ORIGINAL RESEARCH

IJSPT VALIDITY AND RELIABILITY OF VIDEO-BASED

ANALYSIS OF UPPER TRUNK ROTATION DURING
RUNNING
Carolyn F. Weber, DPT, PhD1
Shane McClinton, DPT, PhD, OCS, FAAOMPT 1

ABSTRACT
Background: Two-dimensional (2D) video analysis is a practical tool for assessing biomechanical factors that may
contribute to running-related injury. Asymmetrical or altered coordination of transverse plane trunk movement has
been associated with low back pain, increased vertical and horizontal ground reaction forces, and altered hip abduc-
tion torque and strength. However, the reliability and validity of 2D transverse plane upper trunk rotation (UTR) has
not been assessed.
Study Design: Validity and reliability study
Purpose: To determine the validity and reliability of 2D video-based, transverse plane UTR measurement during
running.
Methods: Sixteen runners ran at self-selected speed on a treadmill while three-dimensional (3D) and 2D motion cap-
ture occurred synchronously. Two raters measured peak UTR for five consecutive strides on two occasions. Interrater
and intrarater reliability and the minimum detectable change was calculated for right and left peak 2D UTR measure-
ment. Concurrent validity and agreement between 2D and 3D measures were determined by calculating Pearson
Product Correlation Coefficients (r) and Bland-Altman plots, respectively.
Results: Using a single UTR measure per runner, intrarater and interrater reliability (ICC2,1) was excellent (intrarater
ICC2,1 range: 0.989-0.999; interrater ICC2,1 range: 0.990-0.995) and the minimum detectable change was 0.39-1.4
degrees. Measurements in 2D and 3D were significantly correlated for peak UTR (all r ≥ 0.986; all p-values < 0.001)
and showed good agreement in Bland-Altman plots.
Conclusion: Two-dimensional video-based measurement of transverse plane peak UTR is valid and reliable.
Clinical Relevance: UTR measurement may provide clinical insight into gait deviations in the transverse plane that
alter angular momentum and increase risk for running-related injury.
Level of Evidence: 2B
Key Words: 2D video analysis; trunk rotation; running; validity; reliability; movement system

CORRESPONDING AUTHOR
Shane McClinton
1
Doctor of Physical Therapy Program,
Des Moines University
3200 Grand Avenue, Des Moines
1
Des Moines University, Des Moines, IA USA IA USA 50312
The authors have no financial or other conflicts of interest to Phone: 515-271-1448, Fax: 515-271-7033
disclose. E-mail: [email protected]

The International Journal of Sports Physical Therapy | Volume 15, Number 6 | December 2020 | Page 910
DOI: 10.26603/ijspt20200910
INTRODUCTION MATERIALS AND METHODS
In the United States, more than 28 million people
Participants
run on a weekly basis and approximately 56% of
Participants included runners who had a past his-
recreational runners become injured each year.1
tory of, or were currently competing on a collegiate
Individuals do not always self-select optimal gait
track and/or cross-country team and recreational
mechanics and demonstrate movement patterns
runners that ran an average of 16.1 kilometers (10
that increase risk for musculoskeletal overload (e.g.,
miles) per week during the previous year (Table 1).
overstriding, increased genu valgum, contralateral
Participant exclusion criteria included having an
pelvic drop, excessive hip adduction).2 Gait analysis
adhesive allergy, a current injury preventing partici-
has become an important clinical tool for identify-
pant from running, a neurological condition or nerve
ing faulty movement patterns and guiding individu-
injury, or being younger than 18 years of age. Prior
alized rehabilitation programs.2 Three-dimensional
to study participation, written informed consent was
(3D) motion analysis remains the gold standard in
obtained from each participant in accordance with
gait analysis, but presents a logistical and financial
the protocol approved by the Des Moines University
burden that prohibits routine use in patient care.
Institutional Review Board.
This, in combination with the availability of numer-
ous inexpensive two-dimensional (2D) video cap-
Treadmill Running Protocol
ture devices, has made 2D video-based analysis of
Each participant was subjected to simultaneous
gait a practical alternative.
3D motion and 2D video capture while running
Most research on gait analysis and retraining has on a PRECOR C962i treadmill (PRECOR USA,
focused on the lower extremity (LE) due to the fre- ­Woodinville, WA, USA) in the Des Moines Univer-
quent occurrence of LE injuries (i.e., medial tibial sity Human Performance Laboratory (Des Moines,
stress syndrome, patellofemoral pain syndrome, IA, USA). All participants wore tight-fitting shorts
iliotibial band syndrome, stress fractures, plantar and preferred running shoes; female runners wore
fasciitis and achilles tendinopathy).e.g.,3,4,5 To date, sports bras. Reflective markers (14 mm diameter)
studies on 2D video-based analysis have focused on were placed on bilateral acromioclavicular joints
determining the reliability and validity of LE param- to define the upper trunk segment and on the left
eters in the frontal and sagittal planes (i.e., peak hip scapulae to assure right-left orientation as a part of
adduction, peak knee abduction, contralateral pelvic a whole-body model (42 total markers). Reflective
drop, foot abduction).8-10 However, no studies have markers were also placed on all four corners of the
determined the reliability or validity of 2D trans- treadmill to define the running surface orientation.
verse plane measures even though some of the most
prevalent running injuries (e.g., patellofemoral syn-
drome, medial tibial stress syndrome, iliotibial band
stress syndrome) have been associated with abnor- Table 1. Characteristics of study participants reported as
mean (± SD).
mal joint loading and movement in the transverse
plane.9-12 Therefore, quantifying movement in the
transverse plane such as upper trunk rotation (UTR)
may provide valuable information in the assessment
and treatment of injured runners and there is a need
for a valid and reliable UTR measurement that is
amenable to routine clinical use.

The purpose of this study was to, 1) assess intra- and

interrater reliability of 2D video analysis of UTR
and, 2) determine the concurrent validity of 2D and
3D measurements of transverse plane UTR during
treadmill running.

The International Journal of Sports Physical Therapy | Volume 15, Number 6 | December 2020 | Page 911
Each participant self-selected a running speed that rear corners of the treadmill was measured; 2) the
he or she would typically sustain for running 8 km. measured angle was subtracted from 90 degrees to
Participants acclimated to this self-selected speed follow convention in reporting UTR with respect to
for three minutes prior to motion capture. The accli- the frontal plane.
matization period was chosen based on previous
findings that three to five minutes were sufficient 2D data acquisition and analysis
for generalization of treadmill running mechan- Two-dimensional video capture (12-15 seconds)
ics to overground running mechanics13 and that in the transverse plane was completed using Hudl
all study participants were experienced treadmill Technique (Hudl; Lincoln, NE, USA) on an Apple
users. Simultaneous 3D and 2D motion capture com- Ipad Air 2 tablet (Apple; Cupertino, CA, USA). The
menced immediately following the acclimatization tablet was positioned in a tripod mount (ishot Prod-
period. ucts, Inc. Woodridge, IL, USA) placed 2.4 m over the
treadmill (Figure 2). The transverse plane video was
3D data acquisition and analysis captured using the tablet’s front camera (60 fps),
Reflective marker trajectories were captured at 120 which enabled viewing of the capture area to assure
frames per second (fps) using a 10-camera Motion that the camera was square to the treadmill during
Analysis System (Motion Analysis Corporation, Santa overhead set-up. Preliminary trials using capture
Rosa, CA). Marker trajectories were calculated and rates of 60 and 120 fps demonstrated that the mag-
then low-pass filtered using a 6 Hz Butterworth filter nitude of transverse plane UTR measured did not
with Cortex software (v6.0.0.1645; Motional Analy- differ between the two capture rates.
sis Corporation, Santa Rosa, CA, USA). As depicted
in Figure 1a and 1b, respectively, right and left trans- Kinovea software (www.kinovea.org) was used to
verse plane UTR were calculated as follows: 1) the identify peak UTR to the left and right for five con-
angle between the line connecting reflective mark- secutive strides beginning at the runner’s first initial
ers on the right and left acromioclavicular joints and foot contact after the 2D-3D synchronization point,
the line connecting markers on the left front and which was defined by the instantaneous movement
of a reflective calibration wand positioned in view of
3D and 2D video capture devices. In one case, a par-
ticipant’s hair obscured the markers on the AC joints
during one of the five consecutive strides and a sixth

Figure 1. Peak trunk rotation to the right (A) and left (B)
were measured as the angular excursion of a line drawn
between reflective markers positioned on left and right acro- Figure 2. Set up for 2D transverse plane video capture in
mion processes with respect to the left side of the treadmill. the clinical setting including, a) the location of the tablet
Trunk rotation angles were reported with respect to the frontal mount above the treadmill and b) tablet front camera field of
plane. view used to assure proper position for video capture.

The International Journal of Sports Physical Therapy | Volume 15, Number 6 | December 2020 | Page 912
stride was analyzed. Once a video frame containing using a single measurement per runner as well as
peak UTR was identified, it was saved as an image the mean of two, three or five UTR measurements
and imported into ImageJ software (https://imagej. per runner. The single, two, or three measurements
net) to measure UTR. Each image was magnified to used in the calculations were randomly selected by
300% and the angle depicted in Figure 1 was mea- using a random number generator (https://www.
sured. UTR was reported relative to the frontal plane random.org) that would select a number 1 through
by subtracting the measured angle from 90 degrees. 10; the UTR measurement(s) in the step correspond-
UTR was recorded along with the time stamp of the ing to the number(s) generated were utilized in the
measured image. calculation.

Using the method described above, two indepen- Bland-Altman plots were generated to assess the
dent raters analyzed all 2D videos on two occasions agreement of 2D and 3D UTR measurements.15 The
separated by seven days to assess intrarater and frequency at which each rater chose the same or
interrater reliability of 2D right and left peak UTR different video frames for rating 1 and rating 2 was
measurement. The two raters were physical therapy determined; when the frames chosen for ratings 1
students; one student had one year of experience and 2 were different, the time difference and the dif-
with 2D video-based gait analysis and the other stu- ference in UTR measurement between the two dif-
dent had no prior experience. For the second rat- ferent time points was determined.
ing, a third party not involved in the rating process
blinded the raters to the original runner identifica- Magnitude of left and right peak UTR for female and
tion numbers on the videos. During both ratings, the male runners was summarized for 2D and 3D mea-
raters were blinded to the results of the other rater. surements using the mean and standard deviation.
Paired T-tests were performed to determine the dif-
Statistical analysis ference in average 2D and 3D UTR measurements
Participant characteristics were summarized using (α =0.05). Mean (SD) stride-to-stride difference in
the mean and standard deviation (Table 1). For 2D right and left peak UTR was determined by calcu-
peak right and left UTR measurements, intrarater lating the average difference in UTR measurement
reliability was determined for each rater and interra- between five consecutive strides for all 16 runners
ter reliability was determined for the first and second (n=4 stride-to-stride differences per runner).
ratings using intraclass coefficients (ICC2,1, 95%CI).
Intraclass coefficients were calculated using a single RESULTS
UTR measurement and the mean of two, three and
Intrarater reliability
five measurements for each runner.
Excellent intrarater reliability was found for right
Measurement responsiveness was assessed using and left peak UTR (Table 2). Including both raters,
the standard error of measurement (SEM) according the ICC2,1 for right and left UTR ranged from 0.987 to
to the equation, SEM = SD x √ (1-ICC) and minimum 0.999 when a single measurement for each runner
detectable change (MDC) at the 95% confidence lim- was utilized (Table 2). The SEM ranged from 0.14
its according to the equation, MDC95% = SEM x 1.96 to 0.5 degrees and the MCD ranged from 0.39 to 1.4
* √2.14 All statistical analyses were completed using degrees (Table 2). Compared to when a single left or
SPSS statistics Version 22 (IBM Corporation Armonk, right peak UTR measurement was used per runner,
NY, USA). All graphical displays were created using the intrarater ICC2,1 was increased by 0.008 at most
KaleidaGraph version 4.5.2 (Synergy Software, Read- when averaging two, three or five measurements of
ing PA, USA). left or right peak UTR per runner (Table 3). Given
the excellent intrarater reliability of the 2D UTR
Concurrent validity of 2D measurements was measurement when using a single measurement per
assessed by calculating the Pearson Product Corre- runner, the mean absolute differences (SD) between
lation coefficients (r) for the association between measurements, SEM and MDC were calculated for
2D and 3D peak left and right UTR measurements right and left UTR using only one measurement

The International Journal of Sports Physical Therapy | Volume 15, Number 6 | December 2020 | Page 913
 Table 2. Intrarater reliability and responsiveness of the 2D transverse plane upper trunk
rotation.

Table 3. Interrater and intrarater coefficients for single, and the average of 2, 3, and 5 trunk rotation measures
per runner (95% CI).

per runner (Table 2). Between ratings, the absolute measurements were averaged per runner, the inter-
differences between measurements for each rater rater ICC2,1 for right and left UTR ranged from 0.986
ranged from 0.25 to 0.73 degrees, the SEM ranged to 0.998 (Table 3). When a single UTR measure was
from 0.14 to 0.5 degrees and the MDC ranged from utilized per runner the mean absolute difference
0.39 to 1.4 degrees (Table 2). between UTR measurements taken by each of the
raters at ratings 1 and 2 ranged from 0.47 to 0.65,
At ratings 1 and 2, 58% and 44% of the video frames
the SEM ranged from 0.31 to 0.41 degrees and the
selected by Rater 1 and Rater 2, respectively, were
MCD ranged from 0.85 to 1.1 degrees (Table 4). For
identical. For ratings 1 and 2, 100% of the frames
ratings 1 and 2, the raters chose the identical frames
selected by Rater 1 and 95% of the frames selected
38.8% and 41.9% of the time, respectively. For rat-
by Rater 2 were separated by 0.05 seconds or less.
ings 1 and 2, 93.8% and 98.8% of the frames selected
Selection of frames differing in time by 0 to 0.05
by the two raters, differed by 0.04 seconds or less.
seconds resulted in UTR measures that differed,
on average, by 0.46 degrees (SD=0.47 degrees;
range=0-2.92 degrees). Concurrent validity and agreement
Peak left and right UTR measurements in 2D and 3D
Interrater reliability were significantly correlated when a single randomly
Excellent interrater reliability was found for both chosen measurement was utilized per runner (all r
right and left peak UTR (Table 4). When a sin- ≥ 0.986, all p-values < 0.001; Table 5). Bland-Altman
gle UTR measurement was utilized, the interra- plots indicate that there is good agreement between
ter ICC2,1 for right and left UTR ranged from 0.990 2D and 3D measures (Figure 3). Average peak right
to 0.995 (Table 4). When two, three or five UTR UTR measured in 3D (n=16) was 15.5 ± 4.1 (SD)

The International Journal of Sports Physical Therapy | Volume 15, Number 6 | December 2020 | Page 914
 Table 4. Interrater reliability and responsiveness of the 2D transverse plane upper trunk rotation.

Table 5. Association between two- and three-dimensional degrees for right UTR and 1.5 (1.3) degrees for left
peak upper trunk rotation measurements. UTR, which did not differ significantly from that
measured in 2D: 1.4 (1.1) degrees for right UTR and
1.5 (1.3) degrees for left UTR (all p > 0.05). The
mean stride-to-stride differences (SD) in right and
left UTR for male and female runners are displayed
in figure 5.

DISCUSSION
Results indicate that 2D measurement of transverse
degrees for males and 17.6 ± 4.7 (SD) degrees for plane peak UTR is reliable, valid, and feasible for
females; average peak left UTR measured in 3D was clinical use. Transverse plane video capture only
15.3 ± 3.6 (SD) for males and 21.5 ± 2.2(SD) degrees requires a device capable of a capture rate of 60 fps
for females (Figure 4). Average 2D and 3D peak UTR and a mount secured over a treadmill. A capture rate
measurements were not significantly different (all of 120 fps has been recommended for video analy-
p-values > 0.05; Figure 4). sis of frontal and sagittal plane lower extremity mea-
sures,16 but the slower speed of UTR makes it possible
Across all runners the mean stride-to-stride differ- to use a lower capture rate. This study utilized the
ence (SD) in UTR measured in 3D was 1.4 (1.0) tablet’s front camera (maximum capture rate 60 fps)

Figure 3. Bland Altman plots for right trunk rotation (a) and left trunk rotation (b).

The International Journal of Sports Physical Therapy | Volume 15, Number 6 | December 2020 | Page 915
 Excellent intrarater and interrater reliability was
achieved using only one measurement of right or
left UTR per runner even by examiners with varied
levels of experience. The lowest confidence interval
for ICC2,1 based on a single measurement with a rela-
tively inexperienced examiner was 0.962. Although
using the average of multiple measurements
achieved slightly greater reliability, the minimal
improvement in ICC2,1 does not appear to warrant
taking additional measurements.

UTR measurements may inform intervention for

injured runners or identify factors associated with
increased injury risk. Transverse plane movement
of the runner’s trunk and upper extremities (UEs)
play an important role in controlling whole body
Figure 4. Mean (± SD) right and left trunk rotation for male angular momentum and, therefore, the loading of
and female runners as measured from 3D motion capture
joints and tissues. During running, each UE alter-
and 2D video analysis (n=16). For simplicity, only 2D mea-
surements from rater 1, rating 1 is displayed. nately swings in the anteroposterior direction from
the shoulder with the contralateral LE while the
pelvis and upper trunk rotate in antiphase coordi-
nation in the transverse plane.17, 18 With symmetri-
to visualize the camera’s field of view and assure it
cal and reciprocal motion of the UEs and LEs in the
was squarely positioned over the treadmill (Figure 2).
transverse plane, upper trunk and UE movement
Reliability of the 2D measurement depends primar- counterbalance angular momentum generated by
ily on, 1) visual selection of the video frame that the pelvis and LE’s.19, 20 When transverse plane angu-
exhibits peak UTR and 2) the manual digitization of lar impulses created by the upper and lower body
measurement landmarks to determine trunk posi- of the runner do not cancel out, the magnitude of
tion relative to the running surface. The amount of the moment that the runner applies to the ground
measurement error attributable to frame selection through his or her foot (i.e., the free moment) must
appears to be minimal. Of the video frames selected increase to maintain running in a straight line.21, 12
at ratings 1 and 2, 93-100% of the frames were sep- Larger free moments create increased tibial torsion
arated by 0.04 seconds or less within or between strain, which has been associated with increased
raters, which corresponded to an average UTR mea- risk for tibial stress fractures in retrospectives stud-
surement difference of 0.44 ± 0.45 (SD) degrees. ies of female runners.22, 23 Likewise, in-phase coor-
Measurement differences attributed to frame selec- dination of the upper trunk and pelvis is associated
tion and the MDC for Rater 1 (Table 2), are lower with increased vertical impact peak and braking
than the average stride-to-stride variability in right impulse,24 which have been associated with plantar
and left UTR that was observed (right UTR: 1.4 ± 1.1 fasciitis, tibial stress fractures and patellofemoral
(SD) degrees; left UTR: 1.5 ± 1.3 (SD) degrees). The pain.2 In-phase coordination has also been observed
MDC for Rater 2 were similar to the mean stride-to- in runners with current or past history of low back
stride differences in UTR (Table 2). Rater 1’s lower pain.25 Low back pain may be due to increased com-
MDC might be explained by having prior experience pressive forces in the lumbar spine when pelvic
with 2D video analysis. Clinically relevant changes transverse plane motion reaches a maximum but
in UTR magnitude remain to be determined, but UTR does not increase appropriately with increased
results indicate that deviations from normal UTR speed to minimize whole body angular momen-
or from contralateral UTR > 1.5 degrees can be tum.26, 27 Lastly, UTR is inversely correlated with iso-
detected with confidence using 2D video analysis. kinetic hip extension and hip abduction torque28 and

The International Journal of Sports Physical Therapy | Volume 15, Number 6 | December 2020 | Page 916
Figure 5. Mean difference (SD) in right and left peak trunk rotation between 5 consecutive strides (n=4) for female runners
(a and b) and male runners (c and d).

hip abduction strength deficit has been associated validity of 2D UTR measurements. Although peak
with iliotibial band syndrome.29 Therefore, a valid UTR provides insight into the total range of motion
and reliable measure of UTR may assist in identify- of the upper trunk in the transverse plane it does
ing injury risk, and guiding clinical decision-making not provide information regarding the timing and
regarding transverse plane impairments in runners. coordination of UTR with respect to key phases of
gait (i.e., initial contact, midstance) and other kine-
Limitations matic measures (i.e., pelvis rotation). The relation-
The generalizability of this study is limited by the ship between upper trunk and pelvis coordination is
small number of young, healthy runners included not always linear as speed is increased30 and, there-
in this study who ran at preferred speed for data col- fore, the UTR measurement studied here may not be
lection. Injured runners seeking clinical assessment an indicator of the timing and coordination of trunk
may display greater gait deviations, particularly in and pelvis rotation, which may be useful in guiding
the frontal plane, that may impact the reliability and intervention.

The International Journal of Sports Physical Therapy | Volume 15, Number 6 | December 2020 | Page 917
CONCLUSION 11. Stefanshyn DJ, Stergiou P, Lun VMY, Meeuwisse
Two-dimensional video-based measurements of WH, Worobets JT. Knee angular impulse as a
peak UTR in the transverse plane is a valid and reli- predictor of patellofemoral pain in runners. Am J
Sports Med. 2006; 34(11): 1844-1851.
able measure to be utilized in 2D video-based analy-
12. Willwacher S, Goetz I, Fischer KM, Brüggeman G-P.
ses of running gait. This study provides a foundation
The free moment in running and its relation to joint
for future research that examines the relationships loading and injury risk. Footwear Sci. 2016; 8(1): 1-11.
between UTR magnitude and asymmetry with spe-
13. Riley PO, Dicharry J, Franz J, Della Croce U, Wilder
cific running injuries as well as the efficacy of inter- RP, Kerrigan, DC. A kinematic and kinetic
ventions that target transverse plane movement in comparison of overground and treadmill running.
managing running injuries. Med Sci Sports Exerc. 2008; 40(6): 1093-1100.
14. Vincent WJ, Weir JP. Statistics in Kinesiology. 4th
REFERENCES Edition. Human Kinetics, Champaign, IL. 2012. Pp:
1. Running USA. 2012 State of the Sport Part II: 222-226. Chapter 13 Quantifying Reliability.
Running Industry Report. Available at: http:// www. 15. Bland JM, Altman D. Statistical methods for
runningusa.org/2012-state-of-sport-part-2. Accessed assessing agreement between two methods of
Nov 14, 2016. clinical measurement. Lancet. 1986; 327(8476):
2. Davis IS, Futrell E. Gait retraining: Altering the 307-310.
fingerprint of gait. Phys Med Rehabil Clin N Am. 2016; 16. Souza RB. An evidence-based videotaped running
339-355. biomechanics analysis. Phys Med Rehabil Clin N Am.
3. Schubert AG, Kempf J, Heiderscheit BC. Influence of 2016; 27(1): 217-236.
stride frequency and length on running mechanics: 17. Ford MP, Wagenaar RC, Newell KM. Arm constraint
a systematic review. Sports Health. 2014; 6(3): and walking in healthy adults. Gait Posture. 2007; 26:
210-217. 135-141.
4. Tate, JJ, Milner, CE. Real-time kinematic, 18. Van Emmerik REA, Wagenaar RC. Effects of walking
temporospatial, and kinetic biofeedback during gait velocity on relative phase dynamics in the upper
retraining in patients: A systematic review. Phys trunk in human walking. J Biomech. 1996; 29(9):
Ther. 2010; 90(8): 1123-1134. 1175-1184.
5. Van Gent RN, Siem D, van Middlekoop M, et al. 19. Arellano CJ, Kram R. The metabolic cost of human
Incidence and determinants of lower extremity running: is swinging the arms worth it? J Expt Biol.
running injuries in long distance runners: a 2014; 217: 2456-2461.
systematic review. Br J Sports Med. 2007; 41: 469-480. 20. Hinrichs R. Upper extremity function in running II:
6. Dingenen B, Staes FF, Santermans L, et al. Are angular momentum considerations. Int J Sports Phys
two-dimensional measured frontal plane angles Ther. 1987; 3: 242-263.
related to three-dimensional measured kinematic 21. Holden JP, Cavanagh PR. The free moment of
profiles during running? Phys Ther Sport. 2017; 29: ground reaction in distance running and its changes
84-92. with pronation. J Biomech. 1991; 24, 887–897.
7. Maykut JN, Taylor-Haas JA, Paterno MV et al. 22. Milner CE, Davis IS, Hamill J. Free moment as a
Concurrent validity and reliability of frontal plane predictor of tibial stress fracture in distance runners.
motion during running. Int J Sports Phys Ther. 2015; J Biomech. 2006; 39(15): 2819-2825.
10(2): 136-146.
23. Pohl MB, Mullineux DR, Milner CE, Hamill J, Davis
8. McClay I, Manal K. The influence of foot abduction IS. Biomechanical predictors of retrospective tibial
on differences between two-dimensional and three- stress fractures in runners. J Biomech. 2008; 41(6):
dimensional rearfoot motion. Foot Ankle Int. 1998; 1160-1165.
19(1): 26-31. 24. Morley JJ and Traum E. The effects of dorso-lumbar
9. Miller CE, Davis IS, Hamill J. Free moment as a motion restriction on the ground reaction force
predictor of tibial stress fracture in distance runners. components during running. J Bodyw Mov Ther.
J Biomech. 2007; 39(15): 2819-2825. 2016; 20:432-440
10. Noehren B, Davis IS and Hamill J. ASB clinical 25. Seay JF, Van Emmerik REA, Hamill J. Low back pain
biomechanics award winner 2006 prospective study status affects pelvis-trunk coordination and
of the biomechanical factors associated with iliotibial variability during walking and running. Clin
band syndrome. Clin Biomech. 2006; 22(9): 951-956. Biomech. 2011; 26: 572-578.

The International Journal of Sports Physical Therapy | Volume 15, Number 6 | December 2020 | Page 918
26. Preece SJ, Mason D, Bramah C. How do elite 29. Mucha MD, Caldwell W, Schlueter EL, Walters C,
endurance runners alter movements of the spine Hassen A. Hip abductor strength and lower
and pelvis as running speed increases? Hum Mov Sci. extremity running related injury in distance
2016; 46: 132-134. runners: a systematic review. J Sci Med Sport. 2017;
27. Kudo S, Fujimoto M, Isaka T, Nagano A. Quantitative 20: 349-355.
assessment of thorax deformation during running. 30. Li Y, Kakar RS, Walker MA, Guan L, Simpson KJ.
J Biomech. 2017; 59: 116-121. Upper trunk-pelvis coordination during running
28. Ford KR, Taylor-Haas JA, Genthe K, Hugentobler J. using the continuous relative fourier phase method.
Relationship between hip strength and trunk motion J App Biomech. 2018; 34: 312-319.
in college cross-country runners. Med Sci Sport Exerc.
2013; 45(6): 1125-30.

The International Journal of Sports Physical Therapy | Volume 15, Number 6 | December 2020 | Page 919