Patellofemoral Kinematics in Patellofemoral Pain Syndrome: The Influence of Demographic Factors

Abstract

Due to the multifactorial nature of patellofemoral pain, it is often difficult to identify an individual patient’s exact cause of pain. Understanding how demographic variability influences these various factors will support improved consensus in regards to the etiology of PF pain. Thus, in this retrospective study, we tested the hypothesis that sex, height, weight, body mass index (BMI), and age influence determining between-groups differences in PF kinematics. We included 41 skeletally mature patients with patellofemoral pain and 79 healthy controls. Three-dimensional patellofemoral kinematics were quantified from dynamic magnet resonance images. We ran multiple regression analyses to determine the influence of demographic covariates (age, sex, height, weight, and BMI) on patellofemoral kinematics. Patellar shift was significantly influenced by weight (p=0.009) and BMI (p=0.009). Patellar flexion was influenced by height (p=0.020) and weight (p=0.040). Patellar tilt and superior displacement were not influence by demographic variables. Age and sex did not influence kinematics. This study supports the hypothesis that demographic parameters influence PF kinematics. The fact that weight, a modifiable measure, influences both patellar shift and flexion has strong implications for future research and clinical interventions. Clinically, weight loss may have a dual benefit of reducing joint stress and maltracking in patients who are overweight and experiencing patellofemoral pain. The influence of key demographics on patellofemoral kinematics, reinforces the clear need to control for population characteristics in future studies. As such, going forward, improved demographic matching between control and patient cohorts or more advanced statistical techniques that compensate for confounding variables are necessary.

1. Introduction

Patellofemoral (PF) pain is a common complaint in the general adult population, with a prevalence of 23% (Smith et al., 2018). This pain is aggravated by various activities that load the joint (e.g., squatting, running, ascending/descending stairs, etc.), greatly limiting the daily activities of individuals with PF (Crossley et al., 2016). PF pain is considered a complex condition, influenced by several factors such as PF kinematics/alignment, bone shape, whole leg alignment, and soft tissue forces (Drew et al., 2016; Lankhorst et al., 2013; Petersen et al., 2014; Sheehan et al., 2012). Due to the multifactorial nature of this condition, it is often difficult to tease out an individual’s exact cause of pain (Powers et al., 2012). The lack of a more granular insight into the etiology of PF pain places an incredible challenge on healthcare providers as it limits their ability to provide patient-specific treatments.

Understanding how demographic variability influences the various factors associated with PF pain will likely support improved consensus across the field in regards to the role these factors play in the etiology of PF pain, by reducing the inter- and intra-study variability. Our recent review (Grant et al., 2020) evaluating the influence of confounding variables on patellofemoral maltracking was unable to investigate the influence of demographic characteristics (e.g., height, weight, etc.) on patellar maltracking due to the varying study designs and different assessment protocols within and across studies. Of note, sixty percent of the studies within the review (Grant et al., 2020) reported information regarding participant demographics, indicating an awareness across the field that demographics play some role in patellofemoral kinematics. Yet, none of the studies directly controlled for demographic variability.

Clinically, ignoring the potential influence of demographic variability on the etiology of PF pain has hampered our ability to better understand PF pain at the individual level. For example, sex likely plays a role in mediating one’s susceptibility to developing PF pain, as this pain in more prevalent in females prevalence (Boling et al., 2010; Taunton et al., 2002). However, the underlying mechanism(s) for this higher has not been fully explored. Similarly, obesity is associated with knee pathology (Uhorchak et al., 2003; Zhai et al., 2007) and increased cartilage stress (Collins et al., 2018), but its actual role in the development of PF pain is not well understood and the current literature offers conflicting results in terms of the direct relationship between BMI and PF pain. In a recent systematic review, Hart and colleagues (2017) found a higher body mass index (BMI) in individuals with PF pain compared to controls. Conversely, a different systematic review (Neal et al., 2019b), found no significant link between gender, height, weight, or age and PF pain development. In both reviews, the conclusions were drawn from a meta-analysis comparing demographic characteristics between the control and PF pain groups. However, the influence of such characteristics on known mechanisms leading to PF pain has not been investigated. As numerous treatments for PF pain either indirectly (e.g., muscle strengthening) or directly (e.g., surgical intervention) aim at restoring normative PF kinematics (Bayoumi et al., 2021; Gulati et al., 2018), an understanding of how demographic factors influence patellar maltracking will provide clinicians deeper insights for designing interventions.

Thus, the aim of this study is to quantify the potential effect of the demographic covariates on determining between-groups differences in PF kinematics. The specific hypothesis is that gender, height, weight, BMI, and age influence PF kinematics.

2. Methods

As part of an IRB-approved protocol, we conducted a retrospective study of previously recruited patients with idiopathic PF pain and healthy controls. All participants were 18 years or older and skeletally mature (i.e. fused tibiofemoral growth plates). Recruitment ran from 2000 to 2019. Clinicaltrials.gov, flyers, word-of-mouth, and direct contact with local orthopedic clinics were used to advertise the study (Table 1). Participants in the PF pain cohort had pain lasting greater than six months prior to enrollment that started after their 18^th birthday. Patients were included within the database only if an in-house physiatrist diagnosed them with idiopathic PF pain. Seventy-nine individuals with no history of any knee issues, recruited from the greater Washington, DC Area, formed the control group. Regardless of cohort, participants were excluded if they had a clinical diagnosis affecting the knee joint (e.g., ligament, meniscus, iliotibial band, or cartilage damage, other lower extremity injuries, any history of patellar dislocation, arthritis, traumatic knee injury and/or knee surgery). A diagnosis of isolated PF pain was not an exclusion for the PF pain cohort. Any individual who was unable to undergo an MR scan was excluded. Morbid obesity (BMI ≥ 40) was an exclusion criterion. Prior to MR imaging, all subjects provided written consent. An in-house physiatrist or nurse practitioner performed a thorough medical history and physical exam, including a focused knee evaluation. PF pain was assessed with two validated scoring methods: a visual analog scale (VAS) (Price et al., 1983) and anterior knee pain scale (Kujala Score) (Kujala et al., 1993). Data inclusion was limited to a single knee per participant. For control subjects this knee was selected at random. For patients with bilateral knee pain, the knee with the highest numeric VAS score was selected. If the pain was equal bilaterally, then the leg with (in order) the largest clinically measured lateral hypermobility, largest Q-angle, or a J-sign was selected. If all of these markers were equal, the decision was based on the clinician’s history and physical (n=3). The demographic covariates assessed in this study included age, sex, height, weight, and BMI.

Table 1.

Population demographic characteristics

	Full dataset		Dataset Model_HW		Dataset Model_BMI
	Control	PFP	Control	PFP	Control	PFP
N	79	41	72	31	74	31
% of female	64.6	73.2	66.67	80.65	64.86	77.42
Age (years)	27.5 ± 7.7	32.0 ± 9.9 *	26.6 ± 5.7	29.6 ± 6.7 *	26.5 ± 5.7	29.9 ± 6.3 *
Weight (Kg)	65.5 ± 12.2	67.0 ± 12.2	63.7 ± 9.8	63.6 ± 9.7	64.5 ± 11.0	64.0 ± 10.3
Height (cm)	170.0 ± 8.8	167.8 ± 9.0	169.9 ± 7.7	166.2 ± 6.7 *	170.4 ± 8.1	166.9 ± 8.2 *
BMI (kg/cm 2 )	22.6 ± 3.4	23.7 ± 3.2	22.0 ± 2.7	23.0 ± 2.6	22.1 ± 2.8	22.9 ± 2.6
Medial shift (mm)	−0.5 ± 2.2	−1.7 ± 3.9 *	−0.6 ± 2.2	−0.9 ± 2.7	−0.6 ± 2.2	−1.1 ± 3.0
Superior shift (mm)	20.7 ± 5.3	22.0 ± 6.0	20.9 ± 5.2	20.2 ± 5.2	20.9 ± 5.2	20.4 ± 5.5
Flexion (degree)	7.3 ± 4.0	9.0 ± 4.1 *	7.3 ± 4.1	9.3 ± 3.4 *	7.2 ± 4.0	8.9 ± 3.1 *
Medial Tilt (degree)	14.4 ± 6.9	9.0 ± 10.0 **	13.9 ± 6.9	10.6 ± 8.3 *	14.0 ± 6.9	10.4 ± 8.4 *
AKPS	NA	69.4 ± 15.9	NA	70.6 ± 15.9	NA	70.5 ± 15.1
VAS (TD)	NA	33.3 ± 20.9	NA	36.4 ± 21.2	NA	36.4 ± 21.2
VAS (PA)	NA	62.1 ± 22.9	NA	63.3 ± 24.2	NA	63.3 ± 24.2

Abbreviations: BMI: Body Mass Index; AKPS = anterior knee pain scale (Kujala score); VAS (TD) & VAS (PA): Pain during a typical day (TD) and during activities that cause pain (PA), as measured using a visual analog scale (VAS), NA: not applicable.

^{* and **}indicates significant differences (p< 0.05 and p<0.001) between the control and patient cohorts, based on a two-tailed Student’s t-test.

For dynamic scanning, we positioned participants supine within the MR scanner with their knee in a custom-built coil holder (Figure 1). Within the holder, a pair of large-flex coils were positioned medial and lateral to the knee. A pair of medium-flex coils were positioned anteriorly. A triangular-shaped cushioned wedge supported the hip and knee at ~45° of flexion, allowing participants to actively extend/flex their knees from ~45° of flexion to full extension. This range was controlled by the fixed space inside the MR scanner. The participant’s toes touching the top of the MR scanner’s inner bore marked terminal extension and their heel contacting a cushion on the bottom of the inner bore marked maximum flexion. Individuals raised the lower leg at a rate of 30 cycles per minute, as dictated by an auditory metronome. As no additional weight was added, participants were required to raise and lower the weight of their lower leg. Throughout this extension-flexion motion, we acquired 3D tibiofemoral and PF kinematics utilizing a complete set of sagittal plane, cine-phase contrast (CPC) MR images. Twenty-five percent of the data was collected prior to 2008 on a 1.5-Tesla MR imager (GE Medical Systems, Milwaukee, WI). After this date a 3.0T scanner was used (Philips Electronics, Eindhoven, The Netherlands).

Figure 1: MRI Positioning.

For dynamic scanning, participants lay supine within the MR scanner. Their knee was placed within a custom-built coil holder, which stabilized paired of flex coils medial-lateral and anterior to the knee. A cushioned wedge supported the hip and knee at ~45° of flexion, allowing participants to actively extend/flex their knees from ~45° of flexion to full extension.

PF displacement was reported relative to a coordinate system fixed within the femur (Figure 2). Positive displacements were medial, superior, and anterior. Flexion, medial tilt, and varus rotation (superior pole moves laterally) were positive rotations (Sheehan and Mitiguy, 1999). The accuracy of tracking kinematics using cine phase contrast has been reported as 0.33mm (Behnam et al., 2011) for 3.0T and 0.55mm for the 1.5T (Sheehan et al., 1998)).

Figure 2:

Coordinate Systems

Multiple regression analyses were run to determine the relative influence of covariates (BMI, weight, height, age, and gender) on determining between-groups differences in PF kinematics. We focused the analysis on medial shift, medial tilt, superior displacement, and patellar flexion, at a single knee angle of 10⁰ in terminal extension. This knee angle, defined using the tibiofemoral kinematics, approximately equates to a full extension, measured clinically (Freedman and Sheehan, 2013), where maltracking is typically most evident. To limit statistical costs, we chose to eliminate PF anterior displacement and spin from the analysis. No studies have associated PF anterior displacement with PF pain and few studies have associated spin with PF pain (Sheehan et al., 2009; Wilson et al., 2009), whereas numerous studies have demonstrated an association with the remaining four variables and PF pain (Grant et al., 2020). Prior to running the analyses, outliers were statistically removed to ensure all the modelling assumptions were met. Participants were identified as outliers if their data points existed outside of the upper and lower limits of the distribution of each individual variable (less than quartile 1 – 1.5*the interquartile range and greater than quartile 3 + 1.5*the interquartile range). For each outcome of interest, BMI was included alternatively to weight and height in the model to avoid redundancy. Thus, Model_HW included group, age, sex, height, and weight. Whilst, Model_B included group, age, sex, and BMI. Given that our interest was to evaluate the effect of the covariates on determining between-groups differences in PF kinematics, group variable (PF pain vs control cohort) was always kept in the model, regardless of its significance. Significance was set at p < 0.05. All results were reported as 95% confidence intervals (CI) and regression coefficient (β) values. Statistical analyses were performed using R statistical software version 3.6.0 (Team, 2019). The packages used within R were: car (Fox and Weisberg, 2019), lmtest (Zeiles and Hothorn, 2002) and MASS (Venables and Ripley, 2002).

3. Results

We removed 17 outliers from model_HW, leaving 31 patients with PF pain and 72 controls populating the model and 15 outliers from model_B, leaving 31 patients and 74 controls (Table 1). By removing outliers, all modeling assumptions were met.

Comparisons between groups for patellar medial shift (Table 2) were influenced by weight in model_HW (β= −0.06, CI= −0.10 to −0.02, p= 0.090) and by BMI in model_B (β= −0.27 CI= −0.43 to −0.10, p= 0.009). Comparisons between groups for patellar flexion was influenced by height (β= −0.16, CI =−0.29 to −0.02, p = 0.002) and weight (β= 0.10, CI = 0.00 to −0.20, p = 0.004) in model_HW, but not influenced by any covariate in model_B. Between-group comparisons of superior displacement and medial tilt were not influenced by any covariates in either model. Age and sex were not significant covariates in either model. Group had a significant effect for flexion (β= 0.17, p = 0.040) and tilt (β= −3.55, p = 0.026) in model_B and tilt in model_HW (β= −3.20, p = 0.030).

Table 2.

Multiple regression analyses results.

	Model_HW
Kinematic parameter		Regression Coefficient (β)	Std. Error	CI {min, max }	P-value
MEDIAL SHIFT	Intercept	3.4	1.51	{0.38, 6.40}	0.027
	Weight	−0.06	0.23	{−0.10, −0.02}	0.009
	Group	−0.30	0.49	{−1.28, 0.68}	0.542
	Adjusted R2= 0.05				0.027
SUPERIOR POSITION	Intercept	20.85	0.62	{19.63, 22.08}	<2e-16
	Group	−0.63	1.13	{−2.867, 1.60}	0.576
	Adjusted R2= 0.006				0.575
FLEXION	Intercept	27.25	9.58	{8.25, 46.26}	0.005
	Weight	0.10	0.05	{0.002, 0.20}	0.004
	Height	−0.16	0.06	{−0.29, −0.02}	0.002
	Group	1.50	0.85	{−0.21, 3.20}	0.080
	Adjusted R2= 0.084				0.008
TILT	Intercept	13.91	0.86	{12.19, 15.63}	<2e-16
	Group	−3.20	1.58	{−6.43, −0.16}	0.030
	Adjusted R2= 0.03				0.039
	Model_B
Kinematic parameter		Regression Coefficient	Std. Error	CI	P-value
MEDIAL SHIFT	Intercept	5.37	1.88	{1.649, 9.10}	0.005
	BMI	−0.27	0.08	{−0.44, −0.10}	0.001
	Group	−0.25	0.51	{−1.25, 0.76}	0.629
	Adjusted R2= 0.081				0.005
SUPERIOR POSITION	Intercept	20.87	0.62	{19.64, 22.10}	<2e-16
	Group	−0.49		{−2.75, 1.76}	0.666
	Adjusted R2= −0.007				0.660
FLEXION	Intercept	7.23	0.44	{6.36, 8.10}	<2e-16
	Group	1.67	0.81	{0.07, 3.28}	0.041
	Adjusted R2= 0.03				0.040
MEDIAL TILT	Intercept	13.98	0.85	{12.29, 15.679}	<2e-16
	Group	−3.55	1.57	{−6.67, −0.44}	0.0256
	Adjusted R2= 0.038				0.0256

Abbreviations: Std Error: standard error and CI: confidence interval.

4. Discussion

This study fulfilled its primary aim, demonstrating that demographic parameters influence PF kinematics in patients with PF pain and controls. The fact that weight, a modifiable measure, influences both patellar shift and flexion has implications for future research and clinical interventions. In our recent review (Grant et al., 2020), 40% of the included papers did not report the average weight or BMI of their cohorts. When weight or BMI was reported, it was either not matched across cohorts or matched on an average basis. Moving forward, demographics need to be more consistently reported and matching demographics across cohorts should be implemented. An alternative to the latter, is to use larger cohorts and more expansive statistical methods that compensate for the confounding effects of demographic variability. Clinically, based on the influence of BMI and weight on patellar kinematics, comprehensive longitudinal studies are needed to determine if weight loss can reduce PF pain (Li et al., 2019) not only by reducing joint stress, but by improving maltracking.

The influence of weight (and BMI) on PF medial shift, but not on PF medial tilt, found in the current study, is supported by the quadriceps’ lines-of-action (Wilson and Sheehan, 2010), the patellar tendon’s mechanical advantage in terminal extension, and current surgical interventions for maltracking (Ford and Post, 1997; Fulkerson, 2002). During extension, three of the four heads of the quadriceps, along with the patellar tendon, exert a lateral force on the patella (Wilson and Sheehan, 2010). The increased demand to lift a heavier leg creates increased lateral force on the patella, resulting in increased lateral shift in terminal extension. As the quadriceps acts primarily to exert an extensor torque on the patella (Wilson and Sheehan, 2010), it would appear opposite to the current findings that increased weight increases flexion. However, in terminal extension, the patellar tendon’s mechanical advantage over the quadriceps’ tendon (Thomeer et al., 2017; Yamaguchi and Zajac, 1989) fosters increased flexion torque exerted on the patella when a greater demand is placed on the quadriceps (e.g., raising a heavier leg). The lack of influence that weight has on tilt fits with the clinical view that tilt is more strongly influenced by passive forces (Ford and Post, 1997; Fulkerson, 2002). For example, after conservative treatment has failed, a lateral release to change passive constrains is recommended for correcting patellar tilt, but a tibial osteotomy (re-alignment of the quadriceps-patellar tendon force) is recommended for correcting excessive lateral shift. Lastly, a previous clinical study documented that patients who were overweight and experiencing PF pain, demonstrated improved tibiofemoral kinematics with weight loss, accompanied by reduced knee pain (Li et al., 2019). This tangentially supports the relationships found between both weight and BMI with PF kinematics.

This study intentionally focused on a single knee angle (10 deg) in active terminal extension, as maltracking is typically most evident during terminal extension with active quadriceps (Grant et al., 2020). In some individuals maltracking is likely a direct cause of PF pain due to the excessive cartilage sheer in terminal extension and the cartilage stress induced when the patella re-engages in the groove. For others, the maltracking measured in terminal extension is indicative of excessive contact loads in mid- to deep flexion. Specifically, once the patella re-engages within the groove, the soft tissue imbalance, leading to maltracking in terminal extension, is compensated for with excessive patellofemoral contact force. This contact force establishes more normative kinematics, but likely leads to pain. Thus, maltracking in terminal extension is likely indicative of higher contact forces in mid- to deep-flexion. Further study exploring this relationship is needed.

The relationship between participant height and PF extension can be explained by a rotation of the quadriceps and the patellar tendon’s line-of-action with height (Figure 3). A recent study (Choudhary et al., 2019) demonstrated that as the tibia and femur lengthen with height, the pelvis does not widen to the same extent, leading to a reduction in the Q-angle in taller individuals. If the same concept holds true in the sagittal plane, increased height would result in a more inferiorly oriented patellar tendon and superiorly oriented quadriceps tendon, promoting a reduction in the patellar tendon’s flexion and the quadriceps tendon’s extension moment arm, relative to the patellar center. As the quadriceps tendon is oriented primarily in the superior direction, with only a minor lateral-posterior component, this rotation towards the superior-inferior axis would be greater for the patellar tendon. This disparate change in orientation promotes a greater decrease in the patellar tendon’s flexor moment, resulting in an increase in the extensor moment acting on the patella with increasing height. Hence, with increasing height, increased patellar extension is seen. This rotation of the quadriceps line-of-action with height is supported by the relationship between height and tibiofemoral extensor torque capacity (Gross et al., 1989; Neder et al., 1999; Pereira et al., 2019). Both Gross et al. and Neder et al. reported that 50% of the maximum tibiofemoral extensor torque is predicted by an individual’s height. Following the above logic, with increased height, the quadriceps orients more superiorly, increasing the superior force it transfers to the patellar tendon, resulting in increased tibial extensor moment. The relationship between increased weight and patellar flexion is opposite to the relationship between height and patellar extension, explaining the lack of a relationship between BMI and flexion.

Figure 3: Force direction of quadriceps and patellar tendons (QT & PT).

Based on the MRI of one study participant of average height and weight, the unit vectors describing the force direction of the QT and PT were proportionally drawn (left is original dimensions and right is an increase in height by 30%). The original line of action (LOA) for the QT and PT are shown with solid blue lines (unit vector proportions shown on left). Assuming the superior component increases by 30%, without any increase in the posterior direction, the quadriceps and patellar tendon (dashed black line unit vector proportions shown on right) rotate 2⁰ and 5⁰ towards the inferior-superior axis. Thus, the QT posteior force reduces from 14% to 11% of its superior force, but patellar tendon posterior force reduces from 41% to 31% of its superior force. As the superior component of the QT and PT crosses through the patellar center, only the posterior component of the QT and PT produces an extensor and flexor moment, respectively, on the patella about its center. Thus, with increasing height, the overall extension moment on the patella increases, creating a positive influence of height on patellar extension.

The lack of influence of sex on PF kinematics agrees with previous PF studies (Seisler and Sheehan, 2007; Varadarajan et al., 2010). These studies, focused solely on controls, used fairly small populations (female and male cohort sizes ranged from 18–20 and 13–14). Studies evaluating whole limb kinematics and kinetics during running (Neal et al., 2019a) and single leg squats (Nakagawa et al., 2012; Willy et al., 2012) did evaluate both controls and patients with PF pain, but few distinctions between males and females were found. A common finding was that females had greater tibiofemoral abduction than males. For the current study, the higher prevalence of PF pain among females (Taunton et al., 2002) led to a smaller cohort of men (~25%) being included within the study, which may have led to underestimating statistical power for evaluating this covariate. Thus, future studies expanding the representation of males in the both controls and patients with PF pain is warranted to test the true influence of sex on PF maltracking. Nonetheless, a lack of influence of sex should not be evidence that men and women have identical PF pain etiologies, as maltracking is likely just one of the various etiological factors that lead to PF pain (Selfe et al., 2016). A combination of factors may lead an individual to pain (Selfe et al., 2016) and this combination may be unique among males and females. Lastly, other sex-related factors might be intricately associated with PF pain through the interaction of biomechanical (Selfe et al., 2016) or biopsychosocial factors (Lack et al., 2018).

Both age (Johnson and Hunter, 2014) and abnormal PF kinematics (Tsavalas et al., 2012) are associated with osteoarthritis (OA); however, our findings demonstrate that PF maltracking is not influenced by age. This suggests that pathological kinematic profiles are likely present prior to the development of OA, echoing recent evidence implicating PF pain as a precursor to OA (Eijkenboom et al., 2018). One question that was not addressed in the current study, and merits further investigation, focusses on the other side of the age spectrum: do PF kinematics change during bone development?

The primary limitation of this study is the reduction in cohort size, resulting from the statistical withdrawal of outliers to ensure compliance with the modeling assumptions. Even after the statistical removal of outliers, this study still represent one of the largest cohort sizes for the study of PF pain maltracking (Grant et al., 2020). This adherence to the modeling assumptions removed patients with the most extreme lateral shift and tilt maltracking. Thus, the lack of influence of cohort on patellar shift profiles reported herein must be viewed with caution. Interestingly, even with the removal of the patients with extreme lateral tilt, cohort did influence tilt in both models. A secondary limitation was its cross-sectional design which prohibits us from assuming that changes in weight (as the only modifiable characteristic evaluated) will affect patellar kinematics at an individual level. Longitudinal studies investigating the association between changes in the individuals’ baseline characteristics with the development of PF pain are needed. In addition, future studies should evaluate certain population characteristics not investigated in this study (e.g., level of physical activity, age-of-pain-onset, and lower limb strength) and their influence on PF kinematics. As the accuracy and precision of dynamic scanning was similar on both platforms (Behnam et al., 2011; Sheehan et al., 1998) and there was no systematic bias in how the scanners were used, it is unlikely that the use of two different platforms affected the results.

In conclusion, our results show that demographic characteristics (height, weight, and BMI) do influence PF kinematics across control and patient populations, but such evidence was not found for sex and age. As such, this influence must be accounted for in future research studies. Further longitudinal work is needed to evaluate if weight loss results in improved PF kinematics and reduced pain in individual patients with PF pain, as weight loss has previously been shown to improve tibiofemoral kinematics and reduce PF pain (Li et al., 2019). While sex does not appear to influence PF kinematics, previous studies have clearly demonstrated that sex likely influences the development of PF pain. Therefore, sex should be considered in future studies of PF pain. Ultimately, going forward, more expansive matching between cohorts and/or advanced statistical techniques must be used to adjust for individual characteristics when investigating the etiology of and interventions for PF pain.

Data Sharing Statement

All data relevant to the study are included in the article or uploaded as supplementary information.