), cancer, respiratory conditions, neurodegenerative disorders, and even mortality [[1], [2], [3], [4], [5]].

Grip strength varies widely due to sociodemographic factors (e.g., age, gender, ethnicity, and employment) and health-related determinants (e.g., physical activity, nutrition, and obesity). A critical limitation of absolute grip strength is its failure to account for variations in body size [6]. For instance, individuals with greater body weight may exhibit higher absolute grip strength due to increased muscle mass; nevertheless, their functional strength relative to body weight may lag than that of lighter individuals with lower absolute grip strength. To address this, the grip strength-to-weight ratio (GSWR) has emerged as a more precise and individualized metric by normalizing grip strength by body weight. Recent analyses, including correlation and regression studies, reveal that after accounting for weight, no significant correlation exists between height and grip strength. This suggests that weight normalization is more effective than height-based adjustments in capturing true muscular capacity [7]. A study based on the Sixth Korea National Health and Nutrition Examination Survey also demonstrated that the GSWR may be a superior indicator of metabolic syndrome (MS) compared to absolute grip strength or grip strength normalized by body mass index (BMI) [8]. The GSWR is increasingly proposed as a “vital sign” due to its ease of measurement.

This study aims to explore the associations between GSWR and key health outcomes, specifically all-cause mortality, cardiovascular mortality, and prevalence of CVD, using data from the US National Health and Nutrition Examination Survey (NHANES) and the China Health and Retirement Longitudinal Study (CHARLS). These datasets provide robust, nationally representative samples, facilitating a comprehensive evaluation of the relationship between GSWR and health outcomes across different populations.

2. Methods

2.1. Study population

The NHANES is a nationally representative survey run by the National Center for Health Statistics (NCHS), a division of the U.S. Centers for Disease Control and Prevention (CDC). It assesses the U.S. population’s sociological, health, and nutritional status through interviews, physical examinations, and laboratory tests. All procedures were approved by the NCHS Research Ethics Review Board (Protocol #2011–17 and its continuation), and all participants provided written informed consent. Data from two NHANES cycles (2011–2012 and 2013–2014) were linked to National Death Index (NDI) mortality data, forming the basis for this population-based cohort study. The CHARLS is another nationally representative survey led by the National School of Development at Peking University that focuses on Chinese residents aged 45 years and older. It includes about 18,000 participants from 150 counties and 450 villages. Launched in 2011, CHARLS collects data on health, socioeconomic status, and aging through biennial interviews, physical examinations, and blood tests. Data collection was approved by the Peking University Biomedical Ethics Review Committee (IRB00001052–11,015). All participants, or their legal representatives, provided written informed consent at baseline and follow-up.

In the NHANES cohort, a total of 10,176 individuals remained for the analysis of the relationship between the GSWR and mortality and 9615 individuals was used to examine the relationship between GSWR and CVD prevalence. In the CHARLS cohort, a total of 11,625 individuals remained for the analysis of the relationship between GSWR and all-cause mortality and 12,762 individuals was used to examine the relationship between GSWR and CVD prevalence. The selection process for study participants is illustrated in Fig. 1, and baseline characteristics of included and excluded participants are compared for NHANES in Table S1 and for CHARLS in Table S2.

Fig. 1.

Flowchart of participants inclusion and exclusion from NHANES (A) and CHALRS (B).

2.2. Grip strength measurement

In NHANES, trained health technicians at the Mobile Examination Center collected anthropometric data. Body weight was measured in kilograms (kg) using a calibrated digital scale. Grip strength was assessed using the TKK5401 Digital Grip Strength Dynamometer (Takei Scientific Instruments, Ltd, Tokyo, Japan), which was pre-calibrated for accuracy. Before the assessment, participants underwent a screening process that included a series of questions to determine eligibility and address potential confounding variables. During the test, participants adopted a standardized posture according to the relevant requirements of Muscle Strength Procedures Manual. Each hand was tested three times in alternating sequence, with a 60-s rest interval between trials for the same hand. Each participant underwent three grip strength trials per hand, totaling six trials. In CHARLS, the measurement method and unit for body weight were consistent with those in NHANES. Grip strength was measured using the Yuejian™ WL-1000 handgrip dynamometer (Nantong Yuejian Physical Measurement Instrument Co., Ltd., Nantong, China), with values recorded in kg. The device was factory-calibrated prior to distribution to ensure measurement accuracy. During the grip strength assessment, participants were instructed to stand upright while firmly grasping the dynamometer with their elbow flexed at 90°. They were required to exert maximum grip force for 3–5 s. A 30-s rest interval was provided between trials for the same hand to minimize fatigue. Each participant underwent two grip strength trials per hand, totaling four trials.

Combined grip strength was calculated by summing the maximum recorded values from each hand and reported in kg. Participants who completed the test with only one hand were excluded from this calculation. The GSWR was determined using the following formula:

$$\text{GSWR}=\frac{\text{Conbined}\text{Grip}\text{Strength}\left(\text{kg}\right)}{\text{Body}\text{Weight}\left(\text{kg}\right)}$$

2.3. Ascertaining the outcome

The NHANES data were linked to mortality outcomes using each participants’ unique identifier (SEQN) through the NDI. Participant vital status was tracked until December 31, 2019. Cardiovascular mortality was determined based on the underlying cause of death, classified according to the International Classification of Diseases, 10th Revision (ICD-10) system. The follow-up period extended from the date of the survey interview to either the participant’s date of death or the end of the available mortality follow-up period. The determination of whether respondents have CVD is based on their responses to a standardized questionnaire about health condition diagnoses during home-based interviews. CVDs included congestive heart failure, coronary artery disease, angina/angina pectoris, heart attack, and stroke.

In CHARLS, participants initially enrolled in the 2011 baseline survey were followed across four subsequent waves. The second wave (2013) recorded exact death dates, while the third (2015) and fourth (2018) waves documented only mortality status without specifying dates. By the fifth wave (2020), exact death dates were available for only a small subset of participants. When the exact date of death was unavailable, survival time was estimated as the interval from baseline to the interview wave preceding the death event, plus the median interval between the preceding and death-confirming interview waves. For participants with a confirmed date of death, survival time was calculated as the period from baseline to the date of death. Participants who remained in the study and survived through the fifth wave had their survival time measured as the interval between the baseline and the fifth wave survey. CVDs were identified based on standardized household survey responses regarding health and functional status. Participants were classified as having CVDs at baseline if they reported a medical diagnosis of heart problems (Heart attack, CAD, angina, congestive heart failure, or other heart problems) and stroke by a healthcare professional or had received treatment for these conditions using traditional Chinese medicine, western medicine, or other therapeutic approaches. The cause of death among participants was only ascertained in the second wave of CHARLS, whereas it was not determined in other waves. Consequently, the CHARLS dataset only permits the study of all-cause mortality and CVD prevalence, but precludes the analysis of cardiovascular mortality.

2.4. Covariates

Sociodemographic characteristics including gender, age, marital status, and education qualification were collected. In the NHANES analysis, additional variables such as race/ethnicity and the family poverty income ratio (PIR) were also considered. In the NHANES cohort, smoking status was determined based on serum cotinine concentrations and classified as follows: values below the limit of detection (LOD) were categorized as non-smokers; values between the LOD and 10 ng/mL as indicative of second-hand smoke exposure; and values exceeding 10 ng/mL as active smokers. Drinking status was determined using self-reported data from the alcohol use questionnaire and defined as the consumption of at least 12 alcoholic beverages within the past year. In the CHARLS cohort, smoking status was defined as having smoked more than 100 cigarettes in a lifetime, while drinking status was categorized based on responses to standardized household surveys regarding health and functional status, with options for drinking alcohol at least once a month in the past year, less than once a month, and not drinking at all. Both cohorts included common laboratory indicators, such as fasting blood glucose and serum lipid profiles. In the questionnaire-based assessment, individuals were classified as hypertensive if they had been diagnosed with hypertension by a doctor or other health professional or were receiving pharmacological treatment for the condition. Additionally, individuals with an average blood pressure of ≥140/90 mmHg across multiple measurements were also considered hypertensive. Diabetes mellitus (DM) was defined according to the following criteria: self-reported physician diagnosis, current use of antidiabetic medication or insulin, fasting blood glucose ≥7.0 mmol/L, 2-h postprandial glucose ≥11.1 mmol/L, and/or glycated hemoglobin (HbA1c) ≥6.5 %. Data were used in the present analysis as covariates.

2.5. Statistical analysis

Both NHANES and CHARLS employed complex sampling designs and applied sampling weights to ensure the representativeness of their respective populations. These design features were rigorously integrated into our statistical analyses.

All analyses were performed using R software (version 4.3.2). Continuous variables are reported as mean ± standard error (SE), and categorical variables as proportions with corresponding 95 % confidence intervals (CIs). The chi-square test was used to analyze categorical variables. Data normality was assessed using the Kolmogorov-Smirnov test. For data with a normal distribution, the independent samples t-test was used to compare the two groups; otherwise, the Mann-Whitney U test was applied. The GSWR was categorized into quartiles (Q1-Q4). Weighted Cox proportional hazards regression models were used to estimate hazard ratios (HRs) and 95 % CIs for mortality across quartiles of the GSWR. The proportional hazards assumption was tested for all models using Schoenfeld residuals. No evidence of violation of this assumption was found (all P > 0.05), confirming the appropriateness of the Cox models for this analysis. The relationships between GSWR and prevalence of CVD were further examined via binary logistic regression, with results expressed as odds ratios (ORs) and 95 % CIs. Three statistical models were constructed. Model 1 adjusted for gender, age, education qualification, and marital status, with an additional adjustment for the race/ethnicity and PIR in the NHANES cohort. Model 2 further adjusted for serum cotinine levels (or smoking status), alcohol consumption, and the presence of hypertension and DM. Model 3 additionally controlled for BMI and laboratory indicators. We assessed for multicollinearity among the covariates in our final adjusted models by calculating the variance inflation factor (VIF). All VIF values were well below 5, indicating that multicollinearity was not a significant concern in our analyses. Multivariable-adjusted restricted cubic spline (RCS) model with four knots was used to explore the dose-response relationship. Non-linearity was assessed using the likelihood ratio test. Following established statistical recommendations for ensuring a good model fit, the knots were placed at the 5th, 35th, 65th, and 95th percentiles of the GSWR distribution. A sensitivity analysis was also performed by varying the number of knots to assess the robustness of our findings. To minimize potential bias arising from missing data, multiple imputation was conducted using a random effects model, which accounts for both within- and between-subject variability. The imputation procedure was iterated multiple times to ensure stability and convergence of the imputed values. Subsequently, the results from each imputed dataset were combined using Rubin’s rules, which appropriately adjust for the uncertainty introduced by the imputation process. This approach allows for valid statistical inference by incorporating both within-imputation and between-imputation variance components. Statistical significance was defined as a two-tailed p-value < 0.05.

3. Results

3.1. Baseline characteristics

A total of 10,176 participants were enrolled in the NHANES cohort, with a weighted mean age of 46.37 ± 0.47 years. Baseline demographic and clinical characteristics of the overall study population, as well as subgroups categorized by survival status, are summarized in Table 1. The all-cause mortality group (0.74 ± 0.01) and the cardiovascular mortality group (0.68 ± 0.02) exhibited significantly lower GSWR compared to the survivor group (0.92 ± 0.01, P < 0.001). Significant differences were also observed between the death and survivor groups in gender, age, race/ethnicity, education levels, marital status, prevalence of hypertension, prevalence of DM, and multiple laboratory indicators. Additionally, 954 participants had CVD, with a weighted mean age of 64.78 ± 0.68 years. Individuals with CVD had significantly lower GSWR (0.76 ± 0.01) compared to those without CVD (0.92 ± 0.01, P < 0.001). Baseline demographic and clinical characteristics of participants with and without CVD are provided in Table S3.

Table 1.

Baseline demographic and clinical characteristics by survival status in the NHANES cohort.

Variables	All (n = 10,176)	Survival (n = 9358 )	All-cause mortality (n = 818)	Cardiovascular mortality (n = 299 )	Pa	Pb
Age (years)	46.37 ± 0.47	44.97 ± 0.48	66.53 ± 0.71	68.92 ± 1.23	<0.001	<0.001
Gender, Male (%)	49.58 (48.63–50.53)	49.34 (48.35–50.32)	53.10 (50.16–56.05)	53.57 (46.35–60.79)	0.021	0.265
Race/ethnicity (%)					<0.001	0.0001
Mexican American	8.24 (5.89–10.58)	8.55 (6.12–10.98)	3.74 (1.81–5.66)	3.57 (0.89–6.25)
Other Hispanic	5.84 (4.13–7.55)	6.03 (4.26–7.80)	3.09 (1.43–4.75)	3.18 (1.00–5.35)
Non-Hispanic White	66.88 (62.00–71.76)	66.16 (61.16–71.17)	77.22 (72.18–82.26)	77.99 (70.47–85.51)
Non-Hispanic Black	11.34 (8.63–14.06)	11.30 (8.59–14.01)	12.01 (8.36–15.65)	12.61 (6.78–18.44)
Other Race	7.70 (6.41–8.98)	7.96 (6.62–9.30)	3.95 (2.62–5.28)	2.66 (0.60–4.72)
Education Qualification (%)					<0.001	<0.001
< High school	15.16 (12.84–17.48)	14.47 (12.03–16.92)	25.09 (20.71–29.47)	26.68 (19.14–34.23)
High school/GED	20.75 (18.94–22.57)	20.44 (18.60–22.28)	25.31 (20.38–30.24)	26.78 (18.06–35.51)
> High school	64.08 (60.71–67.46)	65.09 (61.57–68.61)	49.60 (44.92–54.29)	46.54 (38.84–54.23)
Marital Status (%)					<0.001	<0.001
Divorced/widowed/separated	18.08 (17.00–19.15)	16.70 (15.64–17.76)	38.01 (34.24–41.78)	41.06 (34.88–47.25)
Married/unmarried couple	60.33 (57.91–62.75)	61.05 (58.57–63.54)	49.87 (46.33–53.42)	46.15 (38.22–54.07)
Never married	21.59 (18.89–24.30)	22.25 (19.45–25.05)	12.12 (8.56–15.68)	12.79 (7.86–17.72)
Family PRI					0.099	0.224
<1	16.35 (13.98–18.72)	11.70 (10.74–12.65)	31.63 (28.26–35.00)	19.13 (14.75–23.52)
>1	83.65 (81.28–86.02)	88.30 (87.35–89.26)	68.37 (65.00–71.74)	80.87 (76.48–85.25)
BMI (kg/m2)	28.86 ± 0.14	28.84 ± 0.15	29.25 ± 0.29	30.56 ± 0.48	0.198	0.001
Hypertension (%)	36.67 (34.95–38.39)	34.34 (32.75–35.94)	70.22 (65.81–74.63)	80.66 (75.10–86.21)	<0.001	<0.001
Diabetes Mellitus (%)	12.99 (11.95–14.02)	11.70 (10.74–12.65)	31.63 (28.26–35.00)	45.53 (38.78–52.28)	<0.001	<0.001
Cotinine (ng/mL)
< LOD	33.42 (30.66–36.19)	33.37 (30.52–36.22)	34.14 (30.22–38.06)	34.67 (26.13–43.21)	0.402	0.628
LOD-10	42.12 (40.27–43.98)	42.30 (40.43–44.17)	39.59 (35.93–43.25)	44.74 (38.02–51.45)
> 10	24.46 (22.51–26.40)	24.33 (22.28–26.38)	26.27 (22.16–30.38)	20.59 (12.24–28.94)
TCHO (mmol/L)	4.95 ± 0.02	4.96 ± 0.02	4.79 ± 0.04	4.76 ± 0.08	0.002	0.014
GLU (mmol/L)	5.56 ± 0.03	5.51 ± 0.03	6.32 ± 0.13	6.70 ± 0.25	<0.001	<0.001
TG (mmol/L)	1.71 ± 0.03	1.71 ± 0.03	1.78 ± 0.05	2.00 ± 0.11	0.005	<0.001
SCr (μmol/L)	79.19 ± 0.40	77.80 ± 0.36	99.28 ± 4.29	108.83 ± 7.44	<0.001	<0.001
HDL-C (mmol/L)	1.37 ± 0.01	1.37 ± 0.01	1.36 ± 0.02	1.30 ± 0.03	0.509	0.081
Grip strength (kg)	72.90 ± 0.33	73.92 ± 0.30	58.18 ± 0.94	54.85 ± 1.81	<0.001	<0.001
GSWR	0.91 ± 0.01	0.92 ± 0.01	0.74 ± 0.01	0.68 ± 0.02	<0.001	<0.001

NHANES: National Health and Nutrition Examination Surveys; GED: General Educational Development; PRI: Poverty Income Ratio; BMI: Body Mass Index; LOD: The limit of detection; TCHO: Total Cholesterol; GLU: Glucose; TG: Triglycerides; SCr: Serum Creatinine; HDL-C: High Density Lipoprotein Cholesterol; GSWR: Grip Strength-to-Weight Ratio.

^a. P values for comparisons of variables between Survival and All-Cause Mortality groups.

^b. P values for comparisons of variables between Survival and Cardiovascular Mortality groups.

In the CHARLS cohort, 11,625 participants were included, with a weighted mean age of 59.45 ± 0.17 years. Among them, 1685 deaths were recorded, with a weighted mean age of 69.50 ± 0.40 years. Significant differences between surviving and deceased populations were observed in age, gender, education level, marital status, BMI, smoking status, prevalence of hypertension, prevalence of DM and laboratory indicators including FBG and serum creatinine. The GSWR of the all-cause mortality group (0.99 ± 0.02) was significantly lower than that of the survivor group (1.08 ± 0.01, P < 0.001). These findings are summarized in Table 2. At the baseline survey, 1663 individuals were recorded with CVD, with a weighted mean age of 62.90 ± 0.33 years. The GSWR of this group (0.96 ± 0.01) was significantly lower than that of individuals without CVD (1.08 ± 0.01, P < 0.001), as detailed in Table S4.

Table 2.

Baseline demographic and clinical characteristics by survival status in the CHARLS cohort.

Variables	All (n = 11,625)	Survival (n = 9940)	All-cause mortality (n = 1685)	Pa
Age (years)	59.45 ± 0.17	57.68 ± 0.14	69.50 ± 0.40	<0.001
Gender, Male (%)	47.37 (46.32–48.43)	45.39 (44.25–46.52)	58.64 (55.72–61.57)	<0.001
Education Qualification (%)				<0.001
< High school	88.88 (87.27–90.50)	87.92 (86.09–89.75)	94.31 (92.72–95.91)
High school	9.40 (8.13–10.66)	10.32 (8.89–11.74)	4.16 (3.02–5.30)
> High school	1.72 (1.21–2.24)	1.76 (1.20–2.32)	1.52 (0.59–2.45)
Marital Status (%)				<0.001
Divorced/widowed/separated	18.18 (16.92–19.44)	15.44 (14.10–16.77)	33.75 (30.50–37.00)
Married/unmarried couple	80.93 (79.68–82.19)	83.88 (82.55–85.22)	64.22 (61.03–67.40)
Never married	0.88 (0.67–1.10)	0.68 (0.48–0.88)	2.03 (1.24–2.82)
BMI (kg/m2)	23.53 ± 0.09	23.78 ± 0.10	22.13 ± 0.12
Hypertension (%)	36.80 (35.09–38.50)	34.37 (32.64–36.10)	50.59 (47.40–53.78)	<0.001
Diabetes Mellitus (%)	13.67 (12.79–14.56)	12.55 (11.54–13.56)	19.93 (16.76–23.10)	<0.001
Smoking Status (%)	39.53 (38.28–40.78)	37.33 (35.91–38.74)	52.04 (48.78–55.29)	<0.001
Drinking Status (%)				0.058
Never drink/Rarely drink	24.64 (23.29–26.00)	24.42 (22.87–25.97)	25.94 (23.27–28.62)
Drink less than once a month	8.36 (7.45–9.28)	8.70 (7.69–9.70)	6.53 (5.15–7.91)
Drink less than once a month	66.99 (65.67–68.32)	66.89 (65.41–68.36)	67.53 (64.68–70.37)
TCHO (mmol/L)	4.98 ± 0.02	4.98 ± 0.02	4.91 ± 0.05	0.126
GLU (mmol/L)	6.01 ± 0.03	5.94 ± 0.03	6.37 ± 0.11	<0.001
TG (mmol/L)	1.50 ± 0.02	1.51 ± 0.02	1.49 ± 0.09	0.258
SCr (μmol/L)	69.86 ± 0.40	68.87 ± 0.36	76.40 ± 1.93	<0.001
HDL-C (mmol/L)	1.31 ± 0.01	1.31 ± 0.01	1.33 ± 0.02	0.188
Grip strength(kg)	62.13 ± 0.45	63.61 ± 0.43	53.77 ± 0.81	<0.001
GSWR	1.07 ± 0.01	1.08 ± 0.01	0.99 ± 0.02	<0.001

^a. P values for comparisons of variables between Survival and All-Cause Mortality groups.

3.2. Association between GSWR and all-cause mortality

During a median follow-up of 82.19 months (95 % CI: 80.32–84.05), a total of 818 all-cause deaths were recorded in the NHANES cohort. Participants were stratified into four groups based on quartiles of the GSWR, with the lowest quartile serving as the reference group. Weighted Cox proportional hazards regression models were used to evaluate the relationship between GSWR and all-cause mortality. After adjusting for multiple covariates, higher GSWR levels were significantly associated with lower all-cause mortality, with HRs and 95 % CIs of 0.47 (0.35–0.63), 0.39 (0.26–0.59), and 0.27 (0.16–0.43) for the second, third, and fourth quartiles, respectively. The forest plot illustrating this relationship is presented in Fig. S1.

In the CHARLS cohort, the median follow-up period was 100.07 months (95 % CI: 99.55–100.59), during which 1685 deaths occurred. Similar analyses showed significantly lower all-cause mortality in the second (HR: 0.61, 95 % CI: 0.52–0.70), third (HR: 0.50, 95 % CI: 0.42–0.59), and fourth (HR: 0.53, 95 % CI: 0.43–0.65) GSWR quartiles compared to the first. The corresponding forest plot is presented in Fig. S2.

The results from each adjustment model for both NHANES and CHARLS cohorts are detailed in Table 3. Additionally, dose-response analysis revealed typical l-shaped non-linear relationships between GSWR and all-cause mortality in both cohorts, as illustrated by the RCS curves in Fig. 2.

Table 3.

Hazard ratios for all-cause mortality across GSWR quartiles in the NHANES and CHARLS cohorts under different adjustment models.

Cohortsa	Quartiles of GSWR
	Q1	Q2	Q3	Q4
NHANES
Crude	Ref.	0.44 (0.34–0.55)	0.30 (0.23–0.39)	0.16 (0.12–0.22)
P		<0.001	<0.001	<0.001
Model 1	Ref.	0.53 (0.40–0.68)	0.46 (0.32–0.67)	0.34 (0.22–0.50)
P		<0.001	<0.001	<0.001
Model 2	Ref.	0.54 (0.41–0.70)	0.48 (0.34–0.68)	0.35 (0.24–0.51)
P		<0.001	<0.001	<0.001
Model 3	Ref.	0.47 (0.35–0.63)	0.39 (0.26–0.59)	0.27 (0.16–0.43)
P		<0.001	<0.001	<0.001
CHARLS
Crude	Ref.	0.60 (0.51–0.69)	0.53 (0.45–0.62)	0.58 (0.48–0.70)
P		<0.001	<0.001	<0.001
Model 1	Ref.	0.65 (0.56–0.76)	0.61 (0.52–0.72)	0.72 (0.59–0.88)
P		<0.001	<0.001	0.002
Model 2	Ref.	0.68 (0.59–0.79)	0.63 (0.53–0.74)	0.76 (0.60–0.95)
P		<0.001	<0.001	0.018
Model 3	Ref.	0.61 (0.52–0.70)	0.50 (0.42–0.59)	0.53 (0.43–0.65)
P		<0.001	<0.001	<0.001

Q: Quartile; Ref.: Reference.

^aFor NHANES, each quartile included 2544 participants (Q1-Q4); for CHARLS, the quartiles comprised 3191, 3191, 3189, and 3191 participants (Q1-Q4).

Fig. 2.

Restricted cubic spline (RCS) analysis of nonlinear relationships between grip strength-to-weight ratio (GSWR) and all-cause mortality in the NHANES (A) and CHARLS (B) cohorts. The purple curves depict the adjusted Hazard ratios (HRs) for all-cause mortality across the range of GSWR, with shaded bands representing the 95 % confidence intervals (CIs). The light blue histograms display the population distributions of GSWR, aligned with the right y-axes indicating the percentage of the population. Vertical red dashed lines represent the threshold values of GSWR where it begins to act as a protective factor against mortality risk: 0.89 in NHANES and 1.06 in CHARLS. Both the overall and nonlinear associations were statistically significant (P_overall < 0.0001; P_nonlinear < 0.0001), revealing a typical l-shaped nonlinear relationships between GSWR and all-cause mortality.

3.3. Association between GSWR and cardiovascular mortality

As the CHARLS cohort did not provide information on cause-specific mortality, analysis of cardiovascular mortality was conducted exclusively in the NHANES cohort. A total of 299 cardiovascular deaths were recorded over a median follow-up of 83.66 months (95 % CI: 81.72–85.61). Weighted Cox proportional hazards regression analysis showed that cardiovascular mortality was significantly lower in the second (HR: 0.31, 95 % CI: 0.22–0.44), third (HR: 0.31, 95 % CI: 0.16–0.59), and fourth (HR: 0.21, 95 % CI: 0.09–0.48) GSWR quartiles compared to the first quartile (Fig. S3). Results for each adjustment model are presented in Table 4. Furthermore, RCS curve as shown in Fig. 3 revealed a significant l-shaped non-linear relationship between GSWR and cardiovascular mortality.

Table 4.

Hazard ratios for cardiovascular mortality across GSWR quartiles in the NHANES cohort under different adjustment models.

Cohortsa	Quartiles of GSWR
	Q1	Q2	Q3	Q4
Crude	Ref.	0.27 (0.20–0.35)	0.20 (0.12–0.32)	0.08 (0.04–0.16)
P		<0.001	<0.001	<0.001
Model 1	Ref.	0.30 (0.21–0.42)	0.28 (0.14–0.54)	0.17 (0.09–0.32)
P		<0.001	<0.001	<0.001
Model 2	Ref.	0.33 (0.25–0.44)	0.34 (0.19–0.62)	0.23 (0.12–0.44)
P		<0.001	<0.001	<0.001
Model 3	Ref.	0.31 (0.22–0.44)	0.31 (0.16–0.59)	0.21 (0.09–0.48)
P		<0.001	<0.001	<0.001

^aFor NHANES, the quartiles comprised 2315, 2399, 2453, and 2490 participants (Q1-Q4).

Fig. 3.

RCS analysis of nonlinear relationship between GSWR and cardiovascular mortality in the NHANES cohort. The purple curves depict the adjusted HRs for cardiovascular mortality across the range of GSWR, with shaded bands representing the 95 % CIs. The light blue histograms display the population distributions of GSWR, aligned with the right y-axes indicating the percentage of the population. Two vertical dashed lines mark key GSWR thresholds: red dashed line represent the threshold value (GSWR = 0.90) where it begins to act as a protective factor against mortality risk, while the blue dashed line at GSWR = 1.40 indicates the upper threshold value which mortality risk begins to rise again. Both the overall and nonlinear associations were statistically significant (P_overall < 0.0001; P_nonlinear < 0.0001), revealing a l-shaped nonlinear relationship between GSWR and cardiovascular mortality.

3.4. Association between GSWR and prevalence of CVD

In the NHANES cohort, weighted logistic regression analysis revealed a significantly reduced prevalence of CVD in the second (OR: 0.64, 95 % CI: 0.48–0.86) and third (OR: 0.60, 95 % CI: 0.40–0.88) quartiles of the GSWR compared to the first quartile, as shown in Fig. S4. Similarly, in the CHARLS cohort, the prevalence of CVD was significantly lower in the second (OR: 0.75, 95 % CI: 0.61–0.91), third (OR: 0.72, 95 % CI: 0.59–0.88), and fourth (OR: 0.62, 95 % CI: 0.48–0.81) quartiles, as shown in Fig. S5. Results from the various adjustment models are summarized in Table 5. Furthermore, RCS analysis revealed a significant non-linear association between GSWR and CVD prevalence, as depicted in Fig. 4.

Table 5.

Odds ratios for prevalence of cardiovascular disease across GSWR quartiles in the NHANES and CHARLS cohorts under different adjustment models.

Cohortsa	Quartiles of GSWR
	Q1	Q2	Q3	Q4
NHANES
Crude	Ref.	0.45 (0.34–0.58)	0.31 (0.22–0.42)	0.23 (0.18–0.29)
P		<0.001	<0.001	<0.001
Model 1	Ref.	0.49 (0.37–0.65)	0.39 (0.26–0.57)	0.37 (0.27–0.50)
P		<0.001	<0.001	<0.001
Model 2	Ref.	0.57 (0.44–0.75)	0.50 (0.35–0.73)	0.54 (0.40–0.74)
P		<0.001	<0.001	<0.001
Model 3	Ref.	0.64 (0.48–0.86)	0.60 (0.40–0.88)	0.70 (0.45–1.08)
P		0.003	0.010	0.109
CHARLS
Crude	Ref.	0.59 (0.49–0.72)	0.49 (0.41–0.59)	0.36 (0.29–0.44)
P		<0.001	<0.001	<0.001
Model 1	Ref.	0.66 (0.59–0.80)	0.59 (0.49–0.71)	0.45 (0.36–0.57)
P		<0.001	<0.001	<0.001
Model 2	Ref.	0.73 (0.59–0.89)	0.68 (0.56–0.83)	0.57 (0.45–0.72)
P		0.002	<0.001	0.018
Model 3	Ref.	0.75 (0.61–0.91)	0.72 (0.59–0.88)	0.62 (0.48–0.81)
P		0.004	0.001	<0.001

^aFor NHANES, the quartiles comprised 24,04, 2404, 2403, and 2404 participants (Q1-Q4);; for CHARLS, the quartiles comprised 2907, 2910, 2902, and 2906 participants (Q1-Q4).

Fig. 4.

RCS analysis of nonlinear relationships between GSWR and prevalence of cardiovascular disease (CVD) in the NHANES (A) and CHARLS (B) cohorts. The purple curves depict the adjusted Odds ratios (ORs) for prevalence of CVD across the range of GSWR, with shaded bands representing the 95 % CIs. The light blue histograms display the population distributions of GSWR, aligned with the right y-axes indicating the percentage of the population. Vertical red dashed lines represent the threshold values of GSWR where it begins to act as a protective factor against the prevalence: 0.88 in NHANES and 1.06 in CHARLS.Both the overall and nonlinear associations were statistically significant (P_overall < 0.0001; P_nonlinear < 0.0001 in NHANES and P_nonlinear < 0.05 in CHARLS), revealing nonlinear relationships between GSWR and prevalence of CVD.

3.5. Subgroup analysis

The CHARLS targets individuals aged 45 years and older. To maintain consistency with CHARLS, we stratified the NHANES cohort into two age groups: ≥45 years and <45 years, and conducted separate analyses for each group. Among NHANES participants aged ≥45 years, each successive quartile increase in the GSWR was associated with a significant reduction in all-cause mortality: 52 % (HR = 0.48, 95 % CI: 0.36–0.65, P < 0.001), 64 % (HR = 0.36, 95 % CI: 0.23–0.56, P < 0.001), and 74 % (HR = 0.26, 95 % CI: 0.14–0.48, P < 0.001) for the second, third, and fourth quartiles, respectively. Similarly, cardiovascular mortality decreased by 70 % (HR: 0.30, 95 % CI: 0.20–0.43, P < 0.001), 67 % (HR: 0.33, 95 % CI: 0.17–0.65, P = 0.001), and 87 % (HR: 0.13, 95 % CI: 0.05–0.36, P < 0.001) for the respective quartiles. Furthermore, the prevalence of CVD was reduced by 42 % (OR = 0.58, 95 % CI: 0.45–0.75, P < 0.001) and 39 % (OR = 0.61, 95 % CI: 0.43–0.88, P = 0.007) in the second and third quartiles, respectively. RCS analysis revealed significant nonlinear associations, as depicted in Fig. S6. In contrast, among participants aged <45 years, GSWR was not significantly associated with all-cause mortality, cardiovascular mortality, or CVD prevalence (data not shown).

Given the significant gender-related differences in grip strength and body weight, we conducted subgroup analyses stratified by gender, with detailed results provided in Table S5. Overall, data from both the NHANES and CHARLS cohorts indicate that GSWR is significantly associated with all-cause mortality in both male and female participants, with a stronger association observed in men. In the NHANES cohort, GSWR was significantly associated with cardiovascular mortality only in male participants. Additionally, GSWR was linked to CVD prevalence in both sexes in NHANES, whereas in the CHARLS cohort, a significant association with CVD prevalence was found only among women. This discrepancy may be attributable to differences in the types of CVD assessed in the two surveys.

3.6. Sensitivity analyses

To evaluate the robustness of the non-linear associations identified by the RCS models, we conducted sensitivity analyses by varying the number of knots to three and five. The l-shaped dose-response relationships for all-cause and cardiovascular mortality, as well as the overall curve shape for CVD prevalence, remained essentially unchanged, thereby confirming the stability of our findings (Figs. S7–S9). To address potential reverse causality-where lower GSWR may reflect underlying severe illness rather than act as a true long-term predictor-we repeated the mortality analyses after excluding participants who died within the first two years of follow-up. The association between lower GSWR and increased risks of all-cause and cardiovascular mortality persisted (Table S6), reinforcing the validity of GSWR as a long-term prognostic marker.

4. Discussion

This study is the first to utilize data from both NHANES and CHARLS to explore the associations between GSWR and all-cause mortality, cardiovascular mortality, and CVD prevalence. The findings demonstrated that: 1) with the increase of GSWR, all-cause mortality and cardiovascular mortality significantly decreased, exhibiting an l-shaped non-linear relationship. 2) with the increase of GSWR, the prevalence of CVD showed a decreasing trend, also presenting a non-linear relationship. 3) both age and gender exerted a certain influence on the associations between the GSWR and all-cause mortality, cardiovascular mortality, and prevalence of CVD, with age having a more pronounced effect than gender.

The relationship between grip strength and health has been widely studied and research increasingly highlights grip strength as a valuable biomarker reflecting broader physiological resilience and a predictor of various health outcomes across the lifespan [9,10]. Good grip strength is associated with a reduced risk of CVD, depression, cancer, cognitive decline, and premature death [11]. However, the use of absolute grip strength as the sole measure of an individual’s physical capabilities presents several limitations. Grip strength is often used as a general measure of overall muscle strength and health, especially in older adults. However, this correlation isn't uniform across all muscle groups or functions, such as balance [12,13]. A significant factor influencing absolute grip strength is an individual's body size and weight [7]. Often, heavier individuals tend to exhibit higher absolute grip strength, which can be attributed to a greater overall muscle mass. This dependency on body size can be misleading when comparing the functional strength of individuals with different weights, as a higher absolute value in a heavier person might simply reflect their larger stature rather than superior strength relative to their own body mass. It is necessary to use of normalization methods to facilitate meaningful comparisons across individuals and populations with differing body sizes.

To normalizing the effect of body size and weight on grip strength and better reflect health, several metrics have been proposed, including GSWR, grip strength to height ratio, grip strength to BMI ratio, and even grip strength to height squared ratio[14,15]. However, there is still no consensus on this issue and even widespread controversy [[16], [17], [18]]. Current studies tend to use BMI or body weight to adjust grip strength for normalization. BMI has become a widely recognized and utilized tool in the landscape of health screening and assessment. The use of BMI to normalize grip strength have been widely studied and have yielded many positive results. However, A significant limitation of BMI is its inability to distinguish between fat mass and lean body mass [19]. This means that individuals with a high level of muscle mass, like athletes, may have a high BMI that classifies them as obese. Conversely, older adults who have experienced muscle loss might have a BMI within the normal range, potentially masking underlying health risks. Another limitation of BMI is that it requires measuring both height and weight and also requires calculation, which is relatively complicated. Given the significant covariance between strength capacity and body weight, as well as the direct relationship between strength and physical function and chronic health conditions mediated by the strength-to-weight ratio, it is feasible to normalized grip strength relative to weight [20]. Using body weight to standardize grip strength simplified the complexity of the operation and has been supported by some studies. Chun and co-workers provided evidence that grip strength was not related to metabolic syndrome; nevertheless, the normalization of grip strength by either body weight or BMI can provide its predictive capacity, and concerning this condition, the normalization of grip strength using body weight exhibited greater efficacy than normalization using BMI [8]. Xu et al. demonstrated through correlation and regression analysis that after eliminating the influence of body weight, no significant correlation was observed between height and grip strength [7]. Therefore, it is more preferable to conduct normalizing measurement of grip strength based on body weight rather than on BMI which includes height parameters.

The GSWR is a calculated value representing an individual's maximal handgrip strength (summing the maximum recorded values from each hand) relative to their body weight. Current research investigating the association between GSWR and overall health status is expanding, driven by its potential implications for evaluating physical fitness, functional capacity, and long-term health outcomes. For instance, the Evaluation and Monitoring on School-Based Nutrition and Growth in Shenzhen (EMSNGS) study, initiated in 2021 with 5153 participants, reported a dose-dependent relationship between weight-specific grip strength and cardiometabolic risk [21]. Similarly, analyses of data from the NHANES and the CHARLS identified weight-normalized grip strength as a significant biomarker for cardiometabolic diseases [22]. Gao et al. analyzed the health examination data of 804 Chinese middle-aged community residents aged 40 to 59 years and identified GSWR as a strong predictor of CVD risk factors, including hypertension, arteriosclerosis, hyperuricemia, hyperglycemia, hyperlipidemia, and MS [23]. A meta-analysis further supported these findings, revealing a linear dose-response relationship between GSWR and the prevalence of MS [24]. Additional studies have also reported associations between lower GSWR and increased risks of hypertension and DM [[25], [26], [27], [28]]. In another large-scale study, Whitney and Peterson analyzed data from 4143 participants aged 65 years and older from the National Health and Aging Trends Study (NHATS), with a six-year follow-up, and found GSWR to be associated with mortality and cerebrovascular events [23]. Moreover, Wu et al. demonstrated that GSWR serves as a reliable indicator for reduced muscle mass, as defined by the Global Leadership Initiative on Malnutrition, and can predict survival rates in patients with colorectal cancer in China [24]. In the present study, we are the first to establish a non-linear relationship between GSWR and all-cause mortality, cardiovascular mortality, and CVD prevalence. These findings are consistent with and further supported by the evidence presented in previous studies.

The intricate relationship between GSWR and overall health can be attributed to several underlying physiological mechanisms. A primary mediating factor is muscle mass and quality. This is because muscle tissue plays a crucial role in various bodily functions beyond just strength, including glucose metabolism, immune response, and overall physical resilience [29]. Systemic inflammation also appears to be a significant contributor. Higher levels of chronic, low-grade inflammation have been consistently linked to weaker muscle strength and lower muscle mass. This chronic inflammatory state can have detrimental effects on muscle protein synthesis and contribute to muscle wasting [30]. The nervous system plays a critical role as well. Diminished neural and motor system capacity can lead to decreased grip strength [31]. Metabolic health, particularly insulin resistance and glucose metabolism, is also intertwined with muscle strength [32,33]. Skeletal muscle is a primary site for glucose disposal, and reduced muscle mass and strength can impair glucose uptake, potentially contributing to insulin resistance and metabolic dysfunction [34].

It is important to acknowledge several limitations of this study. First, participants from both NHANES and CHARLS who lacked grip strength or mortality data were excluded, and these individuals differed systematically from the included sample in key baseline characteristics. Although rigorous statistical approaches were applied to minimize potential bias, residual confounding cannot be entirely excluded. Second, the exact date of death was unavailable for some deceased CHARLS participants and was estimated based on intervals between survey waves. While this method is consistent with prior studies [35,36], it inevitably introduces uncertainty into survival time estimation. Third, because standardized measurement protocols for grip strength assessment are lacking globally [37,38], methodological heterogeneity between NHANES and CHARLS limits direct comparison of absolute GSWR values and impedes the establishment of a universal diagnostic threshold. Consequently, although both cohorts consistently exhibited an l-shaped association, the identified GSWR thresholds should be interpreted as cohort-specific rather than broadly applicable to clinical practice. Fourth, differences in the operational definition of CVD may partly explain subtle inconsistencies in the observed associations: NHANES defined CVD as congestive heart failure, coronary artery disease, angina, heart attack, and stroke, whereas CHARLS additionally included other heart conditions. Fifth, CHARLS surveyed only middle-aged and older adults and did not provide cause-specific mortality data, thereby precluding evaluation of GSWR in relation to cardiovascular mortality and limiting applicability to younger populations. Finally, the observational design of our study limits causal inference. Our findings should therefore be interpreted as associations rather than definitive evidence of causation. The possibility of reverse causality, in particular, cannot be fully excluded. Although our sensitivity analysis excluding participants who died within the first two years of follow-up was designed to mitigate this bias, the risk of residual confounding from undiagnosed illness remains. These considerations underscore the need for further research, such as interventional studies, to confirm these relationships.

5. Conclusions

The results of our analyses of the NHANES and CHARLS datasets establish a non-linear relationship between GSWR and all-cause mortality, cardiovascular mortality, and prevalence of CVD. This provides a simple and non-invasive tool for health assessment, facilitating the identification of individuals with elevated mortality and prevalence of CVD. For populations with a lower GSWR, targeted interventions such as strength training and nutritional guidance can be considered to mitigate the anticipated risks.

Central Illustration.

Ethics approval and consent to participate

All data collection procedures of the National Health and Nutrition Examination Survey (NHANES) were approved by the National Center for Health Statistics (NCHS) Research Ethics Review Board (Protocol #2011–17, Continuation of Protocol #2011–17), and written informed consent was obtained from all participants prior to enrollment. Data collection procedures of the China Health and Retirement Longitudinal Study (CHARLS) were approved by the Biomedical Ethics Review Committee of Peking University (IRB00001052–11,015), and all participants or their legal representatives provided written informed consent before enrollment in the baseline and follow-up surveys, ensuring full awareness of the study’s purpose, procedures, and potential risks.

Availability of data and material

The datasets analyzed in this study were obtained from the National Health and Nutrition Examination Survey (NHANES) and the China Health and Retirement Longitudinal Study (CHARLS).

Funding

The study was supported by grants from the Young Taishan Scholars Program of Shandong, China (tsqn202507383) & National Natural Science Foundation of China (81672073) & Shandong Provincial Natural Science Foundation, China (ZR2022MH200).

CRediT authorship contribution statement

Jing Li: Writing – original draft, Formal analysis, Data curation. Jun-Na Sui: Data curation. Lian Tang: Formal analysis. Cong Zhao: Data curation. Ru-Hua Liu: Data curation. Jun-Jie Guo: Data curation. Ting-Ting Zhou: Formal analysis. Deng Pan: Formal analysis. Qing-Wu Tian: Writing – review & editing. Chao Xuan: Writing – original draft, Funding acquisition, Formal analysis, Data curation.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.