Journal of Electromyography and Kinesiology 25 (2015) 175–181
Contents lists available at ScienceDirect
Journal of Electromyography and Kinesiology journal homepage: www.elsevier.com/locate/jelekin
Muscle co-contraction around the knee when walking with unstable shoes Brian Horsak a,b,⇑, Mario Heller c, Arnold Baca c a
Department of Physiotherapy, St. Poelten University of Applied Sciences, St. Poelten, Austria Institute for Sciences and Services in Health, St. Poelten University of Applied Sciences, St. Poelten, Austria c Department of Biomechanics, Kinesiology and Applied Computer Science, University of Vienna, Vienna, Austria b
a r t i c l e
i n f o
Article history: Received 2 April 2014 Received in revised form 10 July 2014 Accepted 31 July 2014
Keywords: Electromyography Co-contraction index Unstable shoes Toning shoes Biomechanics Gait analysis
a b s t r a c t Walking with unstable shoes has been discussed to decrease joint loading. Typical estimates of joint loading using an inverse dynamic approach only account for net joint moments, not considering the potential role of muscular co-contraction. Therefore, the purpose of this study was to compare muscular co-contraction levels when walking with two different unstable shoe constructions (rocker-bottom and toning shoes) compared to walking with regular shoes. For each shoe condition, 12 healthy subjects walked with both, a regular shoe and with an unstable shoe at self-selected walking speed at a 10-m walkway. Surface EMG data of selected muscles were recorded and time normalized for calculating co-contraction indices (CCI) for opposing muscle groups. Results showed an increase of co-contraction primarily for vastii and gastrocnemius muscles for the first and second half of stance when walking with an unstable shoe construction. Therefore, when using an inverse dynamic approach to analyze joint loading differences between regular shoes and unstable shoes, one should be cautious in interpreting the data, as these methods base their estimates of joint moments upon the net joint torque. Ó 2014 Elsevier Ltd. All rights reserved.
1. Introduction In recent years, the influence of footwear on gait biomechanics has been of high interest for research. Typically, shoes and footwear are designed to support the foot in joint stability throughout locomotion and recreational activities, as well as to protect from environmental conditions (Reinschmidt and Nigg, 2000). One special concept of shoe design is the unstable shoe construction. This concept aims to increase instability during standing and locomotion, for example by using a rocker-bottom (e.g. Masai Barefoot Technology, MBT) or balance-pods centered in the forefoot and heel region (e.g. Reebok Easy Tone, ET). The majority of companies, who promote these type of footwear, claim, that these shoes can positively affect health, in terms of increased muscle activation during locomotion and/or a reduction of joint loads of the lower extremities. Research has addressed these claims intensively in the last decade (e.g. Nigg et al., 2006; Romkes et al., 2006; Landry et al., 2012; Buchecker et al., 2013; Horsak and Baca, 2013). The majority of ⇑ Corresponding author at: St. Poelten University of Applied Science, Matthias Corvinus Strasse 15, 3100 St. Poelten, Austria. E-mail address: [email protected] (B. Horsak). http://dx.doi.org/10.1016/j.jelekin.2014.07.015 1050-6411/Ó 2014 Elsevier Ltd. All rights reserved.
research conducted, has used motion analysis and inverse dynamic approaches to analyze effects of unstable shoe constructions on kinematic and kinetic parameters during gait, typically of the lower extremities. Some of the researchers have additionally used electromyography (EMG) to quantify changes in muscle activation patterns (Romkes et al., 2006; Buchecker et al., 2010; Horsak and Baca, 2013). The majority of studies analyzing gait biomechanics during level walking found significant alterations in sagittal and frontal plane kinematics and kinetics for the knee and ankle joints (e.g. Romkes et al., 2006; Landry et al., 2012). Briefly, when walking with rocker-bottom shoes, people tend to walk with reduced hip flexion–extension and ankle adduction–abduction range of motion (Landry et al., 2012), and an increased dorsiflexion angle at initial contact, followed by a continuous plantarflexion movement (Romkes et al., 2006). Researchers also found some notable changes for gait kinetics. Ankle moments tend to be greater for walking with the unstable shoes and at the hip and knee, both increases and decreases in moments in frontal and sagittal plane were observed (e.g. Buchecker et al., 2010; Landry et al., 2012). Linked to the kinematic changes found at the ankle-joint-complex, researchers also identified some significant alterations in muscle activity patterns of the gastrocnemius, vastii and tibialis anterior muscles. Buchecker et al. (2010) reported an increase of vastus
176
B. Horsak et al. / Journal of Electromyography and Kinesiology 25 (2015) 175–181
lateralis and gastrocnemius medialis activity during late stance. This was also observed by Romkes et al. (2006), who reported a significant increase of muscle activity for gastrocnemius, vastus medialis, vastus lateralis and rectus femoris. It is noteworthy that in muscle-driven forward dynamics simulations of normal walking, vastii and gastrocnemius muscles have been shown to be the primary contributors to the tibia-femoral joint forces (intersegmental and compressive forces), with the peak force being even greater than the muscle forces themselves (Sasaki and Neptune, 2010). A major limitation of the inverse dynamic approach to determine joint kinetics is, that only the net effect of muscle activity at a joint can be estimated (Winter, 2005). If co-contraction, the simultaneous activation of various muscles acting at a joint, is taking place, the analysis only yields the net effect of these muscles, hence, only net joint moments. The external knee adduction, for example, is often used as an indicator of internal joint loading. Research has shown, that only low to moderate correlation exists between the external adduction moment and internal compartment loads (e.g. Winby et al., 2013). This, often accepted, inaccuracy in estimating joint forces is even increased, if co-contraction is taking place. One way to overcome this problem is the use of some form of electromyography-driven modeling approach to estimate joint contact force and all muscle forces apparent (e.g. Sasaki and Neptune, 2010; Gardinier et al., 2013). However, this approach is not only time consuming and difficult to apply, it sometimes lacks validity as these calculations are typically handled as a standard minimization problem. Therefore, when reporting joint loading differences between walking with normal and with unstable shoes, it is important to understand, how these shoe constructions contribute to the amount of co-contraction. In respect to the published evidence reporting biomechanical changes when walking with unstable shoes, up to date, only one study could be identified (Buchecker et al., 2010) which reported co-contraction levels. However, the data of Buchecker et al. (2010) are limited to a specific population, as they only recruited overweight men for their study. In contrast to this specific study population, the main population which intends to use such shoe concepts will also include a more healthy and young population. Beside the rockerbottom shoe design, which was analyzed in most of the present studies, several other unstable shoe concepts have emerged in the last years, e.g. the balance pod shoe design. Up to date no scientific literature reported analyses of co-contraction when walking with balance-pod shoe designs, even though these shoes are widely used and popular. Due to the lack of knowledge in the potential role of unstable shoe constructions in increasing the amount of co-contraction around the knee joint, the purpose of this study was to compare co-contraction levels when walking with two different unstable shoe constructions compared to walking with regular shoes in a young healthy population. We hypothesized that walking with unstable shoes will increase the level of co-contraction during stance. Information regarding the influence of unstable shoe constructions on co-contraction levels may help to better interpret already published data on biomechanical changes during walking with unstable shoes. It may also serve as a guide for future research and its underlying methodological approaches to quantify kinetic changes when wearing these shoes.
2. Methods In this study data of two previously conducted studies focusing on gait biomechanics during level walking with regular and with two different types of unstable shoe constructions (Fig. 1), a
Fig. 1. The rocker-bottom shoe (MBT, Mahuta) (A) and the balance-pod shoe (Easy Tone, Reenew) (B) used in this study as unstable shoe constructions.
rocker-bottom and a balance-pod shoe design, were used retrospectively (Horsak and Baca, 2012, 2013). Both data sets were captured within the same methodological framework, as briefly described as follows, and were used separately for analyses in the present study. 2.1. Participants For the rocker-bottom shoe data set, 12 healthy participants (7 male and 5 female) volunteered (age: 25 ± 6 years, height: 174 ± 7 cm, mass: 68 ± 10 kg). The balance-pod shoe data comprised 12 participants (5 male and 7 female; age: 25 ± 4 years, height: 172 ± 11 cm, mass: 67 ± 11 kg). In total for each data set, 14 participants were recruited, but because of technical issues, two participants of each data set had to be removed from analyses. All participants were eligible if they were free from any lower and upper extremity orthopedic or neurological problems and pain. Participants were excluded, if they had used the unstable shoe construction prior to the study. All participants were informed of the aims of this study and signed an informed consent form, which was approved by the Local Ethics Committee. 2.2. Data acquisition and processing Participants were asked to walk with both, a regular shoe (control situation) and with the unstable shoe at self-selected walking speed at a 10-m walkway. Prior to data acquisition, participants were allowed to walk along the walk-way as often as necessary until they felt comfortable during walking with their regular shoes. This was typically reached after five minutes of walking at selfselected walking speed. After this accommodation phase, the data were captured using the different shoes. For each shoe, participants again had approximately five minutes of accommodation time to feel comfortable during walking at the walkway. Surface electromyography (EMG) data were recorded using a telemetry system (DELSYS, Myomonitor IV, Boston, MA, USA) and the software EMGworks version 3.6, which was installed on a stationary computer. Single differential Ag bar-electrodes with a contact dimension of 10.0 1.0 mm, an inter-electrode distance of 10.0 mm and a resulting detection area of 10 mm2 (DELSYS, DE2.1) were attached parallel to the muscle fiber direction over the mid-muscle belly (Hermens et al., 2000). The EMG amplifier system gain was 1000 with an common mode rejection ratio of 92 dB, an input impedance of >1015 O and an overall channel noise 61.2 lV (RMS, R.T.I)2. The analog signals were converted into digital signals using a sampling rate of 1000 Hz and a 16-bit analog–digital converter, using a ±5 V signal input range, which was integrated into to the EMG telemetry handheld device (DELSYS, Myomonitor IV, Boston, MA, USA). Prior to electrode placement, the skin was shaved, slightly abraded and cleaned using skin preparation gel. The ground electrode was placed at the wrist. In total muscle activities of the
B. Horsak et al. / Journal of Electromyography and Kinesiology 25 (2015) 175–181
gluteus maximus (GM), vastus medialis (VM), vastus lateralis (VL), biceps femoris (BF), tibialis anterior (TA), peroneus longus (PL) and gastrocnemius medialis (GMM) of the dominant leg were recorded. Raw data signals were then digitally band-pass filtered using a 4th order Butterworth filter with cut-off frequencies of 10 Hz and 450 Hz. Then, signals were full wave rectified and linear enveloped using another 4th order low-pass Butterworth filter, with a cut-off frequency of 8 Hz. The cut-off frequency was determined using power spectrum analysis according to Nielsen et al. (1994). Briefly, the power spectrum of each trial was calculated using the FFT algorithm, with a Hanning window of 50% overlap and a window length of 2048. This resulted in a resolution of 0.5 Hz for each power spectra. Power spectra were normalized to a maximum value of 1 and then all six trials for each person were averaged to a mean power spectra. The average of the frequencies at which the accumulation of energy reached 95% of total energy (see Nielsen et al., 1994) was defined as the most appropriate cut-off frequency. EMG data were then time normalized to 100% gait cycle. Time instances during each gait cycle for initial contact and toe-off were determined using kinematic and kinetic data (see Horsak and Baca, 2012, 2013). Resulting linear envelopes were then amplitude-normalized to the mean peak muscle activity of all valid control shoe trials (=100%), as described by Buchecker et al. (2010). The co-contraction index (CCI) for each data point of the linear envelope was calculated using Eq. (1) (Rudolph et al., 2000, 2001; Schmitt and Rudolph, 2008):
CCIn¼100 ¼ i¼1
lowEMGi ðhighEMGi þ lowEMGi Þ highEMGi
ð1Þ
The lowEMGi is the normalized value of the EMG signal at one point (i) for the less active muscle. The highEMGi is the corresponding activity of the higher active muscle. The CCI therefore, can range at any time point from zero (no muscle activity) to a maximum of 200 (if both muscle are active 100%; due to the normalization procedure slightly greater values are possible). According to Rudolph et al. (2001), this method provides an estimate of the relative activation of the pair of muscles as well as the magnitude of the co-contraction. CCI was calculated for the following opposing muscle groups: vastus medialis to biceps femoris (VM/BF), vastus lateralis to biceps femoris (VL/BF), vastus medialis to gastrocnemius medialis (VM/GMM), vastus lateralis to gastrocnemius medialis (VL/GMM) and tibialis anterior to gastrocnemius medialis (TA/ GMM). These muscle pairs were selected based on their functional contribution to stabilize the knee joint (in the sagittal plane) and their possible contribution to an increased bone to bone contact force. From the resulting CCI time-series mean amplitude values for the first (0–50% stance) and second half of stance phase (50% stance to toe-off) were calculated. For all CCI parameters, data of six trials under each walking condition were averaged for each participant and then used for further analysis. All data processing steps were performed using self-written scripts in MATLAB (The MathWorks Inc., Ismaning, GER).
177
ANOVA was run separately for each muscle pair to reduce the complexity in interpretation of possible interactions. The described procedure was performed for each unstable shoe condition (rocker-bottom and balance-pod) separately. An alpha level p 6 0.05 was set a priori to indicate significant statistical findings. To estimate practical relevance of observed CCI differences between the analyzed shoe designs Cohen’s d was calculated using Eq. (2). An effect size of 0.2 was considered as a small effect, 0.5 as a moderate effect and values >0.8 as a great effect.
d¼
ðlA lB Þ s
ð2Þ
where lA is the mean value of CCI for the unstable shoe and lB the mean value for the corresponding CCI of the control shoe. The standard deviation (s) was estimated as a pooled standard deviation (spooled) of both groups using Eq. (3).
spooled
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ðnA 1Þs2A þ ðnB 1Þs2B ¼ nA þ nB 2
ð3Þ
3. Results Comparing co-contraction between opposing muscle groups around the knee when walking with the rocker-bottom shoe design revealed some statistically significant alterations (Fig. 2). For VM/BF [F (1, 11) = 6.824, p = 0.024, g2 = 0.38], VM/GMM [F (1, 11) = 11.006, p = 0.007, g2 = 0.50] and VL/GMM [F (1, 11) = 12.962, p = 0.004, g2 = 0.54] significant main effects of shoe type were found, indicating greater CCI values for walking with the rocker-bottom shoe design (Table 1). For all of the analyzed muscle pairs, significant main effects for gait phase were also observed, demonstrating greater CCI during the first half of stance for all of them: for VM/BF [F (1, 11) = 23.217, p = 0.001, g2 = 0.68], VL/BF [F (1, 11) = 23.343, p = 0.001, g2 = 0.68], for VM/GMM [F (1, 11) = 4.959, p = 0.048, g2 = 0.31], VL/GMM [F (1, 11) = 6.840, p = 0.024, g2 = 0.38], TA/GMM [F (1, 11) = 16.559, p = 0.002, g2 = 0.60]. No interactions were found between shoe type and gait phase. Descriptive statistics, effect size as well as 95% CI of the difference between both shoe conditions are presented in Table 1. For the balance-pod shoe design only for the VM/GMM muscle pair, a significant main effect of shoe type was observed [F (1, 11) = 4.976, p = 0.047, g2 = 0.31], also demonstrating greater CCI values for walking with the unstable shoe design (Fig. 3). In addition, except for VL/GMM, all muscle pairs showed a significant main effect of gait phase, showing greater CCI values for the first half of stance: for VM/BF [F (1, 11) = 8.494, p = 0.014, g2 = 0.44], VL/BF [F (1,11) = 13.202, p = 0.004, g2 = 0.55], for VM/GMM [F (1, 11) = 10.690, p = 0.007, g2 = 0.49], TA/GMM [F (1, 11) = 7.413, p = 0.02, g2 = 0.40]. No significant interactions were found. Fig. 4 gives an overview how muscle activity of each muscle contributes to the CCI. For the balance-pod shoe design descriptive statistics, effect size as well as 95% CI of the difference between both shoe conditions are presented in Table 2.
2.3. Statistical analysis 4. Discussion Statistical analyses were performed using IBM SPSS Statistics 21 (Somer, NY, USA). Initially, parameters were tested to comply with needed statistical assumptions. The data were reported using the mean and standard deviation (SD) for the CCI in each shoe condition as well as the 95% CI of the difference between the unstable and the stable shoe. A 2 2 repeated measures ANOVA using the CCI as the dependent variable and the shoe type (unstable and stable condition) and gait phase (early and late stance) as independent variables were used to identify effects of the shoe condition and the gait phase as well as first order interaction effects. The
Up to date there is a potential lack of scientific evidence reporting how unstable shoe constructions (e.g. rocker-bottom or balance-pod shoe design) contribute to co-contraction of opposing muscle groups surrounding the knee joint. The main body of research used inverse dynamic approaches and EMG to analyze if unstable shoe constructions may have the potential to decrease joint loading. When relying on such estimates, one has to know, if co-contraction is taking place, or at least does not change its amount, because inverse dynamic approaches only account for
178
B. Horsak et al. / Journal of Electromyography and Kinesiology 25 (2015) 175–181
Fig. 2. Mean co-contraction index (CCI) wave forms and the corresponding standard deviations for walking with the stable shoes (dashed line) and the rocker-bottom shoes (solid line). The vertical line shows toe-off.
Table 1 Mean (SD) co-contraction index (CCI) for walking with the rocker-bottom shoes and regular shoes. CCI
Rocker-bottom shoe mean (SD)
Control shoe mean (SD)
95% CI of difference
VM/BF 1st half stance 2nd half stance
27.2 6.5
(16.2) (5.3)
22.4 4.6
(12.1) (2.4)
0.4 0.9
10.1 4.7
0.3 0.5
VL/BF 1st half stance 2nd half stance
27.7 6.9
(16.4) (6.0)
26.0 4.8
(15.2) (2.5)
3.0 0.9
6.3 5.1
0.1 0.5
VM/GMM 1st half stance 2nd half stance
15.9 10.1
(6.0) (9.7)
11.1 6.3
(4.0) (2.8)
1.8 0.9
7.8 8.5
0.9 0.5
VL/GMM 1st half stance 2nd half stance
19.4 12.7
(10.2) (9.2)
13.3 7.9
(6.6) (4.6)
2.6 0.3
9.5 9.9
0.7 0.7
TA/GMM 1st half stance 2nd half stance
13.8 7.5
(10.4) (5.8)
13.2 7.2
(8.3) (4.6)
5.5 2.7
6.7 3.5
0.1 0.1
the net joint moments. Therefore, the purpose of this study was to compare co-contraction levels when walking with two different unstable shoe constructions compared to walking with regular shoes. We hypothesized that both unstable shoe constructions will increase co-contraction for opposing muscles surrounding the knee joint. Based on our findings, this was true for the rocker-bottom shoe data as well as for the balance-pod shoe data for one parameter. Besides the many studies analyzing muscle activation patterns between walking with regular and with unstable shoes, only
Cohen’s d
Buchecker et al. (2010) have reported CCI when walking with rocker-bottom shoes. They found a significant increase of co-contraction between vastus lateralis and gastrocnemius medialis for mid and late stance and reported an effect sizes d of greater than 0.7 for both. Though, it has to be outlined that, the data of Buchecker et al. (2010) are limited to a specific population, as they only recruited overweight men for their study. However, their results are very consistent with data observed in our study. Our results for the rocker-bottom shoe data set showed an increased co-contraction for VM/GMM and VL/GMM for early stance, both
179
B. Horsak et al. / Journal of Electromyography and Kinesiology 25 (2015) 175–181
Normalized EMG [%]
Fig. 3. Mean co-contraction index (CCI) wave forms and the corresponding standard deviations for walking with the stable shoes (dashed line) and balance-pod shoes (solid line). The vertical line shows toe-off.
100
A
80 60 40 20 0
Normalized EMG [%]
1st 2nd half of stance
1st 2nd half of stance
1st 2nd half of stance
1st 2nd half of stance
100
1st 2nd half of stance
B
80 60 40 20 0 1st 2nd half of stance VM
1st 2nd half of stance VL
1st 2nd half of stance BF
1st 2nd half of stance GMM
1st 2nd half of stance TA
Fig. 4. Mean (SD) muscle activity during first and second half of stance. Gray bars show muscle activity of walking with regular shoes, white bars walking in the unstable shoe condition: (A) rocker-bottom, (B) balance-pod shoe design.
with a Cohen’s d of greater than 0.7. In slight contrast to Buchecker et al. (2010), we also observed an increase of the VM/BF during first half of stance. Unexpectedly, only the VM/GMM showed an increase of CCI during first half of stance when walking with the balance-pod shoe. The CCI differences found for the rocker-bottom shoe condition may be explained by the altered EMG activation
patterns, already described in other studies (e.g. Romkes et al., 2006). When walking with rocker-bottom shoes, people walk with an increased dorsiflexion angle at initial contact, followed by a continuous plantarflexion movement. These kinematic differences are accompanied by an increased activation of the gastrocnemius during early to mid-stance and an increase of vastii activity during mid
180
B. Horsak et al. / Journal of Electromyography and Kinesiology 25 (2015) 175–181
Table 2 Mean (SD) co-contraction index (CCI) for walking with the balance-pod shoes and regular shoes. CCI
Balance-pod shoe mean (SD)
Control shoe mean (SD)
95% CI of difference
VM/BF 1st half stance 2nd half stance
27.1 6.2
(29.2) (5.0)
20.7 5.5
(12.6) (2.6)
5.0 1.4
18.0 2.9
0.3 0.2
VL/BF 1st half stance 2nd half stance
29.8 5.5
(25.7) (3.3)
24.1 5.3
(14.8) (2.4)
2.7 1.0
14.2 1.5
0.3 0.1
VM/GMM 1st half stance 2nd half stance
13.1 9.0
(7.5) (8.0)
10.8 6.2
(4.7) (3.2)
0.7 0.6
5.4 6.3
0.4 0.5
VL/GMM 1st half stance 2nd half stance
14.3 12.8
(6.3) (13.0)
13.4 8.6
(8.1) (8.0)
1.8 0.1
3.7 8.2
0.1 0.4
TA/GMM 1st half stance 2nd half stance
9.9 6.6
(4.8) (2.6)
11.2 6.7
(8.6) (2.3)
5.0 1.4
2.3 1.1
0.2 0.0
to late stance. The latter was also observed for walking with balance-pod shoes (Horsak and Baca, 2013), and may be a strategy of the neuromuscular system to adopt for increased instability during late stance. The rocker profile of the rocker-bottom shoe and the increased neural drive necessary for gastrocnemius to maintain the continuous plantarflexion movement, may be a reason for the increased CCI when walking with these shoes. As co-contraction was identified as one strategy of the neuromuscular system to stabilize a joint (Lewek et al., 2004), one would also expect a great increase of co-contraction when walking with balance-pod shoes. One reason why only small differences for CCI were found, could be the amount of instability introduced by the balance-pod shoes. These shoes have two balance pods, one centered in the mid of the forefoot, one in the mid of the heel, to increase instability. The pods mainly introduce instability to the medial and lateral directions. Previous research has shown, that when standing with balance-pod shoes, no significant alterations in postural sway was observed (Horsak and Baca, 2013). The instability introduced by the balance-pod shoes might be too small, as it could affect neuromuscular activation patterns associated with stabilization strategies of a joint. A second explanation might be the fact, that muscle pairs selected in this study typically cause motion in the sagittal plane. The balance-pod shoe design destabilizes gait, due to their construction, more in medio-lateral direction. Therefore, the selected muscle pairs may not be appropriate enough to capture co-contraction caused by this shoe concept. Cohen’s d was used in this study to establish an estimate of the practical relevance of CCI differences observed. Cohens’s d is a widely accepted estimate of the effect size, which is often utilized in studies related to sport sciences (Fröhlich et al., 2009). Even though the importance of a found effect can be classified using d, this measure should always be interpreted in the context of the related research area. In our study differences showed relatively high effect sizes of greater than 0.7. Hubley-Kozey et al. (2009), for example have used the same CCI for comparing asymptomatic individuals with those with varying degrees of knee osteoarthritis during walking. They have used an interval of approximately the first 20–25% of gait cycle to calculate a mean CCI value, therefore comparable with our methodology. Their data for the CCI of the vastus medialis and gastrocnemius medialis for asymptomatic controls are relatively consistent with data of our study for the regular shoe condition. For VM/GMM CCI differences between healthy controls and the severe osteoarthritis group was approximately the same size as the differences observed in our study between the rocker-bottom and the regular shoe condition. These results further highlight the potential effect of unstable shoe constructions in increasing co-contraction around the knee joint.
Cohen’s d
This study is subject to some limitations, which have to be recognized. Our study is based on two data sets published earlier. No sample size estimation was performed for the underlying research question. The CCI utilized in this study is an often used estimate of co-contraction (Rudolph et al., 2001; Schmitt and Rudolph, 2008; Hubley-Kozey et al., 2009), but still up to date there is no standard method for calculating and reporting co-contraction. However, a recently published study by Hubley-Kozey et al. (2013) found good to excellent intraclass correlation coefficients for individuals with knee osteoarthritis. Therefore, one could expect that the application of the CCI for healthy individuals is valid too. Another limitation might be a very small difference observed in walking speed between both shoe conditions in both data sets. Anyway, as these differences (
Contents lists available at ScienceDirect
Journal of Electromyography and Kinesiology journal homepage: www.elsevier.com/locate/jelekin
Muscle co-contraction around the knee when walking with unstable shoes Brian Horsak a,b,⇑, Mario Heller c, Arnold Baca c a
Department of Physiotherapy, St. Poelten University of Applied Sciences, St. Poelten, Austria Institute for Sciences and Services in Health, St. Poelten University of Applied Sciences, St. Poelten, Austria c Department of Biomechanics, Kinesiology and Applied Computer Science, University of Vienna, Vienna, Austria b
a r t i c l e
i n f o
Article history: Received 2 April 2014 Received in revised form 10 July 2014 Accepted 31 July 2014
Keywords: Electromyography Co-contraction index Unstable shoes Toning shoes Biomechanics Gait analysis
a b s t r a c t Walking with unstable shoes has been discussed to decrease joint loading. Typical estimates of joint loading using an inverse dynamic approach only account for net joint moments, not considering the potential role of muscular co-contraction. Therefore, the purpose of this study was to compare muscular co-contraction levels when walking with two different unstable shoe constructions (rocker-bottom and toning shoes) compared to walking with regular shoes. For each shoe condition, 12 healthy subjects walked with both, a regular shoe and with an unstable shoe at self-selected walking speed at a 10-m walkway. Surface EMG data of selected muscles were recorded and time normalized for calculating co-contraction indices (CCI) for opposing muscle groups. Results showed an increase of co-contraction primarily for vastii and gastrocnemius muscles for the first and second half of stance when walking with an unstable shoe construction. Therefore, when using an inverse dynamic approach to analyze joint loading differences between regular shoes and unstable shoes, one should be cautious in interpreting the data, as these methods base their estimates of joint moments upon the net joint torque. Ó 2014 Elsevier Ltd. All rights reserved.
1. Introduction In recent years, the influence of footwear on gait biomechanics has been of high interest for research. Typically, shoes and footwear are designed to support the foot in joint stability throughout locomotion and recreational activities, as well as to protect from environmental conditions (Reinschmidt and Nigg, 2000). One special concept of shoe design is the unstable shoe construction. This concept aims to increase instability during standing and locomotion, for example by using a rocker-bottom (e.g. Masai Barefoot Technology, MBT) or balance-pods centered in the forefoot and heel region (e.g. Reebok Easy Tone, ET). The majority of companies, who promote these type of footwear, claim, that these shoes can positively affect health, in terms of increased muscle activation during locomotion and/or a reduction of joint loads of the lower extremities. Research has addressed these claims intensively in the last decade (e.g. Nigg et al., 2006; Romkes et al., 2006; Landry et al., 2012; Buchecker et al., 2013; Horsak and Baca, 2013). The majority of ⇑ Corresponding author at: St. Poelten University of Applied Science, Matthias Corvinus Strasse 15, 3100 St. Poelten, Austria. E-mail address: [email protected] (B. Horsak). http://dx.doi.org/10.1016/j.jelekin.2014.07.015 1050-6411/Ó 2014 Elsevier Ltd. All rights reserved.
research conducted, has used motion analysis and inverse dynamic approaches to analyze effects of unstable shoe constructions on kinematic and kinetic parameters during gait, typically of the lower extremities. Some of the researchers have additionally used electromyography (EMG) to quantify changes in muscle activation patterns (Romkes et al., 2006; Buchecker et al., 2010; Horsak and Baca, 2013). The majority of studies analyzing gait biomechanics during level walking found significant alterations in sagittal and frontal plane kinematics and kinetics for the knee and ankle joints (e.g. Romkes et al., 2006; Landry et al., 2012). Briefly, when walking with rocker-bottom shoes, people tend to walk with reduced hip flexion–extension and ankle adduction–abduction range of motion (Landry et al., 2012), and an increased dorsiflexion angle at initial contact, followed by a continuous plantarflexion movement (Romkes et al., 2006). Researchers also found some notable changes for gait kinetics. Ankle moments tend to be greater for walking with the unstable shoes and at the hip and knee, both increases and decreases in moments in frontal and sagittal plane were observed (e.g. Buchecker et al., 2010; Landry et al., 2012). Linked to the kinematic changes found at the ankle-joint-complex, researchers also identified some significant alterations in muscle activity patterns of the gastrocnemius, vastii and tibialis anterior muscles. Buchecker et al. (2010) reported an increase of vastus
176
B. Horsak et al. / Journal of Electromyography and Kinesiology 25 (2015) 175–181
lateralis and gastrocnemius medialis activity during late stance. This was also observed by Romkes et al. (2006), who reported a significant increase of muscle activity for gastrocnemius, vastus medialis, vastus lateralis and rectus femoris. It is noteworthy that in muscle-driven forward dynamics simulations of normal walking, vastii and gastrocnemius muscles have been shown to be the primary contributors to the tibia-femoral joint forces (intersegmental and compressive forces), with the peak force being even greater than the muscle forces themselves (Sasaki and Neptune, 2010). A major limitation of the inverse dynamic approach to determine joint kinetics is, that only the net effect of muscle activity at a joint can be estimated (Winter, 2005). If co-contraction, the simultaneous activation of various muscles acting at a joint, is taking place, the analysis only yields the net effect of these muscles, hence, only net joint moments. The external knee adduction, for example, is often used as an indicator of internal joint loading. Research has shown, that only low to moderate correlation exists between the external adduction moment and internal compartment loads (e.g. Winby et al., 2013). This, often accepted, inaccuracy in estimating joint forces is even increased, if co-contraction is taking place. One way to overcome this problem is the use of some form of electromyography-driven modeling approach to estimate joint contact force and all muscle forces apparent (e.g. Sasaki and Neptune, 2010; Gardinier et al., 2013). However, this approach is not only time consuming and difficult to apply, it sometimes lacks validity as these calculations are typically handled as a standard minimization problem. Therefore, when reporting joint loading differences between walking with normal and with unstable shoes, it is important to understand, how these shoe constructions contribute to the amount of co-contraction. In respect to the published evidence reporting biomechanical changes when walking with unstable shoes, up to date, only one study could be identified (Buchecker et al., 2010) which reported co-contraction levels. However, the data of Buchecker et al. (2010) are limited to a specific population, as they only recruited overweight men for their study. In contrast to this specific study population, the main population which intends to use such shoe concepts will also include a more healthy and young population. Beside the rockerbottom shoe design, which was analyzed in most of the present studies, several other unstable shoe concepts have emerged in the last years, e.g. the balance pod shoe design. Up to date no scientific literature reported analyses of co-contraction when walking with balance-pod shoe designs, even though these shoes are widely used and popular. Due to the lack of knowledge in the potential role of unstable shoe constructions in increasing the amount of co-contraction around the knee joint, the purpose of this study was to compare co-contraction levels when walking with two different unstable shoe constructions compared to walking with regular shoes in a young healthy population. We hypothesized that walking with unstable shoes will increase the level of co-contraction during stance. Information regarding the influence of unstable shoe constructions on co-contraction levels may help to better interpret already published data on biomechanical changes during walking with unstable shoes. It may also serve as a guide for future research and its underlying methodological approaches to quantify kinetic changes when wearing these shoes.
2. Methods In this study data of two previously conducted studies focusing on gait biomechanics during level walking with regular and with two different types of unstable shoe constructions (Fig. 1), a
Fig. 1. The rocker-bottom shoe (MBT, Mahuta) (A) and the balance-pod shoe (Easy Tone, Reenew) (B) used in this study as unstable shoe constructions.
rocker-bottom and a balance-pod shoe design, were used retrospectively (Horsak and Baca, 2012, 2013). Both data sets were captured within the same methodological framework, as briefly described as follows, and were used separately for analyses in the present study. 2.1. Participants For the rocker-bottom shoe data set, 12 healthy participants (7 male and 5 female) volunteered (age: 25 ± 6 years, height: 174 ± 7 cm, mass: 68 ± 10 kg). The balance-pod shoe data comprised 12 participants (5 male and 7 female; age: 25 ± 4 years, height: 172 ± 11 cm, mass: 67 ± 11 kg). In total for each data set, 14 participants were recruited, but because of technical issues, two participants of each data set had to be removed from analyses. All participants were eligible if they were free from any lower and upper extremity orthopedic or neurological problems and pain. Participants were excluded, if they had used the unstable shoe construction prior to the study. All participants were informed of the aims of this study and signed an informed consent form, which was approved by the Local Ethics Committee. 2.2. Data acquisition and processing Participants were asked to walk with both, a regular shoe (control situation) and with the unstable shoe at self-selected walking speed at a 10-m walkway. Prior to data acquisition, participants were allowed to walk along the walk-way as often as necessary until they felt comfortable during walking with their regular shoes. This was typically reached after five minutes of walking at selfselected walking speed. After this accommodation phase, the data were captured using the different shoes. For each shoe, participants again had approximately five minutes of accommodation time to feel comfortable during walking at the walkway. Surface electromyography (EMG) data were recorded using a telemetry system (DELSYS, Myomonitor IV, Boston, MA, USA) and the software EMGworks version 3.6, which was installed on a stationary computer. Single differential Ag bar-electrodes with a contact dimension of 10.0 1.0 mm, an inter-electrode distance of 10.0 mm and a resulting detection area of 10 mm2 (DELSYS, DE2.1) were attached parallel to the muscle fiber direction over the mid-muscle belly (Hermens et al., 2000). The EMG amplifier system gain was 1000 with an common mode rejection ratio of 92 dB, an input impedance of >1015 O and an overall channel noise 61.2 lV (RMS, R.T.I)2. The analog signals were converted into digital signals using a sampling rate of 1000 Hz and a 16-bit analog–digital converter, using a ±5 V signal input range, which was integrated into to the EMG telemetry handheld device (DELSYS, Myomonitor IV, Boston, MA, USA). Prior to electrode placement, the skin was shaved, slightly abraded and cleaned using skin preparation gel. The ground electrode was placed at the wrist. In total muscle activities of the
B. Horsak et al. / Journal of Electromyography and Kinesiology 25 (2015) 175–181
gluteus maximus (GM), vastus medialis (VM), vastus lateralis (VL), biceps femoris (BF), tibialis anterior (TA), peroneus longus (PL) and gastrocnemius medialis (GMM) of the dominant leg were recorded. Raw data signals were then digitally band-pass filtered using a 4th order Butterworth filter with cut-off frequencies of 10 Hz and 450 Hz. Then, signals were full wave rectified and linear enveloped using another 4th order low-pass Butterworth filter, with a cut-off frequency of 8 Hz. The cut-off frequency was determined using power spectrum analysis according to Nielsen et al. (1994). Briefly, the power spectrum of each trial was calculated using the FFT algorithm, with a Hanning window of 50% overlap and a window length of 2048. This resulted in a resolution of 0.5 Hz for each power spectra. Power spectra were normalized to a maximum value of 1 and then all six trials for each person were averaged to a mean power spectra. The average of the frequencies at which the accumulation of energy reached 95% of total energy (see Nielsen et al., 1994) was defined as the most appropriate cut-off frequency. EMG data were then time normalized to 100% gait cycle. Time instances during each gait cycle for initial contact and toe-off were determined using kinematic and kinetic data (see Horsak and Baca, 2012, 2013). Resulting linear envelopes were then amplitude-normalized to the mean peak muscle activity of all valid control shoe trials (=100%), as described by Buchecker et al. (2010). The co-contraction index (CCI) for each data point of the linear envelope was calculated using Eq. (1) (Rudolph et al., 2000, 2001; Schmitt and Rudolph, 2008):
CCIn¼100 ¼ i¼1
lowEMGi ðhighEMGi þ lowEMGi Þ highEMGi
ð1Þ
The lowEMGi is the normalized value of the EMG signal at one point (i) for the less active muscle. The highEMGi is the corresponding activity of the higher active muscle. The CCI therefore, can range at any time point from zero (no muscle activity) to a maximum of 200 (if both muscle are active 100%; due to the normalization procedure slightly greater values are possible). According to Rudolph et al. (2001), this method provides an estimate of the relative activation of the pair of muscles as well as the magnitude of the co-contraction. CCI was calculated for the following opposing muscle groups: vastus medialis to biceps femoris (VM/BF), vastus lateralis to biceps femoris (VL/BF), vastus medialis to gastrocnemius medialis (VM/GMM), vastus lateralis to gastrocnemius medialis (VL/GMM) and tibialis anterior to gastrocnemius medialis (TA/ GMM). These muscle pairs were selected based on their functional contribution to stabilize the knee joint (in the sagittal plane) and their possible contribution to an increased bone to bone contact force. From the resulting CCI time-series mean amplitude values for the first (0–50% stance) and second half of stance phase (50% stance to toe-off) were calculated. For all CCI parameters, data of six trials under each walking condition were averaged for each participant and then used for further analysis. All data processing steps were performed using self-written scripts in MATLAB (The MathWorks Inc., Ismaning, GER).
177
ANOVA was run separately for each muscle pair to reduce the complexity in interpretation of possible interactions. The described procedure was performed for each unstable shoe condition (rocker-bottom and balance-pod) separately. An alpha level p 6 0.05 was set a priori to indicate significant statistical findings. To estimate practical relevance of observed CCI differences between the analyzed shoe designs Cohen’s d was calculated using Eq. (2). An effect size of 0.2 was considered as a small effect, 0.5 as a moderate effect and values >0.8 as a great effect.
d¼
ðlA lB Þ s
ð2Þ
where lA is the mean value of CCI for the unstable shoe and lB the mean value for the corresponding CCI of the control shoe. The standard deviation (s) was estimated as a pooled standard deviation (spooled) of both groups using Eq. (3).
spooled
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ðnA 1Þs2A þ ðnB 1Þs2B ¼ nA þ nB 2
ð3Þ
3. Results Comparing co-contraction between opposing muscle groups around the knee when walking with the rocker-bottom shoe design revealed some statistically significant alterations (Fig. 2). For VM/BF [F (1, 11) = 6.824, p = 0.024, g2 = 0.38], VM/GMM [F (1, 11) = 11.006, p = 0.007, g2 = 0.50] and VL/GMM [F (1, 11) = 12.962, p = 0.004, g2 = 0.54] significant main effects of shoe type were found, indicating greater CCI values for walking with the rocker-bottom shoe design (Table 1). For all of the analyzed muscle pairs, significant main effects for gait phase were also observed, demonstrating greater CCI during the first half of stance for all of them: for VM/BF [F (1, 11) = 23.217, p = 0.001, g2 = 0.68], VL/BF [F (1, 11) = 23.343, p = 0.001, g2 = 0.68], for VM/GMM [F (1, 11) = 4.959, p = 0.048, g2 = 0.31], VL/GMM [F (1, 11) = 6.840, p = 0.024, g2 = 0.38], TA/GMM [F (1, 11) = 16.559, p = 0.002, g2 = 0.60]. No interactions were found between shoe type and gait phase. Descriptive statistics, effect size as well as 95% CI of the difference between both shoe conditions are presented in Table 1. For the balance-pod shoe design only for the VM/GMM muscle pair, a significant main effect of shoe type was observed [F (1, 11) = 4.976, p = 0.047, g2 = 0.31], also demonstrating greater CCI values for walking with the unstable shoe design (Fig. 3). In addition, except for VL/GMM, all muscle pairs showed a significant main effect of gait phase, showing greater CCI values for the first half of stance: for VM/BF [F (1, 11) = 8.494, p = 0.014, g2 = 0.44], VL/BF [F (1,11) = 13.202, p = 0.004, g2 = 0.55], for VM/GMM [F (1, 11) = 10.690, p = 0.007, g2 = 0.49], TA/GMM [F (1, 11) = 7.413, p = 0.02, g2 = 0.40]. No significant interactions were found. Fig. 4 gives an overview how muscle activity of each muscle contributes to the CCI. For the balance-pod shoe design descriptive statistics, effect size as well as 95% CI of the difference between both shoe conditions are presented in Table 2.
2.3. Statistical analysis 4. Discussion Statistical analyses were performed using IBM SPSS Statistics 21 (Somer, NY, USA). Initially, parameters were tested to comply with needed statistical assumptions. The data were reported using the mean and standard deviation (SD) for the CCI in each shoe condition as well as the 95% CI of the difference between the unstable and the stable shoe. A 2 2 repeated measures ANOVA using the CCI as the dependent variable and the shoe type (unstable and stable condition) and gait phase (early and late stance) as independent variables were used to identify effects of the shoe condition and the gait phase as well as first order interaction effects. The
Up to date there is a potential lack of scientific evidence reporting how unstable shoe constructions (e.g. rocker-bottom or balance-pod shoe design) contribute to co-contraction of opposing muscle groups surrounding the knee joint. The main body of research used inverse dynamic approaches and EMG to analyze if unstable shoe constructions may have the potential to decrease joint loading. When relying on such estimates, one has to know, if co-contraction is taking place, or at least does not change its amount, because inverse dynamic approaches only account for
178
B. Horsak et al. / Journal of Electromyography and Kinesiology 25 (2015) 175–181
Fig. 2. Mean co-contraction index (CCI) wave forms and the corresponding standard deviations for walking with the stable shoes (dashed line) and the rocker-bottom shoes (solid line). The vertical line shows toe-off.
Table 1 Mean (SD) co-contraction index (CCI) for walking with the rocker-bottom shoes and regular shoes. CCI
Rocker-bottom shoe mean (SD)
Control shoe mean (SD)
95% CI of difference
VM/BF 1st half stance 2nd half stance
27.2 6.5
(16.2) (5.3)
22.4 4.6
(12.1) (2.4)
0.4 0.9
10.1 4.7
0.3 0.5
VL/BF 1st half stance 2nd half stance
27.7 6.9
(16.4) (6.0)
26.0 4.8
(15.2) (2.5)
3.0 0.9
6.3 5.1
0.1 0.5
VM/GMM 1st half stance 2nd half stance
15.9 10.1
(6.0) (9.7)
11.1 6.3
(4.0) (2.8)
1.8 0.9
7.8 8.5
0.9 0.5
VL/GMM 1st half stance 2nd half stance
19.4 12.7
(10.2) (9.2)
13.3 7.9
(6.6) (4.6)
2.6 0.3
9.5 9.9
0.7 0.7
TA/GMM 1st half stance 2nd half stance
13.8 7.5
(10.4) (5.8)
13.2 7.2
(8.3) (4.6)
5.5 2.7
6.7 3.5
0.1 0.1
the net joint moments. Therefore, the purpose of this study was to compare co-contraction levels when walking with two different unstable shoe constructions compared to walking with regular shoes. We hypothesized that both unstable shoe constructions will increase co-contraction for opposing muscles surrounding the knee joint. Based on our findings, this was true for the rocker-bottom shoe data as well as for the balance-pod shoe data for one parameter. Besides the many studies analyzing muscle activation patterns between walking with regular and with unstable shoes, only
Cohen’s d
Buchecker et al. (2010) have reported CCI when walking with rocker-bottom shoes. They found a significant increase of co-contraction between vastus lateralis and gastrocnemius medialis for mid and late stance and reported an effect sizes d of greater than 0.7 for both. Though, it has to be outlined that, the data of Buchecker et al. (2010) are limited to a specific population, as they only recruited overweight men for their study. However, their results are very consistent with data observed in our study. Our results for the rocker-bottom shoe data set showed an increased co-contraction for VM/GMM and VL/GMM for early stance, both
179
B. Horsak et al. / Journal of Electromyography and Kinesiology 25 (2015) 175–181
Normalized EMG [%]
Fig. 3. Mean co-contraction index (CCI) wave forms and the corresponding standard deviations for walking with the stable shoes (dashed line) and balance-pod shoes (solid line). The vertical line shows toe-off.
100
A
80 60 40 20 0
Normalized EMG [%]
1st 2nd half of stance
1st 2nd half of stance
1st 2nd half of stance
1st 2nd half of stance
100
1st 2nd half of stance
B
80 60 40 20 0 1st 2nd half of stance VM
1st 2nd half of stance VL
1st 2nd half of stance BF
1st 2nd half of stance GMM
1st 2nd half of stance TA
Fig. 4. Mean (SD) muscle activity during first and second half of stance. Gray bars show muscle activity of walking with regular shoes, white bars walking in the unstable shoe condition: (A) rocker-bottom, (B) balance-pod shoe design.
with a Cohen’s d of greater than 0.7. In slight contrast to Buchecker et al. (2010), we also observed an increase of the VM/BF during first half of stance. Unexpectedly, only the VM/GMM showed an increase of CCI during first half of stance when walking with the balance-pod shoe. The CCI differences found for the rocker-bottom shoe condition may be explained by the altered EMG activation
patterns, already described in other studies (e.g. Romkes et al., 2006). When walking with rocker-bottom shoes, people walk with an increased dorsiflexion angle at initial contact, followed by a continuous plantarflexion movement. These kinematic differences are accompanied by an increased activation of the gastrocnemius during early to mid-stance and an increase of vastii activity during mid
180
B. Horsak et al. / Journal of Electromyography and Kinesiology 25 (2015) 175–181
Table 2 Mean (SD) co-contraction index (CCI) for walking with the balance-pod shoes and regular shoes. CCI
Balance-pod shoe mean (SD)
Control shoe mean (SD)
95% CI of difference
VM/BF 1st half stance 2nd half stance
27.1 6.2
(29.2) (5.0)
20.7 5.5
(12.6) (2.6)
5.0 1.4
18.0 2.9
0.3 0.2
VL/BF 1st half stance 2nd half stance
29.8 5.5
(25.7) (3.3)
24.1 5.3
(14.8) (2.4)
2.7 1.0
14.2 1.5
0.3 0.1
VM/GMM 1st half stance 2nd half stance
13.1 9.0
(7.5) (8.0)
10.8 6.2
(4.7) (3.2)
0.7 0.6
5.4 6.3
0.4 0.5
VL/GMM 1st half stance 2nd half stance
14.3 12.8
(6.3) (13.0)
13.4 8.6
(8.1) (8.0)
1.8 0.1
3.7 8.2
0.1 0.4
TA/GMM 1st half stance 2nd half stance
9.9 6.6
(4.8) (2.6)
11.2 6.7
(8.6) (2.3)
5.0 1.4
2.3 1.1
0.2 0.0
to late stance. The latter was also observed for walking with balance-pod shoes (Horsak and Baca, 2013), and may be a strategy of the neuromuscular system to adopt for increased instability during late stance. The rocker profile of the rocker-bottom shoe and the increased neural drive necessary for gastrocnemius to maintain the continuous plantarflexion movement, may be a reason for the increased CCI when walking with these shoes. As co-contraction was identified as one strategy of the neuromuscular system to stabilize a joint (Lewek et al., 2004), one would also expect a great increase of co-contraction when walking with balance-pod shoes. One reason why only small differences for CCI were found, could be the amount of instability introduced by the balance-pod shoes. These shoes have two balance pods, one centered in the mid of the forefoot, one in the mid of the heel, to increase instability. The pods mainly introduce instability to the medial and lateral directions. Previous research has shown, that when standing with balance-pod shoes, no significant alterations in postural sway was observed (Horsak and Baca, 2013). The instability introduced by the balance-pod shoes might be too small, as it could affect neuromuscular activation patterns associated with stabilization strategies of a joint. A second explanation might be the fact, that muscle pairs selected in this study typically cause motion in the sagittal plane. The balance-pod shoe design destabilizes gait, due to their construction, more in medio-lateral direction. Therefore, the selected muscle pairs may not be appropriate enough to capture co-contraction caused by this shoe concept. Cohen’s d was used in this study to establish an estimate of the practical relevance of CCI differences observed. Cohens’s d is a widely accepted estimate of the effect size, which is often utilized in studies related to sport sciences (Fröhlich et al., 2009). Even though the importance of a found effect can be classified using d, this measure should always be interpreted in the context of the related research area. In our study differences showed relatively high effect sizes of greater than 0.7. Hubley-Kozey et al. (2009), for example have used the same CCI for comparing asymptomatic individuals with those with varying degrees of knee osteoarthritis during walking. They have used an interval of approximately the first 20–25% of gait cycle to calculate a mean CCI value, therefore comparable with our methodology. Their data for the CCI of the vastus medialis and gastrocnemius medialis for asymptomatic controls are relatively consistent with data of our study for the regular shoe condition. For VM/GMM CCI differences between healthy controls and the severe osteoarthritis group was approximately the same size as the differences observed in our study between the rocker-bottom and the regular shoe condition. These results further highlight the potential effect of unstable shoe constructions in increasing co-contraction around the knee joint.
Cohen’s d
This study is subject to some limitations, which have to be recognized. Our study is based on two data sets published earlier. No sample size estimation was performed for the underlying research question. The CCI utilized in this study is an often used estimate of co-contraction (Rudolph et al., 2001; Schmitt and Rudolph, 2008; Hubley-Kozey et al., 2009), but still up to date there is no standard method for calculating and reporting co-contraction. However, a recently published study by Hubley-Kozey et al. (2013) found good to excellent intraclass correlation coefficients for individuals with knee osteoarthritis. Therefore, one could expect that the application of the CCI for healthy individuals is valid too. Another limitation might be a very small difference observed in walking speed between both shoe conditions in both data sets. Anyway, as these differences (
