Abstract
Mediolateral stability during walking requires active control and is complex. Step width, a proxy for stability, follows a curvilinear relationship as gait speeds increase. However, despite the complexity of maintenance for stability, no study has yet investigated the variation across individuals of the relationship between speed and step width. The purpose of this study was to determine if variation between adults affects the estimation of the relationship between speed and step width. Participants walked on a pressurized walkway 72 times. Gait speed and step width were measured within each trial. Mixed effects models assessed the relationship between gait speed and step width, and the variability in the relationship across participants. The relationship between speed and step width followed a reverse J-curve on average, but the relationship was moderated by participants’ preferred speed. Step width response as speed increases is not homogenous in adults. This finding suggests that “appropriate” stability moderation (tested across a range of speeds) differs as a function of an individual’s preferred speed. Mediolateral stability is complex, and further research to elucidate individual factors contributing to variation is needed.
Similar content being viewed by others
Introduction
Maintaining mediolateral stability is necessary to sustain an upright posture during walking1,2,3. Mediolateral stability during steady-state unperturbed walking can be measured by step width2,4,5, which does not linearly relate to walking speed6 but instead follows a reverse J-curve; step width is high at relatively slow and fast speeds and is lowest near preferred speeds2. Despite the nonlinear, complex relationship between speed and step width, step width continues to be used in rehabilitation settings as a simple marker for gait health. Therefore, it is important to understand how step width varies across people.
The purpose of this study was to determine how variability between adults affects the estimation of the relationship between speed and stability. We hypothesized that between-participant variability would not affect the within-participant relationship between speed and stability in adults.
Methods
Participants
Students aged 19–35 from Auburn University were recruited via word-of-mouth and using a college-wide online study recruitment site that offered course extra credit. Data from these participants have been published previously7. Participants did not self-report any lower extremity bone fractures, muscle strains, or joint dislocations in the past six months, nor any chronic pain that affected walking such as pain associated with patellar tendonitis or Osgood-Schlatter Disease. All participants provided written informed consent by reading and signing a consent form approved by the Institutional Review Board of Auburn University. All methods were conducted in accordance with the protocol approved by Auburn University’s Institutional Review Board.
Procedures
As detailed previously, testing occurred on a 5.8-m-long pressurized walkway (GAITRite, CIR Systems Inc., Clifton, NJ) centered within a 12.1-m laboratory space, and participants began and ended each walking trial at least one stride before and after the walkway7. The same research assistant provided 24 different prompts used to elicit a range of walking speeds, participants walked 3 times per prompt, and the order of prompts were randomized across participants (for the list of prompts used, see Brinkerhoff et al.7). The prompt was repeated prior to each walking trial. Therefore, we collected and analyzed kinematic data during 72 walking trials per participant.
Data analysis
Speed and step width were exported from GAITRite software. Figure 1 details how speed and step width were calculated within the GAITRite software. Speed was taken as an average per trial. Step width was measured separately for each foot and average per trial (144 step width observations across 72 speeds, per participant). In total, 15 trials were missing from the sample, across 5 participants. Each participant’s preferred speed was determined as the average of their speeds from the prompts, “walk at your normal speed,” and, “walk at your typical speed” were averaged (6 per participant).
Previously, we explored the effect of prompt and individual variance on gait speed using mixed effects models7. Here, we explored the effect of gait speed on step width using mixed effects models in R8,9,10. Five separate mixed effects models were built to predict step width and were compared for fit to the data using the AIC, such that a smaller AIC by at least 2 units indicated a better fit to the data11. We disaggregated between- and within-participant variation to investigate whether gait speed, as a percent of a participant’s preferred speed, interacted with a participant’s preferred speed to predict step width. To calculate gait speed as a percent of a preferred speed, speed was group-centered within participant on a participant’s preferred speed, and then normalized by calculating trial speeds as percent of preferred speed. Therefore, a participant’s preferred speed was estimated as 0% and speeds above or below their preferred were estimated as a positive or negative percent of preferred speed, respectively. For ease of viewing the table and figure, 100 was added to each speed value so that preferred speed was indicated by 100%.
Results
Fifty-six adults participated in this study (33 females). Mean age was 21 (SD 3) years and ranged from 19 to 35. Since only two participants were over the age of 25, we ran the same analyses with and without these two participants. The best-fitting model remained the same regardless of inclusion of the older participants. Therefore, we kept all 56 participants in the analysis. Mean leg length was 94 (SD 6) cm, and mean mass was 71.5 (SD 17.5) kg. Mean preferred speed was 1.31 (SD 0.14) m/s and mean speed across all trials was 1.37 (SD 0.32) m/s. Mean step width across all trials was 8.11 (SD 3.93) cm.
The first model included a fixed effect for the mean intercept at percent of preferred speed and a random intercept by participant. The second model was cross-level and built upon the first by adding a fixed interaction between the linear slope of percent of preferred speed and preferred speed; this model tested whether the effect of relative speed on step width was moderated by preferred speed. The second model fit the data better than the first (X2(3) = 71.644, p < 0.001). The third model built upon the second by adding a fixed quadratic effect of speed, to test if the within-participant relationship between speed and step width was curvilinear. The third model fit the data better than the second (X2(1) = 110.46, p < 0.001). The fourth model built upon the third by adding a random slope by participant for the percent of preferred speed. The fourth model fit the data better than the third (X2(2) = 146.39, p < 0.001). The variance in the random intercept and the variance in the random slope were free to correlate. The fifth model added fixed covariates for sex and leg length, to determine if these factors affected step width. The fifth model do not fit the data better than the fourth (X2(2) = 0.890, p = 0.641). The results of each model are provided in Table 1. The intra-class correlation for the fourth and best-fitting model suggested that between-participant differences explained 61% of the variance in step width. The fourth model residuals approximated a normal distribution (upon visual inspection)12 based on the QQ plots and plots of the residuals.
According to the fourth model, the estimated mean step width at participants’ preferred speed was 8.02 cm. There was a significant negative linear slope such that as speed increased by 1% of preferred speed, step width decreased by 0.16 cm (p < 0.001). There was a significant quadratic slope such that the relationship between speed and step width was convex (Β = 0.0003, p < 0.001). The negative tilt and convex curve resulted in a reverse J-curve (Fig. 2A). However, the linear slope between percent of preferred speed and step width was moderated by participants’ preferred speed such that those with slower preferred speeds had a larger negative slope between speed and step width, and those with faster preferred speeds had a smaller negative slope (Β = 0.12), p < 0.001) (Fig. 2B). There was significant random variance in the intercept and in the linear slope across participants, and the random variances were correlated (r = −0.59).
Discussion
The purpose of this study was to determine whether a reverse J-curve fit the relationship between gait speed and step width—as a measure of stability—and whether gait speed interacted with a participant’s overall preferred speed in predicting step width. The negative tilt and convex shape of the relationship between percent of preferred speed and step width substantiates that the within-participant relationship between speed and step width, on average, follows a reverse J-curve2 and reaches a minimum around 150% of preferred speed (Fig. 2A). Mediolateral stability is dependent on sensory input, and when instability is recognized, step width tends to increase13,14. It may be that at gait speeds higher than 150% of preferred speed, sensory information about mediolateral stability is reduced (or noise is increased), resulting in increased step width in order to ensure swing foot clearance and mediolateral stability13. While the within-participant relationship between speed and step with is a reverse J-curve, between-participant variability moderates this relationship.
Figure 2B displays the participant-level data in order of increasing preferred speed and demonstrates the model interaction between preferred speed and the linear effect of relative speed. At slower preferred speeds, the relationship between speed and step width follows a reverse J-curve. As preferred speed increases, the slope flattens out and then becomes positive such that as relative speed increases, step width increases. The consequence of this moderation is that some participants reach a minimum step width at their fastest relative speeds (e.g., preferred = 1.22 m/s), some do not change their step width as relative speed changes (e.g., preferred = 1.27 m/s), and some reach a minimum at their slowest relative speeds (e.g., preferred = 1.72 m/s). The considerable variation across people in this sample indicates that “appropriate” stability moderation differs across people as a function of their preferred speed. Crucially, the mixed effects models used in this design allowed us to tease out the within and between subject variability.
Our results support the notion that the relationship between speed and mediolateral stability is complex3,15. Future studies should explore what individual factors contribute to preferred speed and what individual factors affect the relationship between walking speed and mediolateral stability in adults, such as mediolateral pelvis mechanics16 or mediolateral and anteroposterior arm swing17,18. Future studies may also investigate this relationship in different populations, such as older adults, and with external constraints, such as while dual-tasking or in dim lighting, to determine the effect of stability impairments and task constraints on the variation in the relationship between preferred speed, relative speed and step width.
Data availability
The data sets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.
References
Patla, A. E., Adkin, A. & Ballard, T. Online steering: Coordination and control of body center of mass, head and body reorientation. Exp. Brain Res. 129, 0629–0634. https://doi.org/10.1007/s002210050932 (1999).
Helbostad, J. L. & Moe-Nilssen, R. The effect of gait speed on lateral balance control during walking in healthy elderly. Gait Posture 18, 27–36. https://doi.org/10.1016/S0966-6362(02)00197-2 (2003).
Fettrow, T. et al. Interdependence of balance mechanisms during bipedal locomotion. PLoS ONE 14, e0225902. https://doi.org/10.1371/journal.pone.0225902 (2019).
Brach, J. S., Berlin, J. E., VanSwearingen, J. M., Newman, A. B. & Studenski, S. A. Too much or too little step width variability is associated with a fall history in older persons who walk at or near normal gait speed. J. NeuroEngineering Rehabil. 2, 21. https://doi.org/10.1186/1743-0003-2-21 (2005).
Conradsson, D., Paquette, C. & Franzén, E. Medio-lateral stability during walking turns in older adults. PLoS ONE 13, e0198455. https://doi.org/10.1371/journal.pone.0198455 (2018).
Collins, S. H. & Kuo, A. D. Two independent contributions to step variability during over-ground human walking. PLoS ONE 8, e73597. https://doi.org/10.1371/journal.pone.0073597 (2013).
Brinkerhoff, S. A., Murrah, W. M., Hutchison, Z., Miller, M. & Roper, J. A. Words matter: Instructions dictate “self-selected” walking speed in young adults. Gait Posture https://doi.org/10.1016/j.gaitpost.2019.07.379 (2019).
Bates, D., Maechler, M., Bolker, B. & Walker, S. Fitting linear mixed-effects models using lme4. J. Stat. Softw. 67, 1–48. https://doi.org/10.18637/jss.v067.i01 (2015).
R Core Team, R: A language and environment for statistical computing. R Foundation for Statistical Computing (2020). https://www.R-project.org/.
RStudio Team, RStudio: Integrated Development Environment for R (2021). http://www.rstudio.com/.
Burnham, K. P. & Anderson, D. R. Model selection and multimodel inference: A practical information-theoretic approach. J. Wildl. Manag. 67, 655. https://doi.org/10.2307/3802723 (2003).
Pinhiero, J. & Bates, D. Mixed-effects models in S and S-PLUS (Springer, 2006).
Bauby, C. E. & Kuo, A. D. Active control of lateral balance in human walking. J. Biomech. 33, 1433–1440. https://doi.org/10.1016/S0021-9290(00)00101-9 (2000).
Latt, M. D., Menz, H. B., Fung, V. S. & Lord, S. R. Walking speed, cadence and step length are selected to optimize the stability of head and pelvis accelerations. Exp. Brain Res. 184, 201–209. https://doi.org/10.1007/s00221-007-1094-x (2007).
Donelan, J. M., Kram, R. & Kuo, A. D. Mechanical work for step-to-step transitions is a major determinant of the metabolic cost of human walking. J. Exp. Biol. 205, 3717–3727 (2002).
Stimpson, K. H., Heitkamp, L. N., Horne, J. S. & Dean, J. C. Effects of walking speed on the step-by-step control of step width. J. Biomech. 68, 78–83. https://doi.org/10.1016/j.jbiomech.2017.12.026 (2018).
Punt, M., Bruijn, S. M., Wittink, H. & van Dieën, J. H. Effect of arm swing strategy on local dynamic stability of human gait. Gait Posture 41, 504–509. https://doi.org/10.1016/j.gaitpost.2014.12.002 (2015).
Ortega, J. D., Fehlman, L. A. & Farley, C. T. Effects of aging and arm swing on the metabolic cost of stability in human walking. J. Biomech. 41, 3303–3308. https://doi.org/10.1016/j.jbiomech.2008.06.039 (2008).
Acknowledgements
The authors would like to thank Wendi Weimar and the Sports Biomechanics Laboratory for the use of their GAITRite Instrumented Walkway. The authors would also like to thank Matthew Beth, Jaycee Ferguson, Rachel Hall, Brandon Merrill, Trey Moncrief, and William Tucker for their help.
Author information
Authors and Affiliations
Contributions
Study conception and design: S.A.B. and J.A.R. Data analysis: S.A.B. Interpretation: S.A.B., W.M.M., and J.A.R. Drafting manuscript: S.A.B. Critical revision: S.A.B., W.M.M., and J.A.R. All authors read and approved the final manuscript.
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing interests.
Additional information
Publisher's note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Brinkerhoff, S.A., Murrah, W.M. & Roper, J.A. The relationship between gait speed and mediolateral stability depends on a person's preferred speed. Sci Rep 13, 6056 (2023). https://doi.org/10.1038/s41598-023-32948-z
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41598-023-32948-z
- Springer Nature Limited