 Research
 Open Access
 Published:
Combined effects of age and gender on gait symmetry and regularity assessed by autocorrelation of trunk acceleration
Journal of NeuroEngineering and Rehabilitation volume 11, Article number: 109 (2014)
Abstract
Background
The gait of a healthy person is believed to be more regular and symmetrical than those of an individual with a disease. Thus, symmetry and regularity are important indicators of human gait.
Methods
The effects of age and gender on gait symmetry and regularity were investigated in 87 Japanese participants by measuring trunk accelerometry during a 7m walk. The younger group included 26 female and 21 male students, and the elderly group included 24 females and 16 males. Average age for each group was 20 and 70 years, respectively. Gait symmetry and regularity were evaluated on the basis of autocorrelation functions of trunk accelerations of vertical and anteroposterior axes.
Results
The relationship between age and gait symmetry and regularity was statistically significant for both vertical and anteroposterior axes. Elderly participants showed lower symmetry and regularity in their gait than young participants. A significant gender effect was observed for the symmetry index of both axes but not for the regularity index. Male participants showed lower gait symmetry than females. An interaction effect between age and gender was significant in the symmetry index of anteroposterior acceleration. Gender effect was appeared more clearly in elderly than young participants.
Conclusion
Elderly participants showed a more asymmetrical and irregular gait than young participants. In addition to age, a significant gender effect was observed on gait symmetry. However, the effect size of gender was smaller than that of age, and it was not significant for gait regularity. The gait indices obtained by autocorrelation of trunk acceleration can be considered useful to evaluate aging effect on gait.
Background
Gait symmetry and regularity may vary with the individual health condition. Although a healthy person’s gait is more regular and symmetrical than the gait of an individual with a disease. Marked irregularity in gait has been reported in patients with Parkinson’s disease [1], and asymmetrical gait was observed in patients with limblength discrepancy [2], amputees [3], patients with hip osteoarthritis [4], and in patients with chronic stroke [5]. While these studies examined symmetry and/or regularity in patients with orthopedic, neurological, or cerebrovascular diseases, there is little information on the gait characteristics of the normal population.
Age is known to have an effect on gait, with young people walking faster, with longer steps and a higher step frequency compared with elderly people [6]. These changes in gait could be attributed to decline of physical and neurological functions with aging. Similarly, gait symmetry and regularity also are considered to depend on physical and neurological functions [7]; therefore, decreased symmetry and regularity can be expected in the elderly. Furthermore, previous studies have demonstrated different effect of age on some gait characteristics between males and females [8, 9]; thus an interaction of age and gender on gait symmetry and regularity can be assumed.
Gait symmetry and regularity can be calculated from various gait features including spatial [10, 11], temporal [10–12], kinetic [10, 13], or kinematic [14, 15] features. In the late 1990s MoeNilssen suggested a simple new approach to measure gait symmetry and regularity on the basis of autocorrelation of trunk acceleration [16, 17]. In recent years, the utility of this method in gait analysis has been expanded. The present study evaluated gait symmetry and regularity in the normal Japanese population and examined the effect of age and gender on the gait features by autocorrelation of trunk acceleration.
Methods
Participants
Data was gathered from 40 communitydwelling elderly people (24 female, 16 male) and 47 younger university students (26 female, 21 male), with average ages of 70 and 20 years for each group, respectively. The elderly participants were active enough to visit our experimental site in Kahoku City, Ishikawa, Japan. None of the elderly participants were inpatients of hospitals or residents of longterm health care centers. Participant’s age, height, and body mass data are shown in Table 1. The study was approved by the Institutional Ethical Committee of the Ishikawa Prefectural Nursing University, Japan. All the participants provided written informed consent after the aim and procedure of the experiment were explained to them.
Measurements and procedures
Acceleration while walking was measured using a triaxial accelerometer (AC301; GMS Inc., Tokyo, Japan), with a recording interval of 0.02 s (50 Hz). Acceleration was measured within a range of ±2 g. The accelerometer was tightly attached to the waist at the back (at approximately L3–L4). The participants were instructed to walk barefoot at their preferred speed for 7 m and return over a walkway set up in a gymnasium with wooden flooring. All participants walked the 7m walkway in approximately 5–8 s. The entire procedure was consecutively repeated twice. Thus, the participants walked the 7m course four times.
Autocorrelation function
In this study, gait indices were obtained from the autocorrelation function of a series of acceleration measurements. An autocorrelation function defines the correlation between the value at a time t = i and the value at t = i + j, with j representing a time lag. If the autocorrelation is obtained from a finite length signal, it attenuates as the lag increases. Autocorrelation values peak once for each step cycle, and therefore peak values depend on cadence. The autocorrelation function thus follows the equation:
where R(j) is the autocorrelation function with a timelag of j, v(i) is the ith data and N represents the sample size. In the case of j = 0, autocorrelation is calculated using the following equation:
MoeNilssen and Helbostada [18] refer to this equation as an unbiased autocorrelation. In this study, the autocorrelation was normalized by dividing the timelagged value by the R(0), according to the following the equation:
This normalized autocorrelation varies from −1 to +1, becoming 1 when there are no differences between the timelagged correlation and the static correlation. As the nonperiodic component was eliminated from the raw signal data, only a smooth fluctuation in the autocorrelation was detected. The autocorrelation indicated a cyclical change, with the period of the cycle coinciding with the step interval. The first peak of autocorrelation (Ad_{1}, Figure 1) indicates a correlation between steps and is therefore considered the symmetry index. In addition, because the second peak (Ad_{2}, Figure 1) represents a correlation between a stride and the next one, the second peak of the autocorrelation can be considered as the regularity index. Note that in previous studies [19–21], Ad_{1} and Ad_{2} were described as step regularity and stride regularity respectively. Moreover, regardless the axis (vertical, anteroposterior, or mediolateral) used for calculating the autocorrelation function, the symmetry index (Ad_{1} of the autocorrelation) means a correlation between the left and right steps.
The autocorrelation function for anteroposterior acceleration shows an almost similar pattern to that of vertical acceleration, suggesting that symmetry and regularity indices can in principle be obtained either way. However, an autocorrelation of mediolateral acceleration indicates a different pattern. Ad_{1} indicates negative peak, whereas Ad_{2} indicates a positive peak similar to the other axes. In this study, mediolateral acceleration was not used for analysis because it’s autocorrelation indicated unstable form in some cases. Thus, symmetry and regularity indices for the vertical (SIVt and RIVt) and for anteroposterior (SIAp and RIAp) accelerations were analyzed in this study.
The autocorrelation function was computed using data from the 4s mark (after an average of approximately 6 steps) in the middle of each 7m walk. The results of the four measurements were averaged for further analysis. The expected reproducibility of the average of the four measurements (ICC [1, 4]) were 0.860, 0.833, 0.818, and 0.692 for the SIVt, RIVt, SIAp, and RIAp, respectively. These ICC values were almost identical to those found in the previous study [22], thus the measurement of the present study considered to be reliable.
Statistical analysis
All statistical tests were performed using IBM SPSS Statistics 21.0 J for Windows (IBM Corp., Armonk, NY). Effects of age and gender were analyzed by twoway analysis of variance (ANOVA) with a type III sum of squares. The effects were considered significant at an alpha level of 0.05. The effect sizes (η^{2}; not partial η^{2}) were presented for the results of ANOVA. According to Cohen’s guideline [23], η^{2} values of 0.01, 0.06, and 0.14 were interpreted as small, medium, and large effects, respectively.
Results
The effects of age and gender on gait indices are shown in Figure 2 and Table 2. For both genders, young participants showed a higher symmetry and regularity than elderly participants for both vertical and anteroposterior axes. In young groups, male and female participants showed almost similar values of symmetry and regularity; however, in elderly groups, males showed lower symmetry and regularity compared with females. The results of twoway ANOVA are summarized in Table 3. For SIVt, both age and gender effects were significant (p = 0.022, η^{2} = 0.059 and p = 0.041, η^{2} = 0.046), although the combined effect of age and gender together was not statistically significant. The RIVt values were significantly affected by age (p = 0.001, η^{2} = 0.118), but the effect of gender and the interaction effect were not significant. The symmetry and regularity indices of anteroposterior acceleration (SIAp and RIAp) showed results similar to those for vertical acceleration. For SIAp, both age and gender differences were significant (p = 0.001, η^{2} = 0.238 and p = 0.017, η^{2} = 0.052), and the interaction effect between age and gender was statistically significant (p = 0.048, η^{2} = 0.034). The RIAp values were significantly affected by age (p = 0.001, η^{2} = 0.267), but the gender and the interaction effects were not significant.
Discussion
Gait symmetry and regularity of normal adults
Regarding gait indices of anteroposterior acceleration, elderly participants of this study showed approximate values of 0.80  0.86 for symmetry index and 0.84  0.88 for regularity index (see Table 2). Gait symmetry and regularity presented in this study were higher than those reported in the previous studies [7, 19–21]. A possible reason for higher symmetry/regularity in Japanese elderly population may be a difference in physical traits. Japan is known to have the lowest obesity prevalence among developed countries [24]. Most participants of the present study were nonobese (BMI < 30 Kg/m^{2}), with one exception of a young male participant (BMI = 32.6 Kg/m^{2}). Some studies demonstrated that obesity affects gait pattern [25] and balance ability [26]; thus it could be assumed that obesity also affects gait symmetry and regularity.
Effect of age on gait symmetry and regularity
In the present study, the symmetry index was affected by both age and gender, whereas the regularity index was affected only by age (see Figure 2 and Table 3). The effect sizes of age effect (approximately 0.060.27) were larger than those of gender effect (approximately 0.020.05). Previous research has reported a lack of significant effects of age on gait symmetry and regularity [27–30], opposite to the clear findings of this study (Table 3). Differences in age distribution of the sample population may possibly explain this discrepancy. Contrary to previous studies who investigated sample population continuous variation of age [27–30], the present study compared two groups of clearly different ages (20 and 70 years for the young and elderly group, respectively). Himann et al. [31] demonstrated that human gait function remains largely unchanged up to the age of 60 years, after which it is rapidly altered, explaining why studies considering a continuous age variation would fail to find a significant effect of age on gait. In fact, Kobsar et al. [18] after comparing gait symmetry of two distinguished age groups (24 and 72 years in average), found a significantly lower symmetry in elderly participants similarly to our results.
Kobsar et al. [18] also reported that a significant age effect was found only in the gait indices of anteroposterior acceleration but not in vertical and mediolateral accelerations. In this study, significant age effects were found in both vertical and anteroposterior axes. However, the effect sizes of age on the gait indices of anteroposterior acceleration were larger than those of vertical acceleration. In this respect, the present results agreed with those of Kobsar et al. [18]. However intraindividual reproducibility of gait indices of anteroposterior axis were inferior compared with those of vertical axis. Further studies are needed to determine which axis of trunk accelerometry is useful for evaluating gait symmetry and regularity.
Effect of gender on gait symmetry and regularity
In addition to age, gender also showed a significant effect on human gait. Females usually walk slower with shorter stride length than males. Ko et al. [32] showed a significant effect of gender on kinematic features of gait, including range of motion (ROM) of hip and ankle. However, gender effect on symmetry and regularity of gait has not been sufficiently clarified yet. Previous studies [27–30] have reported an insignificant effect of gender on gait symmetry and regularity, although the detailed procedures to calculating of the gait indices differed from the present study. Our results, on the other hand, do show a significant effect of gender on gait symmetry (Table 3). However, the effect sizes of gender on gait symmetry were smaller than those of age. For the normal adult population, age is the main factor affecting gait symmetry rather than gender.
Limitations
The short walking distance represents a limitation of this study. The participants walked 7 m and the autocorrelation function was calculated using the acceleration from the 4s mark in the middle of each 7m walk. The possibility that the analyzed data partially include unstable gait at the beginning and the end of the walk cannot be discarded. However, we believe that if the data would include an unstable portion of walking, the symmetry and regularity of the gait will indicate lower value. As mentioned before, the symmetry and regularity presented in this study were higher than those reported in previous studies. Furthermore, the reproducibility of the gait indices was not inferior to previously published results. This indicates that the possible effect of the unstable portion of walking on the presented results can be regarded as insignificant.
Conclusion
The present study investigated the effects of age and gender on symmetry and regularity of gait using autocorrelation function of trunk acceleration. Significant effects of age and gender on gait symmetry were demonstrated whereas the regularity index was affected only by age. The results suggest symmetry is more sensitive to gender difference than regularity. The results also suggest that age is the main factor affecting the gait indices rather than gender if variation in age is large enough. An interaction between age and gender was significant for SIAP; gender differences tended to be larger in elderly participants than in young participants. Although the interaction was not significant for SIVt, a potential combined effect of age and gender on the symmetry of gait can be presumed.
Abbreviations
 ANOVA:

Analysis of variance
 Ad_{1}:

First peak of the autocorrelation
 Ad_{2}:

Second peak of the autocorrelation
 BMI:

Body mass index
 ICC:

Intraclass correlation coefficient
 RIAp:

The regularity index for anteroposterior acceleration
 RIVt:

The regularity index for vertical acceleration
 SIAp:

The symmetry index for anteroposterior acceleration
 SIVt:

The symmetry index for vertical acceleration.
References
 1.
Paquet JM, Auvinet B, Chaleil D, Barrey E: Analysis of gait disorders in Parkinson's disease assessed with an accelerometer. Rev Neurol (Paris) 2003, 159: 786789.
 2.
Perttunen JR, Anttila E, Södergård J, Merikanto J, Komi PV: Gait asymmetry in patients with limb length discrepancy. Scand J Med Sci Sports 2004, 14: 4956.
 3.
Tura A, Raggi M, Rocchi L, Cutti AG, Chiari L: Gait symmetry and regularity in transfemoral amputees assessed by trunk accelerations. J Neuroeng Rehabil 2010, 7: 4.
 4.
Yamada M, Hirata S, Ono R, Ando H: The assessment of an abnormal gait by gait parameters derived from trunk acceleration in patients with osteoarthritis of the hip: comparison with healthy controls and criterionrelated validity. J Phys Ther Jap Assoc 2006, 33: 1421.
 5.
Helbostad JL, MoeNilssen R: Should trunk movement or footfall parameters quantify gait asymmetry in chronic stroke patients? Gait Posture 2008, 27: 552558.
 6.
Prince F, Corriveau H, Hébert R, Winter DA: Gait in the elderly. Gait Posture 1997,5(2):128135.
 7.
MoeNilssen R, Helbostad JL: Interstride trunk acceleration variability but not step width variability can differentiate between fit and frail older adults. Gait Posture 2005, 21: 164170.
 8.
Callisaya ML, Blizzard L, Schmidt MD, McGinley JL, Srikanth VK: Sex modifies the relationship between age and gait: a populationbased study of older adults. J Gerontol A Biol Sci Med Sci 2008,63(2):165170.
 9.
Oberg T, Karsznia A, Oberg K: Basic gait parameters: reference data for normal subjects, 10–79 years of age. J Rehabil Res Dev 1993,30(2):210223.
 10.
Kim CM, Eng JJ: Symmetry in vertical ground reaction force is accompanied by symmetry in temporal but not distance variables of gait in persons with stroke. Gait Posture 2003, 18: 2328.
 11.
Kodesh E, Kafri M, Dar G, Dickstein R: Walking speed, unilateral leg loading, and step symmetry in young adults. Gait Posture 2012, 35: 6669.
 12.
Hausdorff JM, Ring H: Effects of a new radio frequencycontrolled neuroprosthesis on gait symmetry and rhythmicity in patients with chronic hemiparesis. Am J Phys Med Rehabil 2008, 87: 413.
 13.
Bosch K, Rosenbaum D: Gait symmetry improves in childhood  a 4year followup of foot loading data. Gait Posture 2010, 32: 464468.
 14.
Goswami A: Kinematics quantification of gait symmetry based on bilateral cyclograms. In Proceedings of XIX Cong Int Soc Biomech. Edited by: Milburn P, Wilson B, Yanai T. Dunedin, New Zealand; 2003:3443.
 15.
Shorter KA, Polk JD, Rosengren KS, HsiaoWecksler ET: A new approach to detecting asymmetries in gait. Clin Biomech 2008, 23: 459467.
 16.
MoeNilssen R: A new method for evaluating motor control in gait under reallife environment conditions. Part 1: the instrument. Clin Biomech 1998, 13: 320327.
 17.
MoeNilssen R: A new method for evaluating motor control in gait under reallife environment conditions. Part 2: gait analysis. Clin Biomech 1998, 13: 328335.
 18.
MoeNilssen R, Helbostada JL: Estimation of gait cycle characteristics by trunk accelerometry. J Biomech 2004, 37: 121126.
 19.
Kobsar D, Olson C, Paranjape R, Hadjistravopoulos T, Barden JM: Evaluation of agerelated differences in stridetostride fluctuations, regularity and symmetry of gait using a waistmounted triaxial accelerometer. Gait Posture 2014, 39: 553557.
 20.
Dalton A, Khalil H, Busse M, Rosser A, van Deursen R, Ólaighin G: Analysis of gait and balance through a single triaxial accelerometer in presymptomatic and symptomatic Huntington’s disease. Gait Posture 2013, 37: 4954.
 21.
Yang M, Zheng H, Wang H, McClean S, Hall J, Harris N: A machine learning approach to assessing gait patterns for complex regional pain syndrome. Med Eng Phys 2012, 34: 740746.
 22.
Bautmans I, Jansen B, Van Keymolen B, Mets T: Reliability and clinical correlates of 3Daccelerometry based gait analysis outcomes according to age and fallrisk. Gait Posture 2011, 33: 366372.
 23.
Cohen J: Statistical Power Analysis for the Behavioral Sciences. 2nd edition. Hillsdale, NJ: Lawrence Erlbaum; 1988.
 24.
World Health Organization: Global database on body mass index. An interactive surveillance tool for monitoring nutrition transition. http://apps.who.int/bmi/index.jsp?introPage=intro_3.html
 25.
Ko S, Stenholm S, Ferrucci L: Characteristic gait patterns in older adults with obesity–results from the Baltimore longitudinal study of aging. J Biomech 2010,43(6):11041110.
 26.
Greve JM, Cuğ M, Dülgeroğlu D, Brech GC, Alonso AC: Relationship between anthropometric factors, gender, and balance under unstable conditions in young adults. Biomed Res Int 2013, 2013: 850424.
 27.
Auvinet B, Chaliel D, Barrey E: Accelerometric gait analysis for use in hospital outpatients. Rev Rhum (Engl Ed) 1999, 66: 389397.
 28.
Auvinet B, Berrut G, Touzard C, Moutel L, Collet N, Chaleil D, Barrey E: Reference data for normal subjects obtained with an accelerometric device. Gait Posture 2002, 16: 124134.
 29.
Senden R, Grimma B, Heyligers IC, Savelberg HHCM, Meijer K: Accelerationbased gait test for healthy subjects: reliability and reference data. Gait Posture 2000, 30: 192196.
 30.
Patterson KK, Nadkarni NK, Black SE, McIlroy WE: Gait symmetry and velocity differ in their relationship to age. Gait Posture 2012, 35: 590594.
 31.
Himann JE, Cunningham DA, Rechnitzer PA, Paterson DH: Agerelated changes in speed of walking. Age Ageing 2003, 32: 137142.
 32.
Ko SU, Tolea MI, Hausdorff JM, Ferrucci L: Sexspecific differences in gait patterns of healthy older adults: results from the Baltimore longitudinal study of aging. J Biomech 2011, 44: 19741979.
Acknowledgments
We wish to thank all the participants in the study, all the students who helped us with measurements and analyses, and the staff of the Community Care Center, Ishikawa Prefectural Nursing University. The study was financially partially supported by Grants inAid for Scientific Research from the Japan Society for Promotion of Science (Nos. 15570191 and 19570228). The funders were not involved in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Author information
Additional information
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
HK performed statistical analysis, and drafted the manuscript and study design. WK performed measurements. TK participated in study design and obtained final approval. All authors read and approved the final manuscript.
Rights and permissions
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
About this article
Received
Accepted
Published
DOI
Keywords
 Accelerometry
 Symmetry
 Regularity
 Gait
 Autocorrelation