Introduction

Patellofemoral pain (PFP) is a common lower limb musculoskeletal overuse injury among runners. The prevalence of PFP varies by population and may be as high as 20%–25% in active populations [1, 2]. PFP is characterized by diffuse anterior knee pain and usually associated with activities, such as walking, running, squatting, and stair ascent and descent [3]. It is often persistent and could continue for many years and eventually develop into irreversible patellofemoral osteoarthritis, which in turn negatively influences the quality of life [4].

Although the exact biomechanical pathogenesis of PFP is still not well understood, the possible mechanism underlying the development of PFP is abnormal loading of patellofemoral joint and may be influenced by kinematic factors during gait [4]. Greater knee valgus in the frontal plane during running of individuals with PFP [5, 6] contributes to increased lateral components of patellofemoral joint loading and to pain [7]. Despite claims that abnormal knee flexion is linked to increased patellofemoral joint stress, existing literature remains controversial. Most studies have reported that runners with PFP have increased peak knee flexion compared with asymptomatic runners [8], but other works failed to prove it [6]. Similarly, knee motion in the transverse plane is not consistent across studies; several existing studies showed that knee internal rotation increased in runners with PFP compared with healthy runners [9], whereas others did not [10]. In addition, rearfoot valgus may be one of the biomechanical risk factors causing PFP [11]. The joint motion changes in patients with PFP in different motion planes remain controversial. Runners with PFP may strategize their running strategies to reduce patellofemoral joint stress, release pain, and improve function. Compared with the preferred running cadence, the cumulative patellofemoral load impulse and peak knee flexion decreased by 5.5% and 3.3 degree (7.0%), respectively, in stance phase at 110% step cadence among healthy individuals [12]. Furthermore, Bonacci found that increased running cadence or minimalist footwear have greater variability, which may help distribute loads for patients with PFP [13]. Transitioning from a rearfoot to forefoot strike pattern increased the Lower Extremity Function Scale score and led to greater gastrocnemius and rectus femoris pre-activation during the late swing phase, thereby benefitting runners with PFP [14, 15]. However, in these studies, the effect of a single running strategy change on patients with PFP was compared; research evaluated discrete variables from the dynamic perspective of force and moment.

From the point of view of dynamical system theory, human movement requires the complex organization of multiple degrees of freedom to form coordinated action for coping with external interference or different task conditions. Vector coding technique is a widely used method to quantitatively analyze coordinated action from joint coordination and coordination variability to provide insights into spatial variation between joints with time series and organized flexibility between joints during different motions [16]. The spatial variation information between joints is represented by coordination pattern, which may be dysfunctional if the movement does not conform to anatomical linkages between joints (mechanically unsound) [17]. Mechanically unsound coordination relationship between knee and ankle can increase the risk of joint exposure to abnormal stress and could be related to PFP [18]. In addition, based on variability–overuse injury hypothesis, a lower coordination variability may lead to repeated loads on specific tissues and is more likely to cause overuse injury [19]. Heiderscheit [20] found no differences in coordination variability between PFP and healthy runners when calculating the average across the entire running gait cycle at a self-selected running speed. Conversely, Cunningham indicated that greater coordination variability may be a characteristic of female runners with PFP [19]. Hence, a dynamic system approach is needed to comprehensively determine the coordination relationship and variability between the joints of lower limbs in different planes while running in different combinations of strategies among runners with PFP.

This study aimed to determine the joint coordinative patterns and coordination variability when changing the running cadence and strike pattern in runners with PFP. The authors hypothesized that runners with PFP could display more mechanically sound coordinative patterns and a greater CAV with increasing running cadence and altered strike pattern to forefoot strike pattern.

Methods

Date processing

Joint kinematics were extracted for the ankle, knee and hip. Kinematic data were filtered with a low-pass fourth-order Butterworth filter at a cut-off frequency of 10 Hz [26]. Ankle, knee and hip joint rotations were calculated on the basis of Euler rotation methods (X: flexion/extension, Y: adduction/abduction, Z: internal/external rotation). Vertical ground reaction forces were sample at 1000Hz with a threshold of 20N to detect gait cycle events of the stance phase [27]. Stance phase was time normalised from initial contact (0%) to toe off (100%). Stance was divided into three portions: initial stance (0%–33%), mid stance (34%–66%) and late stance (67%–100%) [18]. Coupling angle (γ) and coupling angle variability (CAV) of hip sagittal versus knee frontal (HsKf), hip sagittal versus knee sagittal (HsKs) and knee transverse versus ankle frontal (KtAf) couples were calculated using circular statistics known as modified vector coding technique [28]. Modified vector coding analysis was performed using a custom-written MatLab code (MatLab 2016a, MathWarks, Inc., Natick, MA).

The coupling angles were calculated on the basis of proximal joint angles (θ_P) and distal joint angles (θ_D) by using the following formula to present values between 0° and 360°: (1) where i is a percent running cycle of the jth trial. And as suggested by Needham, the coupling angles were classified into eight coordination patterns on the basis of coupling angle magnitude [28] (Table 1). The distribution frequency of eight coordination patterns were quantified to reveal the dominant coordination pattern for each coupling pair during the stance phase. Coordination patterns were qualitatively interpreted as mechanically sound or unsound according to the anatomical motion pattern and polarity defined by the right hand rule. Mechanically sound coordination patterns indicated that hip, knee and ankle move as they are anatomically linked [17]. Therefore, during the running stance phase, mechanically sound motion included:

hip extension (EXT)–knee eversion (EV), hip flexion (FLX)—knee inversion (INV);

hip extension (EXT)—knee flexion (FLX), hip flexion (FLX)—knee extension (EXT);

knee internal rotation (IR)—ankle eversion (EV), knee external rotation (ER)—ankle inversion (INV).

Any opposite coordinative pattern was defined as mechanically unsound.

CAV was calculated in accordance with Eqs (2–5) as follows: (2) (3) (4) (5)

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Classification of coordination patterns.

https://doi.org/10.1371/journal.pone.0280477.t001

Results

Coordination patterns

For HsKf couples, coordination patterns were concentrated in in-phase hip EXT-knee EV (mechanically sound) and anti-phase hip EXT-knee INV (mechanically unsound) for FFS and RFS. Specifically, FFS displayed more frequent anti-phase hip EXT-knee EV (mechanically unsound) coordination pattern than RFS under slow10% (p = 0.005, ES = 1.61), normal (p < 0.001, ES = 1.34), and fast10% (p = 0.001, ES = 0.99) running cadence conditions (Fig 1) during the initial stance phase. During the mid and late stance phases, RFS showed more anti-phase hip EXT-knee INV (mechanically unsound) (p = 0.008, ES = 0.67) than FFS. In addition, running cadence exerted no significant effect on the distribution of coordination pattern among FFS runners. For RFS, the influence of cadence on the distribution was detected in anti-phase (mechanically unsound) hip dominancy (p = 0.042, ES = 0.65) and knee dominancy (p = 0.05, ES = 0.70) coordination patterns during the initial stance phase (Fig 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Mean coupling angle in RFS and FFS conditions.

Mean coupling angle for all subjects across stance phase for hip sagittal versus knee frontal (a), hip sagittal versus knee sagittal (b) and knee transverse versus ankle frontal (c) couples in RFS (solid line) and FFS (dash line) during slow10% (left column), normal (middle column) and fast10% (right column) running cadence. Each figure is divided into three parts, which represent the distribution proportion of RFS (slash) and FFS (white) in each coordination pattern during substance phases. * indicates a significant difference between groups (p < 0.05).

https://doi.org/10.1371/journal.pone.0280477.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Mean coupling angle during slow 10%, normal and fast 10% running cadence.

Mean coupling angle for all subjects across stance phase for hip sagittal versus knee frontal (a), hip sagittal versus knee sagittal (b) and knee transverse versus ankle frontal (c) couples in RFS (left column) and FFS (right column) during slow10% (grey line), normal (dash line) and fast10% (black line) running cadence. Each figure is divided into three parts, which represent the distribution proportion of slow10% (white), normal (grey) and fast10% (slash) in each coordination pattern during initial (left column), mid (middle column) and late (left column) stance phases. * indicates a significant difference between groups (p < 0.05).

https://doi.org/10.1371/journal.pone.0280477.g002

For HsKs couples, RFS displayed a large proportional anti-phase hip FLX-knee EXT (mechanically sound), while FFS displayed a large proportional in-phase hip EXT-knee EXT (mechanically unsound) under slow10% (p = 0.003, ES = 1.01), normal (p = 0.001, ES = 0.82), and fast10% (p = 0.001, ES = 0.75) running cadence conditions during the initial stance phase (Fig 1). Hip dominance anti-phase hip EXT-knee FLX (mechanically sound) was the most important coordination pattern for two strike patterns during the mid and late stance phases. However, as the running cadence increased, the coordination pattern gradually transferred from knee dominancy to hip dominancy during the late stance phase for FFS.

For KtAf couples, FFS showed a large proportion in ankle dominance anti-phase knee dominancy knee ER-ankle INV (mechanically sound) under slow (p = 0.012, ES = 0.75) and fast10% running cadence (p = 0.009, ES = 0.77) conditions in the mid stance phase. FFS also showed a large proportion in knee dominance anti-phase knee ER-ankle INV (mechanically sound) under normal (p = 0.008, ES = 0.92) and fast10% running cadence (p = 0.01, ES = 0.79) conditions in the late stance phase (Fig 1). No effect was observed on the distribution of coordination patterns under the three running cadence conditions for RFS and transfer to FFS (Fig 2).

Coordination variability

Modified vector coding analysis revealed no differences in coordination variability across three portions of stance phases in HsKf, HsKs and KtAf couples under six conditions (Table 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. M (P25 and P75) for coordination variability of selected couples at six running conditions during three stance subphases.

https://doi.org/10.1371/journal.pone.0280477.t002

Discussion and implications

This study aimed to determine whether alternating running cadence and strike patterns immediately affect joint coordination patterns and variability during the stance phase in runners with PFP. The results showed that the proportion of each coordination pattern throughout the running gait stance phase was changed in the six running strategies. However, changing strike pattern and running cadence did not show the same trend (mechanically sound/unsound) in the change in the distribution frequency of the coordination pattern, which contradicted our research hypothesis.

On the basis of the modified vector coding technique, the distribution frequency statistics indicated whether a coordination pattern performed a greater proportion of stance phase under different running strategies. For HsKf couples, the results highlighted the important role of the initial 1%–33% period of the stance phase among the six running strategies. Similarly, a previous research revealed that the most important kinematic change by transitioning from RFS to FFS was found in the knee frontal plane movement at initial contact in patients with PFP [14]. Switching to FFS and increasing the running cadence increased the distribution frequency in anti-phase hip EXT-knee INV (mechanically unsound) coordination pattern. This finding is similar to the results of previous research, revealing that the most important kinematic change in the gait retraining of patients with PFP was the change in knee frontal plane movement at initial contact [14]. Considering hip and knee motion during early stance, the knee continues inversion for a period of time and eversion, while the hip remains in extension. Thus, the mechanically unsound coordination pattern is the result of knee inversion after initial contact. The distribution frequency in the in-phase (mechanically sound) coordination pattern for HsKf couples was increased by altered strike pattern to FFS during the mid and late stance phases. From this point of view, we may not be able to simply explain which running cadence is better than the other. The frequency and patterns of couples may have a critical threshold [16]. However, whether the increase in running cadence has no effect on mechanically sound coordination patterns in either RFS or immediately switched to FFS remains unknown.

In this study, the anti-phase coordination pattern indicated mechanically sound motion between hip and knee in the sagittal plane during the stance phase. Proximal or distal dominancy stand for the joint rotating at a faster rate [29]. Under the condition of slow 10% and normal running cadence, simply switching the strike pattern to FFS will increase the distribution frequency in knee-dominated coordination pattern during the mid and late stance phases. As such, switching to FFS will lead to a faster rotating rate of knee flexion than hip extension, indicating that more load was transferred to the knee joint and less to the hip joint [26]. Notably, combining FFS and fast 10% cadence resulted in a change from knee dominancy to hip dominancy and tended to be RFS, a mechanically sound coordination pattern. Hence, hip dominancy coordination pattern may be indicative of another protective strategy in the PFP group in the propulsion phase. Overall, RFS or the combination of FFS and increased 10% cadence showed a greater distribution in hip dominancy coordination pattern, which may have implications for natural FFS runners with PFP. However, a healthy control group and natural FFS runners were not used for comparison, which is a limitation of the present research. Future research should focus on whether a relative angular velocity might produce potential information for the mechanism of runners with PFP.

From the anatomical point of view, knee external rotation-ankle inversion was defined as mechanically sound. No changes were found in the coordination patterns when slow10% or fast10% running cadence for RFS and transfer to FFS conditions. FFS performed a partial transformation from ankle dominancy to knee dominancy, while RFS remained in ankle dominancy during the last two stance phases. As previously discussed, knee dominancy coordination pattern would increase the demand on knee external rotation while simultaneously decreasing the demand on ankle inversion, thereby benefitting runners with PFP [26], despite that knee ER-ankle INV was the mechanically sound coordination pattern.

Inconsistent with the hypothesis, coordination variability did not appear to change with changing running cadence and strike pattern. Although no statistical difference was found in the variability of each condition, the variability tended to decrease with the acceleration of running cadence (Table 1). Our results did not corroborate the finding that 6 weeks of increased cadence gait retraining program increased the joint kinematic and kinetic variability [13]; notably, different statistical methods were used in our study. Care should be taken when using vector coding technique to determine the effect of long-term gait retraining from the perspective of variability.

Several limitations should be noted in this study. First, the treadmill condition could be considered a limitation. Treadmill was used to control the running speed to develop the relationship between gait retraining and joint coordination. Second, only the coordination patterns and variability of three couples were analyzed because they have strong kinematic coupling when running and they are related to PFP. Third, the lack of a health control group led to insufficient explanation on whether the coordination pattern was beneficial for PFP. The primary objective of the present study was to investigate whether changes in running cadence and strike pattern can alter the coordination pattern and variability of PFP. Additionally, the 20 participants included had VAS ranging from 3 to 7, which may affect the results. Future studies could increase the sample size and conduct subgroup analysis based on VAS value.

Conclusions

Our findings suggest that immediately changing the running cadence (± 10%) and transferring strike pattern from RFS to FFS did not change the distribution frequency in mechanically sound coordination patterns and did not affect the coordination variability of runners with PFP. RFS exhibited more hip- and ankle-dominated mechanically sound coordination patterns when changing the running cadence in a short-term, which may be a protective strategy for natural RFS runners with PFP. Furthermore, additional studies are necessary to assess the long-term combination gait retraining effect on runners with PFP.

Acknowledgments

The authors are thankful to all the participants’ voluntary fabution, and complete this study until the end.