A Comparative Analysis of Grip Strength Evaluation Methods in a Large Cohort of Aged Mice

Abstract

Background: Grip strength is a key functional marker of musculoskeletal aging, widely used to assess sarcopenia. In preclinical research, multiple measurement methods are often combined to enhance reliability, but standardization remains challenging. To improve measurement robustness, we previously developed a composite strength score (SS5) that integrates five different grip strength tests into a single variable. While SS5 provides a comprehensive evaluation, its implementation is time-consuming, limiting feasibility in large-scale studies. In this study, we also examine two simplified composite scores, SS2 and SS3, as potential streamlined alternatives. Additionally, although normalizing grip strength to body weight is widely used, its appropriateness in geriatric mouse models has never been formally validated.

Methods: Forelimb grip strength was assessed in a cohort of 160-aged C57BL/6J mice using five methods: Weight Lift Tests (Deacon protocol with sponge weights and a modified version with metal wire weights), the Cage Lift Test and the Grip Strength Meter (trapeze bar and grid). Additionally, a cross-sectional group of 173 mice was analysed to assess the correlation between grip strength and muscle size. Each method was evaluated for its correlation with age, ability to detect sex differences, variability and association with muscle size.

Results: All methods strongly correlated with age (-0.518 ≤ r_s ≤ -0.306). The Grip Strength Meter (trapeze bar) and modified Deacon method were the most effective in detecting sex differences (p < 0.001). While all methods correlated with muscle size (0.153 ≤ r_s ≤ 0.332), the modified Deacon method and Grip Strength Meter showed the strongest associations. The mean coefficient of variation (CV%) ranged from 7% to 17%, demonstrating good repeatability. Notably, despite being widely used, normalization of grip strength to body weight was found to introduce bias in geriatric mice, as age-related weight loss distorts strength assessments. Absolute values proved to be a more reliable measure. To improve efficiency while maintaining reliability, we developed two new composite scores (SS2 and SS3) by integrating a subset of methods from SS5. These scores preserved the strong correlation with age observed in SS5 while reducing the number of required tests, enhancing feasibility.

Conclusions: Combining multiple grip strength assessments improves measurement reliability in aging studies. The newly proposed SS2 and SS3 scores provide a streamlined yet robust alternative to SS5, improving standardization and facilitating future comparisons in preclinical sarcopenia research. Our findings also challenge the routine normalization of grip strength to body weight in geriatric mice, emphasizing the importance of using absolute values to avoid bias.

Keywords: biogerontology; geriatric mouse model; grip strength; musculoskeletal aging; preclinical research.

FIGURE 1

Correlation with age of the five grip strength assessment methods. Correlation was evaluated in a cohort of C57BL/6J mice (n = 160), monitored longitudinally for a total of 782 grip strength measurements. Panels A–E show scatter plots relating age to scores obtained with the Deacon Grip Strength (DGs), Modified Grip Strength (MGs), Cage Lift Strength (CGs), Grid Strength meter (GGs) and Bar Strength meter (BGs) methods, respectively. Each plot displays the full dataset with corresponding linear regression lines and R ² values. Statistical analysis was conducted using two approaches: (1) Spearman's rank correlation coefficient (r_s) with 95% confidence intervals (CIs) and p‐values; and (2) linear mixed‐effects models (LMMs), which account for repeated measurements in a subset of animals, providing slope estimates with 95% CIs and p‐values. All methods demonstrated a significant inverse correlation with age.

FIGURE 2

Sex discrimination of the five grip strength assessment methods. Quantitative estimation of the five strength measurements in a cohort of C57BL/6J mice (n = 160) monitored longitudinally for a total of 782 measurements. Panels A–E show the trajectories of Modified Grip Strength (MG), Bar Strength meter (BG), Grid Strength meter (GG), Deacon Grip Strength (DG) and Cage Lift (CG), respectively, plotted as a function of age (in months) for male (blue line) and female (red line) mice. Values are presented as model‐derived mean estimates with 95% confidence intervals, obtained from a generalized linear mixed model (GLMM) analysis for longitudinal data. The model included sex, age (in months) and experimental batch as fixed effects. P‐values for the fixed effects of sex and age are reported in the insets. Asterisks indicate statistically significant differences by pairwise comparison between sexes at specific timepoints (p < 0.05, p < 0.01, p < 0.001 and p < 0.0001). Among the tested methods, MG and BG demonstrated the highest sensitivity in detecting sex‐related differences in muscle strength during aging.

FIGURE 3

Intragroup and interbatch variability for the five grip strength measurements. (A) Intragroup variability was evaluated by calculating the coefficient of variation (CV%) for each grip strength method at three distinct timepoints: 24, 25 and 26 months of age. The analysis was conducted in a longitudinal cohort of 160 C57BL/6J mice, with 90 mice assessed at 24 months, 150 at 25 months and 126 at 26 months, for a total of 366 measurements. Each dot represents the CV% for a given method at a specific timepoint, and black horizontal bars indicate the mean CV% across the three ages. (B) Interbatch variability was assessed by calculating CV% values separately for two independent experimental batches. Each point represents the CV% of a given method at one of the three timepoints within each batch. Horizontal black bars represent mean CV% for each method within each batch. Statistical analysis was performed using one‐way ANOVA for both intragroup (Panel A) and interbatch (Panel B) comparisons. No significant differences were found among methods (Panel A) or between batches (Panel B) (n.s., not significant). Grip strength was measured using the following five methods: Deacon Grip Strength (DG, orange), Modified Deacon (MG, green), Cage Lift (CG, purple), Grid Strength (GG, red) and Bar Strength (BG, blue). Among all methods, CG and GG showed lower intragroup variability, while BG exhibited good consistency across both intra‐ and interbatch comparisons.

FIGURE 4

Correlation of muscle size and the five grip strength measurements. Muscle size (MS) was measured in a cross‐sectional cohort of C57BL/6J mice, including 100 males and 73 females, each assessed at a single timepoint. (A) Correlation between muscle size and age, shown for the entire population (black line), and separately for male (blue) and female (red) mice. Linear regression (R ²) and Spearman's correlation coefficients (r_s) with significance levels (p‐values) are reported for each subgroup. (B) Correlation between muscle size and grip strength assessed using five methods: Deacon Grip Strength Score (DGs), Modified Grip Strength Score (MGs), Cage Lift Strength Score (CGs), Grid Strength Score (GGs) and Bar Strength Score (BGs). Each scatter plot includes linear regression (R ²) and results of Spearman's correlation analysis, reported with 95% confidence intervals (CIs) and p‐values.

FIGURE 5

Association of age with strength score 5‐methods (SS5), strength score 3‐methods (SS3) or strength score 2‐methods (SS2). C57BL/6J mice (n = 160) were monitored longitudinally, with repeated grip strength measurements collected over time, resulting in a total of 782 observations. Composite strength scores were calculated based on different combinations of grip strength tests: (A) SS5, a five‐method score including Deacon Grip Strength (DG), Modified Grip Strength (MG), Cage Lift (CG), Grid Strength meter (GG) and Bar Strength meter (BG) (as previously defined by Marcozzi et al.); (B) SS3, incorporating three methods (MG, CG and BG); (C) SS2, based on two methods (MG and BG). The scores are plotted as a function of age in male (blue) and female (red) mice. Data are shown as model‐derived mean estimates with 95% confidence intervals, obtained using Generalized Linear Mixed Model (GLMM) analysis for longitudinal data. Sex, age (in months) and batch were included as fixed effects. P‐values for sex and age are reported in the inset boxes. Asterisks indicate statistically significant differences by pairwise comparison between sexes at specific timepoints: *p < 0.05, **p < 0.01 and ***p < 0.001. All three strength scores showed a significant age‐related decline, while SS2 demonstrated the highest sensitivity in detecting sex‐related differences.