Gait & Posture 101 (2023) 106–113

Contents lists available at ScienceDirect

Gait & Posture

journal homepage: www.elsevier.com/locate/gaitpost

Retention of kinematic patterns during a 6-minute walk test in people with

knee osteoarthritis
Stuart C. Millar a, b, *, Kieran Bennett a, Mark Rickman c, a, Dominic Thewlis a, c
a
Centre for Orthopaedic and Trauma Research, Adelaide Medical School, Faculty of Health and Medical Sciences, The University of Adelaide, Adelaide, SA 5000,
Australia
b
Active Vision Lab, School of Psychology, Faculty of Health and Medical Sciences, The University of Adelaide, Adelaide, SA 5000, Australia
c
Department of Orthopaedics and Trauma, Royal Adelaide Hospital, Adelaide, SA 5000, Australia

A R T I C L E I N F O A B S T R A C T

Keywords: Background: Knee osteoarthritis (OA) is a chronic condition affecting the entire joint and surrounding tissue,
Knee osteoarthritis resulting in pain, stiffness and impaired movement. Recent studies have suggested the use of physical perfor­
Gait mance tests, such as the six-minute walk test (6MWT) to assess joint function for those with knee OA. This study
Biomechanics
assessed lower limb sagittal plane joint angles during a 6MWT for people with mild-moderate knee OA.
Performance test
Methods: Thirty-one participants (18 male, 13 female; 62.9 ± 8.4 years) with knee OA were recruited. Gait data
were collected in a single session during which participants completed a 6MWT around a 20 m course. Sagittal
plane joint angles for the hip, knee and ankle were calculated during the first and last minute of the 6MWT.
Statistical parametric mapping (SPM) was used to investigate changes in kinematic traces over the gait cycle.
Results: Mean joint angles for the hip and knee showed no significant differences between the first and last minute
of the 6MWT. Ankle joint kinematic traces indicated there to be a decrease in plantarflexion approaching toe-off
in the last minute of the test – a 1.5◦ reduction from the first minute. No significant differences were calculated
for walking speed or joint range of motion.
Discussion: The lack of significant change in joint kinematic parameters and walking speed suggests the relative
fatigue and pain burden to the participant over the duration of the 6-minute period is insufficient to elicit any
mechanical changes to walking gait.

1. Introduction osteoarthritic burden on the patient in terms of pain and symptoms but
also functional status [5].
Knee osteoarthritis (OA) is a whole joint disease involving several While total knee replacement surgery is clinically effective, with
joint tissues and is mainly characterized by the inhibition of natural symptomatic improvement exceeding 85% [6,7], it is considered a
cartilage repair processes brought about by biochemical and biome­ fallback option for patients under 50 due to its finite lifespan and higher
chanical changes in the joint [1]. Recent estimates indicate that risk of revision [8,9], and is not employed for patients with
approximately 4% of the worldwide population have symptomatic and mild-moderate knee OA [10]. Given the reluctance to intervene surgi­
radiographic disease, with the prevalence increasing with age and pla­ cally upon first signs of osteoarthritic disability it is important to track
teauing beyond 50 years [2]. Critically, there is currently no cure for disease progression from a radiological, symptomatic and functional
knee OA with Osteoarthritis Research Society International (OARSI) standpoint. Previous studies have shown gait analysis to provide a
guidelines for non-surgical management recommending exercise, valuable objective measure of knee function for OA sufferers, assessing
weight management and non-steroidal anti-inflammatory drugs [3,4]. both temporospatial [11] and kinematic parameters [12,13]. However,
While such treatment protocols may serve to slow the progression of the such motion capture systems are expensive and not readily available,
disease, and allow for better pain management for sufferers, there are thus it is necessary to consider alternative solutions for assessing joint
instances where surgical intervention is unavoidable. However, when function in people with knee OA.
considering surgical intervention, it is important to consider the Previous studies have identified a range of performance-based tests

* Correspondence to: Adelaide Medical School, The University of Adelaide, L7 Adelaide Health & Medical Sciences Building, Adelaide, SA 5000, Australia.
E-mail address: [email protected] (S.C. Millar).

https://doi.org/10.1016/j.gaitpost.2023.02.004
Received 21 November 2021; Received in revised form 25 January 2023; Accepted 8 February 2023
Available online 10 February 2023
0966-6362/© 2023 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
S.C. Millar et al. Gait & Posture 101 (2023) 106–113

Fig. 1. Individual scores for each KOOS domain with mean and 95% confidence interval, participants separated based on KL grading. Circles represent KL Grade 2
and triangles KL Grade 3.

Fig. 2. A, C, E – Mean and standard deviation for hip (A), knee (C) and ankle (E) angles during the first (red) and last (black) minute of the 6MWT. B, D, F – The
repeated measures ANOVA test statistic SPM{F} for the within-subjects analysis. At the ankle, the critical threshold F* = 11.010 was exceeded at the supra-threshold
cluster between 57% and 64% with p = 0.007, indicating more plantarflexion within stance phase during the first minute of the 6MWT.

of physical function that can evaluate what individuals are truly capable knee OA found that during a 6-minute treadmill familiarization task
of as opposed to what their perceived capabilities are [14]. OARSI rec­ there were minimal changes to knee biomechanics [20]. Similarly,
ommendations have identified the 6-minute walk test (6MWT) to be a research assessing multiple sclerosis sufferers has suggested the 6MWT
strong measure for assessing physical function in people with knee OA is of insufficient length to elicit biomechanical changes, although was
[14,15]. However, when considering the implementation of such a test it suitable to discern reduction in overall walking distance against a
is important to consider the metabolic cost to the participant associated healthy cohort [21]. While research within an osteoarthritic population
with performance. Gait adaptations associated with knee OA often has assessed knee function over the course of six minutes [20], such
generate less mechanically efficient movement patterns [16,17], which work has only evaluated performance for treadmill walking, this does
in turn increases the metabolic cost associated with walking, thereby not allow for a formal assessment of walking distance as a metric to
exacerbating fatigue [18]. As a time-limited test, the 6-minute walk measure physical function due to the designation of belt speed prior to
enables the participant to adopt a comfortable walking speed ensuring testing. Furthermore, given the inherent differences between treadmill
they can sustain their effort and complete the test without fatiguing and overground walking [22] it would seem necessary that if aiming to
metabolically [19]. Previous research assessing those with moderate implement an overground 6-minute walk assessment it is necessary to

107
S.C. Millar et al. Gait & Posture 101 (2023) 106–113

Table 1 Marker trajectories were reconstructed using Vicon Nexus (V2.12,

Walking speed and kinematic parameters for the first and last minute of the six- Vicon, Oxford, UK).
minute walk test. Kinematic analysis of the data was undertaken in Visual3D (V6, C-
First Minute Last Effect Size P- Motion Inc., Germantown, MD, USA). The kinematic model consisted of
Mean (SD) Minute (Cohen’s d) value seven segments (pelvis, thighs, lower legs and feet) with three rotational
Mean degrees-of-freedom (DoF) at the hip, one rotational DoF at the knee and
(SD)
three rotational DoF at the ankle. Marker trajectories and ground reac­
Walking Speed 1.24 (0.24) 1.21 0.34 0.071 tion forces were low-pass filtered using a zero-lag 2nd order Butterworth
(m/s) (0.25)
filter with cut-off frequencies of 6 Hz and 50 Hz, respectively. Force
Hip ROM (◦ ) 41 (6) 40 (6) 0.43 0.024
Knee ROM (◦ ) 62 (8) 63 (8) -0.23 0.211 platforms were used for initial heel strike event detection with a
Ankle ROM (◦ ) 25 (4) 24 (4) 0.26 0.161 kinematic-based approach [25] using pattern recognition applied
Max Knee 63 (6) 63 (6) -0.16 0.374 thereafter to identify those strikes outside of the platform range. Inverse
Flexion (◦ ) kinematics were used to reconstruct model motion from the gait data
Min Knee 1 (6) 1 (6) 0.15 0.409
Flexion (◦ )
and joint angles were calculated as Euler angles using ISB recommen­
dations [26]. Joint angles for the hip, knee and ankle along with walking
ROM – Range of motion. speed were determined for the first and last minute of the 6MWT. Joint
angles were resampled to 101 data points for each gait cycle for the
initially assess participant performance in such an environment. affected limb. In measuring performance of the affected limb within the
Therefore, the aim of this study was to investigate if lower limb sagittal first and last minute, three consecutive gait cycles for each of the first,
plane kinematics changed during a 6MWT in patients with and final, two laps of the course were reconstructed, thereby accounting
mild-moderate knee OA. We hypothesized that lower limb joint kine­ for a total of twelve gait cycles for each participant to be used for the
matics would remain stable over the course of the 6MWT. Furthermore, statistical analysis.
given the commonality with which patient self-reported outcomes are
utilized both clinically and within the literature as a measure of disease 2.3. Statistical analysis
progression, as a secondary aim we assessed correlations between pa­
tient self-reported outcomes and objective measures of physical function Statistical analyses were performed in MATLAB (R2020a, The
(joint range of motion and walking distance). Mathworks Inc, USA) and SPSS (V27, IBM, New York, USA). Normality
was assessed using Q-Q plots. To compare joint kinematics between the
2. Methods first and last minute, a repeated measures ANOVA was implemented in
MATLAB using SPM1D [27] to determine if there was a difference in
2.1. Participants and protocol sagittal plane joint angles for the hip, knee and ankle across the entire
gait cycle. This was completed by analysing twelve, time-normalised
Participants with knee OA were recruited prospectively as part of a gait cycles for each participant (six from each of the first and last min­
multicentre clinical research trial (trial registration: NCT04124042) ute), where individual changes are determined before calculating group
investigating a new treatment for knee osteoarthritis. The data for this statistics for the first and last minute of the 6MWT. Significance was set
study is taken from the baseline (pre-intervention) assessments. The at α = 0.05. Sphericity corrections were not applied, and equal variance
multicentre trial had a target sample size of 270 participants with our was assumed. The SPM repeated measures ANOVA calculates the
site being the only one to collect functional outcomes incorporating gait traditional F statistic for the within-subjects analysis, subsequently
analysis. No formal sample size calculation was performed for our referred to as SPM{F} over the entire normalized time series. The SPM
analysis. analysis comprised the following steps. (1) Computation of the test
At our site, participants were recruited between May 2020 and statistic, SPM{F}, for the complete time series. (2) Estimation of the
March 2021 and were eligible if they (1) were between 45 and 85 years temporal smoothness of the data based on the average temporal
old; (2) had radiographic tibiofemoral OA defined as Kellgren-Lawrence gradient. (3) Calculation of the critical threshold F* using random field
(KL) grade 2–3; and (3) had a BMI < 40 kg/m2. In participants where theory, this is the threshold above which only 5% of the data would be
both knees were eligible, measurements were taken only on the most expected to reach had the test statistic trajectory resulted from an
symptomatic knee, as designated by the participant. Prior to completing equally smooth random process. (4) Computation of probability values
the gait assessment each participant completed the Knee injury Osteo­ for each “supra-threshold cluster”. For instances where there were
arthritis Outcome Score (KOOS) [23] which is a patient reported multiple consecutive points crossing the critical threshold, a “supra-­
outcome measure specifically designed to assess knee pathologies. threshold cluster” was identified and a cluster-specific p-value was
Ethics approval was granted by the Central Adelaide Local Health calculated using random field theory [28].
Network (CALHN) Human Research Ethics Committee (CALHN refer­ To determine if walking speed and lower limb joint range of motion
ence: R20191105). All participants provided informed written consent. (ROM) changed between the first and last minute of the 6MWT, a paired
samples t-test was performed using SPSS. Statistical significance was set
2.2. Gait analysis at α = 0.05. The level of significance for pairwise comparisons was
adjusted with a Bonferroni correction accounting for multiple compar­
Gait data were collected in a single testing session during which isons (α adjusted to 0.008). Data are presented as the mean and standard
participants were required to complete a 6-minute walk test [14]. Par­ deviation for the first and last minute along with the effect size (Cohen’s
ticipants completed the 6MWT around a marked course of dimensions 9 d). Joint ROM was calculated for each participant across the entire gait
× 1 m, requiring four turns to complete a full 20 m lap (Supplementary cycle and a group mean taken for the first and last minute of the 6MWT.
Figure 1). Over this course a total capture volume of 5 m in length was Finally, a two-tailed bivariate Pearson correlation was used to
utilised to record walking gait using a 10-camera Vicon Vantage system compare scores from KOOS domains with walking distances achieved
(Vicon, Oxford, UK) and two AMTI force platforms (Advanced Medical and total knee ROM, along with the change in walking speed and knee
Technology Inc., Watertown, MA) sampling at 100 Hz and 2000 Hz, ROM during the first and last minute of the 6MWT.
respectively. Thirty-eight retro-reflective markers were placed on
anatomical landmarks of the pelvis and lower limbs to define the joints
along with four rigid clusters attached to the thighs and shanks [24].

108
S.C. Millar et al. Gait & Posture 101 (2023) 106–113

Fig. 3. Pearson correlation between KOOS domains (Pain, Symptoms, ADL, QoL and Sport/Rec) and distance walked during 6MWT. A – Pain. B – Symptoms. C –
ADL. D – QoL. E – Sport/Rec. The line of best fit and 95% confidence interval are presented for each dataset.

3. Results 3.1. Joint kinematics and patient self-reported outcomes

Thirty-one participants (18 M/13 F) were included in this analysis Joint angles averaged across participants, for the first and last minute
with a mean age 62.9 ± 8.4 years, height 1.71 ± 0.10 m, body mass 90 of the 6MWT, are presented in Fig. 2. Mean joint angles at the hip and
± 18.5 kg and body mass index 30.7 ± 4.42 kg/m2. Radiographic knee were similar for the entirety of the first and last minute of the
severity of knee OA was graded using the KL scale (Grade 2 = 7, Grade 3 6MWT, not exceeding the critical threshold in either case (Fig. 2A-D). A
= 24). The mean walking distance during the 6MWT was 372 ± 69 m. single supra-threshold cluster was identified for the ankle (57–64%),
Mean KOOS domains, separated based on KL grading, are presented exceeding the critical threshold of F = 11.010 (p = 0.007) and indi­
within Fig. 1. Participants categorized as KL grade 3 had a greater mean cating there to be more plantarflexion at the ankle in the first minute of
score for each KOOS domain relative to participants graded as KL 2 the 6 MW (Fig. 2F). However, these angles measured less than 1.5◦ in
(Fig. 1). each instance and were therefore considered to be within the potential

109
S.C. Millar et al. Gait & Posture 101 (2023) 106–113

Fig. 4. Pearson correlation between KOOS domains (Pain, Symptoms, ADL, QoL and Sport/Rec) and knee range of motion (first minute) during 6MWT. A – Pain. B –
Symptoms. C – ADL. D – QoL. E – Sport/Rec. The line of best fit and 95% confidence interval are presented for each dataset.

measurement error (Fig. 2E). when considering the change in knee ROM between the first and last
Walking speed showed no significant difference (p = 0.071) between minute there was a statistically significant (p < 0.05) association for
the first and last minute of the 6MWT (Table 1). Similarly, there were no domains; Pain, Symptoms, ADL and Sport/Rec (Fig. 5). Specifically,
significant differences in joint ROM at the hip, knee or ankle at the start those participants indicating a lower score across KOOS domains dis­
and end of the 6MWT (Table 1). Additionally, ROM changes grouped by played an increase in joint ROM between the first and last minute of the
KL grading showed consistency between groups (Supplementary test (Fig. 5). Conversely, those with a higher score across domains
Figure 2). However, this was not considered a formal outcome measure showed a decrease in joint ROM over the course of the 6MWT (Fig. 5).
due to the limitations associated with sample size in the KL 2 group
(n = 7). 4. Discussion
There were no correlations between KOOS domains and distance
walked during the 6MWT (Fig. 3) or total knee ROM (Fig. 4). However, Our study aimed to evaluate the use of a six-minute walk test as an

110
S.C. Millar et al. Gait & Posture 101 (2023) 106–113

Fig. 5. Pearson correlation between KOOS domains (Pain, Symptoms, ADL, QoL and Sport/Rec) and change in knee range of motion during 6MWT. A – Pain. B –
Symptoms. C – ADL. D – QoL. E – Sport/Rec. The line of best fit and 95% confidence interval are presented for each dataset.

indicator of physical function for knee OA sufferers through evaluating six-minute period. Similarly, walking speed showed minimal difference
changes in lower limb joint kinematics over the course of the test. In (0.03 m/s) within the first and last minute of the test, indicating par­
doing this, we assessed sagittal plane joint angles for the hip, knee and ticipants retained consistency in walking speed over the course of the
ankle within the first and last minute of the 6MWT to evaluate consis­ test.
tency. There were no differences measured for hip and knee kinematics Previous research has identified alterations in walking speed [30,31]
across the gait cycle. A statistically significant difference was measured and knee joint kinematics [32,33] to be indicative of osteoarthritic
between 57% and 64% of the gait cycle for ankle kinematics, although severity across individuals. Therefore, when considering the imple­
such measures equated to less than 1.5◦ and were deemed to be within mentation of a physical performance test to assess function, it is
the potential measurement error of the system [29]. Given the similar­ important for such a test to limit fatigue related mechanical changes and
ities in joint angles for the hip, knee and ankle it was shown that the acute pain which would likely inhibit peak performance. The consis­
participants retained consistency in their walking gait across the tencies in walking speed and kinematics identified in this study suggest

111
S.C. Millar et al. Gait & Posture 101 (2023) 106–113

the timing and execution of the 6MWT to be suitable to restrict the coupling the measurement error of the system with the joint angles for
impact of fatigue and exercise induced pain on physical performance, the hip, knee and ankle having an absolute difference of 1◦ is it unlikely
supporting the findings of previous studies indicating the relative fatigue that any measurable difference was calculated between the first and last
burden of a 6MWT to be insufficient to provoke kinematic changes to minute of the 6MWT.
walking gait [21]. Thus, the implementation of such a test can provide
clinicians a means to objectively assess patient disability along with the 5. Conclusion
ability to monitor changes in physical function over time. The findings
from this study align well with similar work assessing function for those Due to the progressive nature of OA, identification of a performance
with mild-moderate knee OA in terms of walking speed [34,35], test capable of limiting sessional fatigue or pain related mechanical
although the total knee joint ROM measured in our study is greater than changes could be invaluable in assessing functional regression associ­
that previously reported [20,36]. However, when evaluating patient ated with knee OA and supplement patient self-reported outcome
disability, it is important to consider aspects of health relating to both measures. Study results show no significant difference in lower limb
the physical functioning of the patient along with more subjective range of motion and walking speed between the first and last minute of
measures of function – assessing pain, symptomatic burden, and quality the 6MWT. Similarly, joint angles for the hip and knee remained
of life. consistent across the entirety of the gait cycle indicating retention of
Within this study, patient self-reported outcome measures were kinematics throughout the full six-minute period. Future work should be
assessed using the KOOS questionnaire. When looking at associations focused longitudinally to investigate if any relationships exist between
between 6MWT distance and knee ROM with scores across KOOS do­ patient self-reported outcome measures and gait performance to deter­
mains for Pain, Symptoms, Activities of Daily Living and Quality of Life mine if long-term functional deficits are also indicative of increased pain
there were no significant correlations identified. This is potentially or symptomatic burden.
influenced by relatively minor differences in pain and symptomatic
burden between mild-moderate stage OA [37], making it difficult to Author contributions statement
truly discern any functional or mechanical discrepancies between the
two stages. Given the significant range in pain and symptoms between SCM, KB, MR and DT were involved in the conception and imple­
mild and severe knee OA [34,37–39], it is plausible to consider that the mentation of the research, collection of data, analysis of results and
inclusion of more severe cases may in fact show more degenerative writing of the final manuscript. All authors have read and approved the
functional changes in walking distance and knee ROM. However, KOOS final submitted manuscript.
metrics measured within this study align more closely with those
considered to have severe knee OA based on KL grading [20,34], sug­ Declarations of interest
gesting that radiographic severity and subjective measures of function
may not align perfectly and may be subject to inherent day-to-day none.
variability in pain and symptoms for the sufferer.
While previous research has identified significant differences across Acknowledgements
patient self-reported outcome measures along with functional gait
characteristics, such differences have been most identifiable when The study received funding from XALUD Therapeutics and A/Prof
comparing asymptomatic or moderately symptomatic knees to more Thewlis receives fellowship funding from the NHMRC (CDF (ID:
severe cases [30–32], which were not assessed in this study. However, 1126229)).
when considering knee joint ROM over the course of the 6MWT the data
presented here suggests there to be a significant association between
Appendix A. Supporting information
change in knee ROM and KOOS domain. The results here indicate a
negative correlation between KOOS domain and change in knee joint
Supplementary data associated with this article can be found in the
ROM, however, the majority of these changes fall within ± 3◦ and
online version at doi:10.1016/j.gaitpost.2023.02.004.
should therefore be interpreted with caution. Given the disparity in
KOOS scores and KL grading presented within this study, along with the
References
associations identified between change in joint ROM and KOOS do­
mains, it is plausible to suggest that the inclusion of functional metrics [1] J.W.J. Bijlsma, F. Berenbaum, F.P.J.G. Lafeber, Osteoarthritis: an update with
such as 6MWT performance may allow for greater agreement as to the relevance for clinical practice, Lancet 377 (9783) (2011) 2115–2126.
stage of disease burden experienced by the individual through truly [2] M. Cross, E. Smith, D. Hoy, et al., The global burden of hip and knee osteoarthritis:
estimates from the Global Burden of Disease 2010 study, Ann. Rheum. Dis. 73 (7)
assessing their physical capability in a normal activity of daily living. (2014) 1323.
The results presented here have identified the 6MWT to be a suitable [3] R.R. Bannuru, M.C. Osani, E.E. Vaysbrot, et al., OARSI guidelines for the non-
performance test to assess functionality in knee OA sufferers, however, surgical management of knee, hip, and polyarticular osteoarthritis, Osteoarthr.
Cartil. 27 (11) (2019) 1578–1589.
there are some limitations. Firstly, this study focused only on those with [4] T.E. McAlindon, R.R. Bannuru, M.C. Sullivan, et al., OARSI guidelines for the non-
mild or moderate knee OA (KL grade 2 or 3), identifying no differences surgical management of knee osteoarthritis, Osteoarthr. Cartil. 22 (3) (2014)
in locomotion across the duration of the six minutes. However, there is 363–388.
[5] L. Gossec, S. Paternotte, C.O. Bingham 3rd, et al., OARSI/OMERACT initiative to
potential that more severe cases (KL grade 4) may exhibit alterations in
define states of severity and indication for joint replacement in hip and knee
the gait cycle across the 6MWT due to the likelihood that the knee has osteoarthritis. An OMERACT 10 Special Interest Group, J. Rheumatol. 38 (8)
degenerated to a greater extent. Secondly, due to the timed nature of the (2011) 1765–1769.
[6] C.M. Callahan, B.G. Drake, D.A. Heck, et al., Patient outcomes following
6MWT, there is a learning effect for participants to ensure they pace
tricompartmental total knee replacement: a meta-analysis. 1994. 271(17): p.
themselves to retain walking speed for the duration of the test and 1349–1357.
maximize the distance walked. Given the importance of ensuring [7] D.A. Heck, R.L. Robinson, C.M. Partridge, et al., Patient outcomes after knee
maximal effort to attain an appropriate measure of disability, it seems replacement 356 (1998) 93–110.
[8] C.J. Dy, R.G. Marx, K.J. Bozic, et al., Risk factors for revision within 10 years of
pertinent that such an assessment be performed on more than a single total knee arthroplasty, Clin. Orthop. Relat. Res. 472 (4) (2014) 1198–1207.
occasion when aiming to establish a baseline measure of function. [9] A.M. Weinstein, B.N. Rome, W.M. Reichmann, et al., Estimating the burden of total
Finally, in testing the proposed hypothesis, the authors application of knee replacement in the United States, J. Bone Jt. Surg. Am. Vol. 95 (5) (2013)
385–392.
the Bonferroni correction makes it more challenging to prove that there [10] W. Zhang, W.B. Robertson, J. Zhao, et al., Emerging TRend In The
were no significant differences between time points. However, when Pharmacotherapy Of Osteoarthritis, Front. Endocrinol. 10 (2019).

112
S.C. Millar et al. Gait & Posture 101 (2023) 106–113

[11] R. Debi, A. Mor, G. Segal, et al., Differences in gait pattern parameters between [26] G. Wu, S. Siegler, P. Allard, et al., ISB recommendation on definitions of joint
medial and anterior knee pain in patients with osteoarthritis of the knee, Clin. coordinate system of various joints for the reporting of human joint motion—part I:
Biomech. 27 (6) (2012) 584–587. ankle, hip, and spine, J. Biomech. 35 (4) (2002) 543–548.
[12] K.R. Kaufman, C. Hughes, B.F. Morrey, et al., Gait characteristics of patients with [27] T.C. Pataky, One-dimensional statistical parametric mapping in Python, Comput.
knee osteoarthritis, J. Biomech. 34 (7) (2001) 907–915. Methods Biomech. Biomed. Eng. 15 (3) (2012) 295–301.
[13] D. Bytyqi, B. Shabani, S. Lustig, et al., Gait knee kinematic alterations in medial [28] R.J. Adler, J.E. Taylor, Random fields and geometry, Springer Science & Business
osteoarthritis: three dimensional assessment, Int. Orthop. 38 (6) (2014) Media, 2009.
1191–1198. [29] L. Chiari, U.D. Croce, A. Leardini, et al., Human movement analysis using
[14] F. Dobson, R.S. Hinman, E.M. Roos, et al., OARSI recommended performance- stereophotogrammetry: Part 2: instrumental errors, Gait Posture 21 (2) (2005)
based tests to assess physical function in people diagnosed with hip or knee 197–211.
osteoarthritis, Osteoarthr. Cartil. 21 (8) (2013) 1042–1052. [30] J.L. Astephen, K.J. Deluzio, G.E. Caldwell, et al., Biomechanical changes at the hip,
[15] G. Coleman, F. Dobson, R.S. Hinman, et al., Measures of Physical Performance, knee, and ankle joints during gait are associated with knee osteoarthritis severity,
Arthritis Care Res. 72 (S10) (2020) 452–485. J. Orthop. Res. 26 (3) (2008) 332–341.
[16] S. Ko, S.M. Ling, C. Schreiber, et al., Gait patterns during different walking [31] K. Mills, M.A. Hunt, R. Ferber, Biomechanical deviations during level walking
conditions in older adults with and without knee osteoarthritis—results from the associated with knee osteoarthritis: a systematic review and meta-analysis,
baltimore longitudinal study of aging, Gait Posture 33 (2) (2011) 205–210. Arthritis Care Res. 65 (10) (2013) 1643–1665.
[17] D.J. Rutherford, C.L. Hubley-Kozey, W.D. Stanish, Knee effusion affects knee [32] J. Favre, J.C. Erhart-Hledik, T.P. Andriacchi, Age-related differences in sagittal-
mechanics and muscle activity during gait in individuals with knee osteoarthritis, plane knee function at heel-strike of walking are increased in osteoarthritic
Osteoarthr. Cartil. 20 (9) (2012) 974–981. patients, Osteoarthr. Cartil. 22 (3) (2014) 464–471.
[18] K.C. Foucher, B. Aydemir, C.-H. Huang, Walking energetics and fatigue are [33] T.L. Heiden, D.G. Lloyd, T.R. Ackland, Knee joint kinematics, kinetics and muscle
associated with physical activity in people with knee osteoarthritis, Clin. Biomech. co-contraction in knee osteoarthritis patient gait, Clin. Biomech. 24 (10) (2009)
88 (2021), 105427. 833–841.
[19] J.F.D. Barbosa, S.S. Bruno, N.S.O. Cruz, et al., Perceived fatigability and metabolic [34] J.E. Naili, E.W. Broström, B. Clausen, et al., Measures of knee and gait function and
and energetic responses to 6-minute walk test in older women, Physiotherapy 102 radiographic severity of knee osteoarthritis – A cross-sectional study, Gait Posture
(3) (2016) 294–299. 74 (2019) 20–26.
[20] D. Rutherford, M. Baker, I. Wong, et al., Dual-belt treadmill familiarization: [35] G.L. Hatfield, W.D. Stanish, C.L. Hubley-Kozey, Three-Dimensional biomechanical
Implications for knee function in moderate knee osteoarthritis compared to gait characteristics at baseline are associated with progression to total knee
asymptomatic controls, Clin. Biomech. 45 (2017) 25–31. arthroplasty, Arthritis Care Res. 67 (7) (2015) 1004–1014.
[21] J.V. McLoughlin, C.J. Barr, B. Patritti, et al., Fatigue induced changes to kinematic [36] I. McCarthy, D. Hodgins, A. Mor, et al., Analysis of knee flexion characteristics and
and kinetic gait parameters following six minutes of walking in people with how they alter with the onset of knee osteoarthritis: a case control study, BMC
multiple sclerosis, Disabil. Rehabil. 38 (6) (2016) 535–543. Musculoskelet. Disord. 14 (1) (2013) 169–175.
[22] M.B. Semaan, L. Wallard, V. Ruiz, et al., Is treadmill walking biomechanically [37] K. Oishi, E. Tsuda, Y. Yamamoto, et al., The Knee injury and Osteoarthritis
comparable to overground walking? A systematic review 92 (2022) 249–257. Outcome Score reflects the severity of knee osteoarthritis better than the revised
[23] E.M. Roos, S.J.H. Toksvig-Larsen, and q.o.l. outcomes, Knee injury and Knee Society Score in a general Japanese population, Knee 23 (1) (2016) 35–42.
Osteoarthritis Outcome Score (KOOS)–validation and comparison to the WOMAC [38] B. Heidari, Knee osteoarthritis prevalence, risk factors, pathogenesis and features:
in total knee replacement. 2003. 1(1): p. 17. Part I, Casp. J. Intern. Med. 2 (2) (2011) 205–212.
[24] A. Cappozzo, F. Catani, U. Della Croce, et al., Position and orientation in space of [39] J. Jackson, R. Iyer, J. Mellor, et al., The burden of pain associated with
bones during movement: anatomical frame definition and determination. Clin. osteoarthritis in the hip or knee from the patient’s perspective: a multinational
Biomech. 10 (4) (1995) 171–178. cross-sectional study, Adv. Ther. 37 (9) (2020) 3985–3999.
[25] S.J. Stanhope, T.M. Kepple, D.A. McGuire, et al., Kinematic-based technique for
event time determination during gait, Med. Biol. Eng. Comput. 28 (4) (1990)
355–360.

113