Journal of Applied Biomechanics, 2015, 31, 250 -257 http://dx.doi.org/10.1123/jab.2014-0138 © 2015 Human Kinetics, Inc.
ORIGINAL RESEARCH
The Effect of Exertion on Joint Kinematics and Kinetics During Running Using a Waveform Analysis Approach Lauren C. Benson and Kristian M. O’Connor University of Wisconsin–Milwaukee About half of all runners sustain a running-related injury every year. Exertion may contribute to risk of injury by altering joint mechanics. The purpose of this study was to examine the effects of exertion on runners’ joint mechanics using principal component analysis (PCA). Three-dimensional motion analysis of the lower extremity was performed on 16 healthy female runners before and after their typical training run. PCA was used to determine exertion-related changes in joint mechanics at the ankle, knee, and hip. Statistical significance for repeated-measures MANOVA of the retained principal components at each joint and plane of motion was at P < .05. Exercise effects were identified at the ankle (greater rate of eversion [PC2: P = .027], and decreased plantar flexion moment [overall: P = .044] and external rotation moment [PC3: P = .003]), knee (increased adduction [overall: P = .044] and internal rotation [PC3: P = .034], and decreased abduction moment [overall: P = .045]), and hip (increased internal rotation [PC1: P = .013] and range of mid- to late-stance rotation [PC2: P = .009], and decreased internal rotation moment [PC1: P = .001]). The observed changes in running mechanics reflect a gait profile that is often linked to running injury. The effects of more strenuous activity may result in mechanics that present an even greater risk for injury. Keywords: principal components analysis, exertion, injury, lower extremity Running is a common mode of exercise, which is important to maintaining good health.1 However, about half of all runners will sustain a running-related injury in a given year,1–3 with most of those injuries occurring at the knee.3–6 Despite research conducted in this field, the injury rates have not dramatically changed.2,3 Understanding the cause of running injuries is necessary for developing methods for prevention and better treatment options. Exposure to multiple impact forces over the course of a run, or many runs, is suspected to play a role in most overuse running injuries. These impact forces may be especially harmful if combined with improper mechanics.6–9 For example, proximal aspects of lower extremity gait may be affected by excessive or prolonged pronation, which has been suspected to contribute to common running injuries at the knee.6,7 Injured runners often complain of a gradual onset of pain as the run progresses,10 suggesting that running in an exerted state could contribute to a pathway to injury. Studies investigating the effects of exertion on running biomechanics typically use a protocol designed to bring runners to the point of maximum exertion to elicit the greatest changes in biomechanics. However, a more ecological protocol for inducing exertion-related changes in runners may be one that closely mimics a typical bout of exercise for a runner, while also providing an objective measure of exertion.10–12 Traditional investigations in biomechanics have focused on discrete variables to describe changes in running mechanics, including peak forces, peak angles, and excursions. However, relying on discrete variables of interest requires an a priori decision about which dependent variables and events in a stride cycle may change as a result of exertion. By examining the full time series, Lauren C. Benson and Kristian M. O’Connor are with the Department of Kinesiology, University of Wisconsin–Milwaukee, Milwaukee, WI. Address author correspondence to Lauren C. Benson at [email protected]. 250
or waveform, principal components analysis (PCA) is an unbiased way to determine relevant differences in joint kinematics.13,14 Previous studies that have examined the effect of exertion on running kinematics have been limited by the design of the fatigue protocol and the a priori identification of discrete dependent variables. The purpose of this study was to examine the influence of running in an exerted state on lower extremity joint kinematics and kinetics using waveform analysis. It was hypothesized that running in an exerted state would result in joint mechanics that could present a risk for running injury. Utilizing a running protocol that mimics a person’s typical training experience will allow for quantification of the ecological effects of running in an exerted state on joint mechanics.
Methods The study protocol was approved by the university institutional review board, and all participants provided informed consent before participating in the study. Sample size estimations were based on a repeated-measures MANOVA design with a medium effect size (ηp2 = .2), 80% power, and α = .05.15 Sixteen recreational runners were recruited through fliers posted on the university campus and with local running organizations. All participants were female runners, 18–45 years of age, who ran a minimum of 15 miles per week for the 6 months before the study. Exclusion criteria included self-reported cardiac risk, any lower extremity pain or runningrelated injury that limited training within the 6 months before the study, any history of major surgery to the lower extremity, the use of orthotics, pregnancy, medical conditions or medications that could impair balance, or a forefoot-strike running pattern.10,11 Additionally, participants were asked to refrain from running in a race in the 48 hours before testing and refrain from all exercise in the 24 hours before testing. Information about the participants’
Effect of Exertion During Running 251
Table 1 Participant information
Downloaded by W VA Univ Library on 09/16/16, Volume 31, Article Number 4
Participant Characteristics
Mean
SD
Height (m)
1.65
0.05
Mass (kg)
58.4
7.0
Age (y)
25
7
Shoe size
8
1
Typical running time (min)
39
10
Typical running distance (miles)
4.5
1.3
Typical running pace (min/mile)
8.85
0.93
Typical weekly mileage (miles)
24
11
typical running habits was collected to determine individualized experiment parameters (Table 1). During one testing session, three-dimensional kinematic data were collected at 200 Hz with a 10-camera Eagle system (Motion Analysis, Inc., Santa Rosa, CA), and ground reaction forces were recorded at 1000 Hz using an AMTI force plate (OR6-5; Advanced Mechanical Technology Inc., Watertown, MA). The participants were fitted with a heart rate monitor (Polar Electro Inc., Woodbury, NY), and the warm-up and run to an exerted state took place on a treadmill (C964i; Precor, Woodinville, WA) with the participants wearing their own training shoes. During the data collection before and after running to an exerted state, all participants ran in the same shoes (NBA-801; New Balance, Brighton, MA) with a mean size of women’s 8 (SD 1) for standardization purposes. This is a shoe without a heel counter, reinforced with tape, to allow for direct observation of rearfoot motion. The participants had a 5-minute warm-up period on the treadmill which consisted of light jogging at 2.2 m/s. Each participant’s pace for the data collection and the treadmill run was self-selected based on their typical pace for a training run.10,11 Retroreflective markers were applied to the participants’ skin to track the motion of the pelvis, thigh, leg, and foot. The tracking markers were placed on the left and right anterior superior iliac spine (ASIS) and posterior superior iliac spine (PSIS), a 4-marker plate was placed on both the thigh and the leg, and a marker triad was attached to the calcaneus. The location of the stand-alone markers on each ASIS and PSIS was marked in ink on the participant’s skin or a piece of tape attached to the participant. A 3-second standing calibration was recorded with additional calibration markers on the left and right iliac crests and greater trochanters, lateral and medial femoral epicondyles, malleoli, and first and fifth metatarsal heads of the right leg. These bony landmarks were chosen as part of the definition of each joint angle, and for their ease of identification. The calibration markers were removed following a 3-second standing calibration. Participants performed as many trials as necessary to record 10 successful running trials at their self-selected pace, ± 5%, in the common laboratory shoes across a 15-m runway containing an embedded force plate. A successful trial was defined as when right leg initial contact and toe-off occurred on the force plate. Kinematic and kinetic data were collected for each trial. Then the participants ran on the treadmill at their self-selected pace in their own training shoes and without the retroreflective markers. To mimic the participants’ typical training run, they were permitted to listen to music via headphones, if they desired. Starting in the first minute of the run and at every 5 minutes during the run, the participants’ heart rate and ratings of perceived exertion (RPE) were recorded.
When the participants reached a state of exertion measured by at least 85% of age-calculated maximum heart rate,16 and a score of at least 17 (very hard) on the RPE scale,10,11,17 they continued running for an additional 2 minutes before ending the run. Their final heart rate and RPE were recorded before they stopped the treadmill. Immediately at the end of the run, participants switched into the laboratory shoes and the tracking markers were reapplied. The participants performed 10 successful running trials overground as kinematic and kinetic data were collected. After recording the running trials, the calibration markers were reapplied and a 3-second standing calibration was recorded for the postrun markers. The methods for applying the tracking and calibration markers were the same as above. The raw coordinate data were filtered using a fourth-order, zero-lag, recursive Butterworth filter with a cutoff at 12 Hz. During the 3-second standing calibration, the pelvis, thigh, leg, and foot coordinate systems were defined as being coincident with one another in that position. The x-axis pointed to the right, the y-axis pointed anteriorly, and the z-axis pointed superiorly. Calculation of hip (H), knee (K), and ankle (A) joint angles in each plane of motion (sagittal [S], frontal [F], and transverse [T]) was done using a joint coordinate system approach.18 Net internal joint moments were calculated using an inverse dynamics analysis and were normalized to body mass.19 All joint angles and moments were time normalized in 1% increments to 101 data points representing 0% to 100% of stance phase, determined from the kinetic data as the period where the ground reaction force in the vertical direction was greater than 20 N. The data processing was done using Visual3D software (v4.75.34; C-Motion, Inc., Rockville, MD). PCA was used to assess changes in joint angles and joint moments before and after the run.20–22 Matrices for each waveform were created. The individual trials populated n rows, and the 101 data points populated p columns in an Xnxp matrix. Eigenvector analysis of the covariance matrix S101 × 101 determined the eigenvector matrix, U101 × 101, by orthonormalizing S101 × 101. The eigenvectors were the coefficients for the principal components (PCs) which represented the original data in new coordinates. The coefficients were the direction cosines that related the new axes to the old axes and were considered one mode of variation describing the variability within the entire original data set. The eigenvalues, L1 × 101, were determined by U’SU = L1 × 101. The eigenvalues represented the relative contribution or the rank of each PC to the total variation. A principal component score, Z nxp, was calculated for each individual waveform by multiplying the individual trial’s variation from the mean of all the trials, x̅ 1 × 101, by the transpose of the eigenvector matrix (Equation 1).
Z n×101 = (X n×101 – (1n×1 × x–1×101 )) × U '101×101 (1)
The PC scores represented the distance from each waveform to the mode of variability described by each principal component. The Z nx101 matrix was reduced to only the PC scores that represented the primary modes of variation. A parallel analysis was performed that retained only those PCs that contributed modes of variation greater than an equivalently-sized input matrix of normally-distributed randomly-generated numbers. The variance not explained by the retained PCs represented random error. The 10-stride means of the retained PC scores for each participant’s waveform before and after the run represented the dependent variables. Custom software was used to perform this PCA (Matlab v8.0.0.783; Mathworks, Inc., Natick, MA).
JAB Vol. 31, No. 4, 2015
252 Benson and O’Connor
Each of the 9 joint angles and moments (ankle, knee, and hip in the sagittal, frontal, and transverse planes) and 3 ground reaction forces (X, Y, Z) had several dependent variables, depending on the number of retained PCs for that waveform. A repeated-measures MANOVA for each waveform was done on the dependent variables. This led to a total of 9 joint angle MANOVAs, 9 joint moment MANOVAs, and 3 ground reaction force MANOVAs. Pre- and postexertion was the independent variable. For each waveform that exhibited a significant pre–post effect with the MANOVA, comparisons for the individual retained PCs was done with dependent t tests. Effect sizes for the MANOVAs and the follow-up t tests are reported using partial eta-squared. All statistical analyses were performed in SPSS with alpha set to .05 (v19.0.0.1; IBM, Inc., Chicago, IL).
to end the run when only one of the criteria was met: 2 met just the heart rate criterion and 5 met just the RPE criterion. For the other 9 participants, both criteria were met before ending the run. There was no obvious effect of this difference in end criteria on the joint kinematics observed before and after the run. At the end of the run, the mean heart rate for all participants was 88% (SD 4) of their age-calculated maximum. The mean final RPE was 18 (SD 1) on the 6–20 scale. The mean time from the end of the run to the end of the postrun data collection (including reapplication of tracking, but not standing, markers) was 10.89 (SD 3.94) minutes. The percent of the variance explained by the retained PCs for the joint angle data ranged from 95% to 98%, and the variance explained by the first PC, which described the most variation in each waveform, ranged from 70% to 92%. A repeated-measures MANOVA of the retained PCs for each joint angle indicated a significant difference in kinematics before and after the run for AF, K F, KT, and HT, with a large exercise effect for AF, KT, and HT, and only a moderate effect for K F (Table 2). The percent of the variance explained by the retained PCs for the joint moment data ranged from 84% to 97%, and the variance explained by the first PC ranged from 47% to 86%. A repeated-measures MANOVA of the
Downloaded by W VA Univ Library on 09/16/16, Volume 31, Article Number 4
Results Participants ran at a mean speed of 3.0 (SD 0.3) m/s for a mean time of 39 (SD 19) minutes until they reached at least 1 of the 2 stopping criteria (heart rate greater than 85% of age-calculated maximum, RPE greater than 17). In 7 cases, the participant asked
Table 2 Percent variance explained for the retained principle components (PCs) and the results of the MANOVAs performed for each waveform Principle Components PC1
PC2
PC3
PC4
AS
78
9
6
4
AF
84
7
4
AT
70
14
6
KS
81
13
4
KF
92
4
PC5
MANOVA
PC6
PC7
PC8
Total
P
ηp2
96
.075
.483
95
.005*
.614
96
.800
.172
98
.070
.408
96
.044*
.361
98
.030*
.634
96
.104
.367
Angle
KT
81
8
4
HS
78
13
6
HF
80
9
7
HT
70
13
9
AS
76
16
5
AF
79
9
6
AT
86
6
3
KS
68
17
KF
73
KT
75
5
3
3
2
96
.747
.087
96
.007*
.663
97
.044*
.452
2
96
.526
.219
95
.009*
.579
5
3
94
.341
.294
7
4
3
87
.045*
.530
12
6
92
.334
.223
93
.970
.070
84
.627
.441
85
.002*
.789
4
Moment
HS
73
9
5
3
3
HF
47
12
6
5
4
HT
62
10
7
4
2
4
4
3
Ground reaction force X
59
14
8
6
4
92
.164
.471
Y
45
27
12
5
3
93
.144
.486
Z
58
25
9
4
96
.165
.395
Note. Within each MANOVA were the mean scores for each retained PC. Effect size is reported as partial eta-squared. *Significantly different between pre and post, P < .05. JAB Vol. 31, No. 4, 2015
Effect of Exertion During Running 253
retained PCs for each joint moment indicated large exercise effects for AS, AT, K F, and HT, as well as significant differences in kinetics (Table 2). The percent of the variance explained by the retained PCs for the ground reaction force data ranged from 92% to 96%, and the variance explained by the first PC ranged from 45% to 59%. A repeated-measures MANOVA of the retained PCs for the ground reaction force in each direction indicated a medium exercise effect but no significant changes in the individual ground reaction forces (Table 2). For each of the above significant MANOVA findings, follow-up t tests were performed (Table 3).
For the AF angle, PC2 identified a greater rate of eversion (P = .027) postexercise (Table 3, Figure 1). For the K F angle, while the MANOVA was significant, there were no significant differences for any of the retained PCs when the subsequent t tests were performed (Tables 2 and 3). The combined effect of the retained PCs suggests an overall increase of about 2° in knee adduction postexercise (Figure 2). For the KT angle, PC3 detected an increase of about 1° in peak internal rotation (P = .034) postexercise (Table 3, Figure 3).
Downloaded by W VA Univ Library on 09/16/16, Volume 31, Article Number 4
Table 3 Results of the post hoc analysis of the retained principle components (PCs) for the waveforms that had significant pre and post differences in the MANOVA PC Scores PC
Pre
Post
P
ηp2
1
–20.70 (57.87)
19.65 (60.22)
.077
.194
2
–4.61 (16.50)
4.46 (13.03)
.027*
.287
3
–1.18 (6.28)
1.41 (13.06)
.448
.039
1
–6.78 (46.40)
7.57 (36.48)
.151
.133
Angle AF
KF KT
HT
2
–1.73 (6.56)
1.61 (9.27)
.094
.176
1
–2.53 (41.81)
2.29 (39.97)
.533
.026
2
1.24 (13.47)
–1.70 (9.57)
.169
.122
3
2.44 (8.61)
–2.22 (7.86)
.034*
.267
4
–1.68 (7.29)
1.56 (7.56)
.112
.159
5
0.97 (5.70)
–1.38 (6.40)
.057
.221
1
–12.59 (40.97)
13.80 (30.07)
.013*
.345
2
–3.59 (16.53)
3.13 (15.49)
.009*
.373
3
0.45 (14.50)
–0.02 (8.07)
.873
.002
4
–0.90 (6.01)
0.96 (5.93)
.371
.054
1
–0.43 (2.77)
0.31 (2.64)
.054
.226
2
0.13 (1.24)
–0.14 (1.17)
.137
.141
3
–0.002 (0.54)
0.01 (0.68)
.928
.001
1
0.53 (1.66)
–0.47 (1.48)
.076
.195
2
0.02 (0.22)
–0.02 (0.50)
.728
.008
3
0.08 (0.28)
–0.08 (0.22)
.003*
.458
1
–0.14 (2.09)
0.20 (2.48)
.213
.102
2
–0.06 (0.70)
0.05 (0.70)
.340
.061
3
0.02 (0.54)
–0.03 (0.41)
.481
.034
4
–0.03 (0.38)
0.06 (0.45)
.338
.061
1
0.20 (0.70)
–0.20 (0.74)
.001*
.550
2
–0.01 (0.32)
0.01 (0.21)
.701
.010
3
–0.03 (0.25)
0.04 (0.22)
.085
.185
4
0.01 (0.18)
–0.001 (0.18)
.664
.013
5
0.02 (0.09)
–0.01 (0.09)
.197
.108
Moment AS
AT
KF
HT
Note. Effect size is reported as partial eta-squared. *Significantly different between pre and post, P < .05.
Figure 1 — Frontal plane angle for the ankle over the duration of the stance phase in the PRE (black) and the POST (gray) conditions. The middle panel corresponds with the second retained principal component (PC) for the ankle frontal plane angle waveform. The waveform for the PRE (black) and POST (gray) conditions are plotted with plus (+) and minus (–) 1 SD of the scores for PC2. The third panel represents the variance explained by PC2 over the duration of stance.
Figure 2 — Frontal plane moment for the knee over the duration of the stance phase in the PRE (thick black) and the POST (thick gray) conditions with the waveform for the combined contributions of all retained principal components PRE (thin black) and POST (thin gray).
JAB Vol. 31, No. 4, 2015
Downloaded by W VA Univ Library on 09/16/16, Volume 31, Article Number 4
254 Benson and O’Connor
Figure 3 — Transverse plane angle for the knee over the duration of the stance phase in the PRE (black) and the POST (gray) conditions. The middle panel corresponds with the third retained principal component (PC) for the knee transverse plane angle waveform. The waveform for the PRE (black) and POST (gray) conditions are plotted with plus (+) and minus (–) 1 SD of the scores for PC3. The third panel represents the variance explained by PC3 over the duration of stance.
For the HT angle, PC1 identified an increase of about 2° in internal rotation throughout stance phase (P = .013) postexercise, and PC2 showed a greater range of rotation in mid- and late-stance (Table 3, Figure 4). There was a moderate effect of the training run on the change in PC scores for the AF, KT, and HT joint angles (Table 3). For the AS moment, while the MANOVA was significant, there were no significant differences for any of the retained PCs when the subsequent t tests were performed (Tables 2 and 3). The combined effect of the retained PCs suggests an overall decrease of less than 1 N∙m/kg in the plantar flexion moment postexercise (Figure 5). For the AT moment, PC3 indicated a greater rate of external rotation moment at the beginning of stance phase postexercise (P = .003) (Table 3, Figure 6). For the K F moment, while the omnibus test was significant, there were no significant differences as a result of the run for any of the retained PCs (Tables 2 and 3). The cumulative effects of these differences in the PCs likely led to an overall significant difference even though no individual PC was different. The reconstructed moment from the retained PCs indicates an overall decrease of about 0.05 N∙m/kg in the abduction moment postexercise (Figure 7). For the HT moment, PC1 indicated a decrease of about 0.05 N∙m/kg in peak hip internal rotation moment (P = .001) after the run (Table 3, Figure 8). There was a large effect of the training run on the change in PC scores for the AT and HT joint moments (Table 3).
Figure 4 — Transverse plane angle for the hip over the duration of the stance phase in the PRE (black) and the POST (gray) conditions. The two middle panels correspond with the first and second retained principal components (PCs) for the hip transverse plane angle waveform. The waveform for the PRE (black) and POST (gray) conditions are plotted with plus (+) and minus (–) 1 SD of the scores for PC1 and PC2. The fourth panel represents the variance explained by PC1 (black) and PC2 (dashed gray) over the duration of stance.
Figure 5 — Sagittal plane moment for the ankle over the duration of the stance phase in the PRE (thick black) and the POST (thick gray) conditions with the waveform for the combined contributions of all retained principal components PRE (thin black) and POST (thin gray).
JAB Vol. 31, No. 4, 2015
Downloaded by W VA Univ Library on 09/16/16, Volume 31, Article Number 4
Effect of Exertion During Running 255
Figure 6 — Transverse plane moment for the ankle over the duration of the stance phase in the PRE (black) and the POST (gray) conditions. The middle panel corresponds with the third retained principal component (PC) for the ankle transverse plane moment waveform. The waveform for the PRE (black) and POST (gray) conditions are plotted with plus (+) and minus (–) 1 SD of the scores for PC3. The third panel represents the variance explained by PC3 over the duration of stance.
Figure 8 — Transverse plane moment for the hip over the duration of the stance phase in the PRE (black) and the POST (gray) conditions. The middle panel corresponds with the first retained principal component (PC) for the hip transverse plane moment waveform. The waveform for the PRE (black) and POST (gray) conditions are plotted with plus (+) and minus (–) 1 SD of the scores for PC1. The third panel represents the variance explained by PC1 over the duration of stance.
Figure 7 — Frontal plane moment for the knee over the duration of the stance phase in the PRE (thick black) and the POST (thick gray) conditions with the waveform for the combined contributions of all retained principal components PRE (thin black) and POST (thin gray).
to differences in exercise protocols and limitations in data analysis methods that require the a priori identification of dependent variables. For example, as a result of a 3200-m maximal effort run, Derrick et al reported significant kinematic changes of the heel contact and maximum angles of the knee and rearfoot.23 By only reporting a table with these discrete values and not providing a figure showing the entire waveform for each joint angle, it is not clear how else the kinematics change for the duration of stance phase. It is possible that the timing of the peak waveform value or the rate of change from heel contact to the peak could also be altered as a result of an exhaustive run. A figure of the waveform, as is provided by Dierks et al, gives a visual representation of these changes, but the changes cannot be quantified with just the reported values of discrete variables.11 Without having to choose a priori a limited set of variables to investigate, PCA provides a way to quantify changes in full waveforms across different conditions. An example of how the PCA approach may result in the detection of changes in waveforms that might go unnoticed with the traditional discrete variable analysis is evident in the AT moment waveform. When observing the mean curves before and after the run in the first panel of Figure 6, it appears the main effect is a difference in the peak external rotation moment. Yet, the only PC that revealed a significant difference between the pre- and postrun conditions was PC3. The effect described by this principle component is shown with the curves representing plus and minus 1
Discussion The purpose of this study was to use PCA to examine the effects of exertion from a typical training run on lower extremity kinematics and kinetics. Since injured runners tend to experience an increase in pain as the time of exercise progresses,10 understanding the changes in joint mechanics during a run in an exerted state can be used to elucidate how these injuries may occur. The effects of exertion on joint mechanics during running have been poorly understood due
JAB Vol. 31, No. 4, 2015
Downloaded by W VA Univ Library on 09/16/16, Volume 31, Article Number 4
256 Benson and O’Connor
SD of the scores for PC3 in the second panel of Figure 6. Adding waveforms that represent the mean PC3 scores for each condition show that the postrun condition had a lower PC3 score. The third panel indicates that the variance explained by the change in PC score occurred around 20% and from 80% to 100% of stance phase. Despite the seemingly large difference in peak external rotation moment at midstance, the change that came as a result of the run was an increase in the rate of external rotation moment at the beginning of stance phase, as well as greater external rotation moment at the end of stance. This study used PCA to identify the relevant ecological changes in all joint kinematics and kinetics for runners in an exerted state. In this study, runners experienced a greater rate of eversion after the run. This is consistent with results from other studies that showed changes in ankle mechanics after a run.10,11,23 Ankle frontal plane mechanics are commonly reported due to the connection between excessive rearfoot eversion—a major component of pronation—and running injury.7 Due to coordination between the segments of the lower extremity, pronation may affect proximal aspects of the lower extremity, leading to common running injuries at the knee.6,7 It is suggested that excessive foot pronation is associated with excessive internal rotation of the tibia, which causes the femur to internally rotate during knee extension. This could result in the knee absorbing more transverse plane rotation, and could cause an increase in lateral patellofemoral joint stress.24 Participants in this study were healthy runners with no major running injuries in the past 6 months. Therefore, running mechanics that could signify a risk for overuse injury were not expected. However, changes in transverse plane mechanics at the knee and hip corresponded with the observed changes in rearfoot kinematics that are often linked to running injury.6,7 A post hoc analysis revealed a moderate correlation between the PC scores representing the increased rate of eversion (AF angle PC2) and the increased rate of ankle external rotation moment (AT moment PC3) after the run (Pearson correlation r = .707, two-tailed P = .002). Since the foot is fixed on the ground during stance phase, an external rotation moment at the ankle indicates a moment acting to cause internal rotation of the tibia from the foot. This increase in tibial internal rotation can also be observed with the increase in knee internal rotation angle, a result also reported by Dierks et al.10,11 Therefore, these results are consistent with the theory that eversion at the foot is related to the internal rotation of the tibia.24 The theory of the effect of the subtalar joint pronation on the patellofemoral joint outlined by Tiberio24 suggests that in cases of excessive pronation, the tibia may not be externally rotated enough for proper knee extension. As a compensatory measure, the hip may internally rotate, which might cause compression on the lateral aspect of the patella, a condition often linked to running injury.24 While rearfoot eversion and tibial internal rotation were not considered excessive in this study, there was greater hip internal rotation throughout stance phase and a decrease in hip internal rotation moment after the run. In addition, there were greater changes in hip transverse plane rotation in mid and late stance. This effect can also be observed in the figure provided by Dierks et al, but was not quantified due to the discrete analysis approach.11 By using PCA, the current experiment was able to show how running in an exerted state influenced control of transverse plane rotation at the hip. While the changes in this study were reported for healthy runners, athletes with abnormal gait patterns might experience pathological running mechanics if they make similar adjustments while running in an exerted state. Running in an even greater state of exertion could amplify the effects observed in this study.
In addition to the coordinated joint mechanics described by Tiberio,24 increased frontal plane knee loading has also been cited as a mechanism for anterior knee pain.25 The current analysis did not identify specific features of the knee frontal plane angle and moment waveforms that changed from before to after the run. However, the overall effect was increased knee adduction and a decrease in knee abduction moment after the run. Taken together, these results appear contradictory, but the effect size for the knee adduction angle is much less than the effect size for the change in knee abduction moment (Table 2). Therefore, the vertical ground reaction force, though not significantly different before and after the run overall, may have been reduced just enough to affect the overall knee frontal plane moment. A similar result occurred in the ankle sagittal plane, where there was an overall decrease in plantar flexion moment. In relation to injury risk, the observed decrease in knee abduction moment and increase in knee adduction angle may actually suggest a protective adaptation undertaken by the subjects in this study, at least as it relates to knee pain. The exercise protocol, slightly modified from the procedures used by Dierks et al10,11 and Bazett-Jones et al,12 was designed to mimic the participants’ typical training run. Participants ran for about the same time and pace as their self-reported information, which suggests that the results indicate what occurs ecologically during running. After the run, participants changed shoes and tracking markers needed to be reapplied. Due to some markers falling off during the postrun data collection, this took longer for some participants than others. While great care was taken to minimize this data collection time window, participants may have recovered from their run as they were not continuously running in an exerted state during this time before the final data collection. Therefore, the differences observed from before and after the run may be modest effects, particularly during exercise at a level of exertion beyond that of a runner’s typical training run. In conclusion, the experimental protocol was designed so that participants ran to a level of exertion similar to that of their typical training run, therefore these results reflect ecological changes in running gait. Increased rate of eversion influences other joint mechanics throughout the kinetic chain, such as knee and hip transverse plane angles and moments, which may be a risky position for runners. These changes may be more pronounced for runners with abnormal gait patterns or at increased levels of exertion. Acknowledgments This work was supported by the University of Wisconsin–Milwaukee, College of Health Sciences Student Research Grant. The authors thank Stephen Cobb, PhD, ATC, CSCS, and Jennifer Earl, PhD, LAT, for their assistance with this project.
References 1. Kaufman KR, Brodine SK, Shaffer RA, Johnson CW, Cullison TR. The effect of foot structure and range of motion on musculoskeletal overuse injuries. Am J Sports Med. 1999;27(5):585–593. PubMed 2. Walter SD, Hart LE, McIntosh JM, Sutton JR. The Ontario cohort study of running-related injuries. Arch Intern Med. 1989;149(11):2561– 2564. PubMed doi:10.1001/archinte.1989.00390110113025 3. Taunton JE, Ryan MB, Clement DB, McKenzie DC, Lloyd-Smith DR, Zumbo BD. A retrospective case-control injuries analysis of 2002 running. Br J Sports Med. 2002;36(2):95–101. PubMed doi:10.1136/ bjsm.36.2.95
JAB Vol. 31, No. 4, 2015
Downloaded by W VA Univ Library on 09/16/16, Volume 31, Article Number 4
Effect of Exertion During Running 257 4. van Gent RN, Siem D, van Middelkoop M, et al. Incidence and determinants of lower extremity running injuries in long distance runners: A systematic review. Br J Sports Med. 2007;41(8):469–480. PubMed doi:10.1136/bjsm.2006.033548 5. Taunton JE, Ryan MB, Clement DB, McKenzie DC, Lloyd-Smith DR, Zumbo BD. A prospective study of running injuries: The vancouver sun run “in training” clinics. Br J Sports Med. 2003;37(3):239–244. PubMed doi:10.1136/bjsm.37.3.239 6. James SL, Jones DC. Biomechanical aspects of distance running injuries. In: Cavanagh PR, ed. Biomechanics of distance running. Champaign, IL: Human Kinetics; 1990:249–269. 7. Hreljac A. Impact and overuse injuries in runners. Med Sci Sports Exerc. 2004;36(5):845–849. PubMed doi:10.1249/01. MSS.0000126803.66636.DD 8. Kannus P. Etiology and pathophysiology of chronic tendon disorders in sports. Scand J Med Sci Sports. 1997;7(2):107–112. PubMed doi:10.1111/j.1600-0838.1997.tb00126.x 9. Nigg BM. Biomechanical aspects of running. In: Nigg BM, ed. Biomechanics of running shoes. Champaign, IL: Human Kinetics; 1986:1–25. 10. Dierks TA, Manal KT, Hamill J, Davis IS. Lower extremity kinematics in runners with patellofemoral pain during a prolonged run. Med Sci Sports Exerc. 2011;43(4):693–700. PubMed doi:10.1249/ MSS.0b013e3181f744f5 11. Dierks TA, Davis IS, Hamill J. The effects of running in an exerted state on lower extremity kinematics and joint timing. J Biomech. 2010;43(15):2993–2998. PubMed doi:10.1016/j.jbiomech.2010.07.001 12. Bazett-Jones DM, Cobb SC, Huddleston WE, Connor KMO, Armstrong BSR, Earl-Boehm JE. Effect of patellofemoral pain on strength and mechanics after an exhaustive run. Med Sci Sports Exerc. 2013;45(7):1331–1339. PubMed doi:10.1249/ MSS.0b013e3182880019 13. Daffertshofer A, Lamoth CJC, Meijer OG, Beek PJ. PCA in studying coordination and variability: A tutorial. Clin Biomech (Bristol, Avon). 2004;19(4):415–428. PubMed doi:10.1016/j.clinbiomech.2004.01.005
14. Deluzio KJ, Wyss UP, Costigan PA, Sorbie C, Zee B. Gait assessment in unicompartmental knee arthroplasty patients: Principal component modelling of gait waveforms and clinical status. Hum Mov Sci. 1999;18(5):701–711. doi:10.1016/S0167-9457(99)00030-5 15. Heiderscheit BC, Hamill J, van Emmerik REA. Variability of stride characteristics and joint coordination among individuals with unilateral patellofemoral pain. J Appl Biomech. 2002;18(2):110–121. 16. American College of Sports Medicine. ACSM’s guidelines for exercise testing and prescription, 8th ed. Philadelphia, PA: Lippincott Williams & Wilkins; 2010. 17. Borg G. Borg’s perceived exertion and pain scales. Champaign, IL: Human Kinetics; 1998. 18. Grood ES, Suntay WJ. A joint coordinate system for the clinical description of three-dimensional motions: Application to the knee. J Biomech Eng. 1983;105(2):136–144. PubMed doi:10.1115/1.3138397 19. Bresler B, Frankel JP. The forces and moments in the leg during level walking. Transactions of the ASME. 1950;72:27–36. 20. Jackson JE. A user’s guide to principal components. Hoboken, NJ: John Wiley & Sons, Inc.; 2003. 21. O’Connor KM, Bottum MC. Differences in cutting knee mechanics based on principal components analysis. Med Sci Sports Exerc. 2009;41(4):867–878. PubMed doi:10.1249/MSS.0b013e31818f8743 22. Wrigley AT, Albert WJ, Deluzio KJ, Stevenson JM. Differentiating lifting technique between those who develop low back pain and those who do not. Clin Biomech (Bristol, Avon). 2005;20(3):254–263. PubMed doi:10.1016/j.clinbiomech.2004.11.008 23. Derrick TR, Dereu D, McLean SP. Impacts and kinematic adjustments during an exhaustive run. Med Sci Sports Exerc. 2002;34(6):998–1002. PubMed doi:10.1097/00005768-200206000-00015 24. Tiberio D. The effect of excessive subtalar JointPronation on patellofemoral mechanics:A theoretical model. J Orthop Sports Phys Ther. 1987;9(4):160–165. PubMed doi:10.2519/jospt.1987.9.4.160 25. Stefanyshyn DJ, Stergiou P, Lun VM, Meeuwisse WH, Worobets JT. Knee angular impulse as a predictor of patellofemoral pain in runners. Am J Sports Med. 2006;34(11):1844–1851. PubMed doi:10.1177/0363546506288753
JAB Vol. 31, No. 4, 2015
ORIGINAL RESEARCH
The Effect of Exertion on Joint Kinematics and Kinetics During Running Using a Waveform Analysis Approach Lauren C. Benson and Kristian M. O’Connor University of Wisconsin–Milwaukee About half of all runners sustain a running-related injury every year. Exertion may contribute to risk of injury by altering joint mechanics. The purpose of this study was to examine the effects of exertion on runners’ joint mechanics using principal component analysis (PCA). Three-dimensional motion analysis of the lower extremity was performed on 16 healthy female runners before and after their typical training run. PCA was used to determine exertion-related changes in joint mechanics at the ankle, knee, and hip. Statistical significance for repeated-measures MANOVA of the retained principal components at each joint and plane of motion was at P < .05. Exercise effects were identified at the ankle (greater rate of eversion [PC2: P = .027], and decreased plantar flexion moment [overall: P = .044] and external rotation moment [PC3: P = .003]), knee (increased adduction [overall: P = .044] and internal rotation [PC3: P = .034], and decreased abduction moment [overall: P = .045]), and hip (increased internal rotation [PC1: P = .013] and range of mid- to late-stance rotation [PC2: P = .009], and decreased internal rotation moment [PC1: P = .001]). The observed changes in running mechanics reflect a gait profile that is often linked to running injury. The effects of more strenuous activity may result in mechanics that present an even greater risk for injury. Keywords: principal components analysis, exertion, injury, lower extremity Running is a common mode of exercise, which is important to maintaining good health.1 However, about half of all runners will sustain a running-related injury in a given year,1–3 with most of those injuries occurring at the knee.3–6 Despite research conducted in this field, the injury rates have not dramatically changed.2,3 Understanding the cause of running injuries is necessary for developing methods for prevention and better treatment options. Exposure to multiple impact forces over the course of a run, or many runs, is suspected to play a role in most overuse running injuries. These impact forces may be especially harmful if combined with improper mechanics.6–9 For example, proximal aspects of lower extremity gait may be affected by excessive or prolonged pronation, which has been suspected to contribute to common running injuries at the knee.6,7 Injured runners often complain of a gradual onset of pain as the run progresses,10 suggesting that running in an exerted state could contribute to a pathway to injury. Studies investigating the effects of exertion on running biomechanics typically use a protocol designed to bring runners to the point of maximum exertion to elicit the greatest changes in biomechanics. However, a more ecological protocol for inducing exertion-related changes in runners may be one that closely mimics a typical bout of exercise for a runner, while also providing an objective measure of exertion.10–12 Traditional investigations in biomechanics have focused on discrete variables to describe changes in running mechanics, including peak forces, peak angles, and excursions. However, relying on discrete variables of interest requires an a priori decision about which dependent variables and events in a stride cycle may change as a result of exertion. By examining the full time series, Lauren C. Benson and Kristian M. O’Connor are with the Department of Kinesiology, University of Wisconsin–Milwaukee, Milwaukee, WI. Address author correspondence to Lauren C. Benson at [email protected]. 250
or waveform, principal components analysis (PCA) is an unbiased way to determine relevant differences in joint kinematics.13,14 Previous studies that have examined the effect of exertion on running kinematics have been limited by the design of the fatigue protocol and the a priori identification of discrete dependent variables. The purpose of this study was to examine the influence of running in an exerted state on lower extremity joint kinematics and kinetics using waveform analysis. It was hypothesized that running in an exerted state would result in joint mechanics that could present a risk for running injury. Utilizing a running protocol that mimics a person’s typical training experience will allow for quantification of the ecological effects of running in an exerted state on joint mechanics.
Methods The study protocol was approved by the university institutional review board, and all participants provided informed consent before participating in the study. Sample size estimations were based on a repeated-measures MANOVA design with a medium effect size (ηp2 = .2), 80% power, and α = .05.15 Sixteen recreational runners were recruited through fliers posted on the university campus and with local running organizations. All participants were female runners, 18–45 years of age, who ran a minimum of 15 miles per week for the 6 months before the study. Exclusion criteria included self-reported cardiac risk, any lower extremity pain or runningrelated injury that limited training within the 6 months before the study, any history of major surgery to the lower extremity, the use of orthotics, pregnancy, medical conditions or medications that could impair balance, or a forefoot-strike running pattern.10,11 Additionally, participants were asked to refrain from running in a race in the 48 hours before testing and refrain from all exercise in the 24 hours before testing. Information about the participants’
Effect of Exertion During Running 251
Table 1 Participant information
Downloaded by W VA Univ Library on 09/16/16, Volume 31, Article Number 4
Participant Characteristics
Mean
SD
Height (m)
1.65
0.05
Mass (kg)
58.4
7.0
Age (y)
25
7
Shoe size
8
1
Typical running time (min)
39
10
Typical running distance (miles)
4.5
1.3
Typical running pace (min/mile)
8.85
0.93
Typical weekly mileage (miles)
24
11
typical running habits was collected to determine individualized experiment parameters (Table 1). During one testing session, three-dimensional kinematic data were collected at 200 Hz with a 10-camera Eagle system (Motion Analysis, Inc., Santa Rosa, CA), and ground reaction forces were recorded at 1000 Hz using an AMTI force plate (OR6-5; Advanced Mechanical Technology Inc., Watertown, MA). The participants were fitted with a heart rate monitor (Polar Electro Inc., Woodbury, NY), and the warm-up and run to an exerted state took place on a treadmill (C964i; Precor, Woodinville, WA) with the participants wearing their own training shoes. During the data collection before and after running to an exerted state, all participants ran in the same shoes (NBA-801; New Balance, Brighton, MA) with a mean size of women’s 8 (SD 1) for standardization purposes. This is a shoe without a heel counter, reinforced with tape, to allow for direct observation of rearfoot motion. The participants had a 5-minute warm-up period on the treadmill which consisted of light jogging at 2.2 m/s. Each participant’s pace for the data collection and the treadmill run was self-selected based on their typical pace for a training run.10,11 Retroreflective markers were applied to the participants’ skin to track the motion of the pelvis, thigh, leg, and foot. The tracking markers were placed on the left and right anterior superior iliac spine (ASIS) and posterior superior iliac spine (PSIS), a 4-marker plate was placed on both the thigh and the leg, and a marker triad was attached to the calcaneus. The location of the stand-alone markers on each ASIS and PSIS was marked in ink on the participant’s skin or a piece of tape attached to the participant. A 3-second standing calibration was recorded with additional calibration markers on the left and right iliac crests and greater trochanters, lateral and medial femoral epicondyles, malleoli, and first and fifth metatarsal heads of the right leg. These bony landmarks were chosen as part of the definition of each joint angle, and for their ease of identification. The calibration markers were removed following a 3-second standing calibration. Participants performed as many trials as necessary to record 10 successful running trials at their self-selected pace, ± 5%, in the common laboratory shoes across a 15-m runway containing an embedded force plate. A successful trial was defined as when right leg initial contact and toe-off occurred on the force plate. Kinematic and kinetic data were collected for each trial. Then the participants ran on the treadmill at their self-selected pace in their own training shoes and without the retroreflective markers. To mimic the participants’ typical training run, they were permitted to listen to music via headphones, if they desired. Starting in the first minute of the run and at every 5 minutes during the run, the participants’ heart rate and ratings of perceived exertion (RPE) were recorded.
When the participants reached a state of exertion measured by at least 85% of age-calculated maximum heart rate,16 and a score of at least 17 (very hard) on the RPE scale,10,11,17 they continued running for an additional 2 minutes before ending the run. Their final heart rate and RPE were recorded before they stopped the treadmill. Immediately at the end of the run, participants switched into the laboratory shoes and the tracking markers were reapplied. The participants performed 10 successful running trials overground as kinematic and kinetic data were collected. After recording the running trials, the calibration markers were reapplied and a 3-second standing calibration was recorded for the postrun markers. The methods for applying the tracking and calibration markers were the same as above. The raw coordinate data were filtered using a fourth-order, zero-lag, recursive Butterworth filter with a cutoff at 12 Hz. During the 3-second standing calibration, the pelvis, thigh, leg, and foot coordinate systems were defined as being coincident with one another in that position. The x-axis pointed to the right, the y-axis pointed anteriorly, and the z-axis pointed superiorly. Calculation of hip (H), knee (K), and ankle (A) joint angles in each plane of motion (sagittal [S], frontal [F], and transverse [T]) was done using a joint coordinate system approach.18 Net internal joint moments were calculated using an inverse dynamics analysis and were normalized to body mass.19 All joint angles and moments were time normalized in 1% increments to 101 data points representing 0% to 100% of stance phase, determined from the kinetic data as the period where the ground reaction force in the vertical direction was greater than 20 N. The data processing was done using Visual3D software (v4.75.34; C-Motion, Inc., Rockville, MD). PCA was used to assess changes in joint angles and joint moments before and after the run.20–22 Matrices for each waveform were created. The individual trials populated n rows, and the 101 data points populated p columns in an Xnxp matrix. Eigenvector analysis of the covariance matrix S101 × 101 determined the eigenvector matrix, U101 × 101, by orthonormalizing S101 × 101. The eigenvectors were the coefficients for the principal components (PCs) which represented the original data in new coordinates. The coefficients were the direction cosines that related the new axes to the old axes and were considered one mode of variation describing the variability within the entire original data set. The eigenvalues, L1 × 101, were determined by U’SU = L1 × 101. The eigenvalues represented the relative contribution or the rank of each PC to the total variation. A principal component score, Z nxp, was calculated for each individual waveform by multiplying the individual trial’s variation from the mean of all the trials, x̅ 1 × 101, by the transpose of the eigenvector matrix (Equation 1).
Z n×101 = (X n×101 – (1n×1 × x–1×101 )) × U '101×101 (1)
The PC scores represented the distance from each waveform to the mode of variability described by each principal component. The Z nx101 matrix was reduced to only the PC scores that represented the primary modes of variation. A parallel analysis was performed that retained only those PCs that contributed modes of variation greater than an equivalently-sized input matrix of normally-distributed randomly-generated numbers. The variance not explained by the retained PCs represented random error. The 10-stride means of the retained PC scores for each participant’s waveform before and after the run represented the dependent variables. Custom software was used to perform this PCA (Matlab v8.0.0.783; Mathworks, Inc., Natick, MA).
JAB Vol. 31, No. 4, 2015
252 Benson and O’Connor
Each of the 9 joint angles and moments (ankle, knee, and hip in the sagittal, frontal, and transverse planes) and 3 ground reaction forces (X, Y, Z) had several dependent variables, depending on the number of retained PCs for that waveform. A repeated-measures MANOVA for each waveform was done on the dependent variables. This led to a total of 9 joint angle MANOVAs, 9 joint moment MANOVAs, and 3 ground reaction force MANOVAs. Pre- and postexertion was the independent variable. For each waveform that exhibited a significant pre–post effect with the MANOVA, comparisons for the individual retained PCs was done with dependent t tests. Effect sizes for the MANOVAs and the follow-up t tests are reported using partial eta-squared. All statistical analyses were performed in SPSS with alpha set to .05 (v19.0.0.1; IBM, Inc., Chicago, IL).
to end the run when only one of the criteria was met: 2 met just the heart rate criterion and 5 met just the RPE criterion. For the other 9 participants, both criteria were met before ending the run. There was no obvious effect of this difference in end criteria on the joint kinematics observed before and after the run. At the end of the run, the mean heart rate for all participants was 88% (SD 4) of their age-calculated maximum. The mean final RPE was 18 (SD 1) on the 6–20 scale. The mean time from the end of the run to the end of the postrun data collection (including reapplication of tracking, but not standing, markers) was 10.89 (SD 3.94) minutes. The percent of the variance explained by the retained PCs for the joint angle data ranged from 95% to 98%, and the variance explained by the first PC, which described the most variation in each waveform, ranged from 70% to 92%. A repeated-measures MANOVA of the retained PCs for each joint angle indicated a significant difference in kinematics before and after the run for AF, K F, KT, and HT, with a large exercise effect for AF, KT, and HT, and only a moderate effect for K F (Table 2). The percent of the variance explained by the retained PCs for the joint moment data ranged from 84% to 97%, and the variance explained by the first PC ranged from 47% to 86%. A repeated-measures MANOVA of the
Downloaded by W VA Univ Library on 09/16/16, Volume 31, Article Number 4
Results Participants ran at a mean speed of 3.0 (SD 0.3) m/s for a mean time of 39 (SD 19) minutes until they reached at least 1 of the 2 stopping criteria (heart rate greater than 85% of age-calculated maximum, RPE greater than 17). In 7 cases, the participant asked
Table 2 Percent variance explained for the retained principle components (PCs) and the results of the MANOVAs performed for each waveform Principle Components PC1
PC2
PC3
PC4
AS
78
9
6
4
AF
84
7
4
AT
70
14
6
KS
81
13
4
KF
92
4
PC5
MANOVA
PC6
PC7
PC8
Total
P
ηp2
96
.075
.483
95
.005*
.614
96
.800
.172
98
.070
.408
96
.044*
.361
98
.030*
.634
96
.104
.367
Angle
KT
81
8
4
HS
78
13
6
HF
80
9
7
HT
70
13
9
AS
76
16
5
AF
79
9
6
AT
86
6
3
KS
68
17
KF
73
KT
75
5
3
3
2
96
.747
.087
96
.007*
.663
97
.044*
.452
2
96
.526
.219
95
.009*
.579
5
3
94
.341
.294
7
4
3
87
.045*
.530
12
6
92
.334
.223
93
.970
.070
84
.627
.441
85
.002*
.789
4
Moment
HS
73
9
5
3
3
HF
47
12
6
5
4
HT
62
10
7
4
2
4
4
3
Ground reaction force X
59
14
8
6
4
92
.164
.471
Y
45
27
12
5
3
93
.144
.486
Z
58
25
9
4
96
.165
.395
Note. Within each MANOVA were the mean scores for each retained PC. Effect size is reported as partial eta-squared. *Significantly different between pre and post, P < .05. JAB Vol. 31, No. 4, 2015
Effect of Exertion During Running 253
retained PCs for each joint moment indicated large exercise effects for AS, AT, K F, and HT, as well as significant differences in kinetics (Table 2). The percent of the variance explained by the retained PCs for the ground reaction force data ranged from 92% to 96%, and the variance explained by the first PC ranged from 45% to 59%. A repeated-measures MANOVA of the retained PCs for the ground reaction force in each direction indicated a medium exercise effect but no significant changes in the individual ground reaction forces (Table 2). For each of the above significant MANOVA findings, follow-up t tests were performed (Table 3).
For the AF angle, PC2 identified a greater rate of eversion (P = .027) postexercise (Table 3, Figure 1). For the K F angle, while the MANOVA was significant, there were no significant differences for any of the retained PCs when the subsequent t tests were performed (Tables 2 and 3). The combined effect of the retained PCs suggests an overall increase of about 2° in knee adduction postexercise (Figure 2). For the KT angle, PC3 detected an increase of about 1° in peak internal rotation (P = .034) postexercise (Table 3, Figure 3).
Downloaded by W VA Univ Library on 09/16/16, Volume 31, Article Number 4
Table 3 Results of the post hoc analysis of the retained principle components (PCs) for the waveforms that had significant pre and post differences in the MANOVA PC Scores PC
Pre
Post
P
ηp2
1
–20.70 (57.87)
19.65 (60.22)
.077
.194
2
–4.61 (16.50)
4.46 (13.03)
.027*
.287
3
–1.18 (6.28)
1.41 (13.06)
.448
.039
1
–6.78 (46.40)
7.57 (36.48)
.151
.133
Angle AF
KF KT
HT
2
–1.73 (6.56)
1.61 (9.27)
.094
.176
1
–2.53 (41.81)
2.29 (39.97)
.533
.026
2
1.24 (13.47)
–1.70 (9.57)
.169
.122
3
2.44 (8.61)
–2.22 (7.86)
.034*
.267
4
–1.68 (7.29)
1.56 (7.56)
.112
.159
5
0.97 (5.70)
–1.38 (6.40)
.057
.221
1
–12.59 (40.97)
13.80 (30.07)
.013*
.345
2
–3.59 (16.53)
3.13 (15.49)
.009*
.373
3
0.45 (14.50)
–0.02 (8.07)
.873
.002
4
–0.90 (6.01)
0.96 (5.93)
.371
.054
1
–0.43 (2.77)
0.31 (2.64)
.054
.226
2
0.13 (1.24)
–0.14 (1.17)
.137
.141
3
–0.002 (0.54)
0.01 (0.68)
.928
.001
1
0.53 (1.66)
–0.47 (1.48)
.076
.195
2
0.02 (0.22)
–0.02 (0.50)
.728
.008
3
0.08 (0.28)
–0.08 (0.22)
.003*
.458
1
–0.14 (2.09)
0.20 (2.48)
.213
.102
2
–0.06 (0.70)
0.05 (0.70)
.340
.061
3
0.02 (0.54)
–0.03 (0.41)
.481
.034
4
–0.03 (0.38)
0.06 (0.45)
.338
.061
1
0.20 (0.70)
–0.20 (0.74)
.001*
.550
2
–0.01 (0.32)
0.01 (0.21)
.701
.010
3
–0.03 (0.25)
0.04 (0.22)
.085
.185
4
0.01 (0.18)
–0.001 (0.18)
.664
.013
5
0.02 (0.09)
–0.01 (0.09)
.197
.108
Moment AS
AT
KF
HT
Note. Effect size is reported as partial eta-squared. *Significantly different between pre and post, P < .05.
Figure 1 — Frontal plane angle for the ankle over the duration of the stance phase in the PRE (black) and the POST (gray) conditions. The middle panel corresponds with the second retained principal component (PC) for the ankle frontal plane angle waveform. The waveform for the PRE (black) and POST (gray) conditions are plotted with plus (+) and minus (–) 1 SD of the scores for PC2. The third panel represents the variance explained by PC2 over the duration of stance.
Figure 2 — Frontal plane moment for the knee over the duration of the stance phase in the PRE (thick black) and the POST (thick gray) conditions with the waveform for the combined contributions of all retained principal components PRE (thin black) and POST (thin gray).
JAB Vol. 31, No. 4, 2015
Downloaded by W VA Univ Library on 09/16/16, Volume 31, Article Number 4
254 Benson and O’Connor
Figure 3 — Transverse plane angle for the knee over the duration of the stance phase in the PRE (black) and the POST (gray) conditions. The middle panel corresponds with the third retained principal component (PC) for the knee transverse plane angle waveform. The waveform for the PRE (black) and POST (gray) conditions are plotted with plus (+) and minus (–) 1 SD of the scores for PC3. The third panel represents the variance explained by PC3 over the duration of stance.
For the HT angle, PC1 identified an increase of about 2° in internal rotation throughout stance phase (P = .013) postexercise, and PC2 showed a greater range of rotation in mid- and late-stance (Table 3, Figure 4). There was a moderate effect of the training run on the change in PC scores for the AF, KT, and HT joint angles (Table 3). For the AS moment, while the MANOVA was significant, there were no significant differences for any of the retained PCs when the subsequent t tests were performed (Tables 2 and 3). The combined effect of the retained PCs suggests an overall decrease of less than 1 N∙m/kg in the plantar flexion moment postexercise (Figure 5). For the AT moment, PC3 indicated a greater rate of external rotation moment at the beginning of stance phase postexercise (P = .003) (Table 3, Figure 6). For the K F moment, while the omnibus test was significant, there were no significant differences as a result of the run for any of the retained PCs (Tables 2 and 3). The cumulative effects of these differences in the PCs likely led to an overall significant difference even though no individual PC was different. The reconstructed moment from the retained PCs indicates an overall decrease of about 0.05 N∙m/kg in the abduction moment postexercise (Figure 7). For the HT moment, PC1 indicated a decrease of about 0.05 N∙m/kg in peak hip internal rotation moment (P = .001) after the run (Table 3, Figure 8). There was a large effect of the training run on the change in PC scores for the AT and HT joint moments (Table 3).
Figure 4 — Transverse plane angle for the hip over the duration of the stance phase in the PRE (black) and the POST (gray) conditions. The two middle panels correspond with the first and second retained principal components (PCs) for the hip transverse plane angle waveform. The waveform for the PRE (black) and POST (gray) conditions are plotted with plus (+) and minus (–) 1 SD of the scores for PC1 and PC2. The fourth panel represents the variance explained by PC1 (black) and PC2 (dashed gray) over the duration of stance.
Figure 5 — Sagittal plane moment for the ankle over the duration of the stance phase in the PRE (thick black) and the POST (thick gray) conditions with the waveform for the combined contributions of all retained principal components PRE (thin black) and POST (thin gray).
JAB Vol. 31, No. 4, 2015
Downloaded by W VA Univ Library on 09/16/16, Volume 31, Article Number 4
Effect of Exertion During Running 255
Figure 6 — Transverse plane moment for the ankle over the duration of the stance phase in the PRE (black) and the POST (gray) conditions. The middle panel corresponds with the third retained principal component (PC) for the ankle transverse plane moment waveform. The waveform for the PRE (black) and POST (gray) conditions are plotted with plus (+) and minus (–) 1 SD of the scores for PC3. The third panel represents the variance explained by PC3 over the duration of stance.
Figure 8 — Transverse plane moment for the hip over the duration of the stance phase in the PRE (black) and the POST (gray) conditions. The middle panel corresponds with the first retained principal component (PC) for the hip transverse plane moment waveform. The waveform for the PRE (black) and POST (gray) conditions are plotted with plus (+) and minus (–) 1 SD of the scores for PC1. The third panel represents the variance explained by PC1 over the duration of stance.
Figure 7 — Frontal plane moment for the knee over the duration of the stance phase in the PRE (thick black) and the POST (thick gray) conditions with the waveform for the combined contributions of all retained principal components PRE (thin black) and POST (thin gray).
to differences in exercise protocols and limitations in data analysis methods that require the a priori identification of dependent variables. For example, as a result of a 3200-m maximal effort run, Derrick et al reported significant kinematic changes of the heel contact and maximum angles of the knee and rearfoot.23 By only reporting a table with these discrete values and not providing a figure showing the entire waveform for each joint angle, it is not clear how else the kinematics change for the duration of stance phase. It is possible that the timing of the peak waveform value or the rate of change from heel contact to the peak could also be altered as a result of an exhaustive run. A figure of the waveform, as is provided by Dierks et al, gives a visual representation of these changes, but the changes cannot be quantified with just the reported values of discrete variables.11 Without having to choose a priori a limited set of variables to investigate, PCA provides a way to quantify changes in full waveforms across different conditions. An example of how the PCA approach may result in the detection of changes in waveforms that might go unnoticed with the traditional discrete variable analysis is evident in the AT moment waveform. When observing the mean curves before and after the run in the first panel of Figure 6, it appears the main effect is a difference in the peak external rotation moment. Yet, the only PC that revealed a significant difference between the pre- and postrun conditions was PC3. The effect described by this principle component is shown with the curves representing plus and minus 1
Discussion The purpose of this study was to use PCA to examine the effects of exertion from a typical training run on lower extremity kinematics and kinetics. Since injured runners tend to experience an increase in pain as the time of exercise progresses,10 understanding the changes in joint mechanics during a run in an exerted state can be used to elucidate how these injuries may occur. The effects of exertion on joint mechanics during running have been poorly understood due
JAB Vol. 31, No. 4, 2015
Downloaded by W VA Univ Library on 09/16/16, Volume 31, Article Number 4
256 Benson and O’Connor
SD of the scores for PC3 in the second panel of Figure 6. Adding waveforms that represent the mean PC3 scores for each condition show that the postrun condition had a lower PC3 score. The third panel indicates that the variance explained by the change in PC score occurred around 20% and from 80% to 100% of stance phase. Despite the seemingly large difference in peak external rotation moment at midstance, the change that came as a result of the run was an increase in the rate of external rotation moment at the beginning of stance phase, as well as greater external rotation moment at the end of stance. This study used PCA to identify the relevant ecological changes in all joint kinematics and kinetics for runners in an exerted state. In this study, runners experienced a greater rate of eversion after the run. This is consistent with results from other studies that showed changes in ankle mechanics after a run.10,11,23 Ankle frontal plane mechanics are commonly reported due to the connection between excessive rearfoot eversion—a major component of pronation—and running injury.7 Due to coordination between the segments of the lower extremity, pronation may affect proximal aspects of the lower extremity, leading to common running injuries at the knee.6,7 It is suggested that excessive foot pronation is associated with excessive internal rotation of the tibia, which causes the femur to internally rotate during knee extension. This could result in the knee absorbing more transverse plane rotation, and could cause an increase in lateral patellofemoral joint stress.24 Participants in this study were healthy runners with no major running injuries in the past 6 months. Therefore, running mechanics that could signify a risk for overuse injury were not expected. However, changes in transverse plane mechanics at the knee and hip corresponded with the observed changes in rearfoot kinematics that are often linked to running injury.6,7 A post hoc analysis revealed a moderate correlation between the PC scores representing the increased rate of eversion (AF angle PC2) and the increased rate of ankle external rotation moment (AT moment PC3) after the run (Pearson correlation r = .707, two-tailed P = .002). Since the foot is fixed on the ground during stance phase, an external rotation moment at the ankle indicates a moment acting to cause internal rotation of the tibia from the foot. This increase in tibial internal rotation can also be observed with the increase in knee internal rotation angle, a result also reported by Dierks et al.10,11 Therefore, these results are consistent with the theory that eversion at the foot is related to the internal rotation of the tibia.24 The theory of the effect of the subtalar joint pronation on the patellofemoral joint outlined by Tiberio24 suggests that in cases of excessive pronation, the tibia may not be externally rotated enough for proper knee extension. As a compensatory measure, the hip may internally rotate, which might cause compression on the lateral aspect of the patella, a condition often linked to running injury.24 While rearfoot eversion and tibial internal rotation were not considered excessive in this study, there was greater hip internal rotation throughout stance phase and a decrease in hip internal rotation moment after the run. In addition, there were greater changes in hip transverse plane rotation in mid and late stance. This effect can also be observed in the figure provided by Dierks et al, but was not quantified due to the discrete analysis approach.11 By using PCA, the current experiment was able to show how running in an exerted state influenced control of transverse plane rotation at the hip. While the changes in this study were reported for healthy runners, athletes with abnormal gait patterns might experience pathological running mechanics if they make similar adjustments while running in an exerted state. Running in an even greater state of exertion could amplify the effects observed in this study.
In addition to the coordinated joint mechanics described by Tiberio,24 increased frontal plane knee loading has also been cited as a mechanism for anterior knee pain.25 The current analysis did not identify specific features of the knee frontal plane angle and moment waveforms that changed from before to after the run. However, the overall effect was increased knee adduction and a decrease in knee abduction moment after the run. Taken together, these results appear contradictory, but the effect size for the knee adduction angle is much less than the effect size for the change in knee abduction moment (Table 2). Therefore, the vertical ground reaction force, though not significantly different before and after the run overall, may have been reduced just enough to affect the overall knee frontal plane moment. A similar result occurred in the ankle sagittal plane, where there was an overall decrease in plantar flexion moment. In relation to injury risk, the observed decrease in knee abduction moment and increase in knee adduction angle may actually suggest a protective adaptation undertaken by the subjects in this study, at least as it relates to knee pain. The exercise protocol, slightly modified from the procedures used by Dierks et al10,11 and Bazett-Jones et al,12 was designed to mimic the participants’ typical training run. Participants ran for about the same time and pace as their self-reported information, which suggests that the results indicate what occurs ecologically during running. After the run, participants changed shoes and tracking markers needed to be reapplied. Due to some markers falling off during the postrun data collection, this took longer for some participants than others. While great care was taken to minimize this data collection time window, participants may have recovered from their run as they were not continuously running in an exerted state during this time before the final data collection. Therefore, the differences observed from before and after the run may be modest effects, particularly during exercise at a level of exertion beyond that of a runner’s typical training run. In conclusion, the experimental protocol was designed so that participants ran to a level of exertion similar to that of their typical training run, therefore these results reflect ecological changes in running gait. Increased rate of eversion influences other joint mechanics throughout the kinetic chain, such as knee and hip transverse plane angles and moments, which may be a risky position for runners. These changes may be more pronounced for runners with abnormal gait patterns or at increased levels of exertion. Acknowledgments This work was supported by the University of Wisconsin–Milwaukee, College of Health Sciences Student Research Grant. The authors thank Stephen Cobb, PhD, ATC, CSCS, and Jennifer Earl, PhD, LAT, for their assistance with this project.
References 1. Kaufman KR, Brodine SK, Shaffer RA, Johnson CW, Cullison TR. The effect of foot structure and range of motion on musculoskeletal overuse injuries. Am J Sports Med. 1999;27(5):585–593. PubMed 2. Walter SD, Hart LE, McIntosh JM, Sutton JR. The Ontario cohort study of running-related injuries. Arch Intern Med. 1989;149(11):2561– 2564. PubMed doi:10.1001/archinte.1989.00390110113025 3. Taunton JE, Ryan MB, Clement DB, McKenzie DC, Lloyd-Smith DR, Zumbo BD. A retrospective case-control injuries analysis of 2002 running. Br J Sports Med. 2002;36(2):95–101. PubMed doi:10.1136/ bjsm.36.2.95
JAB Vol. 31, No. 4, 2015
Downloaded by W VA Univ Library on 09/16/16, Volume 31, Article Number 4
Effect of Exertion During Running 257 4. van Gent RN, Siem D, van Middelkoop M, et al. Incidence and determinants of lower extremity running injuries in long distance runners: A systematic review. Br J Sports Med. 2007;41(8):469–480. PubMed doi:10.1136/bjsm.2006.033548 5. Taunton JE, Ryan MB, Clement DB, McKenzie DC, Lloyd-Smith DR, Zumbo BD. A prospective study of running injuries: The vancouver sun run “in training” clinics. Br J Sports Med. 2003;37(3):239–244. PubMed doi:10.1136/bjsm.37.3.239 6. James SL, Jones DC. Biomechanical aspects of distance running injuries. In: Cavanagh PR, ed. Biomechanics of distance running. Champaign, IL: Human Kinetics; 1990:249–269. 7. Hreljac A. Impact and overuse injuries in runners. Med Sci Sports Exerc. 2004;36(5):845–849. PubMed doi:10.1249/01. MSS.0000126803.66636.DD 8. Kannus P. Etiology and pathophysiology of chronic tendon disorders in sports. Scand J Med Sci Sports. 1997;7(2):107–112. PubMed doi:10.1111/j.1600-0838.1997.tb00126.x 9. Nigg BM. Biomechanical aspects of running. In: Nigg BM, ed. Biomechanics of running shoes. Champaign, IL: Human Kinetics; 1986:1–25. 10. Dierks TA, Manal KT, Hamill J, Davis IS. Lower extremity kinematics in runners with patellofemoral pain during a prolonged run. Med Sci Sports Exerc. 2011;43(4):693–700. PubMed doi:10.1249/ MSS.0b013e3181f744f5 11. Dierks TA, Davis IS, Hamill J. The effects of running in an exerted state on lower extremity kinematics and joint timing. J Biomech. 2010;43(15):2993–2998. PubMed doi:10.1016/j.jbiomech.2010.07.001 12. Bazett-Jones DM, Cobb SC, Huddleston WE, Connor KMO, Armstrong BSR, Earl-Boehm JE. Effect of patellofemoral pain on strength and mechanics after an exhaustive run. Med Sci Sports Exerc. 2013;45(7):1331–1339. PubMed doi:10.1249/ MSS.0b013e3182880019 13. Daffertshofer A, Lamoth CJC, Meijer OG, Beek PJ. PCA in studying coordination and variability: A tutorial. Clin Biomech (Bristol, Avon). 2004;19(4):415–428. PubMed doi:10.1016/j.clinbiomech.2004.01.005
14. Deluzio KJ, Wyss UP, Costigan PA, Sorbie C, Zee B. Gait assessment in unicompartmental knee arthroplasty patients: Principal component modelling of gait waveforms and clinical status. Hum Mov Sci. 1999;18(5):701–711. doi:10.1016/S0167-9457(99)00030-5 15. Heiderscheit BC, Hamill J, van Emmerik REA. Variability of stride characteristics and joint coordination among individuals with unilateral patellofemoral pain. J Appl Biomech. 2002;18(2):110–121. 16. American College of Sports Medicine. ACSM’s guidelines for exercise testing and prescription, 8th ed. Philadelphia, PA: Lippincott Williams & Wilkins; 2010. 17. Borg G. Borg’s perceived exertion and pain scales. Champaign, IL: Human Kinetics; 1998. 18. Grood ES, Suntay WJ. A joint coordinate system for the clinical description of three-dimensional motions: Application to the knee. J Biomech Eng. 1983;105(2):136–144. PubMed doi:10.1115/1.3138397 19. Bresler B, Frankel JP. The forces and moments in the leg during level walking. Transactions of the ASME. 1950;72:27–36. 20. Jackson JE. A user’s guide to principal components. Hoboken, NJ: John Wiley & Sons, Inc.; 2003. 21. O’Connor KM, Bottum MC. Differences in cutting knee mechanics based on principal components analysis. Med Sci Sports Exerc. 2009;41(4):867–878. PubMed doi:10.1249/MSS.0b013e31818f8743 22. Wrigley AT, Albert WJ, Deluzio KJ, Stevenson JM. Differentiating lifting technique between those who develop low back pain and those who do not. Clin Biomech (Bristol, Avon). 2005;20(3):254–263. PubMed doi:10.1016/j.clinbiomech.2004.11.008 23. Derrick TR, Dereu D, McLean SP. Impacts and kinematic adjustments during an exhaustive run. Med Sci Sports Exerc. 2002;34(6):998–1002. PubMed doi:10.1097/00005768-200206000-00015 24. Tiberio D. The effect of excessive subtalar JointPronation on patellofemoral mechanics:A theoretical model. J Orthop Sports Phys Ther. 1987;9(4):160–165. PubMed doi:10.2519/jospt.1987.9.4.160 25. Stefanyshyn DJ, Stergiou P, Lun VM, Meeuwisse WH, Worobets JT. Knee angular impulse as a predictor of patellofemoral pain in runners. Am J Sports Med. 2006;34(11):1844–1851. PubMed doi:10.1177/0363546506288753
JAB Vol. 31, No. 4, 2015
