Introduction

Muscle weakness in older adults independently predicts illness [1], quality of life [35], and all-cause mortality [31]. Historically, scientists have attributed much of the loss of physical function observed during aging to a decline in muscle mass, or sarcopenia. However, longitudinal studies have demonstrated that aging has a considerably greater influence on muscular strength than what can be accounted for by a decline in muscle mass alone [12, 19]. Two common explanations for the loss of muscle strength in the absence of declining muscle mass include intramuscular adipocyte infiltration [18] and impairments in the central nervous system’s ability to recruit and discharge motor units [6, 16, 22]. For example, Clark et al. [7] reported that weaker older adults exhibited a ~ 20% reduction in voluntary activation, as well as lower and higher corticospinal excitability and inhibition, respectively. Other studies carried out in older adults have shown exceptions to the typical inverse relationship between motor unit recruitment thresholds and firing rates [16], as well as inefficient motor unit derecruitment during voluntary contractions [22]. For these reasons, experts have recently called attention to the importance of maintaining neuromuscular function with advanced age [3, 26].

Gait speed is becoming increasingly recognized as an important clinical mobility assessment tool in older adults [11]. While comfortable (or usual) gait speed is commonly studied, fast gait speed declines more so during old age and may therefore be a more sensitive outcome measure [21, 23]. While only a few studies have examined the physiological underpinnings of fast gait speed, evidence suggests that the neuromuscular system’s capacity to generate force/torque rapidly is important. For example, Clark et al. [8] reported noteworthy differences in the rate of force development for the plantarflexors, as well as the rate of muscle activation for the gastrocnemius, in older adults with slow versus fast 10-m gait speed. These authors also reported no differences in muscle cross-sectional area of the triceps surae, quadriceps, and hamstrings muscle groups between slow and fast walkers, suggesting that muscle mass not play a critical role in gait performance [8]. Similarly, using data from the Baltimore Longitudinal Study of Aging, Osawa et al. (2018) reported that the rate of torque development for the knee extensors was a significant, independent predictor of gait speed in adults. Collectively, there is growing evidence to suggest that fast gait speed is an important clinical tool that is mediated by neuromuscular function.

As interest in the ability to evaluate muscle function continues to grow, so too do methodological advancements that allow for scientific inquiry. In particular, dynamometers that provide information about the neuromuscular system’s ability to generate force during large muscle mass, multi-joint movements are becoming increasingly utilized. Palmer et al. [33] recently reported acceptable test–retest reliability statistics for maximal absolute and rapid strength characteristics using a novel isometric squat device. Theoretically, multi-joint strength assessments that allow for analysis of rapid force or torque signals may better reflect function during activities of daily living and therefore provide greater clinical utility.

The objective of this study was to compare the ability of absolute and rapid strength characteristics, as well as leg lean mass, to predict fast gait speed. To determine whether predictability of the independent variables was dependent on age, we examined groups of both younger and older adults. We hypothesized that fast gait speed performance would be best predicted by the rate of torque development, followed by maximal absolute strength and leg lean mass, and that these findings would be unique to older adults.

Methods

Participants

Twenty-four younger adults (12 men, mean ± SD age = 23 ± 3 years, mass = 70.2 ± 17.1 kg, height = 1.72 ± 0.07 m; 12 women, age = 21 ± 2 years, mass = 63.4 ± 11.7 kg, height = 1.64 ± 0.06 m) and 22 older adults (11 men, mean ± SD age = 74 ± 4 years, mass = 77.4 ± 12.5 kg, height = 1.72 ± 0.28 m; 11 women, age = 71 ± 5 years, mass = 69.3 ± 7.6 kg, height = 1.54 ± 0.17 m) volunteered to participate in this study. Preliminary screening was conducted over the telephone to qualify participants based on the following eight inclusion criteria: (1) age between 18 and 30 or 65 and 85 years; (2) did not require the use of an assistive walking device; (3) no surgery on the hip or knee joints within the previous 12 months; (4) had no known neuromuscular or metabolic diseases; (5) had not experienced a myocardial infarction within the previous 12 months; (6) did not regularly engage in resistance training (< three times monthly in the previous 3 months) or other structured aerobic exercise more than 20 min per day, twice per week [24]; (7) did not have uncontrolled hypertension (> 150/90 mm Hg); (8) could not have had their primary physician advise them not to engage in exercise. Once screened for the above qualifications, participants reported to the laboratory where they were further screened for the inclusion criteria of a body mass index between 19 and 33 kg/m2, after which they completed health, medical history, and physical activity questionnaires. All participants gave written informed consent prior to participation in this investigation. This study was approved by the University Institutional Review Board following an expedited review process.

Study Design

Participants reported to the laboratory on two occasions separated by 48–96 h. The initial visit involved completion of study paperwork, body composition analysis, and gait speed testing. The participants were familiarized with isometric squat strength testing during the initial visit to the laboratory. The second visit only involved isometric strength data collection. Participants were asked to refrain from exercise 48 h prior to each laboratory visit.

Dual X-Ray Absorptiometry (DXA)

DXA has been proposed to serve as the reference technique for estimating muscle mass and body composition in older adults [2]. The participants completed one total body scan using the DXA (Lunar Prodigy Primo, GE Healthcare, Madison, WI, USA). All scans were performed by a trained technician that had completed both university radiation training and a training session held by the device’s manufacturer. Automated methods from the manufacturer-provided software (Lunar Radiation Body Composition, version 13.60, GE Healthcare, Wauwatosa, WI, USA) were used to estimate leg lean mass (kg).

Isometric Squat Strength

Isometric squat testing was performed using a commercially-available multi-joint isokinetic strength training device (Exerbotics Squat [eSQ], Exerbotics, LLC, Tulsa, OK, USA). Test–retest reliability statistics for these methods have been reported previously [33, 36]. During testing, the participants stood with their feet shoulder width apart and shoulders positioned under the padded frame. The participants were able to gently hold onto handles directly in front of them for balance and postural support. The researcher then adjusted the frame height such that the knee joint was aligned at a 120° angle. Each participant performed two maximal voluntary contractions separated by 1 min of rest. The participants were specifically instructed to perform the maximal voluntary contractions “as hard and as fast as possible until I instruct you to stop”, with each lasting a total of 4 s.

The torque signal was sampled at 2000 Hz and processed off-line using custom LabVIEW software (version 8.5, National Instruments, Austin, TX, USA). The raw signals were scaled to units (Nm) and filtered using a zero-phase shift, second-order Butterworth filter with a 20 Hz low-pass cutoff frequency. The signal onsets were determined manually via visual inspection as the point when the signal first deflected from the baseline [17], with the aid of a horizontal cursor, which helped the inspector objectively assess the collective baseline level. Peak torque (Nm) was calculated as the highest 500 ms epoch throughout the duration of the contraction. The RTD200 was quantified from the linear slope of the ascending portion of the torque-time curve at 200 ms from onset (Fig. 1). We elected to study RTD200 rather than earlier torque-time intervals because it tends to be more reliable [43], particularly using the equipment of the present investigation [33].

Fig. 1
figure 1

Examples of the isometric squat data collection and analysis procedures. The image on the left shows a younger adult male performing an isometric squat as described herein. The load cell used to acquire the torque data has been circled. On the right, an example torque-time curve has been displayed. The dotted area to the left shows the 200 ms region where the rate of torque development was calculated, whereas the shaded area corresponds to where peak torque was noted

Full size image

Fast Gait Speed

We assessed fast gait speed at distances of both 10-m and 400-m. These distances were selected because they are common in the gait speed literature, as they show high sensitivity for identifying declines in lower-extremity function [8, 23] and are associated with mortality and morbidity rates [32]. The 10-m walk test was conducted according to the procedures of Clark et al. [8], in which the 10-m distance was centered between 5-m acceleration and deceleration zones to help achieve a steady state. Three trials separated by 2-min rest periods were performed. The mean of the three speeds was utilized for analysis (m/s). The 400-m test was performed in accordance with the procedures described by Newman et al. [32], in which the participants walked a total of ten laps between two cones set at 20-m apart (40-m per lap). Participants were instructed to “walk as quickly as you can, without running, at a pace you can maintain.” [32] Only one 400-m test was performed. A stopwatch was used to measure times to complete the tests. All gait speed tests were performed in an indoor, air-conditioned facility.

Statistical Analyses

Differences between younger and older adults were examined with independent samples t tests, Cohen’s d effect size statistics, and 95% confidence intervals (CIs) for mean differences. As both age [9] and height [15] have been shown to be important regulators of gait speed, partial correlations (controlling for age and height) were used to examine the associations among leg lean mass, peak torque, and RTD200 versus 10- and 400-m fast gait speed for each age group. Four separate stepwise regression analyses were performed to examine the ability for leg lean mass, peak torque, RTD200, age, and height to explain variance in 10- and 400-m fast gait speed in younger and older adults. An α level of P ≤ 0.05 was used to determine statistical significance for all data analyses. To monitor multicollinearity, the variance inflation factor (VIF) was calculated using the equation VIF = 1/(1 − R2). A VIF greater than 10 has been suggested to indicate problematic multicollinearity [46]. All statistical analyses were performed using SPSS software (IBM SPSS Statistics, Version 25.0. Armonk, NY: IBM Corp). Univariate scatterplots displaying individual participant data were created using templates provided by Weissgerber et al. [48].

Results

As shown in Fig. 2, the mean difference in leg lean mass between younger and older adults was small [P = 0.246, d = 0.35, 95% CI (−0.74, 2.82 kg)]. In contrast, as shown in Figs. 3 and 4, there were large differences for peak torque [P < 0.001, d = 1.24, 95% CI (250.7, 554.2 Nm)], RTD200 [P < 0.001, d = 1.07, 95% CI (823.9, 2308.1 Nm/s)], 10-m fast gait speed [P = 0.006, d = 0.79, 95% CI (0.08, 0.43 m/s)], and 400-m fast gait speed [P < 0.001, d = 1.43, 95% CI (0.27, 0.50 m/s)]. As shown in Table 1, none of the partial correlations for the younger adults were significant. For the older adults, however, a significant partial correlation between 400-m fast gait speed and RTD200 was demonstrated (r = 0.463, P = 0.040). Similar nonsignificant findings for 10-m fast gait speed and peak torque (r = 0.422, P = 0.064), as well as 400-m fast gait speed and peak torque (r = 0.421, P = 0.065), were demonstrated for the older adults. The results from the stepwise regression analyses indicated that, for the younger adults, none of the independent variables were significant predictors of gait speed. For the older adults, peak torque was a significant predictor of both 10-m (R2 = 0.257; adjusted R2 = 0.220; P = 0.016) and 400-m (R2 = 0.239; adjusted R2 = 0.200; P = 0.021) fast gait speed. Stepwise regression demonstrated that leg lean mass and RTD200 were not useful predictors of fast gait speed.

Fig. 2
figure 2

Individual participant data for leg lean mass with the younger and older adults shown on the left and right, respectively. The thick black symbol corresponds to the mean value. The most relevant statistical results are shown at the top of the graph

Full size image
Fig. 3
figure 3

Individual participant data for (a) squat peak torque and (b) the rate of torque development at 200 ms, with the younger and older adults shown on the left and right, respectively, of each graph. The thick black symbol corresponds to the mean value for each age group. The most relevant statistical results are shown at the top of the graph

Full size image
Fig. 4
figure 4

Individual participant data for (a) 10-m and (b) 400-m fast gait speed, with the younger and older adults shown on the left and right, respectively, of each graph. The thick black symbol corresponds to the mean value for each age group. The most relevant statistical results are shown at the top of the graph

Full size image
Table 1 Partial correlations for younger (top) and older (bottom) adults showing associations between the gait speed outcomes and leg lean mass, peak torque, and the rate of torque development at 200 ms (RTD200) while controlling for age and height
Full size table

Discussion

The primary finding of this study was that measures of absolute and rapid squat strength were associated with short and long-distance fast gait speed in older adults, but leg lean mass was not observed to be an important predictor. More specifically, similar partial correlations were demonstrated for peak torque and RTD200, whereas the stepwise regression analysis showed that peak torque was the best overall predictor of fast gait speed. Furthermore, there were large differences in absolute and rapid squat strength and gait speed between age groups, but there was no difference in leg lean mass between younger and older adults. Combined with the fact that none of the examined variables explained variance in fast gait speed among the younger adults, these findings highlight the importance of maintaining lower-extremity squat strength throughout the lifespan. We believe that these findings may highlight a potential dissociation between leg lean mass and functional performance in older adults.

Our findings demonstrated that leg lean mass did not significantly differ between younger and older adults. Similarity of DXA-derived leg lean mass among younger and older adults has been reported in some [10, 40], but not all [25, 45], studies. These results are also similar in nature to a three-year study in older adults [34], as well as those reported by Clark et al. [8], who reported no differences between slow and fast walkers for MRI-derived muscle cross sectional area of the triceps surae, quadriceps, and hamstrings muscle groups. Given that such dramatic age-related differences were observed for our muscle strength and gait speed measures (see Figs. 3 and 4), but not leg lean mass, these results highlight the dissociation between muscle mass and functional performance in older adults. When one considers that older adults may lose significant muscle strength in the absence of muscle atrophy [12, 19], we suspect that our findings provide further support for the notion that fast gait speed is characterized by the neuromuscular system’s control of muscle force, which may be mediated by parameters such as voluntary activation, corticospinal excitability/inhibition, and/or motor unit firing rates [4]. Clinicians seeking to prevent musculoskeletal declines in aging adults should be concerned with muscle tissue quality, as the most recent sarcopenia guidelines highlight the fact that a single body composition assessment provides little information about a patient’s functional status [11].

Previous studies that sought to characterize gait performance have typically focused on individual muscles or muscle groups. In particular, an emphasis has been placed on understanding the relationship between the gait cycle and ankle plantarflexion performance. These studies have demonstrated the importance of maintaining neuromuscular control of the triceps surae [8, 29], with weakness being directly associated with reduced gait speed in older adults with functional limitations [39]. A unique aspect of our approach to this study, however, was the assessment of absolute and rapid strength characteristics during maximal isometric squats. Isometric multi-joint tests involving the lower extremities have been used to study muscle strength in young adults and athletes [14], but this approach has not been utilized in older adults. We sought to study the isometric squat rather than isolated joint movements because we felt that analysis of a lower-extremity, multi-joint movement involving the activation of dozens of muscles would better characterize fast gait performance. Squatting movements are also commonly utilized in resistance training interventions [37] and physical therapy and rehabilitation clinics [27]. By focusing on muscle performance and lean mass of the entire lower extremity, rather than isolated muscles, we would contend that our approach better reflects measures that may typically be used in a clinical setting. We should concede, however, that while the absolute and rapid strength measurements utilized herein explained roughly 25% of the variance in fast gait speed, other previous studies have demonstrated stronger prediction [9, 27, 47]. For example, Clark et al. [9] reported that the combination of age, muscle and fat mass variables, and the rate of electromyographic rise accounted for 72.4% and 43.7% of the variance in fast gait speed for men and women, respectively. More recently, Mantel et al. [27] was able to explain more than 80% of the variance in fast gait speed utilizing a comprehensive assessment battery focused on both strength and balance. Given the totality of evidence, it seems likely that fast gait speed is best predicted by multiple outcome measures.

Although our study design was not specifically focused on differences in absolute and rapid strength characteristics, we should highlight these findings. As shown in Table 1, the associations between peak torque and RTD200 versus both short- and long-distance fast gait speed were quite similar for the older adults. Stepwise regression analysis revealed that RTD200 was the best single predictor of fast gait speed, with peak torque likely being removed due to high collinearity. While rapid neuromuscular activation should be emphasized in maintaining functional outcomes across the lifespan [8, 9, 42, 44], our findings suggest that absolute strength should not be completely discounted.

We should note that our investigation had limitations. Specifically, it should be noted that the optimal measurement tools for assessing body composition in older adults is still a matter of debate. Like many other previous studies [19, 28, 38, 47], our approach relied on DXA to quantify lean mass, primarily because of its availability and ease of use. Very recently, over 30 experts in the fields of aging and body composition from the European Society for Clinical and Economic Aspects of Osteoporosis and Osteoarthritis working group on frailty and sarcopenia proposed that DXA be universally adopted as the reference standard for measuring muscle mass [2]. These authors noted the benefits of using DXA as a reference measure, which include low radiation, low precision errors, and its rapid assessment of three body compartments [2]. However, this proposal has been met with some skepticism. Clark et al. [5] noted that while DXA has several advantages as a tool to quantify appendicular lean soft tissue mass, it offers lower accuracy than MRI and CT scans relative to cadaveric measurements and is prone to measurement error. Furthermore, studies have reported only a modest relationship between increases in DXA- and MRI- or CT-derived measures of percent change in muscle size [13, 20, 41]. Thus, we concede that there are a variety of methodological techniques for assessing body composition in both younger and older adults, and measurement via ultrasound, MRI, or CT may yield differing conclusions from those of the present study. Other limitations of the present study may include too small of a sample to comprehensively study sex differences, lacking sophisticated neuromuscular measurements (e.g., electromyography, motor unit analysis, transcranial magnetic stimulation, etc.), which may have helped explain variance in fast gait speed, and only studying fast gait speed. Finally, as our investigation utilized a cross-sectional design, these data should not be extrapolated to longitudinal changes that occur throughout the process of aging or interventions that may affect mobility.

In conclusion, our findings indicate that absolute and rapid multi-joint squat strength, but not leg lean mass, are important predictors of both 10- and 400-m fast gait speed in older adults. The fact that leg lean mass was not at all associated with gait speed further highlights the potential dissociation between skeletal muscle mass and function. Additional studies including larger samples, other measures of muscle morphology, and more variables well-grounded in neuromuscular physiology may help replicate these findings, as well as aid in elucidating the unexplained variance in fast gait speed. Understanding the neuromuscular determinants of mobility has important implications for creating assessments and interventions that can be used in practice to prevent functional impairments among older adults.