Professional Documents
Culture Documents
The Knee
a r t i c l e i n f o a b s t r a c t
Article history: Patellofemoral pain (PFP) is a common complaint among female runners. The etiology for PFP is frequent-
Received 12 May 2011 ly associated with increased patellofemoral joint stress (PFJS) and altered hip and knee joint kinematics
Received in revised form 30 August 2011 during running. However, whether PFJS during running is increased among runners with PFP is unknown.
Accepted 12 September 2011
The primary aim of this study was to compare PFJS during running among females with and without PFP.
We also compared hip and knee transverse plane kinematics during running due to their potential influ-
Keywords:
Kinetics
ence on patellofemoral contact area and PFJS. Three dimensional hip and knee running kinematics and
Kinematics kinetics were obtained from 20 females with PFP and 20 females with no pain. Patellofemoral joint stress
Gait during running was estimated using patellofemoral contact area and a sagittal plane patellofemoral joint
Knee model previously described. Patellofemoral joint stress, PFJS-time integral, and hip and knee transverse
Hip plane kinematics at the time of impact peak and peak ground reaction force were compared between
groups using a multivariate analysis of variance. The results show that peak PFJS and PFJS-time integral
were similar between groups. Peak knee flexion angle and net knee extension moment were not different
between groups. However, females with PFP demonstrated hip internal rotation that was 6° greater
(P = 0.04) when ground reaction forces were greatest. The extent these results are influenced by com-
pensations for pain is unclear. However, if increased PFJS contributes to the etiology or exacerbation of
PFP, interventions to minimize altered transverse plane hip kinematics may be indicated among runners
who demonstrate this characteristic.
© 2011 Elsevier B.V. All rights reserved.
0968-0160/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.knee.2011.09.006
704 A.D. Wirtz et al. / The Knee 19 (2012) 703–708
2. Methods were positioned as a cluster of three markers on the rearfoot of the shoe,
a cluster of four markers on the posterior shank, a cluster of four
The study protocol was approved by the university institutional markers on the lateral thigh, and three markers for the pelvis on each
review board and all subjects provided informed consent prior to par- anterior superior iliac spine and at the L5–S1 interspace. The knee
ticipation. Using an alpha level of 0.05 a beta level of 0.2, and joint center was assumed to be the midpoint of a line between the fem-
expected variability of knee extension moment data during running oral condyle markers. Ankle joint center was assumed to be the mid-
in previous studies with similar methodology, 19 participants per point of a line between the two ankle malleoli markers. The hip joint
group were determined to be necessary to identify between-group center was identified using a Newton iterative spherical fitting algo-
differences with effect sizes greater than 0.7 [11,12]. All participants rithm on data recorded during a standing trial where the instrumented
were female runners, 18–35 years old, who ran at least 10 miles per leg was moved in a prescribed fashion prior to the running trials [16].
week and reported their activity level as greater than or equal to 5 During this trial, participants stood on their contralateral leg while mov-
out of 10 on the Tegner activity scale (a measure of regular participa- ing their free leg through two arcs of approximately 80° hip flexion and
tion in recreational sports activities that require running or jumping) two arcs of 50° hip abduction. Following both the standing calibration
[13]. All subjects who were pregnant, reported a known cardiovascu- trial and hip center movement trial, each of the anatomical markers
lar pathology, had surgery to either lower extremity over the last were removed.
12 months or sustained a traumatic injury to either knee joint within All participants were asked to run along a 20 m runway between
6 months of the study were excluded from participation. Subjects in 3.52 and 3.89 m/s as indicated by the forward velocity of the sacral
the control group had to be free of any lower extremity symptoms marker in the lab coordinate system prior to contact with the force
of injury for the last 2 years that prohibited regular participation in platform. After at least five practice trials, five trials were collected
recreational running. for further analysis. During each trial, marker data were collected at
Potential participants with a primary complaint of knee pain dur- 120 Hz using eight Eagle digital cameras (Motion Analysis Corpora-
ing running were screened by a licensed physical therapist for specific tion, Santa Rosa, CA, USA) positioned around the runway. These
criteria to be included in the PFP group. These criteria included a ver- marker trajectories were digitally filtered at 12 Hz using a low pass,
bal pain score of at least a 3 (moderate) on a 10 point verbal pain fourth order Butterworth recursive filter. Ground reaction forces
scale during running and squatting, prolonged sitting, ascending or were recorded at 1080 Hz by a force platform (Model 4080, Bertec
descending stairs, or jumping. Potential participants must have also Corporation, Columbus, OH) flush with the surface of the runway.
described pain behind or adjacent to the patella and not solely at Ground reaction force data used in the calculation of joint moments
the iliotibial band, patellar tendon, or knee-joint line. Knee symptoms using the inverse dynamics approach were digitally filtered at 12 Hz
were required to be of insidious onset and present for at least using a low pass, fourth order Butterworth recursive filter based on
2 months in duration. Participants with PFP had to report that their the recommendations of Bisseling and Hof [17]. The original ground
symptoms were exacerbated with manual compression of the patella reaction force data were then digitally filtered at 50 Hz using a low
into the trochlear groove with the knee in 15° of flexion or with pal- pass, fourth order Butterworth recursive filter in order to identify
pation of the medial or lateral patellar retinaculum against the poste- the peak vertical force impact transient and peak vertical ground re-
rior patellar surface. Lastly, participants with PFP were required to action force for each trial.
score less than 85/100 on the Anterior Knee Pain Scale [14]. Fifteen Hip and knee joint kinematics and joint moments during the
points on this scale has been determined to be the minimum clinically stance phase for each running trial were calculated with Visual 3D
important difference in comparison to healthy controls [15]. All po- software (C-Motion Inc, Rockville, MD) using sequence of rotations
tential participants were excluded if they presented for screening where hip and knee flexion–extension were calculated first, followed
tests with signs and symptoms of meniscus or ligament pathology, by abduction–adduction, and internal–external rotation, respectively.
were currently receiving supervised treatment for PFP, or reported Joint kinematic conventions were defined using the right hand rule.
symptoms in either foot, ankle, hip or low back that were exacerbated Kinematic data were not normalized to the static neutral trial. In
by running. other words, 0° corresponded to an erect posture at the hip and knee.
Screening of potential participants continued until 20 females Kinematic and kinetic variables of interest included peak knee ex-
with PFP and 20 healthy females were identified and agreed to partic- tension moment, peak knee flexion angle, and hip and knee trans-
ipate in this investigation. Forty two potential participants with knee verse plane angle at the vertical impact transient and at the peak
pain were assessed for the above inclusion and exclusion criteria. Ten ground reaction force (GRF) during the stance phase of each running
of the 20 participants with PFP included in this study reported bilat- trial. Hip and knee transverse plane joint angles were compared be-
eral symptoms. For these participants, the most symptomatic lower tween groups at the peak impact transient because this event occurs
extremity was chosen for analysis. The right lower extremity of the early during stance phase and kinematic differences in females with
healthy control group subjects was used for analysis. All participants PFP may be especially apparent in shallow knee flexion angles [18].
wore the same type of shoe (model 629, New Balance, Boston, MA) Transverse plane hip and knee joint kinematics were compared at
during testing in order to reduce variability that may be caused by peak vertical ground reaction force because this event occurs near
different shoe absorption properties. midstance of running when knee flexion and knee extensor moment
values are greatest. Kinematic rotations that decrease patellofemoral
2.1. Procedure contact area during this time may significantly increase PFJS. Kine-
matic dependent variables of interest were identified using custom
Subjects were prepared for 3D motion analysis testing during run- software (LabView 8.6, National Instruments, Austin, TX). The aver-
ning by attaching reflective markers to the leg and pelvis of the involved age of these variables of interest from all five running trials were cal-
limb (PFP group) or right limb (control group). The three-dimensional culated and used for analysis.
coordinates of these markers were used to track the motion of the pel-
vis, femur, shank, and foot, each modeled as a rigid body. Anatomical 2.2. Patellofemoral joint stress
markers used to establish the segmental-coordinate systems were
placed over each iliac crest, the greater trochanters, medial and lateral Patellofemoral joint stress (PFJS) during running was calculated
femoral condyles, medial and lateral proximal tibia, medial and lateral by dividing estimated patellofemoral joint reaction force by estimat-
malleoli, the first and fifth metatarsal heads, and the tip of the shoe. ed patellofemoral contact area. Patellofemoral joint reaction force
Tracking markers, which remained in place for all of the running trials, was estimated using a biomechanical model previously described by
A.D. Wirtz et al. / The Knee 19 (2012) 703–708 705
Salem and Powers [19] which has also been used to estimate PFJS Table 2
during walking, resisted squatting, and stair climbing in other previ- Comparison of mean (SD) estimated patellofemoral joint stress (PFJS) variables be-
tween females with and without PFP during the stance phase of running.
ous studies [5,6,20]. This model uses tibiofemoral sagittal plane
angle and net joint moment data obtained using inverse dynamics PFP group Healthy group P value⁎ Effect size
as input variables. Internal knee extension moment data during the Peak PFJS (MPa) 9.6 (2.5) 8.5 (2.2) 0.14 0.47
stance phase of each running trial was divided by the effective lever PFJS-time integral (MPa s) 236.4 (66.4) 218.5 (87.0) 0.47 0.23
arm for the quadriceps as a function of knee flexion angle to obtain PFP = patellofemoral pain.
quadriceps force. The effective lever arm of the quadriceps is based ⁎ Univariate between-subjects effects (α = 0.05).
on cadaver data presented by van Eijden et al. [21] as represented
by the equation presented by Salem and Powers [19]. Patellofemoral
joint reaction force was then calculated by multiplying calculated
quadriceps force by the ratio of patellofemoral compression force to
quadriceps force as a function of knee flexion angle presented by
van Eijden et al. [21] and represented mathematically by Salem and
Powers [19].
Patellofemoral contact area during weight bearing may differ
among females with PFP compared to females without PFP during
weight bearing activities, particularly at shallow knee flexion angles
[6,18,22]. Thus, the estimated patellofemoral contact area as a func-
tion of knee flexion angle was unique for each group. Patellofemoral
contact area for females with and without PFP was obtained from
data reported by Connolly et al. [22]. In this previous study, contact
area was determined at 15°, 30°, and 45° knee flexion using magnetic
resonance imaging as subjects lay supine with the quadriceps con-
tracted at 10% maximum voluntary force. These data were then line-
Fig. 1. Patellofemoral joint stress for females with and without PFP during the stance
arly interpolated to provide patellofemoral contact area as a function phase of running. Shaded areas represent standard error of the mean at each time
of knee flexion angle for each group. Patellofemoral joint reaction point. The dotted lines represent time of vertical impact peak and peak ground reaction
force was divided by this group-specific patellofemoral contact area force, respectively.
to estimate PFJS during running. Dependent variables of interest
based on the PFJS model included peak PFJS and the PFJS-time inte-
Transverse plane hip kinematics measures were not different between groups at
gral during the stance phase of running.
the time of their vertical impact peak (Table 3, Fig. 2). However, females with PFP dem-
onstrated 6° of greater hip internal rotation at their peak vertical GRF (P = 0.04,
ES = 0.68).
2.3. Statistical analysis Differences in transverse plane knee kinematics between groups were not statisti-
cally significant at either peak vertical force impact transient or peak vertical GRF
Each dependent variable was first compared between groups using (Table 3, Fig. 3). However, female runners with PFP demonstrated nearly 5° greater
knee external rotation in comparison to the control group at the time of their vertical
a multivariate analysis of variance (MANOVA) with a familywise
impact peak (P = 0.13, ES = 0.48) (Table 3).
alpha=0.05. Follow-up univariate tests were performed between groups Peak knee flexion angle was similar between groups (Table 3, Fig. 4). Similarly, no
when statistically significant MANOVA results were identified. All statisti- difference in peak knee extension moment between the PFP group and the control
cal procedures were performed in SPSS (version 13, SPSS Inc., Chicago, IL). group were observed (Fig. 5).
4. Discussion
3. Results
Subjects from both groups were similar with respect to age, height, weight, and The primary purpose of this study was to compare PFJS during
BMI (Table 1). Subjects in the PFP group on average had slightly lower self-reported ac- running between female runners with and without PFP. Transverse
tivity levels and ran five fewer miles per week than the control group (Table 1).
plane hip and knee joint kinematics may also affect PFJS but are not
Significant multivariate effects were found when all dependent variables were con-
sidered together (Wilks'λ = 0.44, F (14,25) = 2.31, P = 0.033). Slightly greater PFJS
included in this PFJS model. Thus, our secondary purpose was to com-
(ES = 0.47) and PFJS-time integral (ES = 0.23) values were found for the PFP group com- pare these kinematic variables between groups. Previous studies have
pared with the control group during running (Table 2, Fig. 1). However, follow-up univar- used these methods to model PFJS among individuals with and
iate analysis revealed these differences were not statistically significant.
Table 3
Comparison of mean (SD) kinematic and kinetic variables between females with and
Table 1 without PFP during the stance phase of running.
Comparison of mean (SD) anthropometric and demographic characteristics for runners
in the PFP and healthy groups. PFP group Healthy group P value⁎ Effect size
Peak knee extensor moment 1.30 (0.2) 1.27 (0.2) 0.70 0.12
PFP group Healthy group
(Nm/kg m)
Age (years) 21.3 (2.6) 21.6 (4.4) Peak knee flexion angle (°) 43.9 (5.0) 41.8 (4.1) 0.16 0.46
Height (m) 1.7 (0.1) 1.7 (0.1) Hip IR at impact peak (°) 3.0 (11.9) − 2.0 (8.7) 0.14 0.48
Weight (kg) 62.9 (7.7) 61.8 (9.2) Hip IR at peak vGRF (°) 3.7 (10.7) − 2.4 (7.2) 0.04 0.68
BMI 22.1 (1.9) 21.7 (1.9) Knee IR at impact peak (°) − 3.2 (11.8) 1.6 (8.0) 0.13 0.49
Miles/week 15.6 (8.1) 20.6 (10.9) Knee IR at peak vGRF (°) 2.3 (9.6) 4.3 (7.5) 0.48 0.23
Years running 4.1 (3.1) 4.8 (3.5)
PFP = patellofemoral pain, IR = internal rotation, vGRF = vertical ground reaction
Tegner score 5.9 (0.8) 6.9 (1.5)
force; negative values represent external rotation.
BMI = body mass index, PFP = patellofemoral pain. ⁎ Univariate between-subjects effects (α = 0.05).
706 A.D. Wirtz et al. / The Knee 19 (2012) 703–708
Fig. 2. Hip internal rotation angle for females with and without PFP during the stance Fig. 5. Knee extension moment for females with and without PFP during the stance
phase of running. Shaded areas represent standard error of the mean at each time phase of running. Shaded areas represent standard error of the mean at each time
point. The dotted lines represent time of vertical impact peak and peak ground reaction point. The dotted lines represent time of vertical impact peak and peak ground reaction
force, respectively. force, respectively.
peak vertical ground reaction force while running may reduce patellofe- and knee transverse plane kinematics during running may increase
moral contact area when patellofemoral joint reaction forces are great- PFJS and appear present among some females with PFP. Future stud-
est, leading to significantly elevated PFJS. However, it is worth noting ies that account for the influence of hip and knee transverse plane ki-
that the minimal detectable difference (MDD) using this methodology nematics on PFJS during dynamic activities appear to be justified.
to determine peak hip transverse plane kinematics during running is es-
timated to be 8.8° [30]. Hip internal rotation differences relative to the
Funding source
control group mean exceeded the MDD for six PFPS participants
(30%). Therefore, although PFP group mean hip internal rotation was
None.
statistically greater than the control group and a moderate effect size
was found, increased hip internal rotation was greater than the range
of potential measurement error among only a minority of females Conflict of interest statement
with PFP.
Females with PFP demonstrated 4.8° greater knee external rota- No authors have any financial or personal relationships with other
tion at the time of the vertical impact peak than females without people or organizations that could inappropriately influence their work.
PFP. These results are consistent with a previous report of 4.3° greater
knee external rotation during running, jumping, and single leg squats Acknowledgements
[12]. Three females with PFPS demonstrated increased knee internal ro-
tation during running which contributed to unexpectedly high The authors acknowledge Kaitlin Strauss and Dan Thour for their
within-group variability and a statistically insignificant between-group assistance with data collection and analysis.
comparison. Using kinematic reliability estimates previously reported,
the MDD for peak knee transverse plane kinematics during running is
4.5° [30]. Seven females with PFP (35%) demonstrated knee external References
rotation values that exceeded the MDD relative to the control group [1] USA Track and Field. Long distance running—state of the sport [Online]. Available
mean. Among these participants with PFP we can be 95% confident at; 2003http://www.usatf.org/news/specialReports/2003LDRStateOfTheSport.asp
that this greater knee external rotation was not due to random mea- [Accessed September 12, 2010].
[2] DeHaven KE, Lintner DM. Athletic injuries: comparison by age, sport, and gender.
surement error. Such females who run with greater knee external ro-
Am J Sports Med 1986;14(3):218–24.
tation magnitudes may experience 10–24% greater PFJS during each [3] Taunton JE, Ryan MB, Clement DB, McKenzie DC, Lloyd-Smith DR, Zumbo BD. A
step [10]. However, greater knee external rotation was not a univer- retrospective case–control analysis of 2002 running injuries. Br J Sports Med
2002;36:95–101.
sal finding among females runners with PFP. This supports the pre-
[4] Boling M, Padua D, Marshall S, Guskiewicz K, Pyne S, Beutler A. Gender differences
mise that there may not be a single predominant mechanism of in the incidence and prevalence of patellofemoral pain syndrome. Scand J Med Sci
injury for PFP. Sports 2010;20(5):725–30.
Sagittal plane differences in knee mechanics during running [5] Brechter JH, Powers CM. Patellofemoral joint stress during stair ascent and de-
scent in persons with and without patellofemoral pain. Gait Posture 2002;16:
among females with PFP were not observed in this study. Peak knee 115–23.
extension moment was similar between groups, which is consistent [6] Brechter JH, Powers CM. Patellofemoral stress during walking in persons with and
with previous studies during walking and running [6,12,31]. Likewise, without patellofemoral pain. Med Sci Sports Exerc 2002;34(10):1582–93.
[7] Ward SR, Powers CM. The influence of patella alta on patellofemoral joint stress
peak knee flexion angle was similar between groups agreeing with during normal and fast walking. Clin Biomech 2004;19:1040–7.
several previous investigations during a variety of weight bearing ac- [8] Salsich GB, Perman WH. Patellofemoral joint contact area is influenced by tibiofe-
tivities [5,6,32,33]. If increased PFJS contributes to the etiology or ex- moral rotation alignment in individuals who have patellofemoral pain. J Orthop
Sports Phys Ther 2007;37(9):521–8.
acerbation of PFP, interventions that focus on sagittal plane knee [9] Lee TQ, Anzel SH, Bennett KA, Pang D, Kim WC. The influence of fixed rotational
mechanics during running may have limited therapeutic effect. Clini- deformities of the femur on the patellofemoral contact pressures in human cadav-
cians may pose interventions that reduce transverse plane rotations er knees. Clin Orthop Relat Res 1994;302:69–74.
[10] Li G, DeFrate LE, Zayontz S, Park SE, Gill TJ. The effect of tibiofemoral joint kinematics
of the hip and knee joints (and increase patellofemoral contact
on patellofemoral contact pressures under simulated muscle loads. J Orthop Res
area) as potentially more valuable in the treatment of PFP. 2004;22:801–6.
There are several notable limitations for this study. First, contact area [11] Ferber R, Davis IM, Williams III DS. Gender differences in lower extremity me-
chanics during running. Clin Biomech 2003;18:350–7.
estimates were based on MRI data obtained for females with and with-
[12] Willson JD, Davis IS. Lower extremity mechanics of females with and without
out PFP as they maintained an isometric contraction of 10% maximal vol- patellofemoral pain across activities with progressively greater task demands.
untary force [22]. Actual contact area for subjects in our study may have Clin Biomech 2008;23:203–11.
been different and would significantly affect PFJS estimates. However, [13] Tegner Y, Lysholm J. Rating systems in the evaluation of knee ligament injuries.
Clin Orthop Relat Res 1985;198:43–9.
the age, height, and weight of the females in our study were very [14] Kujala UM, Jaakkola LH, Koskinen SK, Taimela S, Hurme M, Nelimarkka O. Scoring
similar to the females used for our contact area estimate. Second, the of patellofemoral disorders. Arthroscopy 1993;9(2):159–63.
case–control design used in this study prohibits determination of [15] Watson CJ, Propps M, Ratner J, Zeigler DL, Horton P, Smith SS. Reliability and re-
sponsiveness of the lower extremity functional scale and the anterior knee pain
whether PFP group running mechanics were a cause or effect of pain. scale in patients with anterior knee pain. J Orthop Sports Phys Ther 2005;35(3):
Prospective studies are necessary to indentify running mechanics that 136–46.
contribute to PFP as well as compensatory changes that are most often [16] Hicks JL, Richards JG. Clinical applicability of using spherical fitting to find hip
joint centers. Gait Posture 2005;22:138–45.
employed after the onset of symptoms. Third, all participants ran in [17] Bisseling RW, Hof AL. Handling of impact forces in inverse dynamics. J Biomech
standardized footwear, regardless of their usual footwear or foot type, 2006;39:2438–44.
to minimize potential effects due to differences between shoe types. [18] Souza RB, Draper CE, Fredericson M, Powers CM. Femur rotation and patellofe-
moral joint kinematics: a weight-bearing magnetic resonance imaging analysis.
The change in footwear may have facilitated altered kinematics among
J Orthop Sports Phys Ther 2010;40(5):277–85.
participants not accustomed to the characteristics of the standardized [19] Salem GJ, Powers CM. Patellofemoral joint kinetics during squatting in collegiate
shoes. Fourth, female participants in this study were 18–35 years old, women athletes. Clin Biomech 2001;16:424–30.
[20] Wallace DA, Salem GJ, Salinas R, Powers CM. Patellofemoral joint kinetics while
and our findings may not be generalizable to males or females of differ-
squatting with and without and external load. J Orthop Sports Phys Ther
ent ages and activity levels. Finally, skin movement artifact may have 2002;32(4):141–8.
influenced the position of motion analysis markers during running, lead- [21] van Eijden TMGJ, Kouwenhoven E, Verburg J, Weijs WA. A mathematical model of
ing to random hip and knee joint kinematic error for both groups. the patellofemoral joint. J Biomech 1986;19(3):219–29.
[22] Connolly KD, Ronsky JL, Westover LM, Kupper JC, Frayne R. Differences in patello-
In conclusion, PFJS calculated using a sagittal plane model is simi- femoral contact mechanics associated with patellofemoral pain syndrome. J Bio-
lar among females with and without PFP during running. Altered hip mech 2009;42:2802–7.
708 A.D. Wirtz et al. / The Knee 19 (2012) 703–708
[23] Souza RB, Powers CM. Differences in hip kinematics, muscle strength, and muscle ac- [29] Boling MC, Padua DA, Marshall SW, Guskiewicz K, Pyne S, Beutler A. A prospective
tivation between subjects with and without patellofemoral pain. J Orthop Sports investigation of biomechanical risk factors for patellofemoral pain syndrome: the
Phys Ther 2009;39(1):12–9. joint undertaking to monitor and prevent ACL injury (JUMP-ACL) cohort. Am J
[24] Besier TF, Fredericson M, Gold GE, Beaupre GS, Delp SL. Knee muscle forces during Sports Med 2009;37(11):2108–16.
walking and running in patellofemoral pain patients and pain-free controls. J Biomech [30] Ferber R, McClay Davis I, Williams III DS, Laughton C. A comparison of within- and
2009;42:898–905. between-day reliability of discrete 3D lower extremity variables in runners. J Orthop
[25] Chinkulprasert C, Vachalathiti R, Powers CM. Patellofemoral joint forces and stress Res 2002;20:1139–45.
during forward step-up, lateral step-up, and forward step-down exercises. J Orthop [31] Nadeau S, Gravel D, Hebert LJ, Arsenault AB, Lepage Y. Gait study of patients with
Sports Phys Ther 2011;41(4):241–8. patellofemoral pain syndrome. Gait Posture 1997;5:21–7.
[26] Chen Y, Scher I, Powers CM. Quantification of patellofemoral joint reaction forces [32] Powers CM, Perry J, Hsu A, Hislop HJ. Are patellofemoral pain and quadriceps femoris
during functional activities using a subject-specific three-dimensional model. muscle torque associated with locomotor function? Phys Ther 1997;77(10):1063–78.
J Appl Biomech 2010;26:415–23. [33] Barton CJ, Levinger P, Webster KE, Menz HB. Walking kinematics in individuals
[27] Blond L, Hansen L. Patellofemoral pain syndrome in athletes: a 5.7-year retrospec- with patellofemoral pain syndrome: a case–control study. Gait Posture 2011;33:
tive follow-up study of 250 athletes. Acta Orthop Belg 1998;64(4):393–400. 286–91.
[28] Souza RB, Powers CM. Predictors of hip internal rotation during running: an eval-
uation of hip strength and femoral structure in women with and without patello-
femoral pain. Am J Sports Med 2009;37(3):579–87.