Augmenting strength-to-weight ratio by body weight unloading affects walking performance equally in obese and nonobese older adults

Procedures

Volunteers reported to the laboratory on three separate days, with 2 to 4 days separating visits. On visit 1, informed consent and measurements of height and weight were obtained, followed by completion of the Rapid Assessment of Physical Activity (RAPA) (Topolski et al. 2006) and the Short Physical Performance Battery (SPPB) (Guralnik et al. 1994). Participants were habituated to maximal voluntary isometric strength testing of the knee extensors on a commercial dynamometer (HUMAC Norm, CSMI, Stoughton, MA, USA), and S:W was determined from this initial assessment. The amount of body weight reduction needed to increase the knee extensor S:W by +0.1 Nm kg⁻¹, +0.2 Nm kg⁻¹, and +0.3 Nm kg⁻¹ was calculated according to the algorithm described in Eqs. 1–3. Then, subjects completed a treadmill habituation session which included a 3-min warm-up walk at 0.8 m s⁻¹, walking for 1 min at maximal speed, and walking for 3 min each at four levels of S:W. At each level of S:W, volunteers were asked to self-select preferred walking speed as described in detail later.

During visit 2, knee extensor S:W was again measured and these S:W data were used for analysis in the study and final determination of BWU levels. Subjects completed another BWU treadmill walking habituation session at the four levels of S:W, again self-selecting preferred speed at each level. On visit 3, subjects were prepared for recording the electromyogram (EMG) from the vastus lateralis (knee extensor) and gastrocnemius lateralis (ankle plantarflexor) muscles of each leg and completed a self-selected maximal walking speed trial of 1 min duration to determine peak muscle activation. This was followed by two-minute bouts of walking under normal S:W, S:W + 0.1 Nm kg⁻¹, S:W + 0.2 Nm kg⁻¹, and S:W + 0.3 Nm kg⁻¹ conditions in random order. Two minutes of seated rest were provided between each walking trial. At each level of S:W, participants self-selected preferred walking speed, and spatial, temporal, and kinetic gait parameters were obtained from the instrumented treadmill in the final minute of the walk, while EMG was measured via telemetry.

Strength measurement and S:W manipulation

The analog torque output of the dynamometer was imported into a data acquisition system (BIOPAC MP150, Biopac Systems, Inc., Goleta, CA, USA), was sampled at a rate of 1000 Hz, and smoothed by taking the mean of overlapping 50 sample epochs using the system’s software (BIOPAC AcqKnowledge, Biopac Systems, Inc., Goleta, CA, USA). Subjects performed four maximal, isometric knee extensions, for each leg, for 5 s, with 30 s of recovery between efforts. The first two trials were used to refamiliarize the participant with the required maximal effort, and peak torque was recorded as the average of the last two trials. Peak torque was averaged between contractions and legs, and divided by body mass to obtain the unilateral knee extensor S:W (Nm kg⁻¹). A BWU algorithm was developed to determine the amount of body weight reduction needed to increase S:W in +0.1 Nm kg⁻¹ increments as follows:

The process was repeated for each subject to elicit the four S:W conditions. The weight reduction needed to manipulate S:W was provided by a pneumatic BWU system (PneuWeight, Pneumex, Sandpoint, ID, USA, Fig. ​Fig.1).1). The BWU system applied an unloading force (controlled by microprocessor) to a full-torso harness worn by participants while allowing the normal vertical excursion of the body throughout the gait cycle. While the calculation of unloading weight was based on knee extensor S:W, it was expected that BWU also affected S:W for other lower-extremity muscles in a similar manner. Prior to walking performance testing, the target body weight was verified using the treadmill’s integrated force plates. A sham unloading force of 9.8 N (1 kg) was used for the true S:W condition to partially blind the subjects to the specific S:W condition. Table ​Table22 provides the target S:W, target body weight, and target unloading weight for each of the four S:W conditions by group.

Selection of maximal and preferred walking speed

Immediately prior to the BWU walking trials, with no weight support, maximal walking speed was determined by increasing treadmill speed by 0.015 m s⁻¹ each second until the subjects reported that the speed was the fastest pace they could maintain for a short walk. Next, at each level of unloading, subjects began walking at a speed of 0.8 m s⁻¹, which based on our previous work was below the expected preferred walking speed. Preferred walking speed was then determined by increasing treadmill speed by 0.015 m s⁻¹ each second until the subjects reported that the speed felt faster than they normally walked down the street to the store. Next, the speed was decreased in a similar manner until the subjects reported that the speed was slower than they walked down the street to go to the store. The process was repeated three times, and the three slow speeds and three fast speeds were averaged to obtain preferred walking speed. This method of preferred treadmill speed assessment has been shown to relate well to preferred overground walking speed. (Marques et al. 2013)

Gait measures

To measure gait performance at each S:W condition, participants walked on a motorized, instrumented treadmill (Gaitway II, Kistler Instrument Corp., Amherst, NY, USA) that was positioned directly beneath the unloading system (Fig. ​(Fig.1).1). The treadmill used in-deck force plates to record vertical ground reaction force (vGRF) and center of pressure data. These were then used by the treadmill software (Gaitway v. 2.0.8.50, Kistler Instrument Corp., Amherst, NY, USA) to determine kinetic, spatial, and temporal gait variables as described previously with very good reliability (LaRoche et al. 2011b; LaRoche et al. 2012). Kinetic variables included weight acceptance peak force (first peak of the vGRF) and push-off peak force (second peak of the vGRF); spatial variables included stride length and stride width; and temporal variables included stride frequency, single-limb support time, and double-limb support time. Each gait parameter was calculated independently for right and left feet over 20, sequential steps (ten strides), and the mean of these steps was used for analysis.

EMG

The surface electromyogram was used to assess the activation of the ankle plantarflexors and knee extensors. The skin was prepared to minimize impedance, and silver silver-chloride electrodes (Meditrace 530, Tyco Healthcare, Mansfield, MA, USA) were placed over the vastus lateralis and gastrocnemius lateralis of both legs according to Surface ElectroMyoGraphy for the Non-Invasive Assessment of Muscles (SENIAM) standards. The EMG signal was obtained from the skin using preamplifiers (BN-EMG2, Biopac Systems, Inc., Goleta, CA, USA) at a gain of 2000× and was recorded by the data acquisition system at 1000 Hz. The signal was then bandpass filtered between 20 and 500 Hz, rectified, and integrated every 20 samples. The peak of the integrated EMG signal was recorded for each leg during the maximal speed walking trial and all subsequent EMG data were normalized as percent of peak EMG at maximal speed, and subsequently under each S:W condition were normalized to velocity to account for differences in walking speed between groups and conditions (Hof et al. 2002). The peak integrated EMG from the first five strides of the recording, for each leg, was obtained and averaged prior to statistical analysis.

Walking speed

Knee extensor S:W was significantly correlated with the preferred overground walking speed obtained from the SPPB (Fig. ​(Fig.2A).2A). The interaction between the BMI group and S:W condition for preferred walking speed was not significant (p = 0.645). However, a significant main effect of group (p = 0.028) showed that the obese group preferred speeds 20 % slower than normal weight (p = 0.011) and 16 % slower than overweight (p = 0.043) when averaged across S:W conditions (Fig. ​(Fig.2B).2B). There was no effect of S:W condition on preferred walking speed for any group (p > 0.5).

Open in a separate window

Fig. 2

Relationship between knee extensor strength-to-weight ratio (S:W) and preferred overground walking speed in a cross-sectional comparison among older adult participants (a) and comparing changes in preferred walking speed by weight group in response to acute increases in S:W of +0.1 Nm kg⁻¹, +0.2 Nm kg⁻¹, and +0.3 Nm kg⁻¹, which were elicited by body weight unloading (b). Panel B data are mean ± standard error. a significant main effect of group, different between normal weight and obese, p < 0.05; b significant main effect of group, different between overweight and obese, p < 0.05

Kinetic gait measures

When vGRF kinetic data were expressed in absolute terms (Newtons), the obese group demonstrated 42 % greater weight acceptance peak force (p = 0.001) and 46 % greater push-off peak force (p < 0.001) than normal weight (Fig. ​(Fig.3).3). As S:W increased, obese subjects had a greater reduction in push-off peak force (p = 0.006) than the other groups (Fig. ​(Fig.3).3). Across groups, there was a linear decline in weight acceptance peak force (15 %) and push-off peak force (23 %) from the normal S:W to the +0.3 Nm kg⁻¹ condition (all p < 0.001). When kinetic data were expressed in relative terms (body weights), a significant interaction (p = 0.015) showed that an increase in weight acceptance peak force occurred in overweight (3 %) and obese groups (6 %), as S:W was increased, that did not occur in the normal weight group. Across groups, push-off peak force relative to weight decreased by 9 % as S:W increased (p < 0.001).

Open in a separate window

Fig. 3

Walking vertical ground reaction forces expressed in absolute terms (Newtons, N, left panels) and relative terms (body weights, BW, right panels) at the normal knee extensor strength-to-weight ratio (S:W) and at S:W +0.1 Nm kg⁻¹, +0.2 Nm kg⁻¹, and +0.3 Nm kg⁻¹ for normal weight (a), overweight (b), and obese (c) older adults. Data are displayed as means

Spatiotemporal gait measures

For the spatial gait variables of stride length and stride width, there were no significant interactions, BMI group effects, or S:W condition effects (p > 0.10). When evaluating temporal gait measures, no significant BMI group by S:W condition interactions existed (p > 0.50). Obese subjects had 8 % slower stride frequency than normal weight (p = 0.025) and 12 % slower frequency than overweight (p = 0.001) (Fig. ​(Fig.4A).4A). Single-limb support time was 7 % less in obese compared to normal weight (p < 0.001) and 6 % less than overweight (p < 0.001) (Fig. ​(Fig.4B).4B). Double-limb support time was 27 % greater in obese compared to normal weight (p < 0.001) and 22 % greater than overweight (p = 0.001) (Fig. ​(Fig.4C).4C). When all participants progressed through the S:W conditions, single-limb support time increased 4 % (p < 0.001) and double-limb support time decreased 17 % (p < 0.001).

Open in a separate window

Fig. 4

Temporal gait data showing the effect of increasing knee extensor strength-to-weight ratio (S:W) by +0.1 Nm kg⁻¹, +0.2 Nm kg⁻¹, and +0.3 Nm kg⁻¹ on stride frequency (a), single-limb support time (b), and double-limb support time (c) for normal weight, overweight, and obese older adults. GC gait cycle. Data are mean ± standard error. a significant main effect of group, different between normal weight and obese, p < 0.05; b significant main effect of group, different between overweight and obese, p < 0.05; *p < 0.05, significant change from S:W condition

Lower-extremity forces and preferred walking speed

The increase in knee extensor S:W used in this study was comparable to the knee extensor strength gains (0.3–0.4 Nm kg⁻¹) that occur as a result of resistance exercise programs in older individuals (Suetta et al. 2008; Cannon et al. 2007; Knight and Kamen 2001). However, when the range of S:W elicited by our BWU model is compared to the range of S:W possessed by participants in the study (nearly 2.0 Nm kg⁻¹, see Fig. ​Fig.2A)2A) the interventional increase in S:W may have been inadequate to affect preferred walking speed. Regardless, there is limited research identifying the magnitude of strength gain necessary to elicit improved walking performance in older adults. A recent review by Beijersbergen et al. (2013) demonstrated that in strength training studies, only 21 % of the variance in improved gait speed is explained by leg muscle strength, and concluded that the biomechanical mechanisms of how improved strength capacity transfers to walking performance are unclear. A possible lesson learned from this study is that knee extensor S:W may need to be increased by more than 0.3 Nm kg⁻¹ to affect the preferred walking speed of older adults.

The vGRF is related to the sum of the hip, knee, and ankle extensor torques, represents the net propensity of the lower-limb to push away from the ground, and constitutes the largest component of the ground reaction force vector (Winter 1980). The vGRF acts in opposition to gravity to support body weight, its peak value relates to knee and ankle S:W in older adults, and the vGRF increases linearly with walking speed (LaRoche et al. 2011b). We theorized participants would maintain a constant muscular effort across S:W conditions, vGRF relative to body weight would increase as weight was reduced, and gait speed would increase concurrently. In contrast, the strategy that was employed by this study’s participants in response to BWU was to reduce lower-limb forces while maintaining a constant speed, despite having the capacity to walk faster. As knee extensor S:W was increased via BWU, the absolute vGRF declined in proportion to weight reduction (Fig. ​(Fig.33 left panels) resulting in nearly identical vGRF relative to body weight across conditions (Fig. ​(Fig.33 right panels). Maintenance of an equivalent walking speed in the presence of reduced lower-extremity strength demand suggests that other factors, such as habit, balance confidence, effort, and cardiorespiratory capacity, may mediate preferred walking speed in older adults.

Muscle activation

Peak rectified EMG increases linearly with walking pace and thus speed must be considered when comparing EMGs obtained from different walking conditions (Hof et al. 2002). The muscle activation data in this study were normalized to speed to allow comparison between S:W conditions and BMI groups as walking speed differed between them. As such, the data portray what proportion of peak activation was utilized to walk at a speed of 1.0 m s⁻¹. With this interpretation, it is apparent that obese individuals require a higher level of motor unit activity of the lower-extremity muscles to walk. The consequence of compensating for low S:W through greater muscle activation is that muscular force cannot be sustained as high-threshold motor units fatigue more readily (Maffiuletti et al. 2007). Thus, by maintaining walking speed across S:W conditions, our older participants reduced muscle activation to a level that may help limit fatigue during walking, and appear to have prioritized reduced muscular effort over increased speed. This phenomenon could partially explain the lack of increased walking speed, despite strength gain, seen in some prospective resistance exercise studies.

Similar to the present study, Fischer and colleagues (Fischer et al. 2015) demonstrated approximately 30 % lower activation of tibialis anterior, lateral gastrocnemius, and vastus lateralis muscles in healthy, young participants when walking with 15 % BWU when speed was held constant. During BWU treadmill walking in older women, Thomas et al. (Thomas et al. 2011) showed no difference in vastus lateralis or biceps femoris activation between walking at comfortable speed (mean = 0.92 m s⁻¹) with 0 % body weight support and at fast walking speed (mean = 1.13 m s⁻¹) with 40 % body weight support. Thomas’ findings highlight the importance of normalizing EMG data to speed as these results would likely have demonstrated a lower level of activation at the faster speed, 40 % BWU condition.

Spatial and temporal gait parameters

Walking speed is the product of stride frequency and stride length, both of which have been shown to decrease with age and obesity (Lai et al. 2008; Samson et al. 2001; LaRoche et al. 2011a). Increasing S:W via BWU did not affect stride frequency or stride length in any group, explaining the lack of change in walking speed. These findings are in agreement with Thomas et al. who showed no change in stride length or stride frequency with either 20 or 40 % of body weight unloading (Thomas et al. 2007a), yet van Hedel et al. showed a decrease in stride frequency and increase in stride length, but only at the most extreme level of BWU (75 % of body weight) (van Hedel et al. 2006). The maximal BWU in this study was only 18 % of body weight, and the lack of effect on these spatiotemporal variables may have occurred since the BWU model directly affected S:W, but did not affect inertial properties of the limbs. It is interesting to consider that the mass of the limbs, their moments of inertia, and the muscular work needed to accelerate them may explain the slower stride frequency seen in the obese subjects (Browning et al. 2007), and perhaps is another reason why when body weight was unloaded participants selected equivalent preferred walking speeds.

Temporal gait parameters are influenced by age, body anthropometrics, and walking speed. Double-limb support time occupies a greater proportion of the gait cycle when comparing old to young walkers (Winter et al. 1990), elderly fallers to non-fallers (Mbourou et al. 2003; Maki 1997), and when comparing obese to normal weight walkers (LaRoche et al. 2011a; Lai et al. 2008; Browning and Kram 2007). Spending more time in double-limb support is a strategy that may increase stability and allow a greater extent of the work of weight support to be shared between limbs. As some of the burden of weight support was removed to increase S:W, double-limb support time decreased and single-limb support time increased, mitigating a portion of the age-associated and obesity-associated effects on these variables. These gait cycle shifts occurred independent of walking speed as it remained consistent across levels of S:W. This suggests that qualitative improvements to older adults’ gait, such as improved gait timing and balance control, could occur with increased S:W in the absence of increased speed. The changes in gait timing under BWU conditions are similar to those reported by van Hedel et al. (van Hedel et al. 2006) and Fischer et al. (Fischer and Wolf 2015) and suggest that deficits in the capacity for single-limb support may be partially mediated by S:W. Single-limb and double-limb support times are sensitive to changes in S:W, making these temporal gait variables potentially useful markers of gait quality in older adults.

Practical applications

The lower muscle activation and metabolic cost of BWU treadmill walking (Thomas et al. 2007a) may make it an appropriate adjunct exercise therapy for frail or obese older adults with low S:W, poor balance, mobility limitation, or orthopedic limitations who have difficulty walking for more than a few minutes without assistance. While a multitude of studies have used BWU treadmill walking after stroke, spinal cord injury, or in patients with cerebral palsy, multiple sclerosis, or Parkinson’s disease, only a few recent studies have used BWU to combat the age-associated decline in mobility. Thomas et al. successfully used a BWU treadmill walking intervention to improve the maximal overground walking speed of older women by 13 % in comparison to a 3 % increase in controls (Thomas et al. 2007b), and Peterson and colleagues similarly demonstrated an 18 % increase in usual gait speed and 14 % increase in rapid gait speed compared to baseline measurements (Peterson et al. 2014). While reduced muscle activation with BWU may limit exertion and prolong endurance, lower activation could lead to neuromuscular detraining if it is used frequently in place of unsupported walking.

Although the prevalence of osteoarthritis was not evaluated in the current study, obese older adults are seven times more likely to have osteoarthritis in the lower extremity (Ackerman and Osborne 2012) and could benefit from exercising with lower joint forces. In this study, the absolute weight acceptance peak force and push-off peak forces were reduced by 13 and 23 %, respectively, in the obese group at the highest level of BWU (18.3 kg reduction, 18 % body weight). The reduced ground reaction forces that occur with BWU should elicit lower skeletal loads, possibly contributing to improved exercise tolerance in older adults with osteoarthritis or other orthopedic limitations. In fact, Messier and colleagues report that for each 9.8 N weight loss (1 kg), compressive forces at the knee decrease by 40 N, a factor of 1:4 (Messier et al. 2005).

D.P. LaRoche was supported by the National Center for Advancing Translational Sciences of the National Institutes of Health under award number L30 TR000588-02. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

N.R. Marques was supported by the Fundação de Âmparo à Pesquisa do Estado de São Paulo (FAPESP) process number: 2011/11639-7

M.H Morcelli-Gotardo was supported by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) process number: BEX 17264/12-0