Tibia and rearfoot motion and ground reaction forces in subjects with patellofemoral pain syndrome during walking

1. Introduction 

Abnormal subtalar joint motion is linked to knee injuries as it may affect knee mechanics [36], [38]. Excessive or prolonged pronation can delay the external rotation of the tibia and alter the desirable timing between knee extension and rearfoot supination [13], [35], [36]. It has been suggested, therefore, that abnormal rearfoot motion can lead to patellofemoral pain syndrome (PFPS) [7], [36].

Previous studies have shown no significant deviation in the rearfoot peak angle in subjects with PFPS [2], [29]. However, there were significant deviations in the timing of peak eversion [2]. Moreover, no significant differences in the timing and magnitude of tibial internal rotation were found [29]. The inconsistency in the literature could be related to methodological issues, such as modelling the whole foot rather than the rearfoot [29], and the possibility of altered motion in other planes as the subtalar joint motion is inherently triplanar in nature [6], [33]. A three-dimensional approach thus, would be required to produce reliable results [2].

Subjects with PFPS with abnormal subtalar joint triplanar motion during the stance phase of walking would also demonstrate altered GRFs. For example, pronation in early stance is thought to be a loading response function of the foot therefore increased or delayed eversion may be reflected in the first peak vertical and anterior GRFs. Equivocal changes in the vertical and anterior-posterior GRFs, however, have been reported in subjects with PFPS compared to healthy subjects [2], [28]. Messier et al. [23] also suggested a possible association between the medio-lateral GRF and foot pronation thus, affecting the amount and rate of subtalar joint pronation. Investigating the GRFs in combination with rearfoot kinematics in subjects with PFPS may contribute to further understanding the rearfoot function and its effect on patellofemoral joint mechanics during walking.

The purpose of this study was to investigate the three-dimensional motion of the rearfoot, the tibial transverse plane rotation and the GRFs during the stance phase of walking in subjects with PFPS. It was hypothesized that there would be no differences in the magnitude nor the timing for the rearfoot motion relative to the tibia in the three planes of motion, the tibial transverse plane rotation and the GRFs during the stance phase of walking in subjects with PFPS, when compared to healthy subjects.

2. Methods 

The study design was a cross sectional comparative investigation of independent variables between a symptomatic group (PFPS) and a control group (asymptomatic). Prior to participation, all subjects were informed about the nature of the study and signed an informed consent, which was approved by the Human Ethics Committee of Southern Cross University (ECN-02-101).

Power analysis for subject numbers was performed prior to data collection using the data from Callaghan and Baltzopoulos [2]. The power calculation showed that a minimum of n=8 was required. The symptomatic group consisted of 13 females with PFPS. The PFPS group had a mean±S.D. age, mass and height of 38.4±10.11 years, 70.6±18.16kg and 166.3±5.97cm. PFPS subjects had symptoms affecting their right knee for a mean of 11 years (range of 1.5–30 years). The diagnosis of PFPS was determined by an independent physiotherapist based on the subjective complaint of retropatellar pain and physical examination. Subjects were excluded if their signs and symptoms were suggestive of other conditions including swelling, clicking, giving way and locking. The PFPS subjects either had not received any recent treatment prior to testing and were still symptomatic.

The predominant symptoms of PFPS included patellar pain, which was aggravated during weight bearing activities such as running, squatting, kneeling, ascending or descending stairs as well as after prolonged sitting [10]. Physical examination consisted of active and passive range of motion, palpation of the patella and the related structures, quadriceps contraction in supine position, patella provocation test and passive patella movements. Subjects were excluded if the following signs were positive: valgus/varus stress tests, Lachmann's test, anterior and posterior drawer test, pivot shift and McMurray's test. Subjects with traumatic injury to the knee joint or patella, patellar tendonitis (jumper's knee), previous surgery, ligamentous or meniscal disorders, severe knee deformities, such as, genu valgum/genu varum, severe foot deformities, such as, pes cavus and pes planus or hallux valgus were also excluded from the study. The control group consisted of 14 asymptomatic females with no history of congenital or traumatic deformity to their lower limbs. The control group had a mean age, mass and height of 25.07±8.67 years, 61.3±7.55kg and 166.3±6.53cm. Similar exclusion criteria to the PFPS group were used for the asymptomatic subjects. The subjects from both groups were physically active and participated in similar recreational sporting activities for a mean of 3.0h per week for the PFPS and 4.1h for the control group.

Four video cameras (Panasonic WV-CL830/G colour CCTV) were used to record the rearfoot relative to the tibia and tibia segment motion during the stance phase of gait at 50Hz with a shutter speed of 1/2000s. Prior to each test session the data collection area was calibrated and defined using a 16-point calibration object. Direct linear transformation (DLT) was used to obtain 3D coordinate data from multiple 2D views. An error of less than 0.5% of the three-dimensional DLT percent object space calibration was considered acceptable (Peak Performance Technologies, Peak Motus, version 7, user manual, Englewood, USA). Walking took place on a 10m walkway, which had an embedded force platform (1000Hz, Kistler, type 9287, Winterthur, Switzerland) centrally placed which was used to define the stance phase.

Motion of the rearfoot relative to the tibia in the three planes and rotation of the tibia segment in the transverse plane relative to the laboratory coordinate system were investigated by attaching external retro-reflective markers to an individually moulded tibia shell [20] and the calcaneus. With the subject prone and the foot perpendicular to the floor the examiner drew vertical midlines on the posterior distal one third of the leg and on the posterior calcaneus. The vertical line on the posterior lower leg was drawn midway between the medial and lateral borders without relating to the Achilles tendon [8]. The calcaneus and lower leg bisection midlines were determined by using sliding calipers. The bisection line of the calcaneus was used for the marker placement for the walking trials. The rearfoot segment was defined by three 6-mm diameter external markers on the calcaneus: two markers on the line on the posterior aspect of the calcaneus, which bisected the heel in the frontal plane, one marker on the upper ridge (upper heel) and the second on the lower ridge (lower heel) [22]. A third marker was positioned on the lateral aspect of the calcaneus, directly inferior to the lateral malleolus and approximately mid point between the heel markers (lateral calcaneus) [22] (Fig. 1, Fig. 2). All the foot markers were attached directly to the calcaneus during weight bearing (resting standing) [21]. An individual tibia shell (20.5cm×9cm, 0.08kg), made of heated polyform material, similar to Manal et al. [20], was located at the lateral distal one third of the shank length while the subject was sitting with the tibia perpendicular to the floor. Sports tape was placed around the shank over the shell in order to maintain the position of the tibia shell. Four 1.2-cm diameter reflective markers were attached to the shell similar in position to Manal et al. [20]. The first marker was located 30% of the shank length proximal to the lateral malleolus. The three remaining markers were then positioned with 20% of the shank length defined as the vertical and horizontal spacing between the four markers in lateral and anterior positions [20] (Fig. 2). The reference posture for calculation of the stance phase kinematic data was defined when the drawn bisection lines of the calcaneus and the lower leg were vertically aligned during standing. This posture was achieved while the subjects elevated or lowered their medial longitudinal arch following instructions by the examiner. Vertical alignment of the bisection lines was confirmed visually.

View Large Image Download to PowerPoint

Fig. 1. Marker set on the calcaneus and on the tibia shell.

View Large Image Download to PowerPoint

Fig. 2. Posterior view of the marker set on the calcaneus.

Prior to testing, subjects completed a visual analogue scale (VAS) to report pain level prior to testing and in the last 24h before testing. Subjects were allowed to practice walking along the walkway. Following multiple practice trials, five acceptable barefoot walking trials at self selected speed were recorded. In order to maintain natural gait, subjects were instructed not to look down while walking but to maintain visual contact with a marker located at eye level on the far wall. Trials were considered acceptable when the subjects’ right foot landed on the force plate without any disturbance to their gait. Gait velocity was monitored during the walking session by using two light gates (Swift Performance Equipment) placed approximately 6m apart. After each trial the gait velocity was noted.

Peak Motus (version 7) software was used to capture and initially process the kinematic and kinetic data. The 3D trajectories of each marker were optimally filtered using a Quintic spline with cut off determined by the Jackson knee method (Peak Motus, version 7). The angle of the rearfoot relative to the tibia and the tibia relative to the laboratory coordinate system were calculated according to Grood and Suntay [12] and similar to Manal et al. [20] and Soutas-Little et al. [34]. The upper, lower and lateral calcaneus markers were used to develop a rearfoot segment coordinate system including a virtual marker for the ankle joint centre [34]. The virtual ankle joint centre was calculated as a proportion of foot width at the malleoli, and foot length similar in method to that used by Peak Motus. The vertical axis (irearfoot) was determined by upper and lower heel markers. The anterior–posterior axis (jrearfoot) was determined by the lower heel and ankle joint centre markers and the mediolateral axis (krearfoot) was the cross product of irearfoot and jrearfoot.

Tibia markers including lower-anterior, lower-lateral, upper-anterior and upper-lateral were placed on the tibial shell [22] and used to develop a segment coordinate system. The vertical axis (itibia) was determined by ankle joint centre and lower lateral markers. The mediolateral axis (ktibia) was the cross product between itibia and a vector running between lower lateral and lower anterior tibia markers. The anterior–posterior axis (jtibia) was the cross product of ktibia and itibia. Angular motion of the tibia in space was calculated using the segment coordinate system with respect to the laboratory coordinate system with external rotation considered positive.

To develop a joint coordinate system for the rearfoot relative to the tibia motion a floating axis (l jointant) was constructed from the cross product of ktibia and irearfoot.

Plantarflexion/dorsiflexion motion was around ktibia, abduction/adduction occurred around irearfoot and inversion/eversion axis motion was around l jointant. Dorsiflexion, adduction and eversion were all positive.

The GRFs were normalised according to the subject's body weight (GRFBW−1). Using the vertical GRF, the stance phase data was time normalised such that heel strike was 0% and toe off was 100%. The time of peak motions and GRFs for each of the five trials were expressed as a percentage of stance phase. The mean values of the five trials were used for between groups statistical analysis.

3. Results 

A significant difference was found for the mean age (p<0.01) between the groups but no significant difference for mass or height (p=0.10, 0.99, respectively). The VAS for the PFPS group showed mild pain prior to testing (1/10) and in the last 24h before testing (2/10). Velocity for the control and PFPS groups was not significantly different (p=0.28, mean 1.36±0.10 and 1.40±0.08ms−1, respectively). The inter-trial reliability of measuring velocity showed ICC (3, 1)=0.91 and the PCA showed 75% of the values were within 0.09 and 0.08ms−1 of agreement for the control and PFPS groups, respectively.

3.1. Kinematic measurements 

For rearfoot motion relative to the tibia and tibia segment motion, the multivariate test showed no significant differences between the groups (F=0.79, p=0.63, details in Table 1). The multivariate test for the timing of peak angles showed significant difference between the groups (F=5.52, p<0.01). The univariate F tests showed significant differences for time of peak eversion (p=0.02) with mean 39±7% for the control and 46±6.5% for the PFPS (Table 2). A significant difference was also found for the time of peak dorsiflexion (p=0.02) with mean 73±4% and 66±8% for the control and PFPS groups, respectively (Table 2). The inter-trial reliability for the peak angles of the rearfoot relative to the tibia is presented in Table 3.

Table 1. Mean±S.D. of the peak angles of the rearfoot relative to the tibia and tibia segment for the control and the PFPS groups

	Peak angle	Control	PFPS	F value	p value
	Rearfoot motion relative to the tibia
	Eversion	7.1±5.0°	7.5±2.9°	0.06	0.79
	Inversion	8.5±7.3°	9.8±5.6°	0.27	0.60
	Dorsiflexion	3.8±5.6°	3.7±3.9°	<0.01	0.96
	Plantarflexion	7.6±5.0°	8.9±5.6°	0.53	0.47
	Adduction	4.1±2.9°	1.2±4.2°	4.34	0.05
	Abduction	7.4±2.7°	8.5±5.0°	0.51	0.47
	Tibia motion
	External rotation	2.5±6.5°	0.2±7.7°	0.67	0.41
	Internal rotation	11.4±4.4°	12.1±6.2°	0.09	0.75

Table 2. Mean±S.D. of the % time of the stance phase of rearfoot relative to the tibia and tibia segment peak angles for the control and the PFPS groups with univariate F and p values

	Peak angle	Control	PFPS	F value	p value
	Rearfoot motion relative to the tibia
	Eversion	39±7%	46±6.5%	6.04	0.02*
	Inversion	97±2%	97±2%	0.02	0.86
	Dorsiflexion	73±4%	66±8%	6.04	0.02*
	Plantarflexion	9±6%	10±8.3%	0.03	0.85
	Adduction	85±6%	79±8%	4.02	0.05
	Abduction	15±6%	21±8%	3.93	0.05
	Tibia motion
	External rotation	96±7%	94±7%	0.43	0.51
	Internal rotation	16±7%	16±7%	0.001	0.97

* Significant at p≤0.05.

Table 3. Trial to trial reliability for rearfoot relative to the tibia and tibia segment peak angles for the control and PFPS groups

	Motion	Control		PFPS
		ICC	PCA (75%)	ICC	PCA (75%)
	Rearfoot motion relative to the tibia
	Eversion	0.70	3.4°	0.64	2.3°
	Inversion	0.68	5.0°	0.68	3.9°
	Dorsiflexion	0.74	3.8°	0.63	2.9°
	Plantarflexion	0.62	3.8°	0.45	3.3°
	Abduction	0.74	1.7°	0.88	1.8°
	Adduction	0.71	2.1°	0.79	2.3°
	Tibia motion
	Internal rotation	0.82	2.3°	0.92	2.0°
	External rotation	0.95	1.7°	0.89	2.5°

3.2. GRF measurements 

The multivariate test for the GRF peak values showed a significant difference between the groups (F=3.76, p=0.01). The univariate F tests showed significant difference between the groups for peak medial GRF magnitude (p=0.03) with a mean of 0.04±0.01BW for the PFPS group and 0.06±0.01BW for the control group. The vertical GRF minimum trough (p=0.02) and the peak vertical 2 GRF (p=0.01) were significantly different between the groups, with mean 0.71±0.06 and 0.58±0.17BW for minimum peak vertical trough and mean 1.10±0.05 and 0.91±0.24BW for the peak vertical 2 GRF for the control and PFPS groups, respectively (Table 4). No significant difference was found in the multivariate test for the timing of GRF (F=1.03, p=0.44, details in Table 5). Inter-trial reliability for peak GRF for both groups is presented Table 6.

Table 4. Mean±S.D. peak vertical, anterior-posterior and medio-lateral GRF as a proportion of body weight for the control and PFPS groups with univariate F and p values

	Peak GRF	Control	PFPS	F value	p value
	Posterior	0.22±0.0	0.19±0.06	1.63	0.23
	Anterior	0.25±0.02	0.21±0.06	2.52	0.14
	Medial	0.06±0.01	0.04±0.01	4.96	0.03*
	Lateral	0.05±0.02	0.04±0.02	0.42	0.52
	Vertical1	1.07±0.07	0.97±0.29	1.53	0.25
	Minimum trough	0.71±0.06	0.58±0.17	6.72	0.02*
	Vertical 2	1.10±0.05	0.91±0.24	7.66	0.01*

* Significant at p≤0.05.

Table 5. Mean±S.D. time, as a percentage of the stance phase, of peak vertical, anterior–posterior and medio-lateral GRF for the control and PFPS groups

	GRF	Control	PFPS	F value	p value
	Posterior	15±1.1%	14±3.4%	0.84	0.38
	Anterior	87±1.4%	87±2.6%	0.01	0.92
	Medial	26±5.7%	28±7.2%	0.80	0.37
	Lateral	3±0.8%	3±1.1%	0.51	0.47
	Vertical1	22±2.8%	22±2.6%	0.03	0.84
	Minimum trough	48±2.5%	47±2.8%	0.73	0.39
	Vertical 2	77±1.3%	77±2.0%	0.14	0.71

Table 6. Inter-trial reliability for peak GRFs for the control and PFPS groups

		Control		PFPS
	GRF	ICC	PCA (75%)	ICC	PCA (75%)
	Posterior	0.80	0.05	0.91	0.06
	Anterior	0.81	0.03	0.97	0.03
	Medial	0.60	0.01	0.78	0.02
	Lateral	0.84	0.01	0.87	0.02
	Vertical 1	0.74	0.10	0.97	0.13
	Minimum trough	0.74	0.09	0.96	0.08
	Vertical 2	0.70	0.10	0.96	0.09

4. Discussion 

Subjects with PFPS exhibited peak eversion significantly later in the stance phase but no difference in the magnitude of eversion compared to the control group. These findings were similar to Callaghan and Baltzopoulos [2] but in contrast to Powers et al. [29]. However, Powers et al. [29] modelled the foot as a single segment and rearfoot differences in the frontal plane may not have been detected. In contrast to the present study, Powers et al. [29] reported a significant reduction of walking velocity in the PFPS group, which may have influenced foot and lower limb motion [28], [35].

Foot eversion during the stance phase has been suggested to function shock absorption during floor impact [25], [33]. The delay in peak eversion for the PFPS may be an attempt to attenuate the shock during the braking stage at early stance. Differences in the anterior, lateral and first peak vertical forces, however, were not observed for the PFPS subjects. Although not significantly different, the PFPS group demonstrated a trend towards lower mean magnitude of both the anterior and vertical GRFs (Table 4). The standard deviation of the PFPS group, however, was higher compared to the control group, indicating that a high variability between subjects existed in these forces and may have precluded statistical difference.

Subjects with PFPS exhibited earlier peak dorsiflexion. Since dorsiflexion occurred at the second half of stance phase, differences in this motion may be related to altered function of the foot during the whole foot motion of supination [14] as well as abnormal function of the subtalar joint [17], [33]. This in turn may affect the foot from becoming a rigid and efficient lever during the propulsive stage at late stance [17], [33]. Evidence of poor propulsion was seen in the reduced second peak vertical GRFs. Abnormal foot function during walking may create insufficient propulsion, and inability to generate sustained plantar force on the ground [17], [33]. It is also possible that the delay in peak eversion affects rearfoot function during the propulsive stage. This would result in abnormal pronation [33] and inappropriate timing of supination during propulsion [17].

Abnormal timing of foot pronation is thought to disrupt the temporal sequence of the lower extremity joint motion [37]. In the current study, prolonged eversion was not translated into altered internal tibial rotation possibly due to the lack of association between the time of peak foot eversion and tibial internal rotation [31]. The between subjects variability in subtalar joint axis orientation as well as the amount of transverse plane motion between the rearfoot and the tibia may also explain the degree to which eversion can affect the magnitude of tibial transverse rotation [27], [31].

The magnitude of the medial, minimum vertical trough and the second peak vertical GRF were significantly lower in the PFPS group. This may represent a compensation by the non-injured leg to prevent pain, as suggested by Messier et al. [23]. The association between patellofemoral joint pain and foot function during walking in the current study is unclear. Subjects in the current study complained of only mild pain prior to testing and in the last 24h (2/10 VAS) before testing. Furthermore, pain was not evident during walking trials. It is unclear if the altered rearfoot motion and GRF in the PFPS group reflected a gait modification to avoid pain or an inherent causative factor [30]. As patients with PFPS mainly complain of pain during functional activities requiring intense quadriceps activity, such as running, fast walking [1] and ascending stairs [11], the effect of pain at natural walking speed may be limited. The influence of pain should be further investigated during highly demanding tasks, which would involve higher forces through the patellofemoral joint.

Although differences in the mean age were found between groups, it is unlikely that the kinematic and GRF differences found were age related. Changes due to age are thought to occur at approximately 60 years [4], [32]. Age-related disease process, such as osteoarthritis, affect females mainly over 50 years of age [18], [24]. Thereby it is unlikely that the differences found between the groups would be related to the age difference. In addition, although there was no significant difference in weight between the groups, the PFPS group was slightly heavier than the controls. It is unknown, however, what level of mass increase would contribute to PFPS. It is acknowledged that, in PFPS subjects, chondral lesions occur at earlier age than osteoarthritis. However, it is unknown whether these changes are also age related.

The rearfoot kinematics and the GRFs parameters during gait are important in the assessment of subjects with PFPS to identify abnormal patterns. In the current study, PFPS subjects demonstrated altered rearfoot motion and reduced GRF during walking. The aetiology of PFPS is multifactorial in nature and affects both genders. Since the subjects of this study were all female, these findings may not be representative of the entire PFPS population. Based on our study results, however, rearfoot abnormal function may be an important factor that requires further assessment in order to better understand the aetiology of PFPS. We suggest that assessment of PFPS should be made on an individual basis and clinicians should include rearfoot examination during gait as part of the clinical evaluation.

The inter-trial reliability of peak rearfoot motion relative to the tibia in the three planes and tibia segment transverse plane rotation showed moderate to excellent reliability in line with previous reports [3], [5], [9]. In the present study, three-dimensional rearfoot motion relative to the tibia and tibial transverse rotations were similar to those in the literature [5], [15], [19]. The peak GRF showed moderate to excellent trial-to-trial reliability for both groups, as in previous reports [9], [16].

5. Conclusion 

The current study demonstrated no difference between PFPS subjects and healthy individuals with respect to tibial transverse rotation, however, there was a prolonged rearfoot eversion during the stance phase of walking which could affect transfer of loading forces to the knee. Earlier dorsiflexion of the rearfoot relative to the tibia and reduced vertical and medial GRFs may also indicate an altered propulsive function of the foot during the second half of stance phase, which could also affect the knee joint. Thus, the null hypothesis suggesting differences in the vertical and medial GRFs and the timing of rearfoot eversion and dorsiflexion between the groups was rejected.