A Systematic Review and Meta-Analysis of Lower Limb Neuromuscular Alteration Associates With Knee...
-
Upload
itai-izhak -
Category
Documents
-
view
217 -
download
0
Transcript of A Systematic Review and Meta-Analysis of Lower Limb Neuromuscular Alteration Associates With Knee...
-
8/10/2019 A Systematic Review and Meta-Analysis of Lower Limb Neuromuscular Alteration Associates With Knee Osteoarthri
1/12
Review
A systematic review and meta-analysis of lower limb neuromuscularalterations associated with knee osteoarthritis during level walking
Kathryn Mills a, Michael A. Hunt b, Ryan Leigh a, Reed Ferber a,a Faculty of Kinesiology, University of Calgary, Calgary, Alberta, Canadab Department of Physical Therapy, University of British Columbia, Vancouver, British Columbia, Canada
a b s t r a c ta r t i c l e i n f o
Article history:
Received 21 February 2013Accepted 16 July 2013
Keywords:
Knee osteoarthritis
Neuromuscular alterations
Level walking
Systematic review
Background: Neuromuscular alterations are increasingly reported in individuals with knee osteoarthritis (KOA)
during level walking. We aimed to determine which neuromuscular alterations are consistent in KOA individualsand how these may be inuenced by osteoarthritis severity, varus alignment and/or joint laxity.
Methods:Electronic databases were searched up to July 2012. Cross-sectional observational studies comparing
lower-limb neuromuscular activity in individuals with KOA, healthy controls or with different KOA cohorts
wereincluded. Two reviewers assessed methodological quality. Effect sizes were used to quantify the magnitude
of observed differences. Where studies were homogenous, effect sizes were pooled using a xed-effects model.
Findings: Fourteen studies examining neuromuscular alterations in indices of co-contraction, muscle amplitude
and muscle activity duration were included. Data pooling revealed that moderate KOA individuals exhibit in-
creased co-contraction of lateral knee muscles (ES 0.64 [0.3 to 0.97]) and moderately increased rectus femoris
(ES 0.73 [0.23 to 1.22]), vastus lateralis (ES 0.77 [0.27 to 1.27]) and biceps femoris (ES 1.18 [0.67 to 1.7]) mean
amplitude. Non-pooled dataindicated prolonged activity of these muscles.Increased medial kneeneuromuscular
activity was prevalent for those exhibiting varus alignment and medial knee joint laxity.
Interpretation: Individuals with KOA exhibited increased co-contraction, amplitude and duration of lateral knee
muscles regardless of disease severity, limb alignment or medial joint laxity. Individuals with severe disease,
varus alignment andmedial joint laxitydemonstrate up-regulationof medialknee muscles. Future research inves-
tigating the ef
cacy of neuromuscular rehabilitation programs should consider the effect of simultaneous up-regulation of medial and lateral knee muscles on disease progression.
2013 Elsevier Ltd. All rights reserved.
1. Introduction
Knee osteoarthritis (KOA) is a progressive disease resulting in the
breakdown of joint cartilage and bone that is characterized by joint
pain, stiffness and swelling (Bombardier et al., 2011). Recently, several
studies have examined neuromuscular control of the lower limb in indi-
viduals with KOA during walking. The results of these studies provide
compelling evidence that neuromotor control of the lower limb is al-
tered in this population. Altered neuromuscular control in individuals
with KOA is concerning due to the deleterious effects on joint loading
and stability.
Co-ordinated agonistic and antagonistic muscle activities play major
roles in distributing load between the medial and lateral tibiofemoral
joints during walking. Altered muscle activity, either an increase in me-
dial activity or decrease in lateral activity, at the knee may increase the
demand on lateral soft tissue to resist the external knee adduction mo-
ment (the biomechanical proxy for medial knee joint load) (Schipplein
and Andriacchi, 1991). A high external knee adduction moment in an
individual with varus alignment and/or lateral joint laxity could lead
to a condition where the entire external knee adduction moment, and
potentially the dynamic load, occurs through the medial compartment
of the tibiofemoral joint (Schipplein and Andriacchi, 1991). This is be-
lieved to substantially increase in compressive forces and accelerate de-
generation of the medial compartment.
As the medial compartment degenerates, the distance between the
medialligamentinsertions is reduced andthe medialsoft tissuecontrib-
uting to joint stability can become lax (Lewek et al., 2004; Sharma et al.,
1999). Impairment in this passive stabilization system increases the de-
mand for coordinated neuromuscular activity to compensate. An inabil-
ity to adequately perform this task has been theorized to lead to
recurrent episodes of instability and further degenerative changes
(Lewek et al., 2004; Sharma et al., 1999). Thus, knee joint stability and
load distribution is achieved through an interaction between active
and passive strategies. As such, it is not surprising that increased neuro-
muscular activity of the medialknee muscles hasrecentlybeen correlat-
ed with rate of knee cartilage volume loss in individuals with medial
Clinical Biomechanics 28 (2013) 713724
Corresponding author at: Faculties of Kinesiology and Nursing, University of Calgary,
2500 University Drive NW, Calgary, Alberta, T2N 1N4 Canada.
E-mail address:[email protected](R. Ferber).
0268-0033/$ see front matter 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.clinbiomech.2013.07.008
Contents lists available atScienceDirect
Clinical Biomechanics
j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / c l i n b i o m e c h
http://dx.doi.org/10.1016/j.clinbiomech.2013.07.008http://dx.doi.org/10.1016/j.clinbiomech.2013.07.008http://dx.doi.org/10.1016/j.clinbiomech.2013.07.008mailto:[email protected]://dx.doi.org/10.1016/j.clinbiomech.2013.07.008http://www.sciencedirect.com/science/journal/02680033http://crossmark.crossref.org/dialog/?doi=10.1016/j.clinbiomech.2013.07.008&domain=pdfhttp://www.sciencedirect.com/science/journal/02680033http://dx.doi.org/10.1016/j.clinbiomech.2013.07.008mailto:[email protected]://dx.doi.org/10.1016/j.clinbiomech.2013.07.008 -
8/10/2019 A Systematic Review and Meta-Analysis of Lower Limb Neuromuscular Alteration Associates With Knee Osteoarthri
2/12
compartment KOA and varus lower limb alignment (Hodges et al.,
2012).
A growing number of studies are reporting neuromuscular out-
comes during walking in individuals with KOA. As such, a review of
how neuromuscular control of the lower limb is altered, with respect
to amplitude, duration and antagonistic activity, in individuals with
KOA is timely. To accomplish this task we conducted a systematic re-
view of cross-sectional observational studies comparing neuromuscular
activity in individuals with KOA with either similar healthy controls oracross KOAsubgroups. Our aim was to determinewhichneuromuscular
alterations are consistently observed in individuals with KOA and how
these alterations may be different in the presence of knee varus align-
ment and/or knee joint laxity. As there is evidence to suggest that the
nature and magnitude of neuromuscular alterations may be inuenced
by the severity of KOA (Astephen et al., 2008), we also examined the
effect of disease severity on neuromuscular alterations during level
walking. Our ndings will assist clinicians in designing conservative
rehabilitation programs that incorporate neuromuscular retraining as
well as highlight areas for future research.
2. Methods
2.1. Search strategy
We devised a search strategy for the following electronic databases:
MEDLINE, EMBASE, CINAHL, SportDiscus, PubMed and Web of Science
and included all references listed until the 30th July 2012. The search
terms and strategy were identical forall databases: (1) Knee osteoarthr*
OR gonarth*, (2) Walking OR gait, (3) Combined 1 AND 2, (4)
Neuromotor OR neuromuscular OR electromyography OR EMG OR mus-
cle activity and (5) Combined 3 AND 4. No restrictions were placed on
the year, status or language of publication. All titles returned by the
search strategy were reviewed by a single reviewer (KM), with abstracts
of those papers potentially meeting the selection criteria retrieved for
further consideration. Full text versions of studies meeting the selection
criteria, as determined by two reviewers(KM, RL), were obtained for in-
clusion in the review. Reference lists of included papers and published
knee osteoarthritis reviews were hand-searched (KM) to ensure alleligible data were included.
2.2. Selection criteria
Publications were required to be human-based, cross-sectional ob-
servational studies examining neuromuscular activity in a KOA cohort
and a comparator group during level walking. Diagnosis of KOA was re-
quired to be made using radiographic or clinical criteria. No restriction
was placed on disease severity, involved compartment or lower limb
alignment. Papers were excluded if they contained: (a) participants
with radiographically conrmed OA in other weight-bearing joints,
(b) participants who had undergone a total joint arthroplasty or joint
preservation surgery in the study limb or (c) participants who required
the assistance of walking aids. These exclusion criteria were imposeddue to potential confounding of results (Ouellet and Moffet, 2002;
Rudolph et al., 2007; Simic et al., 2011).
Included publications were juxtaposed for author names, afliation
and participant characteristics to reduce the risk of bias introduced
through duplicate data. Where multiple studies, authored by identical
authors, presented outcome variables from the same participant num-
bers, age, weight and sex ratio, only results of the publication with
higher methodological quality were included.
2.3. Methodological quality
Methodological quality of included papers was assessed using a
modied version (Munn et al., 2010) of a validated quality index for
non-randomized trials (Downs and Black, 1998). This version included
16 items assessing reporting quality, external validity, and internal
validity (bias and confounding). It does not include items relating to
an intervention but still includes items relating to the blinding of
observers. Two independent reviewers (KM, MAH) assessed all
papers, with the second reviewer blinded to author names, title,
afliation and journal name. Disagreements in initial ratings were
discussed at a consensus meeting. Publications scoring less than 50%
on the quality index were excluded from further review (Coombes
et al., 2010).
2.4. Data synthesis
Inter-rater reliability of the modied quality index was evaluated
with the kappa () statistic. The magnitude of agreement was quanti-ed usingHopkins (2000)system of very small (0 to 0.1), small (0.1
to 0.3), moderate (0.3 to 0.5), high (0.5 to 0.7), very high (0.7 to 0.9),
and almost perfect to perfect (0.9 to 1.0).
Two independent reviewers (KM, RL) extracted publication details
(authors, year, publication source and type), sample characteristics
(sample size, sampling technique, source of participants and selection
criteria), participants' characteristics (age, gender, severity of KOA,
limb alignment, knee joint laxity and method of diagnosis), neuromotor
characteristics (electrode type, size, shapeand placement,sampling fre-
quency and amplitude normalization) and point estimates of effect di-
rectly from included papers. Differences were settled by consensus.
Authors were contacted to provide any missing data and, if these were
not provided, they were qualitatively extracted from graph or result
sections. Point estimates of effect were used to calculate mean differ-
ences and 95% condence intervals (CI) between groups. An effect size
(ES = mean difference/pooled standard deviation) was used to quanti-
fy the magnitude of the differences and permit a common metric be-
tween studies. The magnitude of ES was interpreted using Hopkins
et al. (2009)criteria: trivial (0.0 to 0.2), small (0.21 to 0.6), moderate
(0.61 to 1.2) and large (N1.2). Qualitative descriptionsof differences be-
tween groups were also extracted from principle component analysis of
electromyography (EMG) waveforms.
Meta-analysis was performed using the ES in a xed-effects model
within Cochrane Review Manager (V5.1), using theI2 index to measureinconsistency (the percentage of total variation due to heterogeneity)
across included papers and referenced to thresholds of low (N25%),
moderate (50%) and high (75%) evidence of heterogeneity (Higgins
et al., 2003). The criteria for pooling were: (a) participants exhibiting
the same KOA severity, (b) neuromuscular activity being examined
over thesame time periods of thegaitcycle, (c) thesame normalization
methods and (d) co-contraction indices formulated using identical
methods. As this review did not include randomized control trials, a
combination of ES and heterogeneity index was used to quantify levels
of evidence based on those proposed byvan Tulder et al. (2003). Evi-
dence of neuromotor alterations associated with KOA was interpreted
as strong (large ES with low evidence of heterogeneity), moderate
(moderate ES and low evidence of heterogeneity), limited (small ES
with low heterogeneity or moderate/large ES with moderate evidenceof heterogeneity), conicting (high evidence of heterogeneity) and no
evidence (95% CI of ES crossed zero).
3. Results
3.1. Search strategy and study characteristics
The search strategy retrieved 489 papers. Fifteen papers met the se-
lection criteria and underwent quality assessment. No studies met the
criteria for duplicate data, although several studies were published
from thesame laboratory. Onepaper wasexcluded based on poor meth-
odological quality, thus 14 papers were included in the review (Fig. 1).
For 11 papers, inclusion in the KOA group was based on radiographic
and clinical criteria while the remaining three papers used only
714 K. Mills et al. / Clinical Biomechanics 28 (2013) 713724
-
8/10/2019 A Systematic Review and Meta-Analysis of Lower Limb Neuromuscular Alteration Associates With Knee Osteoarthri
3/12
radiographiccriteria. KOAseverity was reported by the authors of seven
papers and interpreted from the radiographic and clinical ndings
(e.g., joint space narrowing, osteophyte formation, or scheduled for
high tibial osteotomy) from a further two. From these studies, there
were nine cohorts of individuals with moderate KOA and four cohorts
with severe disease. Four studies included individuals with a range ofseverities within a single KOA group and were subsequently classied
as KOA.Schmitt and Rudolph (2007)also included individuals with a
range of KOA severities, however as over 60% of their cohort exhibited
mild radiographic changes and symptoms we considered them to be a
predominantly mild KOA cohort. Five studies examined individuals
with varus mechanical alignment. Individuals from three of these stud-
ies also exhibited increased medial knee joint laxity, quantied by stress
radiographs, compared with controls (Table 1).
Samplesizes ranged from 8 to 60 individuals in KOA groupsand 12 to
63 individuals in healthycontrol groups. Neuromuscular variables exam-
inedwere: (a) indicesof muscle co-contraction, (b) magnitude of muscle
activation and (c) muscle activity duration. The muscles examined were
vastus medialis (VM), vastus lateralis (VL), rectus femoris (RF),
biceps femoris (BF), semimembranosis (SM), tibialis anterior (TA),
soleus (Sol), medial gastrocnemius (MG) and lateral gastrocnemius
(LG). Electromyography data were primarily obtained using surface
bipolar, circular Ag/AgCl electrodes. The inter-electrode distances
ranged from 18 to 30 mm, which coincides with the distance between
peaks of the propagating action potential (Kamen and Caldwell, 1996).
All studies positioned electrodes over the muscle belly parallel withmuscle bers. Sampling rates ranged from 1000 to 2000 Hz and data
were normalized to maximum voluntary isometric contraction (MVIC)
in 12 papers and to maximum amplitude during walking (% Max) in
the remaining two (Liikavainio et al., 2010; Schmitt and Rudolph, 2007).
3.2. Methodological quality
The quality indices of included papers ranged from nine to 13 (of a
possible 17) and were interpreted as being, on average, moderate quality
(Table 2). Initial inter-rater reliability between the two reviewers was
very high (= 0.877) with items 18 (= 0.435) and 25 (= 0.562)
differing most. For the remaining items, there was almost perfect
agreement (Hopkins et al., 2009)(Table 2). Consensus was reached
on all differing items during the rst discussion. Overall, papers did
Literature search: knee osteoarthr*, gonarth*, walking, gait,
neuromotor, neuromuscular, electromyography, EMG, muscle activity
Databases:
MEDLINE: 39
EMBASE: 94
CINAHL: 31
Sportdiscus: 63
PubMed: 89
Web of Science: 153
Hand search: 20Search results combined: 489
Excluded n = 450
Non-human
Not knee OA
Duplicates
Post surgical
Experimental design
Review paper
Abstracts screened on the basis of title
n = 39
Full text paper retrieved on basis of
abstract n = 19
Excluded n = 20
No neuromotor variables
Modeling study
Not level walking
Papers considered for inclusion
n = 15
Excluded n = 4
Within group design (no
comparison group)
Papers included in review
n = 14
Excluded n = 1
Quality index score
-
8/10/2019 A Systematic Review and Meta-Analysis of Lower Limb Neuromuscular Alteration Associates With Knee Osteoarthri
4/12
-
8/10/2019 A Systematic Review and Meta-Analysis of Lower Limb Neuromuscular Alteration Associates With Knee Osteoarthri
5/12
Lewek et al.
(2006)
OA n = 15 (6/9) Age: 47.7 (7.4), height 1.75 m (0.09),
mass 91.9 kg (17.4)
Moderate Radiographic Medial compartment and PFJ OA excluded
Clinical Scheduled for high tibial osteotomy
Control n = 15 (6/9) Age: 48.4 (6.3), height 1.71 m (0.09),
mass 83.8 kg (17.3)
Radiographic medial joint space
width = 1.6 (1.1) mm
Medial joint laxity = 4.9 mm (1.8)
Lateral joint laxity = 3.5 mm (1.5)
Weight bearing line = 28.9% (12.7)
Liikavainio et al.
(2010)
OA n = 54 (men only) Age: 59 (5.3), BMI 29.7 kg/m2 (4.7) General OA Radiographic Most symptomatic limb used in analysis
Control n = 53 (men only) KL grade 1 to 4
Varus ranged from 3.2 (2.5) for
KL = 1 to 9.5 (2.5) for KL = 4
Rudolph et al.
(2007)
OA n = 15 (8/7) 0.1 (1.58)
varus
Age: 49.2 (4.5), BMI 30.7 kg/m2 (4.8) Moderate Radiographic Scheduled for high tibial osteotomy
Control n = 15 (8/7) Age: 49.2 (4.25), BMI 28.7 kg/m2 (5.5) Clinical 6.33 (2.39) varus
Rutherford et al.
(2010)
OA n = 17 (10/7) Age: 56 (8.8), BMI 29.8 kg/m2 (6.5) Moderate Radiographic Unil ateral knee OA
Control n = 20 (7/13) Age: 46.5 (7), BMI 25.9 kg/m2 (4.8) Clinical Able to walk a city block, reciprocally
negotiate
10 stairs and jog 5 m
Rutherford et al.
(2011)
Moderate n = 16 (8/8) Age: 61 (6), BMI 31.3 kg/m2 (3.6) Moderate Radiographic Predominantly medial compartment
Severe n = 15 (10/5) Age: 61 (9), BMI 30.7 kg/m2 (5.4) Severe Clinical Severe group scheduled for TKA
Control n = 16 (8/8) Age: 56 (96), BMI 24.6 kg/m2 (3.9) All OA groups able to walk N1.0 m/s
Schmitt and
Rudolph
(2007)
OA n = 28 (14/14) Age: 60.4 (9.75), Height 1.7 m (0.11),
mass 92.91 kg (16.16)
Median =
mild
Radiographic Unilateral and bilateral knee OA
(most symptomatic knee analyzed)
Control n = 26 (13/13) Age: 58.5 (9.5), height 1.68 m (0.07),
mass 83.93 kg (1.85)
Medial compartment
(KL grade N2 in lateral and PFJ excluded)
KL grades 2 to 4 (61% exhibited KL grade =
2)
5.15 (3.43) varus Medial joint laxity = 4.23 mm (1.8)
Lateral joint laxity = 2.77 mm (1.35)
Zeni et al. (2010) Moderate n = 16 (6/10) Age: 62.8 (10) Moderate Radiographic KL grades 2 to 3 for moderate and 4 for
severe.
Severe n = 8 (3/5) Age: 62.2 (8) Severe Medial compartment
Control n = 18 (8/10) Age 61 (11)
-
8/10/2019 A Systematic Review and Meta-Analysis of Lower Limb Neuromuscular Alteration Associates With Knee Osteoarthri
6/12
not adequately report the sampling method used to recruit participants
(item 11) or provide the proportion of participants who were asked to
participate compared with those who agreed to participate (item 12).
No study reported if they blinded assessors to group allocation (KOAor control) duringanalysis of primary outcomemeasures, and few stud-
ies provided adequate information regarding the source of participants
included in the study or the time period for recruitment. These issues
are possible sources of confounding and bias.
3.3. Muscle co-contraction
Ten studies examined the magnitude of co-contraction associated
with KOA. Co-contraction indices were grouped into those specically
comparing medial thigh and shank muscles (VM:SM, VM:MG), those
comparing lateral thigh and shank co-activity (VL:BF, VL:LG, TA:LG)
and those comparing medial muscles with lateral muscle (VM:BF and
VL:SM). In order to examine the inuence of disease severity, compari-
sons were partitioned into those examining general, mild, moderatedand severe KOA cohorts when compared with healthy controls and
when compared with other KOA severities.
3.3.1. KOA cohorts versus controls
Two studies, examining individuals with moderate KOA, met the
criteria for pooling (Hubley-Kozey et al., 2009; Lewek et al., 2004).
Data pooling revealed the only co-contraction index that was systemat-
ically increased was VL:BF, where there was moderate evidence of a
greater co-contraction (ES 0.64 [0.3 to 0.97] I2 = 0%) in individuals
with moderate KOA, compared with healthy controls (Fig. 2A).
Non-pooled data of lateral muscle co-contraction also indicate indi-
viduals with KOA walk with increased lateral muscle co-contraction re-
gardless of disease severity, varus alignment and knee laxity (Childs
et al., 2004; Hortobgyi et al., 2005; Lewek et al., 2006; Schmitt and
Rudolph, 2007)(Table 3). Further,Heiden et al. (2009)reported that
lateral muscles provided the greatest contribution to an increased co-
contraction between knee extensors and knee exors. However, lateral
thigh muscle co-contraction was only signicantly increased duringmidstance (Lewek et al., 2006; Schmitt and Rudolph, 2007). This sug-
gests that while lateral shank muscle co-contraction is increased in
KOA during all phases of the stance phase, increased thigh muscle co-
contraction only occurs when the osteoarthritic limb is fully loaded.
Non-pooled data of medial co-contraction indices suggests that me-
dial knee joint laxity may inuence the magnitude of medial co-
contraction in KOA. Studies examining KOA participants with medial
knee joint laxity report signicantly increased medial shank and thigh
muscle co-contraction regardless of disease severity (Lewek et al.,
2004, 2006; Schmitt and Rudolph, 2007) (Table 3). In studies by
Lewek et al. (2006, 2004) moderate increases (ES 0.84 and 0.83) in
VM:MG were observed during preparation and weight acceptance.
Thus, the presence of greater medial muscle co-contraction may reect
an attempt to increase medial stability by increasing compressive forcesacross the knee.
Non-pooled data of medial and lateral thigh muscle co-contraction
suggested disease severity might inuence the magnitude of co-
contraction (Table 3).Zeni et al. (2010) reported VL:SM co-contraction
was only consistently increased across a range of walking speed in
those with moderate KOA. In contrast, co-contraction indices were not
increased, compared with healthycontrols, in general and severe KOAco-
hortswhentheywalked at a self selected,or fast speeds (Liikavainio et al.,
2010; Zeni et al., 2010).
3.3.2. Between KOA cohorts
Two studies compared co-contraction between individuals with
moderate and severe KOA (Hubley-Kozey et al., 2009; Zeni et al.,
2010). No data pooling was possible. Over similar phases of the gait
Table 2
Modied quality index.
Reporting External validity Internal validity - bias Internal validity - confounding
Publication
1.Hypothesis
clearlydescribed?
2.Mainoutco
mesclearly
described?
3.C
haracteristicsofthepatients
includedclearlydescribed?
5.D
istributionofprinciple
confoundero
feachgroupclearly
described?a
6.Mainfindingsclearly
described?
7.Estimateso
frandom
variabilityprovidedforthemain
outcomes?
10.Actualprobabilityvalues
reportedformainoutcomes?
11.Werethe
subjectsaskedto
participatere
presentativeofthe
entirepopula
tion?
12.Werethe
subjectswhowere
preparedtop
articipate
representativ
eoftheentire
15.Wasthereanattempttoblind
thosemeasur
ingthemain
outcomes?
16.Wasitcle
ariftheresults
werebasedondatadredging
18.Werethe
statisticaltests
appropriate?
20.Werethe
mainoutcome
measuresvalidandreliable?
21.Wereallp
atientsandcontrols
recruitedfrom
thesame
population?
22.Wereallp
atientsand
controlsrecruitedoverthesame
timeperiod?
25.Wastheir
adequate
adjustmentforconfounding?
b
Total(17)
Astephen et al. (2008) 1 1 0 2 0 1 1 0 0 0 1 1 1 0 0 0 9
Childs et al. (2004) 1 1 1 2 1 1 1 0 0 0 1 1 0 0 0 0 10
Heiden et al. (2009) 1 1 1 1 1 1 1 0 0 0 1 1 1 0 0 1 11
Hortobgyi et al. (2005) 1 1 1 1 1 1 0 0 0 0 1 1 1 0 0 0 9
Hubley-Kozey et al. (2006) 1 1 1 2 1 1 1 0 0 0 1 1 1 1 0 0 12
Hubley-Kozey et al. (2009) 1 1 1 2 1 0 0 0 0 0 1 1 1 0 1 0 10
Lewek et al. (2004) 1 1 1 2 1 1 1 0 0 0 1 1 1 0 0 1 12
Lewek et al. (2006) 1 1 1 2 1 1 1 0 0 0 1 0 1 0 0 1 11
Liikavanio et al. (2010) 1 1 1 2 1 1 0 0 0 0 1 1 0 1 0 1 11
Rudolph et al. (2007) 1 1 1 2 1 1 1 0 0 0 1 1 1 0 0 1 11
Rutherford et al. (2010) 1 1 1 1 1 1 1 0 0 0 1 1 1 0 0 0 10
Rutherford et al. (2011) 1 1 1 2 1 1 1 0 0 0 1 1 1 0 1 1 13
Schmitt and Rudolph (2007) 1 1 1 2 1 1 1 0 0 0 1 1 1 1 0 0 12
Zeni et al. (2010) 1 1 0 1 1 1 1 0 0 0 1 1 1 0 0 1 10
Kappa levels of agreement 0.98 1.0 0.63 1.0 0.98 1.0 1.0 0.98 0.97 0.97 1.0 0.44 1.0 1.0 1.0 0.56 0.877
Allitems,except item5, were scored1 forfullling thecriterionor 0 ifthe criterion were notlled. Publications that didnot providesufcient detailsto fulll thecriterionwere alsogivena
0 for unable to be determined as perinstructions of theoriginal index. Kappa scores indicate level of initialagreementbetween reviewers prior to theconsensus meeting. Thetotal kappa
score indicates overall level of agreement.a Thiscategorywas interpretedas theKOA diagnosis being clearlydescribedwithrespect to radiographic severity, clinicalseverityand mechanicalalignment.If allthreeof thecriteria were
described, 2 points were awarded for yes. If two of the criteria were described, 1 point was awarded forpartially.b A full mark was awarded in this category if authors demonstrated no statistically signicant difference in walking speed or pain between KOA group(s) and control or if analyses of
covariance were conducted. Publications that did not investigate differences in walking speed or pain or did not adjust for these potential confounders were awarded unable to be
determined.
718 K. Mills et al. / Clinical Biomechanics 28 (2013) 713724
-
8/10/2019 A Systematic Review and Meta-Analysis of Lower Limb Neuromuscular Alteration Associates With Knee Osteoarthri
7/12
cycle,Hubley-Kozey et al. (2009) reported individuals with severe KOA
exhibited moderately greater co-contraction indices in both medial and
lateral muscle groups (VL:BF: ES 0.62, VL:LG: 0.68 and VM:MG: 0.61)
than those with moderate disease. In contrast, Zeni et al. (2010)found
no difference in VL:SM co-contraction, regardless of whether
participants walked at 1.0 m/s, self-selected or fast walking speeds
(Table 3). This difference in resultmay be dueto thestudies using differ-
ent muscles in their calculations of co-contraction or due to calculating
co-contraction using different equations.
3.4. Muscle amplitude
Ten studiesexamined alterations in amplitude of the gastrocnemius,
soleus, quadriceps and hamstrings associated with KOA during level
walking. Alterations in mean, peak and net (the sum of all normalized
muscle activity (Heiden et al., 2009)) muscle amplitude were investi-
gated. The potential inuence of disease severity was the focus of sever-
al publications and walking data from those with general, mild,
moderate and severe KOA were examined. ES, calculated from differ-
ences in principal components between individuals with moderate
KOA and healthy controls (Rutherford et al., 2010, 2011), was pooled
for gastrocnemius, quadriceps and hamstring amplitudes.
3.4.1. Between KOA and controls
Pooled data revealed no evidence of gastrocnemius overall ampli-
tude alteration associated with moderate KOA(Fig. 2B). While this nd-
ings was supported by results of studies investigating individuals with
severe KOA (Astephen et al., 2008; Rutherford et al., 2011), data from
non-pooled studies investigating those with moderate KOA suggest
that results may be inconclusive. Two studies, that were not pooled,
supported the nding from pooled data (Astephen et al., 2008;
Rudolph et al., 2007). In contrast,Hubley-Kozey et al. (2006)reported
a decrease in mean MG amplitude andHeiden et al. (2009)reported a
net increase in muscle amplitude (Table 3). This suggests that if individ-
uals with moderate KOA exhibit altered gastrocnemius amplitude, there
is no consistency in the direction or magnitude of that alteration.
Moderate increases in VL (ES 0.77 (0.27 to 1.27) I2 = 72%) and RF
(ES 0.73 (0.23 to 1.22)I2 = 0%) amplitude were observed from pooleddata of individuals with moderate KOA compared with healthy controls
(Fig. 2B). Of the eight studies, that were not pooled, only one (Rudolph
et al., 2007) did not report an increase in mean or peak quadriceps am-
plitude in individuals with KOA regardless of disease severity or joint
laxity. This suggests that increases in VL and RF amplitude are consis-
tently associated with KOA regardless of disease severity or joint laxity.
VM mean and net amplitude was only reported to be increased in stud-
ies examining general and predominantly mild KOA cohorts (Heiden
et al., 2009; Liikavainio et al., 2010; Schmitt and Rudolph, 2007).
There was no increasein VM mean amplitude in moderate or severe co-
horts compared with controls (Hubley-Kozeyet al.,2006; Rudolphet al.,
2007; Rutherford et al., 2011).
Pooled data reveal a systematic, moderate increase in amplitude of
BF in those with moderate KOA compared with healthy controls (ES1.18 (0.67 to 1.7) I2 = 0%)(Fig. 2B). This nding was supported by a
Schmitt and Rudolph (2007),investigating differences in mean ampli-
tude between a predominantly mild KOA cohort and controls. For gen-
eral andsevere KOAcohorts, there were also trends(Zeni et al., 2010) or
signicant increases (Astephen et al., 2008; Heiden et al., 2009;
Liikavainio et al., 2010) in hamstring mean amplitude (Table 3). While
SM mean and peak amplitude moderately increased compared with
controls, at a range of walking speeds (Zeni et al., 2010), its mean ampli-
tude was consistently lower than BF (Hubley-Kozey et al., 2006;
Rutherford et al., 2010, 2011).
3.4.2. Between KOA cohorts
Three studies examined differences in muscle amplitude between
individuals with moderate and severe KOA (Astephen et al., 2008;
Rutherford et al., 2011; Zeni et al., 2010). As with KOA-control compar-
isons, alterations in gastrocnemius mean amplitude were inconsistent
in direction and magnitude between those with moderate and severe
KOA(Astephen et al., 2008; Rutherford et al., 2011). There wasno differ-
ence in quadriceps mean amplitude between moderate and severe KOA
groups (Astephen et al., 2008; Rutherford et al., 2011; Zeni et al., 2010),
except during fast walking where mean and peak VL activity was mod-
erately increased in individuals with moderate KOA (ES 0.73 and 0.86)
(Zeni et al., 2010). There was a large reduction in BF mean amplitude(ES 1.46) in those with moderate disease, but only during midstance
(Rutherford et al., 2011) (Table 3).
3.5. Muscle activity duration
Five studies reported on differences in muscle activity duration be-
tween individuals withKOA and controls. Comparisons of gastrocnemi-
us, quadriceps and hamstrings muscles were made between individuals
with moderate and severe KOA and controls (Table 3). No data pooling
was performed.
3.5.1. Between KOA and controls
Data extracted from three studies, which analyzed gastrocnemius du-
rations using principal component analysis, revealed the duration of MG
activity did not differ in individuals with moderate KOA compared with
healthy controls (Astephen et al., 2008; Hubley-Kozey et al., 2006; Ruth-
erford et al., 2011). This nding was not supported by a study examining
duration using discrete variables, which reported a moderate increased in
MG duration (ES 0.97) (Childs et al., 2004). Two studies compared gas-
trocnemius duration between individuals with severe KOA and controls
reported those with severe KOA exhibited longer duration with MG acti-
vating later than LG (Astephen et al., 2008; Rutherford et al., 2011). This
was a different activation pattern than controls.
Waveform and discrete analysis indicate that the activity of the VL
and RF was prolonged in both moderate and severe KOA, compared
with controls (Childs et al., 2004; Hubley-Kozey et al., 2006; Rutherford
et al., 2011). There was no change in VM temporal characteristics for
those with moderate KOA (Hubley-Kozey et al., 2006).
With the exception of onestudy (Rutherford et al., 2011), hamstringactivity was observed to be prolonged in individuals with KOA. While
both BF and SM durations were increased in those with moderate
KOA, compared with controls (Rutherford et al., 2010), BF activity was
prolonged compared with SM in the KOA group. Analysis of discrete
values suggests a large effect (SM: ES 1.38) (Zeni et al., 2010)(Table 3).
3.5.2. Between KOA groups
Two studies examined differences in muscle activity and duration
between individuals with moderate and severe KOA (Astephen et al.,
2008; Rutherford et al., 2011). Compared with individuals with moder-
ate KOA, those with severe disease exhibited delayed onset of MG
(Rutherford et al., 2011) and longer duration (Astephen et al., 2008).
RF activity was observed to be prolonged for both groups but more so
in the severe group and there was no difference in hamstring duration(Rutherford et al., 2011).
4. Discussion
The primary nding of this review is that many individuals with KOA
exhibit altered neuromuscular activity across co-contraction, amplitude
and temporal domains, when compared with healthy controls. Specical-
ly, pooled resultsdemonstrated that individuals withmoderate KOA con-
sistently exhibited moderately increased lateral thigh co-contraction
compared with healthy controls over the period of 100 ms prior to heel
contact until the peak external knee adduction moment (approximately
25% of stance). Pooled data also revealed those with moderate KOA
walk with consistent moderate increases in RF and BF amplitude and a
moderate, although heterogeneous, increase in VL amplitude. While
719K. Mills et al. / Clinical Biomechanics 28 (2013) 713724
-
8/10/2019 A Systematic Review and Meta-Analysis of Lower Limb Neuromuscular Alteration Associates With Knee Osteoarthri
8/12
non-pooled data cannot provide the precision of meta-analysis, the num-
ber of studies reporting elevated lateral muscle co-contraction and
prolonged VL,RF and hamstring duration suggests thatthese arealso con-
sistently associated with KOA regardless of disease severity, knee joint
laxity or varus alignment. Medial knee joint laxity may have an important
inuence on medial knee muscleco-contractionas it wasonlyconsistent-
ly observed in individuals with laxity, regardless of disease severity. Dis-
ease severity does appear to inuence the presence of increased medial:
lateral muscleco-contraction, as it was only signi
cantlyincreased, acrossa range of walking speeds,in those with moderate KOA (Zeni et al., 2010).
Concurrent increases in medial and lateral muscle amplitudes, as ob-
served inthe general KOA group(Liikavainio et al.,2010), could potential-
ly explain the lack of increase in the co-contraction index. However, for
the severe KOA group, who only exhibited an increased co-contraction
when there wasa concurrent increase in medialandlateralmuscleampli-
tudes (Zeni et al., 2010), it could be due to the observed slower walking
speeds or reect an arthrogenic muscle inhibition component in the
disease process (Hurely, 1999).
Several authors have hypothesized that the primary reason why
individuals with KOA adopt neuromuscular alterations is to unload
the medial knee joint (Heiden et al., 2009; Hortobgyi et al., 2005;
Hubley-Kozey et al., 2009). Further, they suggest that lateral muscle
activity has a protective effect on the knee and preserves knee function.
Findings from this review demonstrate that increased lateral muscle
activity is a prevalent adaptation, particularly when the knee is
fully loaded. Recently, a delay in lateral knee muscle co-contraction
was correlated with increased lateral cartilage loss in KOA patients
with varus malalignment over a 12-month period ( Hodges et al.,
2012). However, further research is needed to determine whether
increased lateral muscle activity is a neuromuscular adaptation that
should be encouraged as a protective mechanism in individuals
with KOA.
A second, frequently postulated reason for increased neuromuscular
activity is the need for increased knee joint stability (Heiden et al., 2009;
Hortobgyi et al.,2005; Hubley-Kozey et al., 2009). Specically, increased
coordinated muscle activity is required when the passive stabilization
provided by soft tissue is compromised (Schipplein and Andriacchi,
1991). The ndings of this review support this statement. Medial co-
contraction indices were signicantly greater than controls for individ-
uals with medial knee joint laxity (Hubley-Kozey et al., 2009; Lewek
et al., 2004, 2006; Schmitt and Rudolph, 2007). This suggests that onlythose individuals who have a lack of medial passive knee support, poten-
tially through an approximation of medial ligament insertions resulting
from the loss of cartilage and bone height (Lewek et al., 2004; Sharma
et al., 1999), exhibit increased medial muscle activation.
Even though individuals without knee joint laxity exhibited similar
medial knee muscle neuromuscular activity as healthy control, they
exhibited elevated and prolonged activity of RF regardless of disease se-
verity. This could also be an attempt to protect and stabilize the knee.
However, current evidence questions the efcacy of increased neuro-
muscular activity of the RF and medial knee muscles in accomplishing
this task (Hodges et al., 2012; Schipplein and Andriacchi, 1991;
Schmitt and Rudolph, 2008). Increased neuromuscular activation of
muscles that cross the knee increases compressive forces. While this
would increase stability, it would also increase load on the joint and
could accelerate degeneration of the medial compartment (Schipplein
and Andriacchi, 1991). Further, increased duration medial knee muscle
co-contraction hasrecently been correlated with increasedloss of medi-
al knee cartilage volume (Hodges et al., 2012). Interestingly, it is possi-
ble for increased neuromuscular activity of medial and lateral knee
muscles as well as RF to co-exist as demonstrated in a severeKOA cohort
(Hubley-Kozey et al., 2009) and in a predominantly mild KOA cohort
with varus alignment and medial knee joint laxity (Schmitt and
Rudolph, 2007). However, due to the cross-sectional nature of studies
Fig. 2. Forest plot of data pooling. Filled diamonds indicate pooled data ES and 95% condence intervals. Condence intervals that do not cross zero indicate a systematiceffect. The con-
sistencyof neuromuscularalterations wasdeterminedbasedon thepresence of a systematiceffectanda low I2 index. (A)PooledES forco-contraction indices.(B) PooledES foramplitude.
Amplitude ES was calculated from mean (SD) factor scores of principal components so mean differences were not calculated.
720 K. Mills et al. / Clinical Biomechanics 28 (2013) 713724
-
8/10/2019 A Systematic Review and Meta-Analysis of Lower Limb Neuromuscular Alteration Associates With Knee Osteoarthri
9/12
Table 3
Results of included papers.
Authors Variable Knee OA Comparator Result/mean difference (95% CI) and effect size (ES)
Co-contraction index
Childs et al. (2004) VL:BF (% MVIC) General OA Control 13.0 (2.37 to 23.63) ES 0.68
TA:MG (% MVIC) 9.0 (4.17 to 13.83) ES1.04
Heiden et al. (2009) Muscles: lateral muscles
(SM, VM, MG:BF, VL, LG % MVIC)
General OA Control OA group exhibited signicantly greater co-contraction during loading and early
stance and signicantly less during midstance. Lateral muscle activation was the
greatest contributor to the co-contraction index
SM:BF (% MVIC) OA group exhibited greater co-contraction during loading, early and midstance.BF activation was the major contributor to the co-contraction index.
Knee exors: knee extensors
(SM, BF, MG, LG:VL, VM, RF %
MVIC)
OA group exhibited greater co-contraction during midstance. There was no
difference between groups during loading and early stance.
Hortobgyi et al. (2005) 200 ms prior to IC to toe-off General OA Control
BF:VL (% MVIC) 41.0 (23.85 to 58.15) ES 1.26
LG:TA (% MVIC) 40 (15.95 to 64.05) ES 0.9
Hubley-Kozey et al.
(2009)
100 ms prior to IC to peak KAM
VL:BF (% MVIC) Moderate Control 10.17 (4.39 to 15.94) ES 0.66
Sever e Control 23 .49 (17 .07 to 2 9.9) ES 1.52
Moderate Severe 13.32 (21.52 to 5.12) ES 0.62
VL:LG (% MVIC) Moderate Control 2.16 (0.69 to 5.02) ES 0.28
Severe Con trol 9. 57 (5.72 to 13.42) ES 1.01
Moderate Severe 7.41 (11.68 to 3.14) ES 0.68
VM:SM (% MVIC) Moderate Control 4.46 (0.73 to 9.65) ES 0.32
Severe Con trol 7. 6 (3.26 to 11.94) ES 0.32
Moderate Severe
3.14 (
9.05 to 2.77) ES
0.2VM:MG (% MVIC) Moderate Control 1.22 (1.27 to 3.71) ES 0.18
Severe Control 6.88 (10.21 to 3.64) ES 0.85
Moderate Severe 5.66 (9.26 to 2.06) ES 0.61
Lewek et al. (2004) 100 ms prior to IC to peak KAM
VL:BF (% MVIC) Moderate
(varus and lax)
Con tro l 5.7 (2.55 to 13.95) ES0.53
VL:LG (% MVIC) 2.8 (3.21 to 8.81) ES 0.36
VM:SM (% MVIC) 0.4 (7.13 to 6.33) ES 0.05
VM:MG (% MVIC) 6.3 (0.5 to 12.1) ES 0.84
Lewek et al. (2006) Preparation Moderate
(varus and lax)
Control
VL:BF (% MVIC) 4.6 (2.63 to 11.83) ES 0.44
VL:LG (% MVIC) 2.9 (2.9 to 8.7) ES 0.35
VM:SM (% MVIC) 0.7 (6.24 to 4.84) ES 0.09
VM:MG (% MVIC) 6.2 (1.01 to 11.39) ES 0.83
Liikavainio et al. (2010) VM: BF (% Max) G eneralOA (var us) C ontrol No d ifferenc e between groups
Rudolph et al. (2007) 100 ms prior to IC to peak KAM Moderate (varus) Control No difference between groups
VL:BF (% MVIC)VL:LG (% MVIC)
VM:SM (% MVIC)
VM:MG (% MVIC)
Schmitt and Rudolph
(2007)
Preparation and weight
acceptance
Mild
(varus and lax)
Control
VL:BF (% Max) No difference between groups
VL:LG (% Max) OA group exhibited a trend towards greater co-contraction during preparation
and signicantly higher co-contraction during weight acceptance.
VM:SM (% Max) OA group exhibited a trend towards greater co-contraction during preparation
and signicantly higher co-contraction during weight acceptance.
VM:MG (% Max) No difference between groups
Midstance OA group exhibited higher co-contraction for all muscles
Zeni et al. (2010) VL:SM (% MVIC) (1.0 m/s) Moderate Control 12.4 (5.81 to 18.99) ES 1.29
Severe Control 11.5 (2.21 to 20.79) E S 1.33
Moderate Severe 0.9 (9.91 to 11.71) ES 0.07
VL:SM (% MVIC) (PW) Moderate Control 11.7 (4.81 to 18.59) E S 1.16
Severe Control 6.8 (2.51 to 16.11) ES 0.75Moderate Severe 4.9 (5.85 to 15.65) ES 0.37
VL:SM (% MVIC) (FW) Moderate Control 14.4 (5. 27 to 23.53) ES 1.08
Severe Control 6.2 (2.24 to 14.64) ES0.72
Moderate Severe 8.2 (3.41 to 19.81) ES 0.5
Muscle amplitude
Astephen et al. (2008) Gastrocnemius (LG, MG % MVIC) Moderate Control No difference between groups
Severe C ontrol No d if ferenc e between groups
Moderate Severe Severe OA group exhibited higher mean MG amplitude in early stance and in
swing phase and lower mean amplitude in late stance.
Quadriceps (VL, VM, RF % MVIC) Moderate Control Mean RF amplitude was higher in moderate OA group throughout the majority
of stance and late swing.
Severe Control Mean RF amplitude was lower in severe OA group during early stance and early
swing but higher mean amplitude in mid to late stance.
Moder ate Sever e No d if ferenc e between groups
(continued on next page)
721K. Mills et al. / Clinical Biomechanics 28 (2013) 713724
-
8/10/2019 A Systematic Review and Meta-Analysis of Lower Limb Neuromuscular Alteration Associates With Knee Osteoarthri
10/12
Table 3(continued)
Authors Variable Knee OA Comparator Result/mean difference (95% CI) and effect size (ES)
Hamstrings (BF, SM % MVIC) Moderate Control No difference between groups
Severe Control Mean SM and BF amplitude was higher in severe OA group during stance.
Moderate Severe No difference between groups
Heiden et al. (2009) Net muscle activity of
BF, SM, VL, RF, VL, MG, LG (%
MVIC)
General OA Control OA group exhibited signicantly greater net muscle activation during weight
acceptance and early stance compared with controls
Hortobgyi et al.
(2005)
BF (% of MVIC) General OA Control 47 (28.18 to 65.82) ES 1.27
VL (% of MVIC) 47 (20.53 to 59.47) ES 1.09
Hubley-Kozey et al.
(2006)
Gastrocnemius (MG, LG % MVIC) Moderate Control Mean MG amplitude was decreased in OA group and peak amplitude occurred
earlier in gait cycle. The control group recruited MG more than LG, whereas OA
group recruited both muscles equally.
Quadriceps (RF, VL, VM % MVIC) Mean VL amplitude was higher in OA group during initial stance and there was a
trend towards higher mean RF amplitude. Mean VM amplitude was similar
between OA and controls.
Hamstrings (BF, SM % MVIC) Mean B F amplitude w as h igher a nd p eak a mplitude o ccurred p rior t o heel s trike i n
OA group compared with controls. Mean BF amplitude was higher than SM for
both groups.
Liikavainio et al.
(2010)
BF (% Max) General OA Control OA group exhibited signicantly higher mean BF amplitude at IC, except when
walking fast.
VM (% Max) Mean VM amplitude in late stance and early swing were signicantly greater in
the OA group for all gait speeds
Rudolph et al. (2007) VL, VM, BF, SM LG, MG (% MVIC) Moderate
(varus)
Control Tendency towards greater mean MG activity in the OA group. No other difference
between groups.
Rutherford et al.
(2010)
Gastrocnemius (MG, LG % MVIC) Moderate Control OA group exhibited moderately less difference between early and late stance
activity than controls for MG activity (ES 0.68) and a moderate tendency ofsmaller difference in LG activity (ES 0.63)
Quadriceps (VL, VM, RF % MVIC) OA group exhibited large reduction in difference between mid and late stance
amplitude for RF (ES 1.2) and small tendency towards smaller difference
between mid and late stance amplitude for VL and VM (ES 0.59, 0.55). RF
differential was moderately less than VM in OA group (ES 0.64), and large
reduction compared with VL (ES 1.8) and VM (ES 1.71) in control group.
Ha mstrings ( BF, SM % MVI C) OA group exhib ited which w as m od erately great er overall a mp litu de of B F than
SM (ES 1.09). OA group SM amplitude differential between early stance and late
swing versus midstance was moderately greater than control group (ES 0.86)
Rutherford et al.
(2011)
Gastrocnemius (MG, LG % MVIC) Moderate Control OA group exhibited moderately smaller difference between early and late stance
amplitude (ES 0.83)
Severe Control OA group exhibited large reductions in late stanceearly stance amplitude
differential for MG (ES 1.26) and LG (ES 1.21)
Severe Moderate Moderate OA group exhibited moderately greater mean MG amplitude (ES 0.74)
and moderately larger difference between early and latestance amplitude (ES 0.91).
Quadriceps (RF, VM, VL % MVIC) Severe Control DifferenceinRFamplitudebetweenearlyandmidlatestancewas moderatelylower
in OA group (ES 0.89). Difference between mid
late stance and swing phaseamplitude was moderately higher in OA group for VL (ES 0.82) and RF (ES 0.83).
Moderate Severe No difference in mean amplitude. Severe group exhibited moderately greater
difference between midlate stance and swing phase VL amplitude compared
with moderate group (ES 0.83)
Hamstrings (BF, SM % MVIC) Moderate Control OAgroup exhibited moderate reduction inmean SMamplitude duringmidstance
and late swing burst (P= 0.05, ES 0.69).
Severe Control OAgroup exhibited large reductionsin mean BFamplitude duringmidstance and
late swing burst (P= 0.00, ES 1.46).
Moderate Severe Moderate group exhibited moderately lower BF mean amplitude during
midstance and late swing burst (P= 0.04, ES 0.73)
Schmitt and Rudolph
(2007)
VL (% Max) Mild (varus and
lax)
C ontrol Signicantly greater mean amplitude in OA group during midstance
VM (% Max) Signicantly greater mean amplitude in OA group during weight acceptance and
midstance
BF (% Max) Trend towards greater mean amplitude in OA group during preparation,
signicantly greater mean amplitude during midstance
SM (% Max) Signicantly greater mean amplitude in OA group during weight acceptance
LG (% Max) Trend towards greater mean amplitude in OA group during preparation,
signicantly greater mean amplitude during weight acceptance
MG (% Max) No difference between groups
Sol (% Max) No difference between groups
Zeni et al. (2010) VL (mean % MVIC) (1.0 m/s) Moderate Control 16 (7.06 to 24.94) ES 1.24
Severe Control 10.3 (1.5 to 19.1) E S 1.29
Moderate Severe 5.7 (6.46 to 17.86) ES 0.34
V L (mean % MVI C) ( PW) Mod erate C ontrol 15 .5 ( 6.06 to 24.4 9) ES1 .14
Severe Control 8.9 (1.02 to 18.82) ES 1.02
Moderate Severe 6.6 (6.78 to 19.98) ES 0.37
V L (mean % MVI C) ( FW ) Mod erate C ontrol 18.4 ( 5.21 to 31 .5 9) ES 0 .9 7
Severe Control 1.7 (5.7 to 9.1) ES 0.27
Mod erate Severe 16.7 ( 2.28 to 31 .1 2) ES0 .73
VL (peak % MVIC) (1.0 m/s) Moderate Control 26.8 (10.99 to 42.61) ES 1.18
Severe Control 21.2 (6.29 to 36.11) ES 1.55
Moderate Severe 5.6 (15.45 to 26.65) ES 0.19
Muscle amplitude
Astephen et al. (2008)
722 K. Mills et al. / Clinical Biomechanics 28 (2013) 713724
-
8/10/2019 A Systematic Review and Meta-Analysis of Lower Limb Neuromuscular Alteration Associates With Knee Osteoarthri
11/12
included in this review, no conclusions can be drawn regarding the in-
teraction between the potentially negative effects of increased medial
knee muscle and RF activity and the potential positive effects of in-creased lateral activity. Future research is needed to determine how
this interaction affects KOA progression.
Along with the inability to draw conclusions regarding inuenceson
the progression of KOA, there are several additional limitations of this
review. The primary methodological concern of the studies included
in this review was poor external validity specically a lack of
reporting sampling methods and demonstrating that the people who
participated in the study are representative of the population they
were selected from. This makes generalizing results beyond individuals
who meet specic inclusion criteria difcult. Another methodological
issue was the variety of formulas used to calculate co-contraction indi-
ces highlighting the numerous methods to achieve a single number.
As such, interpretation of this index is difcult unless there is also an in-
vestigation into how this number is reached. The relative contributions
of muscles involved in any co-contraction formula are of particular in-
terest to clinicians, particularly if the ratio contains both medial and lat-
eral muscle contributions. Some studies explore this (Heiden et al.,2009; Hubley-Kozey et al., 2009; Zeni et al., 2010) and future studies
would benet from similar exploration.
An interesting issue in interpreting the validity of thendings of this
review is that the majority of studies normalized their neuromotor data
to MVIC. Normalizing EMGdata to MVIC provides information regarding
the relative degree of muscle activation and facilitates comparisons be-
tween groups. However, in injured and post-operativepatients, the abil-
ity to perform an MVIC can be affected by pain (Arvidsson et al., 1986;
Benoit et al., 2003). In KOA cohorts, higher pain levels and lower physi-
cal function have been associated with increased neuromotor activity
independent of radiographic disease severity (Heiden et al., 2009;
Wilson et al., 2011). Thus, individuals with KOA may not be able to pro-
duce an accurate MVIC and subsequent signicant differences in neuro-
muscular activity may be erroneous. Conversely, reported increases in
Table 3(continued)
Authors Variable Knee OA Comparator Result/mean difference (95% CI) and effect size (ES)
VL ( peak % MVIC) (PW) Moder ate Control 23 .7 (7.31 to 40 .0 2) ES 1 .0 1
Severe Control 16.1 (1.69 to 33.89) ES 1.0
Moderate Severe 7.6 (15.75 to 30.95) ES 0.24
VL ( peak % MVIC) (FW) Moder ate C ontrol 40 .20 (14 .31 to 6 6.09 ) ES 1.08
Severe Control 1.4 (11.54 to 14.34) ES 0.09
Moder ate Sever e 38 .8 (11.21 to 6 6.39) ES 0.86
SM (mean % MVIC) (1.0 m/s) Moderate Control 8.9 (1.64 to 16.16) ES 0.84
Severe Control 10.9 (0.09 to 21.89) ES 1.1
Moderate Severe 2.0 (14.63 to 10.63) ES 0.13
SM (mean % MVI C) ( PW) Moder ate C ontrol 8.0 (2 .18 to 1 3.82 ) ES 0 .92
Severe Control 10.0 (5.02 to 25.02)
Moderate Severe 2.0 (17.49 to 13.49) ES 0.13
SM (mean % MVI C) ( FW) Moder ate Control 8.6 (1.3 9 to 1 5.81 ) ES 0 .81
Severe Control 8.8 (1.96 to 19.56) ES 0.84
Moderate Severe 0.2 (12.19 to 11.79) ES 0.01
SM (peak % MVIC) (1.0 m/s) Moderate Control 16.5 (4.7 to 28.3) ES 0.96
S ev ere Control 17.5 (2.78 to 32.22) ES 1.25
Moderate Severe 1.0 (18.84 to 16.84) ES 0.04
SM (peak % MVI C) ( PW) Mod er ate Control 14 .7 (2.53 to 26 .8 7) ES 0.8 1
Severe Control 18.8 (8.19 to 45.79) ES 0.76
Moderate Severe 4.1 (32.11 to 23.91) ES 0.15
SM (peak % MVIC) (FW) Moderate Control 12.7 (1.0 to 26.40) ES 0.62
Severe Control 13.3 (1.66 to 28.26) ES 0.75
Moderate Severe 0.6 (17.7 to 16.5) ES 0.03
Muscle activity duration
Astephen et al. (2008) Gastrocnemius (MG) Moderate Control MG was active for the majority of the gait cycle in severe OA group whereas
moderate
Sever e OA group and c ontr ols exhibited MG ac tivity during late stance.
Childs et al. (2004) VL (ms) General OA Control 165 (82.98 to 247.02) ES 1.12
SM (ms) 167 (99.89 to 234.11) ES 1.38
TA (ms) 158 (108.92 to 207.08) ES 1.79
MG (ms) 140 (59.27 to 220.73) ES 0.97
Hubley-Kozey et al.
(2006)
Gastrocnemius (MG, LG) Moderate Control There was no difference in temporal characteristics of LG or MG.
Quadriceps (RF, V L, VM) VL and RF a ctivities wer e prolonged during mid- to lat e-sta nce. T here wa s no
change in temporal characteristics for VM.
Hamstrings (BF, S M) BF activity was highe r and prolon ged in OA group compared with con trols. BF
activity was higher than SM for both groups.
Rutherford et al. (2010) Gastrocnemius (MG, LG) Moderate Control No phase differences between groups for MG and LG
Quadriceps (VL, VM, RF) No phase shift between groups for VL, VM and RF
Hamstrings (BF, SM) Duration o f BF and SM activities w as moderately p rolonged in O A group compared
with controls (ES 1.16 and 0.91)
Rutherford et al. (2011) Gastrocnemius (MG, LG % MVIC) Moderate Control MG onset was moderately earlier than LG onset for both groups (OA:ES 1.18.
Control:ES 1.01).
Severe Control OA group exhibited moderately later MG onset compared with controls (ES 1.1)
Moderate Severe Severe group exhibited largely later MG onset than moderate group (ES1.2)
BF: biceps femoris, LG: lateral gastrocnemius, MG: medial gastrocnemius, RF: rectus femoris, SM: semimembranosis, Sol: soleus, TA: tibialis anterior, VL: vastus medialis, VM: vastus
medialis.
IC: initial contact, ms: milliseconds, MVIC: maximum isometric voluntary contraction, m/s: meters per second, FW: fast walking, PW: preferred walking speed, KAM: peak external knee
adduction moment.
Preparation: 100 ms prior to heel strike until initial contact.
Weight acceptance: initial contact until single leg stance.
Muscle amplitude
Zeni et al. (2010)
723K. Mills et al. / Clinical Biomechanics 28 (2013) 713724
-
8/10/2019 A Systematic Review and Meta-Analysis of Lower Limb Neuromuscular Alteration Associates With Knee Osteoarthri
12/12
neuromotor activity may be due to those with more severe KOA
reporting higher pain levels than moderate or healthy counterparts
rather than the radiographic severity of their diagnosis. Further investi-
gation is needed into the association between pain, KOA severity and
neuromuscular control during gait.
5. Conclusion
Pooled and non-pooled data from moderate quality observationalstudies demonstrate individuals with KOA exhibit altered lower limb
neuromuscular activity during gait compared with healthy controls. In-
dividuals with KOA exhibited increased co-contraction, amplitude and
duration of lateral knee muscles regardless of their disease severity,
lower limb alignment or medial knee joint laxity. Neuromuscular activ-
ity of RF was also increased and prolonged for many KOA cohorts and
medial knee muscle activity was also increased. Based on current evi-
dence, this suggests thatindividuals with KOAare adopting neuromotor
patterns during gait that have the potential to protect the medial knee
joint. However, those with severe disease or medial knee joint laxity
and/or varus alignment are also exhibiting neuromuscular patterns
that increase the risk of accelerated medial compartment degeneration.
In some instances, both potentially protective and potentially degener-
ative neuromuscular pattern occur simultaneously. Therefore, while re-
sults of this review suggest neuromuscular rehabilitation focusing on
increasing lateral muscle activity and down-regulating medial muscle
and RF activity shows promise as way of potentially slowing this pro-
gression, future research into the efcacy of such a program is needed.
Such research needs to further investigate how simultaneous up-
regulation in medial and lateral knee muscle activities inuences dis-
ease progression and should also consider metrics of pain and function
when comparing individuals with KOA.
Acknowledgments
Kathryn Mills is supported by Alberta Innovates Health Solutions
Team in Osteoarthritis #200700596. Ryan Leigh is supported by Alberta
Innovates Health Solutions Clinical Fellowship #201200131. The au-
thors declare no conicts of interest.
References
Arvidsson, I., Eriksson, E., Knutsson, E., Arner, S., 1986. Reduction of pain inhibition onvoluntary muscle activation by epidural analgesia. Orthopedics 9, 14151419.
Astephen, J.L., Deluzio, K.J., Caldwell, G.E., Dunbar, M.J., Hubley-Kozey, C.L., 2008. Gait andneuromuscular pattern changes are associated with differences in knee osteoarthritisseverity levels. J. Biomech. 41, 868876.
Benoit, D.L., Lamontagne, M., Cerulli, G., Liti, A., 2003.The clinical signicance of electro-myography normalisation techniques in subjects with anterior cruciate ligamentinjury during treadmill walking. Gait Posture 18, 5663.
Bombardier, C., Hawker,G., Mosher, D., 2011.The impactof arthritis in Canada:today andover the next 30 years (no. fall). Arthritis of Canada Alliance.
Childs, J.D., Sparto, P.J., Fitzgerald, G.K., Bizzini, M., Irrgang, J.J., 2004.Alterations in lowerextremity movement and muscle activation patterns in individuals with knee osteo-arthritis. Clin. Biomech. 19, 4449.
Coombes, B.K., Bisset,L., Vicenzino, B.,2010. Efcacy andsafetyof corticosteroid injectionsand other injections for management of tendinopathy: a systematic review ofrandomised controlled trials. Lancet 376, 17511767.
Downs, S.H., Black, N., 1998. Thefeasibility of creatinga checklist forthe assessmentof themethodological quality both of randomised and non-randomised studies of healthcare interventions. J. Epidemiol. Community Health 52, 377384.
Heiden, T.L., Lloyd, D.G., Ackland, T.R., 2009. Knee joint kinematics, kinetics and muscleco-contraction in knee osteoarthritis patient gait. Clin. Biomech. 24, 833841.
Higgins, J.P.T., Thompson, S.G., Deeks, J.J., Altman, D.G., 2003. Measuring inconsistency inmeta-analyses. BMJ 327, 557560.
Hodges, P.W., vander Hoorn,W., Bennell, K.,Wrigley, T.V., 2012. Rate of cartilage loss in me-dial knee osteoarthritis is faster in patients with increased duration of cocontraction ofmedial knee muscles. JOSPT A41.
Hopkins, W.G., 2000. A New View of Statistics [WWW Document]. wwwsportsciorg/
resource/stat(URL (accessed 10.6.12)).Hopkins, W.G., Marshall, S.W., Batterham, A.M., Hanin, J., 2009.Progressive statistics forstudies in sports medicine and exercise science. Med. Sci. Sports Exerc. 41, 313.
Hortobgyi, T., Westerkamp, L., Beam, S., Moody, J., Garry, J., Holbert, D., Devita, P., 2005.Altered hamstring-quadriceps muscle balance in patients with knee osteoarthritis.Clin. Biomech. 20, 97104.
Hubley-Kozey, C., Deluzio, K., Landry, S., McNutt, J., Stanish, W., 2006. Neuromuscular alter-ations during walking in persons with moderate knee osteoarthritis. J. Electromyogr.Kinesiol. 16, 365378.
Hubley-Kozey,C.L., Hill,N.A., Rutherford,D.J., Dunbar, M.J., Stanish, W.D., 2009. Co-activationdifferences in lower limb muscles between asymptomatic controls and thosewith varying degrees of knee osteoarthritis during walking. Clin. Biomech. 24,407414.
Hurely, M.V., 1999. The role of muscle weakness in the pathogenesis of osteoarthritis.Rheum. Dis. Clin. N. Am. 25, 283298.
Kamen, G., Caldwell, G.E., 1996. Physiology and interpretation of the electromyogram.J. Clin. Neurophysiol. 13, 366384.
Lewek, M.D., Rudolph, K.S., Snyder-Mackler, L., 2004.Control of frontal plane knee laxityduring gait in patients with medial compartment knee osteoarthritis. Osteoarthr.
Cartil. 12, 745751.Lewek, M.D., Scholz, J., Rudolph, K .S., Snyder-Mackler, L., 2006.Stride-to-stride variability
of knee motion in patients with knee osteoarthritis. Gait Posture 23, 505511.Liikavainio, T., Bragge, T., Hakkarainen, M., Karjalainen, P.A., Arokoski, J.P., 2010.Gait and
muscle activation changes in men with knee osteoarthritis. Knee 17, 6976.Munn, J., Sullivan, S.J., Schneiders, A.G., 2010.Evidence of sensorimotor decits in functional
ankle instability: a systematic review with meta-analysis. J. Sci. Med. Sport. 13, 2 12.Ouellet, D., Moffet, H.L.N., 2002. Locomotor decits before and two months after knee
arthroplasty. Arthritis Rheum. 47, 484493.Rudolph, K.S., Schmitt, L.C., Lewek, M.D., 2007.Age-related changes in strength, joint lax-
ity, and walking patterns: are they related to knee osteoarthritis? Phys. Ther. 87,14221432.
Rutherford, D.J., Hubley-Kozey, C.L., Stanish, W.D., 2010.The neuromuscular demands ofaltering foot progression angle during gait in asymptomatic individuals and thosewith knee osteoarthritis. Osteoarthr. Cartil. 18, 654661.
Rutherford, D.J., Hubley-Kozey, C.L., Stanish, W.D., Dunbar, M.J., 2011. Neuromuscular al-terations exist with knee osteoarthritis presence and severity despite walking veloc-ity similarities. Clin. Biomech. 26, 377383.
Schipplein, O.D., Andriacchi, T.P., 1991. Interaction between active and passive kneestabi-lizers during level walking. J. Orthop. Res. 9, 113119.
Schmitt, L.C., Rudolph, K.S., 2007. Inuences on kneemovement strategies during walkingin persons with medial knee osteoarthritis. Arthritis Rheum. 57, 10181026.
Schmitt, L.C., Rudolph, K.S., 2008. Muscle stabilization strategies in people with medialknee osteoarthritis: the effect of instability. J. Orthop. Res. 26, 11801185.
Sharma, L., Hayes, K.W., Felson, D.T., Buchanan, T.S., Kirwan-Mellis, G., Lou, C., Pai, Y.C.,Dunlop, D.D., 1999.Does laxity alter the relationship between strength and physicalfunction i n knee osteoarthritis? Arthritis Rheum. 42, 2532.
Simic, M., Hinman, R.S., Wrigley, T.V., Bennell, K.L., Hunt, M.A., 2011. Gait modicationstrategies for altering medial knee joint load: a systematic review. Arthritis CareRes. 63, 405426.
van Tulder, M., Furlan, A., Bombardier, C., Bouter, L., Editorial Board of the CochraneCollaboration Back Review Group, 2003.Updated method guidelines for systematicreviews in the Cochrane Collaboration Back Review Group. Spine 28, 12901299.
Wilson, J.L.A., Deluzio, K.J., Dunbar, M.J., Caldwell, G.E., Hubley-Kozey, C.L., 2011.Theassociation between knee joint biomechanics and neuromuscular control andmoderate knee osteoarthritis radiographic and pain severity. Osteoarthr. Cartil.19, 186193.
Zeni, J.A., Rudolph, K., Higginson, J.S., 2010.Alterations in quadriceps and hamstrings co-ordination in persons with medial compartment knee osteoarthritis. J. Electromyogr.Kinesiol. 20, 148154.
724 K. Mills et al. / Clinical Biomechanics 28 (2013) 713724
http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0160http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0160http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0160http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0160http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0010http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0010http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0010http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0010http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0010http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0015http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0015http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0015http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0015http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0015http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0015http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0015http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0165http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0165http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0020http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0020http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0020http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0020http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0020http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0025http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0025http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0025http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0025http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0025http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0025http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0025http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0030http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0030http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0030http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0030http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0030http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0035http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0035http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0035http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0035http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0040http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0040http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0040http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0040http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0170http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0170http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0170http://wwwsportsciorg/resource/stathttp://wwwsportsciorg/resource/stathttp://refhub.elsevier.com/S0268-0033(13)00164-2/rf0050http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0050http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0050http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0050http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0055http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0055http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0055http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0055http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0060http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0060http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0060http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0060http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0060http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0065http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0065http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0065http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0065http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0065http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0065http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0070http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0070http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0070http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0070http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0075http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0075http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0075http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0075http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0080http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0080http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0080http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0080http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0080http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0085http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0085http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0085http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0085http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0085http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0090http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0090http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0090http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0090http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0095http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0095http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0095http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0095http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0095http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0095http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0100http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0100http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0100http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0100http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0100http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0100http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0105http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0105http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0105http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0105http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0105http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0110http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0110http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0110http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0110http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0110http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0115http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0115http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0115http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0115http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0115http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0120http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0120http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0120http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0120http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0125http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0125http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0125http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0125http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0125http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0125http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0130http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0130http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0130http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0130http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0135http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0135http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0135http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0135http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0140http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0140http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0140http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0140http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0140http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0140http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0140http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0180http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0180http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0180http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0180http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0150http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0150http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0150http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0150http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0150http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0150http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0155http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0155http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0155http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0155http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0155http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0155http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0155http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0155http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0150http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0150http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0150http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0150http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0180http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0180http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0140http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0140http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0140http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0135http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0135http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0130http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0130http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0125http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0125http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0120http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0120http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0115http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0115http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0115http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0110http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0110http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0110http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0105http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0105http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0105http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0100http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0100http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0095http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0095http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0090http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0090http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0085http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0085http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0080http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0080http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0080http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0075http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0075http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0070http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0070http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0065http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0065http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0065http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0065http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0060http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0060http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0060http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0055http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0055http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0050http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0050http://wwwsportsciorg/resource/stathttp://wwwsportsciorg/resource/stathttp://refhub.elsevier.com/S0268-0033(13)00164-2/rf0170http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0170http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0170http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0040http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0040http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0035http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0035http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0030http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0030http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0030http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0025http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0025http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0025http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0020http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0020http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0020http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0165http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0165http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0015http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0015http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0015http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0010http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0010http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0010http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0160http://refhub.elsevier.com/S0268-0033(13)00164-2/rf0160