Quantifying Cause-Effect Relations Between Walking Speed, Propulsive Force, and Metabolic Cost

Walking speed is a useful surrogate for health status across the population. Walking speed appears to be governed in part by propulsive force (FP) generated during push-off and simultaneously optimized to minimize metabolic cost. However, no study to our knowledge has established empirical cause-effect relations between FP, walking speed, and metabolic cost, even in young adults. To overcome the potential linkage between these factors, we used a self-paced treadmill controller and real-time biofeedback to independently prescribe walking speed or FP across a range of condition intensities. Walking with larger and smaller FP led to instinctively faster and slower walking speeds, respectively, with about 80% of variance explained between those outcomes. We also found that comparable changes in either FP or walking speed elicited predictable and relatively uniform changes in metabolic cost, each explaining about ~53% of the variance in net metabolic power and ~15% of the variance in cost of transport, respectively. These findings build confidence that interventions designed to increase FP will translate to improved walking speed. Repeating this protocol in other populations may identify additional cause-effect relations that could inform the time course of gait decline due to age and disease.


17
Walking speed serves as a simple surrogate for human health status. For example, faster 18 walking speeds associate with a host of positive health factors, including increased muscle 19 strength, better cognitive function, greater independence, and reduced healthcare costs. [1][2][3][4][5][6][7] By 20 understanding the mechanistic pathways that contribute to slower walking speeds we may identify 21 avenues to maintain and restore independence and pedestrianism for safe and effective recreation, 22 transport, and health in our population. 23 We often attribute the selection of walking speed to the minimization of metabolic cost. 24 The cost of transport (CoT, i.e., net metabolic cost per unit distance traveled) during walking is U- 25 shaped, with increasing costs as walking speed deviates from preferred. This suggests that our 26 movement biomechanics and underlying muscle actions are tuned to minimize metabolic cost at 27 our preferred speeds. Unfortunately, compared to young adults or unimpaired controls, numerous 28 walking studies in older adults or people with gait limitations document higher CoT 8-12 and slower 29 preferred walking speeds. 11,12 A variety of factors likely explain the higher CoT in these 30 individuals, including systemic factors (e.g., reduced cardiopulmonary function), local muscle and 31 tendon factors (e.g., reduced muscle metabolic efficiency, lower tendon stiffness), and altered 32 neural control or gait biomechanics (e.g., wider steps, increased co-activation, redistributing 33 mechanical work to more proximal leg joints/muscles). However, the often-simultaneous 34 presentation of slower speeds and higher CoT challenges our ability to fully understand the time 35 course of gait decline due to aging or gait pathology. 36 Biomechanically, walking speed is understood to be regulated by the magnitude of the 37 peak anterior component of the ground reaction force 13 -namely, the peak propulsive force (FP). 38 We generate FP from the trailing limb during push-off, which acts to accelerate and redirect the 39 body's center of mass forward and upward, thereby regulating walking speed. Humans typically 40 generate a vigorous FP via ankle plantarflexion using a combination of well-timed calf muscle 41 contraction, elastic energy returned from the Achilles tendon, and effective limb orientation for 42 mechanical advantage. 14-18 Because the ankle plantarflexor muscles and tendons account for about 43 60% of the work performed in typical gait, 14,19 it is no anomaly that plantarflexor pathologies affect 44 both walking speed and walking economy. 8,11,20 This suggests that FP, walking speed, and 45 metabolic cost are inextricably linked, posing a longstanding scientific challenge with significant 46 potential for improved clinical countermeasures.

47
Before scientists and clinicians can design and implement strategies to improve walking 48 speed and lower metabolic cost in older adults or in individuals with gait pathology, we need to 49 better understand exactly how FP impacts walking speed in the context of walking metabolic cost.

50
To our knowledge, no study has established empirical cause-effect relations between FP, walking 51 speed, and walking metabolic cost, even in unimpaired young adults. Thus, our purpose was to:

52
(1) investigate how FP governs the selection of walking speed, and (2) quantify how the selection 53 of FP or walking speed impacts walking economy. Exploring these relations may build confidence 54 that restoring FP will lead to improvements in walking speed. Additionally, our results may be 55 useful when designing interventions or devices that seek to improve walking ability and economy. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted October 19, 2021. ; https://doi.org/10.1101/2021.10.18.21265129 doi: medRxiv preprint M a n u s c r i p t

58
Participants 59 We recruited a convenience sample of twenty young adults who provided informed consent 60 prior to any activities in this study. The University of North Carolina at Chapel Hill institutional 61 review board approved all research procedures. All participants were free of current lower 62 extremity injuries, neuromuscular complications, and walking assistive devices that might prevent 63 protocol completion. On average, participants were 24.7 ± 5.2 years old (mean ± standard 64 deviation), stood at a height of 1.77 ± 0.11 m, had a mass of 75.6 ± 13.7 kg, and thus had a BMI 65 of 24.0 ± 3.4 kg/m 2 .

82
In some trials, we also used real-time visual biofeedback to display the average peak FP 83 from the previous two steps on a screen in front of the participant with a target line representing 84 the prescribed FP according to our study protocol (see below). We instructed participants to "match 85 their push off force to the target". The biofeedback line turned green when participants' 86 instantaneous FP for that step was within 5% of the target value (in newtons), but was otherwise

91
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted October 19, 2021. ; https://doi.org/10.1101/2021.10.18.21265129 doi: medRxiv preprint M a n u s c r i p t

100
Experimental Protocol 101 We determined participants' preferred walking speed via the average speed from four 30-102 m passes in a hallway where we instructed them to "walk normally, as if you were walking down 103 a sidewalk". Participants completed a 3-minute warm-up walk at their preferred walking speed 104 followed by a 3-minute walk to become familiar with the self-pacing treadmill mode and targeted 105 FP biofeedback. During the self-pace and targeted biofeedback warm-up, we ensured all 106 participants could increase and decrease their FP on command and could regulate their walking 107 speed at will using the self-pace mode.
108 Figure 2 summarizes our experimental protocol. Participants walked at a fixed speed (speed 109 clamp) for five 5-minute trials at their preferred speed (Norm) and at ±10% and ±20% of Norm in 110 randomized order. We extracted the average peak FP from each speed clamp trial to use as targets 111 for the ensuing biofeedback trials. Participants then completed a randomized series of 5-minute 112 walking trials with biofeedback to target their average peak FP from each of the speed clamp trials.

113
These targeted biofeedback trials used the self-paced treadmill controller, thus prescribing a target 114 FP while allowing walking speed to vary (aptly-named: FP clamp). Participants rested in a seated 115 position for at least 1 minute between trials, and were allowed to take longer rests ad libitum 116 (average time between trials: 167±105 s). 117 We recorded treadmill speed and ground reaction forces from the real-time treadmill CoT. When we found a significant main effect or interaction, we used Tukey's post-hoc tests to 149 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

182
Effects of Clamp Type 183 We found significant main effects of clamp type, where, on average across all condition 184 intensities, the FP clamp elicited 2.6% faster speeds, 1.4% greater FP, 8.9% higher net metabolic 185 power, and 6.2% greater CoT compared to the speed clamp (Fig. 5, p≤0.01, ɳp 2 ≥0.298). We also 186 found significant interactions between condition intensity and clamp type for walking speed and 187 FP. The interactions revealed that the difference between clamp types became larger with slower 188 speed and with larger FP (Fig. 5A-B). At the lowest condition intensity, participants walked faster 189 (0.07±0.03 m/s (mean difference ± standard deviation), p=0.008, d=0.887) and with higher net    Fig. 5A-B). Net 206 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted October 19, 2021. ; https://doi.org/10.1101/2021.10.18.21265129 doi: medRxiv preprint M a n u s c r i p t metabolic power increased and decreased along with changes in intensity for all conditions 207 (p≤0.049, d≥0.642) except when prescribing the -10% condition intensity for both clamp types 208 (p≥0.050, d≤0.639, Fig. 5C). Conversely, CoT only significantly increased from Norm when 209 prescribing +20% condition intensity for both speed and FP clamps (Fig. 5D, p≤0.013, d≥0.827).

211
Our goal was to objectively establish cause-effect relations between FP, walking speed, and 212 walking metabolism across a range of intensities through a unique experimental paradigm 213 designed to separately prescribe both FP and walking speed in a cohort of healthy young adults. Our secondary goal was to establish the metabolic consequences of the interplay between 237 FP and walking speed. We found two key outcomes regarding walking metabolism. First, we found 238 that net metabolic power was more strongly associated with changes in walking speed and FP 239 (R 2 ≈0.53, Fig. 4 B&D) compared to CoT (R 2 ≈0.15, Fig. 4 C&E), at least across a prescribed 240 change of ±20% in condition intensity. Previously, our group found that when walking at fixed 241 speeds, both net metabolic power and CoT (inferred from equivalent speeds) increased by ~20% 242 when targeting 20% larger FP and increased by ~30% when targeting 20% smaller FP. 34

243
Conversely, in this study, net metabolic power increased by ~40% and decreased by ~25% on 244 average when targeting 20% larger and smaller FP, respectively. Additionally, when participants 245 in this study could instinctively select their own walking speed, CoT increased by ~16% on average 246 when targeting 20% larger FP but did not differ from usual walking when targeting 20% smaller 247 FP. This relative lack of sensitivity of CoT to changes in condition intensity agrees with the "broad 248 minimum" theory, wherein a range of walking speeds neighboring the local minimum of the CoT 249 curve may share similar metabolic costs. 35 For normal-and lower-intensity conditions, our 250 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted October 19, 2021. ; https://doi.org/10.1101/2021.10.18.21265129 doi: medRxiv preprint M a n u s c r i p t participants adjusted FP or speed in a manner that maintained a relatively invariant CoT, and 251 thereby operated within their "broad minimum" CoT. Thus, under certain circumstances, walkers 252 may exploit the interaction between FP and walking speed to preserve or even reduce walking CoT. 253 Ultimately, changing FP or walking speed predictably alters net metabolic power but need not 254 impact CoT. 255 Second, in further support of the strong relation between FP and walking speed, we 256 discovered that changing the magnitude of either yielded relatively similar effects on measures of 257 walking metabolism. In other words, whether we prescribed a change in FP or walking speed, 258 effects on walking metabolic cost were nearly indistinguishable. We noted this also from our 259 correlations (Fig. 4), which revealed quantitatively similar R 2 values and regression coefficients 260 as well as qualitatively similar trendlines between clamp types. However, this is not to say that 261 there were no meaningful differences between clamp types. Indeed, we found a significant main 262 effect of clamp type on metabolic cost; FP-clamp trials tended to require 9% higher net metabolic 263 power and 6% CoT compared to speed-clamp trials on average (Fig. 5C-D). We have previously 264 shown that walking with FP biofeedback at a fixed treadmill speed does not itself exact a metabolic 265 penalty. 34 Nevertheless, we suspect that the greater metabolic energy cost associated with FP clamp 266 trials may arise from a cognitive "tax" levied to simultaneously regulate step-to-step adjustments 267 in response to biofeedback and to treadmill self-pacing. Such a tax may allude to additional 268 cognitive processing and/or neuromuscular costs associated with a shift toward supraspinal control 269 of walking patterns rather than primarily relying on central pattern generators in the spinal cord. Because FP and walking speed are inextricably linked, we would expect our protocol to 277 yield highly similar biomechanical and metabolic outcomes across both clamp types. Indeed, this 278 was true for most outcomes across most condition intensities (Fig. 5). However, when targeting 279 20% smaller than normal FP, our participants could have selected a slower walking speed and 280 lower net metabolic power when producing the requisite FP, but they chose not to. Rather, we 281 identified a naturally-emergent discrepancy between clamp types, in which participants exerted 282 similar FP, but selected faster speeds at a higher net metabolic power during the FP clamp than the 283 speed-clamp. The instinctive selection of faster speeds at a metabolic penalty despite an 284 indistinguishable FP across clamp types demonstrates that humans do not always seek to minimize 285 metabolic cost. In our daily lives, we may prioritize factors other than walking economy when we 286 rush, become excited, or feel threatened or scared. However, we can see evidence of this even in 287 laboratory environments. For example, healthy young adults sometimes select walking speeds 288 somewhat faster than their most economical speed, even though they spend more energy when 289 doing so. 35 In another example, young healthy subjects have been shown to prioritize stability 290 rather than take advantage of gravity-aided propulsion when walking down a gentle slope. 36 These 291 phenomena may explain the discrepancy we identified at low condition intensities, potentially 292 optimizing a cost function other than the most economical gait patterns. We plan to further 293 investigate how participants regulated their speed on a step-to-step basis when interacting with the 294 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted October 19, 2021. ; https://doi.org/10.1101/2021.10.18.21265129 doi: medRxiv preprint M a n u s c r i p t self-pace controller, and the role that lower extremity muscles serve in generating and regulating 295 FP.

296
Translational implications 297 It is unclear whether these direct relations between FP, walking speed, and walking 298 metabolism will hold in populations who may be candidates for clinical countermeasures to 299 enhance gait performance or mitigate walking-related fatigue. For example, older adults typically 300 walk slower, with smaller FP, and at higher metabolic costs compared to young adults. It is actually 301 not well known whether or not older adults have movement biomechanics and underlying muscle 302 actions that are tuned to minimize metabolic cost at their preferred speeds. Older adults also have 303 the capacity to generate propulsive forces comparable to those measured in younger adults, but 304 typically choose not to utilize that additional force capacity to increase walking speed. 23 Future 305 studies may consider enrolling older adults to participate in a similar design to determine whether 306 age influences the cause-effect relation between FP and walking speed, or if the metabolic 307 consequences of that relation are altered by hallmark changes in muscle morphology and 308 composition, cardiopulmonary function, sensorimotor integration, or executive processing.

310
One limitation in this study is that our FP magnitudes and walking speeds were limited to 311 a relatively small range compared to other studies that quantify walking metabolism. Our span 312 across condition intensities deviated 20% from typical gait, yielding walking speeds between 1 313 and 2 m/s. Our protocol was informed by the magnitude of changes we would deem clinically 314 meaningful. However, speed-dependent increases in CoT in otherwise healthy young adults do not 315 typically arise until ≤1 m/s 9,10 . Another limitation is that we analyzed average profiles over the 316 final 2 minutes of each 5-minute walking trial. Although subjects were provided an exploration 317 period and reported being comfortable with biofeedback and self-pacing, further practice could 318 have affected walking metabolic cost. Finally, we did not quantify individual determinants of 319 propulsive force (i.e., trailing limb angle and peak ankle moment 18,22 ).

321
Using a unique clamp protocol, we confirm a direct causal relation between FP and walking 322 speed. We provide strong empirical evidence that young adults walking with larger/smaller FP 323 yields faster/slower walking speeds, demonstrating that the peak anterior component of the ground 324 reaction force during push-off (FP) is an independent factor that governs walking speed. We also 325 quantify the metabolic implications of altering walking speed and FP, finding that comparable 326 changes in either FP or walking speed elicit predictable and relatively uniform changes in walking 327 metabolic cost.