The results of our study provide support for our hypothesis, as we demonstrated that hip adduction and pelvic tilt toward the swing limb are important kinematic factors for HAM impulse during gait. The body weight and stance phase duration may explain 61% of the variance in HAM impulse. Nevertheless, this study showed significant contributions by hip and pelvic kinematics during gait to the HAM impulse. Although these findings are theoretically understandable [18] and the relationship between peak HAM and kinematics of the pelvis and trunk have been reported in patients with gluteal tendinopathy [19], it is clinically significant that this study has, for the first time, explained the variance in HAM impulse by noting the variation in the gait in actual patients with hip OA. We have chosen to include patients with acetabular dysplasia and hip OA because in a context where the focus of treatment for OA has shifted from palliation to prevention [20], it is reasonable to include patients with acetabular dysplasia as it is regarded as a pre-osteoarthritis condition in the course of hip OA progression.
Hip adduction angle had the greatest influence on HAM impulse among gait kinematics, explaining 9.4% of the variance in HAM impulse. The association between hip adduction angle during gait and HAM impulse (adjusted R2: 0.700) is comparable to, or slightly stronger than, the bivariable association between the knee adduction angle and peak external knee adduction moment (KAM) during gait in healthy individuals (adjusted R2: 0.566) [6], asymptomatic individuals with normally or varus-aligned knees (adjusted R2: 0.467) [21], and patients with knee OA (adjusted R2: 0.489) [22]. As a motion pattern, most of the patients’ hip joints showed a pattern of movement in the adduction direction during the loading response after initial contact, in the same manner as that in healthy individuals [23]. However, there were many inter-individual variations in the hip joint position of hip abduction/adduction during the stance phase (Fig. 1). Relatively medially shifted contact of the lower limb with the floor during hip adduction can displace the origin of the GRF medially, relative to the center of the hip joint. Thus, larger hip adduction would increase the HAM impulse by increasing the lever arm between the hip joint center and GRF vector. In muscle-driven gait simulations, an increase in hip adduction was shown to be the most influential kinematic affecting the increase in peak HAM and the hip contact force [18]. In gait analysis in the context of hip joint overloading, excessive hip adduction should be given substantial attention.
We observed that pelvic tilt toward the swing limb was also a factor related to the larger HAM impulse. The excessive pelvic tilt toward the swing limb during gait is widely known as Trendelenburg gait. The Trendelenburg gait has been confirmed not only in patients with hip OA but also in patients with gluteal tendinopathy [19]. Allison et al. [19] reported that the contralateral pelvic drop at each time point were related to the variation of the first and second peaks of HAM which were greater in patients with gluteal tendinopathy than in the controls. The mean pelvic tilt angle of patients with hip OA was 3.2°, which was not vastly different from that of healthy individuals (approximately 4°, the value of healthy individuals at almost the same gait speed as in the patients in our study) [24]. However, some patients showed excessive pelvic tilt toward the swing limb with a maximum angle of 7.3°. The pelvic tilt toward the swing limb can displace the mass of the upper body to the swing side even if it is not accompanied by a trunk lean. Consequently, the center of mass of the body is displaced to the swing side, and the GRF vector is also displaced away from the hip joint. The larger HAM impulse would attribute to the change in the direction of the GRF vector associated with pelvic tilt. Given that the excessive pelvic tilt toward the swing limb is commonly observed in patients with hip disease, pelvic tilt toward the swing limb is a key point in clinical gait observation and modification in the sense that it can lead to an increase in the HAM impulse.
On the other hand, trunk lean toward the stance limb was associated with smaller HAM impulse, although its contribution was as small as 3.2%. We hypothesized that trunk lean toward the swing limb is associated with larger HAM impulse. Nevertheless, there were only a few patients in this study who showed trunk lean toward the swing limb (Fig. 1). Trunk lean toward the stance limb moves the GRF vector closer to the hip joint by displacing the center of mass of the body to the stance limb. Consequently, HAM impulse can be reduced. In patients with knee OA, trunk lean toward the stance limb has a significant effect on the reduction in KAM [6, 25]. Trunk lean is used as one of the gait modification strategies for reducing KAM [26]. In patients with hip OA, trunk lean toward the affected side tends to increase during gait [9, 27, 28], especially in patients with hip abductor muscle weakness that show significant trunk lean [29]. Trunk lean may be an adaptive strategy to reduce hip loading and pain by decreasing HAM impulse However, although it was observed only in a few cases, HAM impulse was generally large in patients whose trunk leaned toward the swing limb during the stance phase (Fig. 2). For such rare patients, clinicians need to carefully observe trunk lean toward the swing limb as a potential factor affecting HAM impulse.
Interestingly, hip pain was not a factor in explaining the variance in the HAM impulse. The results are inconsistent regarding the relationship between hip pain and kinetic variables in the hip joint during gait. Hurwitz et al. [30] reported that an increased level of hip pain correlated with decreased hip extension moment. Conversely, Zeni et al. [29] have confirmed that there were no gait parameters, including hip moment, associated with hip pain. Moreover, previous prospective cohort studies have shown that the HAM impulse is associated with radiographical (i.e., structural) hip OA progression but not with worsening of hip pain [2, 31]. The relationship between hip pain and gait biomechanics may be modified by various factors such as the condition of joint structures and difference in compensatory strategies for gait using parts other than the hip joint such as the trunk; therefore, hip pain might not be directly related to the HAM impulse. Furthermore, in the present study, the inclusion of younger patients with little hip pain and disability may account for the lack of correlation between hip pain and the HAM impulse.
It is notable that the magnitude of HAM is not necessarily larger in patients with hip OA compared with healthy individuals. Previous studies have shown that HAM is not significantly different between the patients with hip OA and healthy individuals [9, 32], or rather HAM of patients with hip OA is smaller than that of the healthy individuals [30, 33]. However, joint degeneration in patients with secondary hip OA who have morphologic abnormalities such as dysplasia may be adversely affected even if the magnitude of hip loading is equal to or inferior to that of healthy individuals. This is because patients with hip OA generally have a decreased cartilage contact area than do healthy individuals [34], as well as damaged articular cartilage and labrum [35]. Indeed, in the secondary hip OA group, a higher cumulative hip loading related to a large HAM impulse is a risk factor for narrowing of the hip joint space [2]. Thus, identifying the gait parameters related to the increase in HAM impulse is critical to establish gait modification training aimed at preventing hip OA progression.
This study has several limitations. Patients with terminal-stage hip OA were excluded from this study in order to render the findings applicable to the prevention of hip OA progression. However, gait kinematics and kinetics are different between patients with mild-to-moderate hip OA and patients with severe hip OA [36]. Therefore, the results of this study may not be generalizable to patients with terminal-stage hip OA. It is difficult to elucidate causal relationships as this study was a cross-sectional study. Thus, it remains to be firmly established whether HAM impulse during gait in daily life is reduced by improving hip joint adduction and pelvic tilt with gait modification. On the basis of a prospective cohort study showing an association between hip OA progression and gait biomechanics [2], HAM impulse was adopted as a dependent variable. However, other variables such as peak or mean of HAM during the stance phase may also be important factors associated with hip joint loading. Additionally, HAM is an indirect measure of hip joint loading, although HAM is strongly correlated to hip contact force [18]. It is necessary in the future to analyze hip joint force in the model including patient-specific bone morphology, tension of the muscles and ligaments, and muscle recruitment patterns.