Putting the best foot forward: Relationships between indicators of step quality and cadence in three gait vulnerable populations

Abstract

BACKGROUND:

Aging and neurological conditions like Multiple Sclerosis (MS) and Parkinson’s disease (PD) make people vulnerable for gait impairments, limit function, and restrict sustained walking needed for health promotion. Walking to meet physical activity guidelines requires adequate cadence which is difficult to achieve for gait vulnerable populations.

OBJECTIVE:

The objective of this study is to estimate, for seniors and people with MS or PD, the extent to which cadence is associated with heel-to-toe stepping pattern (good steps), angular velocity of ankle at heel-strike and its variability.

METHODS:

A cross-sectional regression analysis was performed on data collected during walking tests using the Heel2Toe sensor.

RESULTS:

Health condition (MS = 57, PD = 27, seniors = 56) had an association with cadence, independent of age and sex. Only angular velocity showed a significant relationship with cadence such that every – 50° difference in angular velocity (more negative is better) was associated with a difference of ≈3.5 steps per minute.

CONCLUSION:

Adequate angular velocity occurs with an optimal heel-to-toe movement. This heel-to-toe gait can easily be targeted during therapy but technology would be an asset to sustain the relearned movement during everyday activities, Technology that provides real-time feedback for steps with adequate angular velocity at heel strike could be a valuable therapeutic adjunct.

Keywords

Cadence angular velocity coefficient of variation Multiple Sclerosis Parkinson’s disease seniors

1. Background

Many key gait parameters are required to produce a gait that is rhythmic, reproducible and sustainable. Gait quality is affected by neurological integrity, balance, power and is manifested in ability to produce consistent steps. The ability to walk safely for function, recreation and health promotion depends on producing good quality steps. Two key parameters of good quality steps are angle of ankle at heel strike and consistency. The former is a function of stride length which is shortened if balance is poor and contributes to foot scuffing while walking. The latter (consistency) depends on gait automaticity which when lost increases the attentional requirements of walking (Hollman, Kovash, Kubik, & Linbo, 2007; Lajoie, Teasdale, Bard, & Fleury, 1996). Aging and neurological conditions like Multiple Sclerosis (MS) and Parkinson’s disease (PD) make people vulnerable for gait impairments and this is manifested by inconsistent stepping, wide base of support, lack of heel-to-toe stepping, among others (Crenshaw, Royer, Richards, & Hudson, 2006a; Hausdorff, Cudkowicz, Firtion, Wei, & Goldberger, 1998). Lack of heel to toe stepping pattern leads to dangerous shuffling gait that is often seen with seniors and people with PD (Alcock, Galna, Lord, & Rochester, 2016). For people with MS, walking for a sustained duration results in fatigue in leg muscles particularly in antigravity foot muscles (ankle dorsiflexors) that produces foot drop and increases work of walking and risk of falls (Benedetti et al., 1999; Martin et al., 2006; Matsuda et al., 2011; Schwid, Covington, Segal, & Goodman, 2002; Scott, van der Linden, Hooper, Cowan, & Mercer, 2013). People with MS and PD are prone to gait vulnerability and less likely to meet physical activity guidelines for health promotion because of unsafe, effortful and inefficient gait.

Physical activity guidelines recommend 150 minutes of moderate to vigorous intensity physical activity over a week in bouts of 10 minutes (http://www.csep.ca/en/guidelines/get-the-guidelines). This translates to cadence of 100 steps a minute for bouts of 10 minutes, twice a day. Another indication of activity is steps per day and 10,000 or more steps per day would classify a person as ‘active’ whereas individuals who take >12,500 steps per day are classified as ‘highly active’ (Tudor-Locke & Bassett, 2004). It is well established that in order to gain health benefits from walking, intensity of walking performance in crucial. Conventionally, people are asked to walk faster but walking faster could be achieved by either increasing the cadence or step length or both (Grieve & Gear, 1966; Lamoreux, 1970; Murray, Kory, & Clarkson, 1969). Cadence, a gait parameter that is easily understood by the patients, is increasingly used as a treatment outcome and rehabilitation goal. However, a high cadence with short stride length is often a feature of Parkinsonian gait so there is an optimal ratio between cadence and stride length. That optimal ratio may that when the stride length permits adequate heel-to-toe gait. This would lead to a hypothesis that the greater the degree of heel strike, shown by large angular velocity, the greater will be the cadence, but that his may not hold for people with PD. An optimal heel strike could only be achieved with sufficient step length and ankle dorsiflexion during the preceding swing phase in a gait cycle. The foot flat event succeeding a heel strike is a result of eccentric control of ankle dorsiflexor muscles. The rate at which foot slaps the walking surface is directly related to the extent of ankle dorsiflexion and eccentric control of dorsiflexor muscles. The eccentric control of ankle dorsiflexors at heel strike to foot flat is recorded as angular velocity degrees per second (°/sec). Typical values of angular velocity have been estimated to range from –260°/sec to –650°/sec (Figueiredo, Félix et al. 2018). Literature is scant with respect to typical values of angular velocity for walking and the methodology for obtaining these values differs. CV of gait parameters that are related to AV are under 8% (Brach, Berlin et al. 2005).

Adequate cadence is only one feature of health promoting walking, sustained walking is another. Irregular gait pattern could interfere with sustainability. Some degree of variability is physiologic to human biology and is demonstrated in heart beats, respiration, blood flow and gait. Systems are designed work efficiently within a limit of variability as too much or too little variability is counterproductive. Gait variability has been a focus of scientific enquiry in last the decade. Typically in healthy subjects, walking at slower gait speeds, and consequently slower cadence, increases variability of kinematic and kinetic parameters of motion at knee, hip and ankle joint. In other words, faster steps are more consistent than slower steps. Gait that is externally paced is also more consistent that self-paced gait, demonstrated by less variability among lower limb joint co-ordination when walking on treadmill compared to when walking over ground (Wheat, Milner, & Bartlett, 2003).

Variability in gait parameters has been shown to be associated with falls in seniors and is a feature of gait in people with MS and PD (Beauchet, Dubost, Herrmann, & Kressig, 2005; Brach, Berlin, VanSwearingen, Newman, & Studenski, 2005; Bryant et al., 2011; Crenshaw, Royer, Richards, & Hudson, 2006b; Owings & Grabiner, 2004; Schrager, Kelly, Price, Ferrucci, & Shumway-Cook, 2008; Socie & Sosnoff, 2013). Variability is also tested in other gait parameters. Increased variability in stride time and length, swing and stance time, and base width has been associated with falls in seniors, Alzheimer’s and PD (Lamoth et al., 2011; Schaafsma et al., 2003; Sheridan, Solomont, Kowall, & Hausdorff, 2003). Coefficient of variation, the ratio of variability (SD) to the mean, is an indicator of variability (Abdi, 2010).

In this study, we used the Heel2Toe sensor to capture angular velocity at heel strike and cadence. The Heel2Toe sensor is a device that is a combination of an accelerometer and a gyroscope constructed from off the shelf components comprising a Shimmer 2r motion module with six degree of freedom sensor comprising a 3 axis accelerometer (Freescale MMA7361) and a 3 axis gyroscope (InvenSense 500 series MEMS Gyros). The module also incorporates a microcontroller, 8 channels of 12 bit A/D. The sensing module is attached to the side of the subject’s right foot using a strap or clip and sensor signals are streamed via Bluetooth to the biofeedback module that runs a real-time algorithm that discriminates good from poor steps using an algorithm based on an angular velocity boundary. When this boundary is crossed, the appropriate real-time feedback is generated. The sensor, shown in Fig. 1, has been shown to be valid and reliable tool in detecting good and bad heel strike events (Vadnerkar et al., 2014, 2017). The output from the sensors include cadence, proportion of good steps, and angular velocity among other gait parameters. Figure 1 shows the position of Heel2Toe sensor on the foot. It runs a real-time algorithm that discriminates good from poor steps with 94% accuracy (Vadnerkar et al., 2014, 2017) and generates an auditory feedback signal via a Bluetooth connection to a smartphone. No auditory feedback is generated if the step does not pass the angular velocity of = 50° per second. The relationship between cadence, good steps and angular velocity in people with seniors, MS, and PD is yet to be established.

Fig.1

Shows the position of Heel2Toe sensor on the foot.

2. Objective

The objective of this study is to estimate, across three populations defined by health condition (seniors, and people with MS or PD), the extent to which cadence is associated with heel-to-toe stepping pattern (good steps), angular velocity and CV of angular velocity at heel strike using Heel2Toe sensor. We hypothesize that the relationship between cadence and proportion of good steps, angular velocity and coefficient will be different among the three populations, with the PD sample showing different relationships than the senior or MS samples.

3. Methods

3.1. Study design

This is a cross-sectional analysis of data collected with Heel2Toe sensor during clinical walking tests.

3.2. Participants

Data on seniors was available (n = 40) from a previous validation study on Heel2Toe sensor. Additional data (n = 17) on seniors were added. People with MS (n = 57) were recruited for a randomized controlled trial on Role of Exercise in Modifying Outcomes for People with MS (Mayo et al., 2013). To be included in the trial, diagnosis of MS had to be done after 1994, age between 19 to 65 years and be independent in ambulation without use of walking aid. Participants were excluded if they: (i) were already exercising three or more time per week; (ii) had any additional illness that restricted their function; (iii) had experienced a relapse during the past 30 days; and (iv) showed difficulty reading, understand or speaking either French or English (Jacobs et al., 2000; Marriott, Miyasaki, Gronseth, & O'Connor, 2010; Polman et al., 2006). Data from people with PD (n = 26) were collected as a part of pilot study on a novel Approach to Clinical Assessment and Personalized Intervention in Early Parkinson’s Disease. Heel2Toe was deployed during clinical walk tests. For people with MS, the sensor was incorporated during six-minute walk test. For seniors and people with PD the sensor was used during a 2 minute walk test.

3.3. Data extraction and statistical methods

The raw data from the sensor was exported and variables of interested extracted using MATLAB 2017b (The Math Works, Inc., Natick, MA, USA). The variables analysed from the Heel2Toe sensor were cadence (steps per minute), good steps (%), angular velocity (degrees per second). The values for angular velocity are negative representing the extent of ankle dorsiflexion and subsequent plantarflexion, more negative values are better. Patients’ characteristics were summarized using means and standard deviations (SD) and proportions, where appropriate. Pearson product moment correlations were performed, separately for each population, between pairs of continuous variables: proportion of good steps, angular velocity, coefficient of variation, and cadence. Multiple linear regression was carried out with cadence as the outcome variable. The other variables in the model included health condition, age and sex as adjustment variables, and each of the gait quality variables separately. The interaction with health condition and age was tested. The results are displayed as beta (β), standard error (se), and 95% confidence intervals (CI).

Diagnostics, including Kolmogorov-Smirnov test, residual-by-predicted plots, and scatter plots were generated to verify the assumptions of normality, homoscedasticity, linearity. All the analysis were performed using Statistical Analysis System^® 9.4 software (Institute, 2014).

4. Results

Data on a total of 140 people were available for analyses: 57 people with MS, 27 with PD and 56 seniors. Table 1 displays the characteristics of the three sample populations with mean and standard deviation (SD) for cadence, proportion of good steps and angular velocity.

Table 1
Characteristics of study participants

Variables Mean±SD or N (%)

MS PD Seniors

No. of subjects 57 27 56

Age (years) 48.1±9.9 71.0±6.9 82.6±8.7

Sex (Women/Men) 39/17 10/12 6/13

Time since diagnosis 7.4 (5.9) 6.0±4.1 –– –

Cadence (no. of steps per minute) 120.1 (12.6) 110.1 (10.2) 97.4 (14.5)

% good steps 89.9 (5.1) 73.7 (29.5) 72.6 (29.5)

Angular Velocity (°/sec) –220.6 (52.1) –159.6 (57.3) –141.3 (57.2)

Variables	Mean±SD or N (%)
No. of subjects	57	27	56
Age (years)	48.1±9.9	71.0±6.9	82.6±8.7
Sex (Women/Men)	39/17	10/12	6/13
Time since diagnosis	7.4 (5.9)	6.0±4.1	–– –
Cadence (no. of steps per minute)	120.1 (12.6)	110.1 (10.2)	97.4 (14.5)
% good steps	89.9 (5.1)	73.7 (29.5)	72.6 (29.5)
Angular Velocity (°/sec)	–220.6 (52.1)	–159.6 (57.3)	–141.3 (57.2)

Table 2 shows the Pearson product moment correlation coefficients between angular velocity, CV of angular velocity and proportion of good steps with cadence across three health conditions. For the seniors group the correlation coefficients between cadence and angular velocity, CV of angular velocity, and % good steps were –0.49, 0.25 and 0.41, respectively. For people with PD and MS, these coefficients were somewhat lower. Age was correlated moderately with cadence (r = –0.59) and ankle angular velocity (r = 0.57), less strongly with % good steps (0.34) and weakly with CV of angular velocity (r = –0.17).

Table 2

Correlation of Angular Velocity, Coefficient of Variation of Angular Velocity and Proportion of Good Steps with Cadence across three Health Conditions

Group	MS (n = 57)	PD (n = 27)	Seniors (n = 56)	Age
	r (95% CI)
Angular Velocity	–0.35 (–0.56, –0.10)	–0.06 (–0.33, 0.43)	–0.49 (–0.67, –0.26)	0.57 (0.45, 0.67)
CV angular velocity	0.16 (–0.40, 0.10)	–0.19 (–0.54, 0.21)	0.25 (–0.01, 0.48)	–0.17 (–0.33, –0.005)
Good steps (%)	0.16 (–0.40, 0.10)	–0.17 (–0.52, 0.23)	0.41 (0.16, 0.61)	–0.34 (–0.48, –0.19)

CI: Confidence interval; those that exclude.

Table 3 shows the results of multiple linear regression analysis on the average cadence as the outcome. The preliminary model included only age and sex and, while age was significantly associated with cadence (β: – 5.4; se; 0.7), sex was not (β: – 1.2; se: 2.3). Model 1 retained age and sex and added health condition with the senior group as the referent category. Three other models were tested. Models 2, 3 and 4 also included age and sex and each of the other explanatory gait variables separately (% good steps, angular velocity and CV of angular velocity). All of the four models showed a statistically significant main effect for health condition with no interaction with age. Of the variables related to gait quality, only angular velocity showed a significant relationship with cadence. Every – 50° difference in angular velocity (more negative is better) was associated with an increase of ≈3.5 steps per minute.

Table 3

Linear regression models with explanatory variables for cadence

	β (se)	(95% CI)
Model 1: R²		0.45
Health condition
Seniors	Referent
MS	20.4 (5.7)	14.7, 26.1
PD	14.9 (4.2)	10.7, 19.1
Age (per decade)	1.8 (2.1)	–0.3, 3.9
Sex (men v/s women)	–0.2 (0.1)	–0.3, –0.1
Model 2^*: R²		0.46
Health condition
Seniors	Referent
MS	20.1 (5.7)	14.4, 25.8
PD	15.1 (4.2)	10.9, 19.3
% good Steps per 10%	0.4 (0.6)	–0.2, 1.0
Model 3^*: R²		0.51
Health condition
Seniors	Referent
MS	18.7 (5.5)	13.2, 24.2
PD	14.8 (3.9)	10.9, 18.7
Angular velocity (°/sec) per 10°	–0.7 (0.2)	–0.5, –0.9
Model 4^*: R²
Health condition		0.47
Seniors	Referent
MS	20.2 (5.6)	14.6, 25.8
PD	14.3 (4.1)	10.2, 18.4
CV of angular velocity per 0.1	1.9 (1.1)	0.8, 3.0

^*Adjusted for age and sex in a multiple linear regression model; Significant results (where 95% CI exclude the value of 1.0) are shown in bold.

5. Discussion

The mean values for cadence, % good steps and angular velocity were highest for people with MS (mean age 48 years), next highest for people with PD (mean age 71 years) and lastly seniors (mean age 82 years). Age and health condition were strongly related and, as shown in Table 3, age was a significant predictor of cadence, but not when health condition was in the model. This suggests that health condition has an association with cadence that is independent of age.

The results of linear regression analysis show that cadence was not predicted by % good steps or by CV of angular velocity (see Table 3), but was predicted by degree of angular velocity. Previous research suggests that humans prefer to walk at certain speed and cadence so as to optimize the metabolic cost of walking (Donelan, Kram, & Kuo, 2001). In this study, it seems that this preference for optimal cadence holds across health conditions and is not influenced by consistency in the quality of the steps (% good steps or consistency and CV of angular velocity). However, it is influenced by angular velocity which is linked to cadence through stride length. With an increase stride length, comes a sharper angle at heel strike. Heel strike would seem to be a primary therapeutic target followed by consistency in the quality of the steps.

Of the three populations, people with MS, who were younger and who had already signed up to join an exercise study, had the best values for cadence and for gait quality. However, they still showed the same high degree of variability in angular velocity suggesting that consistency in heel strike would be a therapeutic target even for those with good values on the other parameters.

Targeting gait quality would be important preparation for reaching targets for health promoting walking. This study suggests that therapy to improve walking should target angular velocity of the ankle joint. Practicing heel-to-toe gait pattern is often incorporated into gait training. There are some technologies that are commercially available that provide feedback for one or more of these gait quality parameters. Technology alone cannot not translate to improve walking unless accompanied by educational material to improve the persons capacity to benefit from the technology. For example, for people with MS, balance, core and peripheral muscle strength, and endurance are affected. It is important one walks well before walking long.

6. Limitations

This study has several limitations. The three populations were selected as convenience samples as participants were enrolled in other studies. Nevertheless this is the largest sample to date that has information on these gait parameters. The Heel2Toe technology is relatively new and is only one way of extracting gait parameters. Future studies could compare outputs from Heel2Toe with other motion capture systems and validate this relationship in different samples including healthy populations of different ages.

7. Conclusion

Of the three indicators of gait quality studied here, only angular velocity had a relationship with cadence in these three populations. Angular velocity is an easily targeted parameter during therapy, but to sustain the relearned movement during everyday activities, technology would be an asset. A technology preference would be for real-time feedback for steps with adequate angular velocity at heel strike.

Conflict of interest

None of the authors have any conflict of interest.

References

Abdi, Hervé. (2010). Coefficient of variation. Encyclopedia of Research Design, 1, 169–171.

Afiah, I. N., Nakashima, H., Loh, P. Y., & Muraki, S. (2016). An exploratory investigation of changes in gait parameters with age in elderly Japanese women. Springerplus, 5(1), 1069. doi: 10.1186/s40064-016-2739-7

Alcock, L., Galna, B., Lord, S., & Rochester, L. (2016). Characterisation of foot clearance during gait in people with early Parkinson’s disease: Deficits associated with a dual task. Journal of Biomechanics, 49(13), 2763–2769.

Beauchet, O., Dubost, V., Herrmann, F. R., & Kressig, R. W. (2005). Stride-to-stride variability while backward counting among healthy young adults. J Neuroeng Rehabil, 2, 26. doi: 10.1186/1743-0003-2-26

Benedetti, M. G., Piperno, R., Simoncini, L., Bonato, P., Tonini, A., & Giannini, S. (1999). Gait abnormalities in minimally impaired multiple sclerosis patients. Mult Scler, 5(5), 363–368. doi: 10.1177/135245859900500510

Brach, J. S., Berlin, J. E., VanSwearingen, J. M., Newman, A. B., & Studenski, S. A. (2005). 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 Neuroeng Rehabil, 2, 21.

Bryant, M. S., Rintala, D. H., Hou, J. G., Charness, A. L., Fernandez, A. L., Collins, R. L.,… Protas, E. J. (2011). Gait variability in Parkinson’s disease: Influence of walking speed and dopaminergic treatment. Neurol Res, 33(9), 959–964. doi: 10.1179/1743132811y.0000000044

Crenshaw, S. J., Royer, T. D., Richards, J. G., & Hudson, D. J. (2006a). Gait variability in people with multiple sclerosis. Multiple Sclerosis Journal, 12(5), 613–619. doi: 10.1177/1352458505070609

Crenshaw, S. J., Royer, T. D., Richards, J. G., & Hudson, D. J. (2006b). Gait variability in people with multiple sclerosis. Mult Scler, 12(5), 613–619. doi: 10.1177/1352458505070609

10.

Donelan, J. M., Kram, R., & Kuo, A. D. (2001). Mechanical and metabolic determinants of the preferred step width in human walking. Proc Biol Sci, 268(1480), 1985–1992. doi: 10.1098/rspb.2001.1761

11.

Figueiredo

, F’ix

, Costa

, Moreno

J. C.

, & Santos

C. P.

(2018). Gait event detection in controlled and real-life situations: Repeated measures from healthy subjects. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 26(10), 1945–1956.

12.

Grieve, D. W., & Gear, Ruth J. (1966). The relationships between length of stride, step frequency, time of swing and speed of walking for children and adults. Ergonomics, 9(5), 379–399.

13.

Hausdorff, J. M., Cudkowicz, M. E., Firtion, R., Wei, J. Y., & Goldberger, A. L. (1998). Gait variability and basal ganglia disorders: Stride-to-stride variations of gait cycle timing in Parkinson’s disease and Huntington's disease. Movement Disorders, 13(3), 428–437.

14.

Heidenfelder, J., Sterzing, T., Bullmann, M., & Milani, T. L. (2008). Heel strike angle and foot angular velocity in the sagittal plane during running in different shoe conditions. Journal of Foot and Ankle Research, 1(Suppl 1), O16. doi: 10.1186/1757-1146-1-S1-O16

15.

Hollman, J. H., Kovash, F. M., Kubik, J. J., & Linbo, R. A. (2007). Age-related differences in spatiotemporal markers of gait stability during dual task walking. Gait & Posture, 26(1), 113–119.

16.

Institute, SAS. (2014). SAS 9.4 Language Reference: Concepts: SAS Institute Inc.

17.

Jacobs, L. D., Beck, R. W., Simon, J. H., Kinkel, R. P., Brownscheidle, C. M., Murray, T. J.,… & Sandrock, A. W. (2000). Intramuscular interferon beta-1a therapy initiated during a first demyelinating event in multiple sclerosis. CHAMPS Study Group. NEngl J Med, 343(13), 898–904. doi: 10.1056/nejm200009283431301

18.

Lajoie, Y., Teasdale, N., Bard, C., & Fleury, M. (1996). Atten-tional Demands for Walking: Age-Related Changes.

In Ferrandez

A.-M.

Teasdale

(Eds.) Advances in Psychology (Vol. 114, pp. 235–256): North-Holland.

19.

Lamoreux, L. W. (1970). Kinematic measurements in the study of human walking. Bulletin of Prosthetics Research, 10(15), 3–84.

20.

Lamoth, C. J., van Deudekom, F. J., van Campen, J. P., Appels, B. A., de Vries, O. J., & Pijnappels, M. (2011). Gait stability and variability measures show effects of impaired cognition and dual tasking in frail people. Journal of Neuroengineering and Rehabilitation, 8(1), 2.

21.

Marriott, J. J., Miyasaki, J. M., Gronseth, G., & O'Connor, P. W. (2010). Evidence Report: The efficacy and safety of mitoxantrone (Novantrone) in the treatment of multiple sclerosis Report of the Therapeutics and Technology Assessment Subcommittee of the American Academy of Neurology. Neurology, 74(18), 1463–1470.

22.

Martin, C. L., Phillips, B. A., Kilpatrick, T. J., Butzkueven, H., Tubridy, N., McDonald, E., & Galea, M. P. (2006). Gait and balance impairment in early multiple sclerosis in the absence of clinical disability. Mult Scler, 12(5), 620–628. doi: 10.1177/1352458506070658

23.

Matsuda, P. N., Shumway-Cook, A., Bamer, A. M., Johnson, S. L., Amtmann, D., & Kraft, G. H. (2011). Falls in multiple sclerosis. PM R, 3(7), 624–632; quiz 632. doi: 10.1016/j.pmrj.2011.04.015

24.

Mayo, N. E., Bayley, M., Duquette, P., Lapierre, Y., Anderson, R., & Bartlett, S. (2013). The role of exercise in modifying outcomes for people with multiple sclerosis: A randomized trial. BMC Neurol, 13, 69. doi: 10.1186/1471-2377-13-69

25.

Murray, M. P., Kory, R. C., & Clarkson, B. H. (1969). Walking patterns in healthy old men. Journal of Gerontology, 24(2), 169–178.

26.

Owings, T. M., & Grabiner, M. D. (2004). Variability of step kinematics in young and older adults. Gait Posture, 20(1), 26–29. doi: 10.1016/s0966-6362(03)00088-2

27.

Polman, Chris H, O'Connor, Paul W, Havrdova, Eva, Hutchinson, Michael, Kappos, Ludwig, Miller, David H,… Wajgt, Andrzej. (2006). A randomized, placebo-controlled trial of natalizumab for relapsing multiple sclerosis. New England Journal of Medicine, 354(9), 899–910.

28.

Schaafsma, Joanna D, Giladi, Nir, Balash, Yacov, Bartels, Anna L, Gurevich, Tanya, & Hausdorff, Jeffrey M. (2003). Gait dynamics in Parkinson's disease: Relationship to Parkinso-nian features, falls and response to levodopa. Journal of the Neurological Sciences, 212(1-2), 47–53.

29.

Schrager, M. A., Kelly, V. E., Price, R., Ferrucci, L., & Shumway-Cook, A. (2008). The effects of age on medio-lateral stability during normal and narrow base walking. Gait Posture, 28(3), 466–471. doi: 10.1016/j.gaitpost.2008.02.009

30.

Schwid, Steven R, Covington, MMSB, Segal, Benjamin M, & Goodman, Andrew D. (2002). Fatigue in multiple sclerosis: Current understanding and future directions. Journal of Rehabilitation Research and Development, 39(2), 211–224.

31.

Scott, S. M., van der Linden, M. L., Hooper, J. E., Cowan, P., & Mercer, T. H. (2013). Quantification of gait kinematics and walking ability of people with multiple sclerosis who are new users of functional electrical stimulation. J Rehabil Med, 45(4), 364–369. doi: 10.2340/16501977-1109

32.

Sheridan, Pamela L, Solomont, Judi, Kowall, Neil, & Hausdorff, Jeffrey M. (2003). Influence of executive function on locomotor function: Divided attention increases gait variability in Alzheimer’s disease. Journal of the American Geriatrics Society, 51(11), 1633–1637.

33.

Socie, M. J., & Sosnoff, J. J. (2013). Gait variability and multiple sclerosis. Mult Scler Int, 2013, 645197. doi: 10.1155/2013/645197

34.

Tudor-Locke, C., & Bassett, D. R., Jr. (2004). How many steps/day are enough? Preliminary pedometer indices for public health. Sports Med, 34(1), 1–8.

35.

Vadnerkar, A., Figueiredo, S., Mayo, N. E., & Kearney, R. E. (2014). Classification of gait quality for biofeedback to improve heel-to-toe gait. Conf Proc IEEE Eng Med Biol Soc, 2014, 3626–3629. doi: 10.1109/embc.2014.6944408

36.

Vadnerkar, A., Figueiredo, S., Mayo, N., & Kearney, R. (2017). Design and Validation of a Biofeedback Device to Improve Heel-to-toe Gait in Seniors. IEEE J Biomed Health Inform. doi: 10.1109/jbhi.2017.2665519

37.

Wheat, J., Milner, C. E., & Bartlett, R. M. (2003). Joint coordination variability during overground and treadmill running. Hip, 22(16.7), 8.5.