Human Movement Science 40 (2015) 154–162
Contents lists available at ScienceDirect
Human Movement Science journal homepage: www.elsevier.com/locate/humov
Alterations in stride-to-stride variability during walking in individuals with chronic ankle instability Masafumi Terada a,⇑, Samantha Bowker b, Abbey C. Thomas c, Brian Pietrosimone d, Claire E. Hiller e, Martin S. Rice f, Phillip A. Gribble a a
University of Kentucky, 900 South Limestone Street, Lexington, KY 40536-0200, United States Kent State University, MAC Center, P.O. Box 5190, Kent, OH 44242, United States c University of North Carolina at Charlotte, 9201 University City Blvd, Charlotte, NC 28223-0001, United States d University of North Carolina at Chapel Hill, 26D Fetzer Hall, Chapel Hill, NC 25599, United States e University of Sydney, 75 East St Lidcombe, NSW 2141, Australia f University of Toledo, 2801 W Bancroft Street, Toledo, OH 43606, United States b
a r t i c l e
i n f o
PsycINFO classification: 2221 Keywords: Joint injury Movement pattern Sensorimotor control Gait
a b s t r a c t The aim of this study was to evaluate stride-to-stride variability of the lower extremity during walking in individuals with and without chronic ankle instability (CAI) using a nonlinear analysis. Twenty-five participants with self-reported CAI and 27 healthy control participants volunteered for this study. Participants walked on a motor-driven treadmill for 3 min at their selected speed. Lower extremity kinematics in the sagittal and frontal planes were recorded using a passive retroreflective marker motion capture system. The temporal structure of walking variability was analyzed with sample entropy (SampEn). The CAI group produced lower SampEn values in frontal-plane ankle kinematics compared to the control group (P = .04). No significant group differences were observed for SampEn values of other kinematics (P > .05). Participants with CAI demonstrated less stride-to-stride variability of the frontal plane ankle kinematics compared to healthy controls. Decreased variability of walking patterns demonstrated by participants with CAI indicates that the presence of CAI may be associated with a less adaptable sensorimotor system to environmental changes. The altered sensorimotor function associated with CAI
⇑ Corresponding author at: 206A Charles T. Wethington Building University of Kentucky, 900 South Limestone Street, Lexington, KY 40536-0200, United States. Tel.: +1 859 218 0594; fax: +1 859 323 6003. E-mail address: [email protected] (M. Terada). http://dx.doi.org/10.1016/j.humov.2014.12.004 0167-9457/Ó 2014 Elsevier B.V. All rights reserved.
M. Terada et al. / Human Movement Science 40 (2015) 154–162
155
may be targets for clinical interventions, and it is critical to explore how interventions protocols affect sensorimotor system function. Ó 2014 Elsevier B.V. All rights reserved.
1. Introduction The ankle is the most common site for initial and recurrent injuries related to sport (Fernandez, Yard, & Comstock, 2007; Swenson, Yard, Fields, & Comstock, 2009). Many physically active individuals suffer from ankle sprains (Hootman, Dick, & Agel, 2007; Waterman, Owens, Davey, Zacchilli, & Belmont, 2010), and it has been estimated that up to 73.6% of those with a history of ankle sprain will go on to experience recurrent ankle sprains and repeated bouts of perceived feeling of giving-way (Anandacoomarasamy & Barnsley, 2005; Gerber, Williams, Scoville, Arciero, & Taylor, 1998; Konradsen, Bech, Ehrenbjerg, & Nickelsen, 2002). Chronic ankle instability (CAI) is a significant orthopedic concern in physically active populations and characterized by a recurrent perception of the ankle giving away (Hertel, 2002). People with CAI demonstrate decreased physical activity levels (Hiller et al., 2012; Verhagen, van Mechelen, & de Vente, 2000) and quality of life (Arnold, Wright, & Ross, 2011), as well as a higher risk of developing posttraumatic ankle osteoarthritis (Hirose, Murakami, Minowa, Kura, & Yamashita, 2004; Valderrabano, Hintermann, Horisberger, & Fung, 2006). Elucidating the underlying mechanism of CAI is critically important to improve current care for CAI. It has been suggested that inappropriate alterations in sensorimotor control play a significant role in perpetuating the recurrent perception of ankle instability (Hertel, 2008; Wikstrom, HubbardTurner, & McKeon, 2013). Movement patterns during gait have been examined with biomechanical measures to estimate sensorimotor function in individuals with CAI, with altered lower extremity kinematics and kinetics observed in those with CAI. (Brown, 2011; Drewes, McKeon, Kerrigan, & Hertel, 2009; Drewes et al., 2009; Hass, Bishop, Doidge, & Wikstrom, 2010; Herb et al., 2014; Wikstrom, Bishop, Inamdar, & Hass, 2010). Individuals with CAI have demonstrated less joint coupling variability compared to healthy controls during gait (Herb et al., 2014). Applying dynamical system theory of motor control, a healthy sensorimotor system self-organizes in multiple ways by adapting organismic, environmental, and task constraints to find the most stable solutions to achieve movement goals (Davids, Glazier, Araujo, & Bartlett, 2003; Hoch & McKeon, 2010). As constraints on the sensorimotor system increase, the sensorimotor system becomes unstable and switches to use a new, more stable movement strategies (Davids et al., 2003). The presence of CAI may increase organismic constraints on the sensorimotor system and diminish its ability to reorganize movement strategies and adjust changes in task demands or environmental conditions, thereby producing rigid and inflexible movement patterns (Herb et al., 2014; McKeon, 2012; Wikstrom et al., 2013). Stergiou and Decker (2011) suggested that ideal variability of healthy motor control has chaotic characteristics that are not random but have a deterministic pattern. In a healthy state, the sensorimotor system is flexible and adaptable to stresses placed on the human body and changes in tasks demands or environmental conditions (Stergiou, Harbourne, & Cavanaugh, 2006). Herb et al. (2014) has calculated an intersegmental variability coefficient to assess the magnitude of the stride-to-stride variability in shank-rearfoot coupling in patients with CAI. During this variability analysis, kinematic data from a few strides are averaged to generate a mean ensemble curve and accompanied by time normalization (Buzzi, Stergiou, Kurz, Hageman, & Heidel, 2003). However, the time normalization masks the temporal variations of the gait pattern since it stretches or pulls the original data (Buzzi et al., 2003). Alternatively, a chaotic gait pattern can be determined using a nonlinear mathematical approach that quantifies the temporal structure of stride-to-stride variability (Buzzi et al., 2003; Stergiou & Decker, 2011). The nonlinear variability analysis focuses on understanding how a movement pattern changes over multiple gait cycles and determines whether a chaotic structure and complexity are present in movement (Buzzi et al., 2003; Stergiou & Decker, 2011).
156
M. Terada et al. / Human Movement Science 40 (2015) 154–162
Gait variability with nonlinear analysis in patients with anterior cruciate ligament (ACL) injury (Moraiti, Stergiou, Ristanis, & Georgoulis, 2007) and knee osteoarthritis (Tochigi, Segal, Vaseenon, & Brown, 2012) have found decreased optimal state of variability compared to healthy controls. However, it is currently unclear what effect CAI has on the structure of movement variability in a time series, such as changes observed in gait fluctuations over multiple gait cycles. Using a nonlinear dynamical approach for exploring movement variability in CAI population may provide additional theoretical estimates of the sensorimotor and neuromuscular mechanisms under CAI. Therefore, the purpose of the study was to evaluate stride-to-stride variability of the lower extremity during walking in individuals with and without CAI using a nonlinear analysis. Our hypothesis was that individuals with CAI would demonstrate decreased stride-to-stride variability in the lower extremity compared to healthy controls. 2. Method The single-blinded, case-control study involved two groups of individuals including: (1) CAI; and (2) healthy control participants. Participants were screened by an investigator for inclusion criteria and group allocation. Two other investigators who measured and analyzed the selected outcome measures were blinded to group membership. 2.1. Participants Fifty-two physically active participants from the university community volunteered for this study. All participants read and signed an informed consent approved by a University Institutional Review Board. All participants had no: (1) diagnosed balance or vestibular disorders; (2) history of low back pain in the previous 6 mon; (3) history of surgery in the lower extremity; (4) history of seizures; (5) history of a concussion in the past 6 months; (6) history of neurological injuries or diseases; and (7) history of any self-reported musculoskeletal and neurovascular injuries and disorders in the lower extremity other than lateral ankle sprains in the previous 2 years. Participants’ physical activity levels were determined by asking how many days per week and how much time each day they spent doing moderate-to-vigorous-intensity sports, fitness or recreation activities. Twenty-five participants (14 M, 11 F; 22.48 ± 3.98 years; 171.40 ± 8.68 cm; 76.19 ± 14.77 kg) were included in the CAI group based on recommendations from the International Ankle Consortium (Gribble et al., 2013, 2014a, 2014b). Participants in the CAI group had to report: (1) a previous history of at least two acute lateral ankle sprains that caused pain, swelling, and temporary loss of function; (2) at least two repeated episodes of ‘‘giving-way’’ (perceived instability) in the past 6 months; and (3) self-reported ankle dysfunction assessed as scoring P4 the Ankle Instability Instrument (AII) and P10 on Identification of Functional Ankle Instability (IdFAI). No participant with CAI had acutely sprained their ankle in the 3 months before testing. In the event participants reported a bilateral history of ankle sprains, the limb with the greatest reported functional impairments on the AII and IdFAI was included in the study. Twenty-seven participants (10 M, 17 F; 21.56 ± 3.15 years; 166.42 ± 8.09 cm; 66.61 ± 12.99 kg) were included in the control group. The control group participants were required to have no history of ankle sprain and score 0 on both the AII and IdFAI. A test limb for the control group was randomly selected. Additionally, all participants completed the Foot and Ankle Ability Measure (FAAM) activities of daily living subscale and FAAM sports subscale (Martin, Irrgang, Burdett, Conti, & Van Swearingen, 2005) to better understand their self-reported ankle disability levels. 2.2. Test procedures Gait variability was assessed by examining the time series of sagittal and frontal plane kinematics of the lower extremity and trunk during walking. Participants were asked to walk on a motor-driven treadmill (Cambridge Model 3050, Quinton Instruments Inc., Seattle, MA) for 3 min at their
M. Terada et al. / Human Movement Science 40 (2015) 154–162
157
self-selected speed while kinematic data were captured using 12 Eagle digital cameras (Motion Analysis Corporation, Santa Rosa, CA) and Cortex 3.6.1 motion capture/processing software (Motion Analysis Corporation, Santa Rosa, CA). A self-selected speed was determined during an 8-min warm-up (Grabiner & Troy, 2005; Yentes et al., 2013). The inclination of the treadmill was set to 0. Prior to data collection, fifty-six retroreflective markers with double-sided adhesive tape were placed on selected anatomical landmarks of the lower leg, the pelvis, the shoulder, and the neck. Markers were affixed bilaterally at: (1) iliac crest; (2) anterior superior iliac spine; (3) posterior superior iliac spine; (4) greater trochanter; (5) thigh clusters consisting of four lateral markers; (6) anterior distal thigh; (7) lateral femoral epicondyle; (8) medial femoral epicondyle; (9) patella; (10) tibial tuberosity; (11) anterior distal shank; (12) shank cluster consisting four lateral markers; (13) lateral malleolus; (14) medial malleolus; (15) anterior distal lower leg; (16) 1st metatarsal (MT) head; (17) 2nd MT head; (18) 5th MT base; (19) dorsal surface of navicular; (20) posterior aspect of calcaneus; and (21) acromioclavicular joints, as well as a single marker on the manubrium, sacrum, and C7. Following marker placement, participants were asked to stand in a neutral position to create a kinematic model of skeletal segments. All gait data were sampled at 200 Hz. 2.3. Data processing The sagittal and frontal plane kinematics at the hip, knee, ankle, and trunk were processed in Visual 3D (C-Motion, Germantown, MD). Lower extremity joint and trunk rotations were defined based on the neutral stance of each participant and aligned with a three-dimensional laboratory coordinate system. A kinematic model consisting of eight skeletal segments (bilateral foot, shank, and thigh segments, pelvis, and trunk) with 27 degrees of freedom was created using the static standing trial (McLean, Su, & van den Bogert, 2003). The pelvis and trunk were defined with respect to the global coordinate system, while the hip, knee, and ankle joints of each limb were defined locally (McLean et al., 2007). The hip joint center was defined according to Bell, Pedersen, and Brand (1990). The knee joint center was defined according to Vaughan, Davis, and O’Connor (1992). The ankle joint center was defined as the midpoint between the lateral and medial malleoli. Each lower extremity joint was assigned with three rotational degrees of freedom. Joint angles during walking were calculated using the Cardan rotation sequence (Cole, Nigg, Ronsky, & Yeadon, 1993). Sagittal and frontal kinematics of the trunk were calculated as the angle between the trunk segment and the laboratory coordinate system. All kinematic data were expressed relative to each participant’s neutral stance trial. Stride-to-stride variability of lower extremity motion patterns during walking was examined using the nonlinear measure of the Sample Entropy (SampEn). The SampEn for each gait variable was calculated with a custom MATLAB file (Mathworks, Inc., Natick, MA) using the mathematical algorithms previously described in detail by Richman and Moorman (2000). The SampEn quantifies the uncertainty or unpredictability of kinematic time series (Yentes et al., 2013), with a smaller value of SampEn indicating a more periodic and regular pattern (Tochigi et al., 2012). It has been suggested that filtering the data may eliminate important information and provide a skewed view of inherent variability within the locomotor system (Mees & Judd, 1993). Therefore, the kinematic data were unfiltered for SampEn calculation to produce a more accurate representation of the variability within the sensorimotor system. Each kinematic time series from the testing trials contains 3600 data points. Input parameters for our SampEn calculation were (1) a series length (m) of 2 data points and (2) a tolerance widow (r) normalized to .2 times the standard deviation of individual time series (Yentes et al., 2013). 2.4. Statistical analysis Because of the ordinal nature, physical activity levels (the amount of hours per day and days per week) were compared between the CAI and control groups using Mann–Whitney U tests. Self-reported ankle dysfunction (AII and IdFAI) and disability (FAAM and FAAM Sport) were compared between the CAI and control group using independent t-tests to verify group inclusion. Based on analysis of the data using a Kolmogorov–Smirnov Z test for normality, all outcome variables were found not to be normally distributed (P < .05). Therefore, Mann–Whitney U tests were performed to examine differences in SampEn values between the CAI and control groups. Additionally, self-selected walking
158
M. Terada et al. / Human Movement Science 40 (2015) 154–162
speed was compared using an independent t-test between the groups. Cohen’s d effect sizes using the means and pooled standard deviations were calculated, along with 95% confidence interval (CI) to determine the magnitude of differences in dependent variables between groups (Cohen, 1988). The strength of effect sizes was interpreted as weak (d < 0.40), moderate (0.40 6 d < 0.80), and strong (d P 0.80) (Cohen, 1988). An a priori alpha level was set at P < .05 using SPSS 21.0 (SPSS, Inc. Chicago, IL.) for Windows for all statistical tests. 3. Results The amount of hours per day and days per week of physical activity did not differ between the CAI and control groups (P > .05). The CAI group scored significantly higher on the AII and IdFAI and lower on the FAAM compared to the control group (P < .01). Additional ankle injury characteristic information is provided in Table 1. No significant differences between groups were noted for the self-selected walking speed (t56 = 0.374, P = .71). The CAI group demonstrated a significantly lower SampEn value in frontal plane ankle kinematics compared to the control group (z = 2.08, P = .04 d = 0.59, 95% Cis: 1.14, 0.03). There were no differences in any other SampEn values between the CAI and control groups (P > .05, Table 2), with small effect sizes (d < 0.04). 4. Discussion The SampEn data in the current study partially support the hypothesized loss of optimal variability in participants with CAI. The CAI group demonstrated less variability of the frontal plane ankle kinematics compared to the control group. The finding of this current study indicates that the presence of CAI may make the sensorimotor system more rigid. Decreased frontal plane variability at the ankle joint observed in the current study may be an effort of the rigid sensorimotor system to minimize ankle giving-way by eliminating extra movements while walking. However, the increased rigidity of the sensorimotor system beyond the healthy range observed in this current study may decrease adaptability and flexibility to different perturbations and constraints (Stergiou & Decker, 2011), possibly having an association with mechanism of CAI and contributing to future degenerative pathology. Thus, it is critical to explore how interventions and rehabilitation protocols can develop proper movement strategies to cope with their ankle pathology. It has been suggested that gait variability analysis with nonlinear measures can provide useful information of the sensorimotor mechanisms under both healthy and pathological conditions (Stergiou & Decker, 2011). According to the ‘‘optimal movement variability’’ theoretical model, a healthy state is associated with chaotic temporal variation that has deterministic structure as well as reflects flexibility and adaptability of the motor control system to change in environmental
Table 1 Ankle injury characteristics for chronic ankle instability (CAI) and control groups (Mean ± SD).
n Physical activity (days/week) Physical activity (hours/day) AII IdFAI FAAM (%) FAAM sport (%) # Of lateral ankle sprain Time since last ankle sprain (month) # Of giving-way in past 6 months
CAI
Control
P-value
25 (14 male, 11 female) 3.96 ± 2.05 1.39 ± 0.89 5.96 ± 1.40 18.76 ± 3.56 89.97 ± 8.50 79.20 ± 11.34 3.76 ± 2.28 (range: 2–15) 35.54 ± 30.47 9.17 ± 19.81 (range: 2–50)
27 (10 male, 17 female) 3.83 ± 1.64 1.08 ± 0.51 0.00 0.00 99.96 ± 0.23 99.65 ± 1.81 – – –
– .82 .11
Contents lists available at ScienceDirect
Human Movement Science journal homepage: www.elsevier.com/locate/humov
Alterations in stride-to-stride variability during walking in individuals with chronic ankle instability Masafumi Terada a,⇑, Samantha Bowker b, Abbey C. Thomas c, Brian Pietrosimone d, Claire E. Hiller e, Martin S. Rice f, Phillip A. Gribble a a
University of Kentucky, 900 South Limestone Street, Lexington, KY 40536-0200, United States Kent State University, MAC Center, P.O. Box 5190, Kent, OH 44242, United States c University of North Carolina at Charlotte, 9201 University City Blvd, Charlotte, NC 28223-0001, United States d University of North Carolina at Chapel Hill, 26D Fetzer Hall, Chapel Hill, NC 25599, United States e University of Sydney, 75 East St Lidcombe, NSW 2141, Australia f University of Toledo, 2801 W Bancroft Street, Toledo, OH 43606, United States b
a r t i c l e
i n f o
PsycINFO classification: 2221 Keywords: Joint injury Movement pattern Sensorimotor control Gait
a b s t r a c t The aim of this study was to evaluate stride-to-stride variability of the lower extremity during walking in individuals with and without chronic ankle instability (CAI) using a nonlinear analysis. Twenty-five participants with self-reported CAI and 27 healthy control participants volunteered for this study. Participants walked on a motor-driven treadmill for 3 min at their selected speed. Lower extremity kinematics in the sagittal and frontal planes were recorded using a passive retroreflective marker motion capture system. The temporal structure of walking variability was analyzed with sample entropy (SampEn). The CAI group produced lower SampEn values in frontal-plane ankle kinematics compared to the control group (P = .04). No significant group differences were observed for SampEn values of other kinematics (P > .05). Participants with CAI demonstrated less stride-to-stride variability of the frontal plane ankle kinematics compared to healthy controls. Decreased variability of walking patterns demonstrated by participants with CAI indicates that the presence of CAI may be associated with a less adaptable sensorimotor system to environmental changes. The altered sensorimotor function associated with CAI
⇑ Corresponding author at: 206A Charles T. Wethington Building University of Kentucky, 900 South Limestone Street, Lexington, KY 40536-0200, United States. Tel.: +1 859 218 0594; fax: +1 859 323 6003. E-mail address: [email protected] (M. Terada). http://dx.doi.org/10.1016/j.humov.2014.12.004 0167-9457/Ó 2014 Elsevier B.V. All rights reserved.
M. Terada et al. / Human Movement Science 40 (2015) 154–162
155
may be targets for clinical interventions, and it is critical to explore how interventions protocols affect sensorimotor system function. Ó 2014 Elsevier B.V. All rights reserved.
1. Introduction The ankle is the most common site for initial and recurrent injuries related to sport (Fernandez, Yard, & Comstock, 2007; Swenson, Yard, Fields, & Comstock, 2009). Many physically active individuals suffer from ankle sprains (Hootman, Dick, & Agel, 2007; Waterman, Owens, Davey, Zacchilli, & Belmont, 2010), and it has been estimated that up to 73.6% of those with a history of ankle sprain will go on to experience recurrent ankle sprains and repeated bouts of perceived feeling of giving-way (Anandacoomarasamy & Barnsley, 2005; Gerber, Williams, Scoville, Arciero, & Taylor, 1998; Konradsen, Bech, Ehrenbjerg, & Nickelsen, 2002). Chronic ankle instability (CAI) is a significant orthopedic concern in physically active populations and characterized by a recurrent perception of the ankle giving away (Hertel, 2002). People with CAI demonstrate decreased physical activity levels (Hiller et al., 2012; Verhagen, van Mechelen, & de Vente, 2000) and quality of life (Arnold, Wright, & Ross, 2011), as well as a higher risk of developing posttraumatic ankle osteoarthritis (Hirose, Murakami, Minowa, Kura, & Yamashita, 2004; Valderrabano, Hintermann, Horisberger, & Fung, 2006). Elucidating the underlying mechanism of CAI is critically important to improve current care for CAI. It has been suggested that inappropriate alterations in sensorimotor control play a significant role in perpetuating the recurrent perception of ankle instability (Hertel, 2008; Wikstrom, HubbardTurner, & McKeon, 2013). Movement patterns during gait have been examined with biomechanical measures to estimate sensorimotor function in individuals with CAI, with altered lower extremity kinematics and kinetics observed in those with CAI. (Brown, 2011; Drewes, McKeon, Kerrigan, & Hertel, 2009; Drewes et al., 2009; Hass, Bishop, Doidge, & Wikstrom, 2010; Herb et al., 2014; Wikstrom, Bishop, Inamdar, & Hass, 2010). Individuals with CAI have demonstrated less joint coupling variability compared to healthy controls during gait (Herb et al., 2014). Applying dynamical system theory of motor control, a healthy sensorimotor system self-organizes in multiple ways by adapting organismic, environmental, and task constraints to find the most stable solutions to achieve movement goals (Davids, Glazier, Araujo, & Bartlett, 2003; Hoch & McKeon, 2010). As constraints on the sensorimotor system increase, the sensorimotor system becomes unstable and switches to use a new, more stable movement strategies (Davids et al., 2003). The presence of CAI may increase organismic constraints on the sensorimotor system and diminish its ability to reorganize movement strategies and adjust changes in task demands or environmental conditions, thereby producing rigid and inflexible movement patterns (Herb et al., 2014; McKeon, 2012; Wikstrom et al., 2013). Stergiou and Decker (2011) suggested that ideal variability of healthy motor control has chaotic characteristics that are not random but have a deterministic pattern. In a healthy state, the sensorimotor system is flexible and adaptable to stresses placed on the human body and changes in tasks demands or environmental conditions (Stergiou, Harbourne, & Cavanaugh, 2006). Herb et al. (2014) has calculated an intersegmental variability coefficient to assess the magnitude of the stride-to-stride variability in shank-rearfoot coupling in patients with CAI. During this variability analysis, kinematic data from a few strides are averaged to generate a mean ensemble curve and accompanied by time normalization (Buzzi, Stergiou, Kurz, Hageman, & Heidel, 2003). However, the time normalization masks the temporal variations of the gait pattern since it stretches or pulls the original data (Buzzi et al., 2003). Alternatively, a chaotic gait pattern can be determined using a nonlinear mathematical approach that quantifies the temporal structure of stride-to-stride variability (Buzzi et al., 2003; Stergiou & Decker, 2011). The nonlinear variability analysis focuses on understanding how a movement pattern changes over multiple gait cycles and determines whether a chaotic structure and complexity are present in movement (Buzzi et al., 2003; Stergiou & Decker, 2011).
156
M. Terada et al. / Human Movement Science 40 (2015) 154–162
Gait variability with nonlinear analysis in patients with anterior cruciate ligament (ACL) injury (Moraiti, Stergiou, Ristanis, & Georgoulis, 2007) and knee osteoarthritis (Tochigi, Segal, Vaseenon, & Brown, 2012) have found decreased optimal state of variability compared to healthy controls. However, it is currently unclear what effect CAI has on the structure of movement variability in a time series, such as changes observed in gait fluctuations over multiple gait cycles. Using a nonlinear dynamical approach for exploring movement variability in CAI population may provide additional theoretical estimates of the sensorimotor and neuromuscular mechanisms under CAI. Therefore, the purpose of the study was to evaluate stride-to-stride variability of the lower extremity during walking in individuals with and without CAI using a nonlinear analysis. Our hypothesis was that individuals with CAI would demonstrate decreased stride-to-stride variability in the lower extremity compared to healthy controls. 2. Method The single-blinded, case-control study involved two groups of individuals including: (1) CAI; and (2) healthy control participants. Participants were screened by an investigator for inclusion criteria and group allocation. Two other investigators who measured and analyzed the selected outcome measures were blinded to group membership. 2.1. Participants Fifty-two physically active participants from the university community volunteered for this study. All participants read and signed an informed consent approved by a University Institutional Review Board. All participants had no: (1) diagnosed balance or vestibular disorders; (2) history of low back pain in the previous 6 mon; (3) history of surgery in the lower extremity; (4) history of seizures; (5) history of a concussion in the past 6 months; (6) history of neurological injuries or diseases; and (7) history of any self-reported musculoskeletal and neurovascular injuries and disorders in the lower extremity other than lateral ankle sprains in the previous 2 years. Participants’ physical activity levels were determined by asking how many days per week and how much time each day they spent doing moderate-to-vigorous-intensity sports, fitness or recreation activities. Twenty-five participants (14 M, 11 F; 22.48 ± 3.98 years; 171.40 ± 8.68 cm; 76.19 ± 14.77 kg) were included in the CAI group based on recommendations from the International Ankle Consortium (Gribble et al., 2013, 2014a, 2014b). Participants in the CAI group had to report: (1) a previous history of at least two acute lateral ankle sprains that caused pain, swelling, and temporary loss of function; (2) at least two repeated episodes of ‘‘giving-way’’ (perceived instability) in the past 6 months; and (3) self-reported ankle dysfunction assessed as scoring P4 the Ankle Instability Instrument (AII) and P10 on Identification of Functional Ankle Instability (IdFAI). No participant with CAI had acutely sprained their ankle in the 3 months before testing. In the event participants reported a bilateral history of ankle sprains, the limb with the greatest reported functional impairments on the AII and IdFAI was included in the study. Twenty-seven participants (10 M, 17 F; 21.56 ± 3.15 years; 166.42 ± 8.09 cm; 66.61 ± 12.99 kg) were included in the control group. The control group participants were required to have no history of ankle sprain and score 0 on both the AII and IdFAI. A test limb for the control group was randomly selected. Additionally, all participants completed the Foot and Ankle Ability Measure (FAAM) activities of daily living subscale and FAAM sports subscale (Martin, Irrgang, Burdett, Conti, & Van Swearingen, 2005) to better understand their self-reported ankle disability levels. 2.2. Test procedures Gait variability was assessed by examining the time series of sagittal and frontal plane kinematics of the lower extremity and trunk during walking. Participants were asked to walk on a motor-driven treadmill (Cambridge Model 3050, Quinton Instruments Inc., Seattle, MA) for 3 min at their
M. Terada et al. / Human Movement Science 40 (2015) 154–162
157
self-selected speed while kinematic data were captured using 12 Eagle digital cameras (Motion Analysis Corporation, Santa Rosa, CA) and Cortex 3.6.1 motion capture/processing software (Motion Analysis Corporation, Santa Rosa, CA). A self-selected speed was determined during an 8-min warm-up (Grabiner & Troy, 2005; Yentes et al., 2013). The inclination of the treadmill was set to 0. Prior to data collection, fifty-six retroreflective markers with double-sided adhesive tape were placed on selected anatomical landmarks of the lower leg, the pelvis, the shoulder, and the neck. Markers were affixed bilaterally at: (1) iliac crest; (2) anterior superior iliac spine; (3) posterior superior iliac spine; (4) greater trochanter; (5) thigh clusters consisting of four lateral markers; (6) anterior distal thigh; (7) lateral femoral epicondyle; (8) medial femoral epicondyle; (9) patella; (10) tibial tuberosity; (11) anterior distal shank; (12) shank cluster consisting four lateral markers; (13) lateral malleolus; (14) medial malleolus; (15) anterior distal lower leg; (16) 1st metatarsal (MT) head; (17) 2nd MT head; (18) 5th MT base; (19) dorsal surface of navicular; (20) posterior aspect of calcaneus; and (21) acromioclavicular joints, as well as a single marker on the manubrium, sacrum, and C7. Following marker placement, participants were asked to stand in a neutral position to create a kinematic model of skeletal segments. All gait data were sampled at 200 Hz. 2.3. Data processing The sagittal and frontal plane kinematics at the hip, knee, ankle, and trunk were processed in Visual 3D (C-Motion, Germantown, MD). Lower extremity joint and trunk rotations were defined based on the neutral stance of each participant and aligned with a three-dimensional laboratory coordinate system. A kinematic model consisting of eight skeletal segments (bilateral foot, shank, and thigh segments, pelvis, and trunk) with 27 degrees of freedom was created using the static standing trial (McLean, Su, & van den Bogert, 2003). The pelvis and trunk were defined with respect to the global coordinate system, while the hip, knee, and ankle joints of each limb were defined locally (McLean et al., 2007). The hip joint center was defined according to Bell, Pedersen, and Brand (1990). The knee joint center was defined according to Vaughan, Davis, and O’Connor (1992). The ankle joint center was defined as the midpoint between the lateral and medial malleoli. Each lower extremity joint was assigned with three rotational degrees of freedom. Joint angles during walking were calculated using the Cardan rotation sequence (Cole, Nigg, Ronsky, & Yeadon, 1993). Sagittal and frontal kinematics of the trunk were calculated as the angle between the trunk segment and the laboratory coordinate system. All kinematic data were expressed relative to each participant’s neutral stance trial. Stride-to-stride variability of lower extremity motion patterns during walking was examined using the nonlinear measure of the Sample Entropy (SampEn). The SampEn for each gait variable was calculated with a custom MATLAB file (Mathworks, Inc., Natick, MA) using the mathematical algorithms previously described in detail by Richman and Moorman (2000). The SampEn quantifies the uncertainty or unpredictability of kinematic time series (Yentes et al., 2013), with a smaller value of SampEn indicating a more periodic and regular pattern (Tochigi et al., 2012). It has been suggested that filtering the data may eliminate important information and provide a skewed view of inherent variability within the locomotor system (Mees & Judd, 1993). Therefore, the kinematic data were unfiltered for SampEn calculation to produce a more accurate representation of the variability within the sensorimotor system. Each kinematic time series from the testing trials contains 3600 data points. Input parameters for our SampEn calculation were (1) a series length (m) of 2 data points and (2) a tolerance widow (r) normalized to .2 times the standard deviation of individual time series (Yentes et al., 2013). 2.4. Statistical analysis Because of the ordinal nature, physical activity levels (the amount of hours per day and days per week) were compared between the CAI and control groups using Mann–Whitney U tests. Self-reported ankle dysfunction (AII and IdFAI) and disability (FAAM and FAAM Sport) were compared between the CAI and control group using independent t-tests to verify group inclusion. Based on analysis of the data using a Kolmogorov–Smirnov Z test for normality, all outcome variables were found not to be normally distributed (P < .05). Therefore, Mann–Whitney U tests were performed to examine differences in SampEn values between the CAI and control groups. Additionally, self-selected walking
158
M. Terada et al. / Human Movement Science 40 (2015) 154–162
speed was compared using an independent t-test between the groups. Cohen’s d effect sizes using the means and pooled standard deviations were calculated, along with 95% confidence interval (CI) to determine the magnitude of differences in dependent variables between groups (Cohen, 1988). The strength of effect sizes was interpreted as weak (d < 0.40), moderate (0.40 6 d < 0.80), and strong (d P 0.80) (Cohen, 1988). An a priori alpha level was set at P < .05 using SPSS 21.0 (SPSS, Inc. Chicago, IL.) for Windows for all statistical tests. 3. Results The amount of hours per day and days per week of physical activity did not differ between the CAI and control groups (P > .05). The CAI group scored significantly higher on the AII and IdFAI and lower on the FAAM compared to the control group (P < .01). Additional ankle injury characteristic information is provided in Table 1. No significant differences between groups were noted for the self-selected walking speed (t56 = 0.374, P = .71). The CAI group demonstrated a significantly lower SampEn value in frontal plane ankle kinematics compared to the control group (z = 2.08, P = .04 d = 0.59, 95% Cis: 1.14, 0.03). There were no differences in any other SampEn values between the CAI and control groups (P > .05, Table 2), with small effect sizes (d < 0.04). 4. Discussion The SampEn data in the current study partially support the hypothesized loss of optimal variability in participants with CAI. The CAI group demonstrated less variability of the frontal plane ankle kinematics compared to the control group. The finding of this current study indicates that the presence of CAI may make the sensorimotor system more rigid. Decreased frontal plane variability at the ankle joint observed in the current study may be an effort of the rigid sensorimotor system to minimize ankle giving-way by eliminating extra movements while walking. However, the increased rigidity of the sensorimotor system beyond the healthy range observed in this current study may decrease adaptability and flexibility to different perturbations and constraints (Stergiou & Decker, 2011), possibly having an association with mechanism of CAI and contributing to future degenerative pathology. Thus, it is critical to explore how interventions and rehabilitation protocols can develop proper movement strategies to cope with their ankle pathology. It has been suggested that gait variability analysis with nonlinear measures can provide useful information of the sensorimotor mechanisms under both healthy and pathological conditions (Stergiou & Decker, 2011). According to the ‘‘optimal movement variability’’ theoretical model, a healthy state is associated with chaotic temporal variation that has deterministic structure as well as reflects flexibility and adaptability of the motor control system to change in environmental
Table 1 Ankle injury characteristics for chronic ankle instability (CAI) and control groups (Mean ± SD).
n Physical activity (days/week) Physical activity (hours/day) AII IdFAI FAAM (%) FAAM sport (%) # Of lateral ankle sprain Time since last ankle sprain (month) # Of giving-way in past 6 months
CAI
Control
P-value
25 (14 male, 11 female) 3.96 ± 2.05 1.39 ± 0.89 5.96 ± 1.40 18.76 ± 3.56 89.97 ± 8.50 79.20 ± 11.34 3.76 ± 2.28 (range: 2–15) 35.54 ± 30.47 9.17 ± 19.81 (range: 2–50)
27 (10 male, 17 female) 3.83 ± 1.64 1.08 ± 0.51 0.00 0.00 99.96 ± 0.23 99.65 ± 1.81 – – –
– .82 .11
