Sex-Based Differences and Asymmetry in Hip Kinematics During Unilateral Extension From Deep Hip Flexion

Abstract

The purpose of this study was to identify side-to-side and sex-based differences in hip kinematics during a unilateral step-up from deep flexion. Twelve (eight men, four women) asymptomatic young adults performed a step ascent motion while synchronized biplane radiographs of the hip were collected at 50 images per second. Femur and pelvis position were determined using a validated volumetric model-based tracking technique that matched digitally reconstructed radiographs created from subject-specific computed tomography (CT) bone models to each pair of synchronized radiographs. Hip kinematics and side-to-side differences were calculated and a linear mixed effects model evaluated sex-based differences. Women were on average 10.2 deg more abducted and 0.2 mm more medially translated than men across the step up motion (p < 0.001). Asymmetry between hips was up to 14.1 ± 12.1 deg in internal rotation and 1.3 ± 1.4 mm in translation. This dataset demonstrates the inherent asymmetry during movements involving unilateral hip extension from deep flexion and may be used provide context for observed kinematics differences following surgery or rehabilitation. Previously reported kinematic differences between total hip arthroplasty and contralateral hips may be well within the natural side-to-side differences that exist in asymptomatic native hips.

Introduction

Hip pathologies such as femoroacetabular impingement (FAI) or developmental dysplasia of the hip (DDH) cause movement-based deficiencies, pain, and contribute to long-term degenerative joint changes [1–3]. Rehabilitation programs often focus on strengthening and increasing hip stability in both unilateral and bilateral postures with the goal of minimizing asymmetry in strength, kinematics, and range of motion between the affected and contralateral hip [4]. However, knowledge of the expected kinematic asymmetry in healthy individuals is lacking, especially during positions of deep hip flexion that challenge the stability of the hip and may be associated with pain and dysfunction in individuals with pathology.

Previous studies have explored the relationship between hip pathology/surgical intervention and kinematics during squatting [3,5–7] and while performing a step up motion [8–11] utilizing skin-mounted marker-based motion capture techniques. However, kinematics derived from conventional motion capture are affected by soft tissue artifact which can be up to 21.8 deg in rotation and 5.4 cm in translation during low flexion activities [12,13]. Dual fluoroscopy and biplane radiography are highly accurate techniques (bias and precision of 0.2 deg and 0.8 deg in rotation and 0.2 mm and 0.3 mm in translation, respectively) [14] that have been utilized to examine in vivo hip kinematics, however, those studies focused only on differences in kinematics between operated or pathological hips and healthy hips [15–17]. Therefore, the inherent asymmetry in asymptomatic hip kinematics remains unknown. Filling this knowledge gap will provide context for previous and future studies that assess side-to-side differences in hip kinematics due to surgery or pathology.

Clinical studies that indicate men have a higher prevalence of cam FAI while women have higher prevalence of pincer-type FAI [18], and studies that report the prevalence of hip osteoarthritis (OA) is higher in women compared to men [19] suggest that the etiology of hip pathology may differ between men and women. For this reason, sex-based kinematic differences and kinematics asymmetry during treadmill walking, bodyweight squatting [20], and during a static weight-bearing apprehension position [21] have been measured. However, it is unknown if those sex-based differences persist during unilateral extension from deep hip flexion (i.e., a step ascent motion). It is important to characterize hip kinematics during deep hip flexion–extension movements because impingement-related pain often occurs during functional positions where the hip is flexed at and beyond 90 deg, potentially introducing kinematic abnormalities [22,23]. Determining native asymmetry in hip kinematics in asymptomatic individuals performing movements requiring unilateral extension from deep hip flexion, such as ascending a high step, will provide a valuable reference database for clinicians to compare to populations with hip pathology or to evaluate function following surgical intervention.

The purpose of this study was to identify side-to-side and sex-based differences in hip kinematics during a unilateral extension from deep flexion (step ascent) in a cohort of asymptomatic young adults utilizing dynamic biplane radiography.

Methods

Participants.

The data analyzed as part of this study is a subset of a larger study investigating hip kinematics in healthy young adults [20,21]. Inclusion criteria for participants were no history of severe injury to the hip, lower extremities, or lumbar spine, age between 18 and 26 years, and a body mass index (BMI) less than 25 kg/m² [20,21]. Exclusion criteria were existing chronic hip-related pathology, prior hip surgery, or pregnancy. Twenty-four young adults provided written informed consent to participate in this institutional review board (IRB)-approved study (IRB protocol STUDY19050254) (eight men, four women; Age: 22.2 ± 2.1 years (men: 22.4 ± 2.4 years, women: 21.8 ± 1.5 years); BMI: 21.8 ± 2.3 kg/m² (men: 22.5 ± 2.5 kg/m², women: 20.6 ± 1.3 kg/m²), but complete datasets where available for only 12 individuals included in this analysis (Table 1). Hip morphology was assessed using the anterior-posterior scout CT image (Table 1). Briefly, a board-certified orthopedic surgeon measured each hip's alpha angle, lateral center edge angle (LCEA), and Tönnis angle from an anterior-posterior scout CT scan and identified hips with asymptomatic cam FAI as having an alpha angle >60 deg and hips with asymptomatic DDH as having an LCEA between 20 and 25 deg and/or Tönnis angle >10 deg. These morphological variations have been previously published and were not a criterion for exclusion from study participation [20,21].

Table 1.

Computed tomography-based morphological measurements for the cohort

		Left hip			Right hip
Subject	Sex	LCEA (deg)	Tönnis angle (deg)	Alpha angle (deg)	LCEA (deg)	Tönnis angle (deg)	Alpha angle (deg)
1	M	30	3	57	30	3	57
2	M	39	−4	51	34	−2	54
3	M	26	5	42	21—DDH	11	50
5	F	32	−1	35	31	−1	31
9	M	31	0	52	29	1	61
12	F	28	1	40	26	3	41
16	F	31	−2	42	33	−5	42
19	M	32	−1	64—CAM	30	1	52
20	F	27	4	39	26	6	38
22	M	29	3	67—CAM	31	1	71—CAM
23	M	34	−3	53	34	−3	56
24	M	32	2	65—CAM	31	3	67—CAM

Asymptomatic developmental dysplasia of the hip (DDH) was defined as lateral center edge angle (LCEA) between 20 and 25 deg and/or Tönnis angle >10 deg and asymptomatic cam morphology was defined as alpha angle >60 deg.

Data Collection.

The data collection procedures for this cross-sectional observational study have been previously described in reference to treadmill walking and bodyweight squatting [20]; CT collection, camera configurations and X-ray parameters remain the same for the present study. Briefly, participants performed a step ascent motion using an 18–38 cm step placed within a dynamic biplane radiography imaging system (Figs. 1(a) and 2). Six motion trials were performed for each participant (three trials per side) while synchronized biplane radiographs of the hip were collected at 50 images/sec for 1 s (maximum exposure: 90 kV, 500 mA, 4 ms pulse duration) (Fig. 1(b)). Image acquisition was manually triggered by the researcher when the participant initiated the step-up motion starting from their point of maximum hip flexion. A small platform placed on the ground was used to support the stance limb and adjust the step height from 18–38 cm to achieve approximately 90 deg hip flexion for participants of varying heights. Participants stood with their foot placed on a step with their hip flexed to approximately 90 deg; foot placement of either limb was not controlled beyond ensuring the imaged hip achieved a deep flexion position within the field of view of the biplane radiography system. The flexed hip was imaged as the participant stepped up onto the step at a self-selected speed (Fig. 2). Participants were free to rest their hand on a bar located to the side of the step for support if they needed assistance balancing during the task, but they were instructed to not put weight on the bar as they stepped up and to use it solely for balancing purposes. An additional static standing trial (50 images/sec for 0.1 s) was collected for each hip and used to normalize step ascent kinematics to static standing posture [20,21]. Radiation exposure from the biplane radiography system was estimated to be 6.0 millisieverts (mSv) or less (six dynamic trials, two static trials) (PCXMC 2.0, STUK, Helsinki, Finland). Technical specifications of the camera system, distortion correction, and calibration are previously described in Ref. [14].

Fig. 1.

The workflow utilized to acquire and analyze biplane radiographic images of dynamic hip joint motion during step ascent: (a) Participants ascended a step from a position of unilateral deep hip flexion on an instrumented treadmill while (b) synchronized biplane radiographs were collected (50 images/s). (c) CT scans were segmented to create subject-specific three-dimensional proximal femur and hemi-pelvis bone models for each participant with coordinate systems determined from bony landmarks (d). (e) Digitally reconstructed radiographs, created from segmented femur and pelvis bone tissue, were matched to the radiographic images in the virtual lab space using a validated model-based tracking technique. (f) Six DOF kinematics of the hip joint during the step-up motion were generated for further analysis.

Fig. 2.

An example participant performing the step ascent motion within the dynamic biplane radiography imaging space. (a) shows an anterior view of a participant preparing to perform a step ascent motion, starting from a position of approximately 90 deg hip flexion. (b) shows the corresponding lateral view.

Computed tomography scans of the proximal femurs, full pelvis, and bilateral knee slices were collected for each participant as previously described in Johnson et al. and Ruh et al. (average 0.37 × 0.37 mm in-plane resolution, 0.625 mm slice thickness; LightSpeed Pro 16; GE Medical Systems, Waukesha, WI). The average radiation exposure from the CT scan (full pelvis and proximal femurs, with bilateral knee slices) was 16.6 mSv. Bone tissue of the bilateral pelvis and proximal femurs were identified from each participants' CT images using a semi-automated segmenting methods (Mimics software, Materialize Inc, Ann Arbor, MI) and reconstructed into three-dimensional bone models for later use in model-based tracking [24] (Fig. 1(c)).

Data Processing.

Anatomical coordinate systems were established for each CT-based bone model of the proximal femur and hemi-pelvis (Fig. 1(d)) [20,21,25]. Femur and pelvis pose were determined using a validated volumetric model-based tracking technique that matched the digitally reconstructed radiographs created from subject-specific CT bone models to each pair of synchronized radiographs (bias: 0.2 deg, 0.2 mm; precision: 0.8 deg, 0.3 mm) (Fig. 1(e)) [14]. Six degree-of-freedom (DOF) hip joint kinematics during the step ascent motion were calculated [25], filtered (4th order Butterworth filter, 4 Hz cutoff frequency), and normalized to static standing pose using a custom MATLAB script (Mathworks, Natick, MA) (Fig. 1(f)).

Kinematics during step ascent were interpolated to every 2 deg of hip flexion using matlab to allow comparisons at corresponding hip flexion angles. The average kinematics waveform for each hip for each individual during the step up motion was used for statistical analysis. Absolute side-to-side differences (SSD_A) in kinematics were calculated within each individual at corresponding hip flexion angles. SSD_A values for all participants were then averaged to provide a single SSD_A value for the cohort during step ascent [20].

Statistical Analysis.

We fit a linear mixed effects model with normalized 6DOF kinematic variables (abduction, internal rotation, anterior translation, superior translation, and medial translation) as the outcome variables, with fixed effects of sex, hip flexion (every 2 deg), and the interaction of flexion*sex, with participants treated as random effects within the linear mixed model. The linear mixed model design is robust to variable amounts of data between participants [20] and is therefore a good fit for analyzing datasets with inherent among-participant movement variability. The covariance matrix was estimated using the restricted maximum likelihood (REML) method and a Satterthwaite approximation estimated denominator degrees-of-freedom (IBM SPSS statistics, version 27.0). If significant main effects were found, then posthoc tests were performed with the Benjamin-Hochberg Procedure (false discovery rate: 5%) to account for comparisons at multiple flexion angles. Significance was set at p < 0.05 for all tests. Based on a power analysis of simulated data fit to a linear mixed model, 12 total hips (six male hips, six female hips) were needed to detect medium sex-based differences (Cohen's f = 0.5) in step ascent kinematics with at least 80% power [26].

Results

A total of 57 step ascent trials from 12 participants (24 hips) were included in the analysis. If the participant moved outside the imaging volume during the data capture or tissue density created poor image quality that hindered model-based tracking, the affected frames of biplane radiographic images were excluded from analysis, resulting in limited availability of trackable data at the end ranges of the movement (Fig. 3) [20]. Average maximum hip flexion during step ascent was 87.8 ± 6.8 deg. The range of hip flexion captured from all trials was from 100 deg to 64 deg (Figs. 3 and 4).

Fig. 3.

Kinematics data available for analysis during step ascent at every 2 deg hip flexion

Fig. 4.

Six degree-of-freedom-kinematics during step ascent, expressed over hip flexion (deg). Solid lines indicate average kinematics for the cohort and shaded areas indicate standard deviations. The step ascent motion proceeds along the x-axis from left to right, with the participants beginning in their position of greatest hip flexion and extending at the hip throughout the captured motion.

There was a significant effect of sex on abduction (F = 18.82, p < 0.001) (Table 2); on average, women were 10.2 deg more abducted than men. No significant sex differences in abduction were revealed when comparing at 2 deg intervals of hip flexion. Medial-lateral translation was also affected by sex (F = 7.59, p = 0.01) (Table 2); on average, women were 0.2 mm more medially translated than men. When comparing at 2 deg intervals of hip flexion, women were approximately 1.8 mm more medially translated than men at 80 deg hip flexion (p = 0.003) (Fig. 5). No significant effects of flexion angle or interactions between flexion angle and sex were found in this cohort (all p > 0.05) (Table 2).

Table 2.

Type III tests of fixed effects for all kinematic variables during step ascent, and mean difference in kinematics between men (M) and women (W)

	Internal rotation		Abduction		Anterior translation		Superior translation		Medial translation
Fixed effects	F	p	F	p	F	p	F	p	F	p
Sex	3.75	0.06	18.82	* <0.001	1.07	0.30	0.42	0.52	7.59	* 0.007
	Mean diff (W–M): −5.8 deg		Mean diff (W–M): 10.2 deg		Mean diff (W–M): 0.5 mm		Mean diff (W–M): 0.1 mm		Mean diff (W–M): 0.2 mm
Flexion angle	0.20	>0.99	1.25	0.24	0.30	>0.99	0.12	0.34	0.66	0.85
Flexion angle*sex	0.42	0.95	0.23	0.99	0.23	>0.99	0.40	0.96	0.79	0.66

^*Indicates significance (p < 0.050). Mean difference is shown as women minus men (W–M).

Fig. 5.

Six degree-of-freedom-kinematics for men and women during step ascent, expressed over hip flexion (deg). Solid lines indicate average kinematics for the male or female cohort and shaded areas indicate standard error. The step ascent motion proceeds along the x-axis from left to right, with the participants beginning in their position of greatest hip flexion and extending at the hip throughout the captured motion. Significant effects of sex across the motion are indicated with gray shading along the bottom of the graph. Significant effects of sex at specific hip flexion angles are indicated with vertical gray shading on the applicable regions of the graph.

SSD_A values averaged 14.1 ± 12.1 deg in internal rotation and 5.3 ± 2.8 deg in abduction (Fig. 6). SSD_A in translation were 1.3 mm or less for all components during step ascent (Fig. 6). No trends in SSD_A over flexion angle were observed.

Fig. 6.

Side-to-side differences in six DOF-kinematics during step ascent expressed over hip flexion (deg). Solid lines indicate average absolute side-to-side differences for the cohort and shaded areas indicate ±1 standard deviation. The mean value for each kinematic variable is visible on each graph and expressed as mean ± standard deviation and the 95% confidence interval for the average SSD_A is shown as (lower bound, upper bound). The step ascent motion proceeds along the x-axis from left to right, with the participants beginning in their position of greatest hip flexion and extending at the hip throughout the captured motion.

Discussion

The goals of this study were to identify sex-based differences in hip kinematics during a step ascent motion in a cohort of young asymptomatic adults and to characterize inherent kinematic asymmetry. The main finding of this study was that women were more abducted and medially translated than men during the step-up motion and that the normative asymmetry in our asymptomatic cohort was up to 14.1 ± 12.1 deg in rotation and 1.3 ± 1.4 mm in translation.

Our finding that women were more abducted and medially translated than men across the step ascent motion (Table 1, Fig. 5) differs from a previous analysis of sex-based kinematic differences during bodyweight squatting that failed to find any effects of participant sex [20]. This could indicate that sex-based differences in pelvic morphology or strength may have a greater influence on hip kinematics during unilaterally loaded deep flexion in comparison to bilaterally loaded deep flexion. Law et al. utilized marker-based motion capture to show that women had greater peak hip adduction and abduction-adduction range of motion than men during a stair climbing task; however, sex-based differences in peak hip abduction were less than 1 deg [27]. Differences in data collection technology, step height, and participant demographics (i.e., age, height, and weight) could contribute to the greater magnitude of sex-based differences in hip abduction found in the present study. Furthermore, a recent systematic review has described sex-based differences in hip muscle strength, showing that men have greater peak torque than women during concentric hip flexion–extension, hip abduction-adduction, and hip internal-external rotation across a variety of isokinetic strength tests [28]. Similarly, a study analyzing 1000 adults and children to determine normative isometric muscle strength and flexibility values across a wide range of ages has shown that men aged 20–59 years had significantly higher hip internal rotation, external rotation, and abductor strength than their female counterparts [29]. These baseline differences in muscle strength, combined with potential differences in hip muscle moment arm length due to sex-based differences in pelvis morphology [30] leading to differences in muscle torque production, may contribute to the sex-based differences in hip kinematics observed in the present study.

Our analysis of asymmetry revealed that kinematic differences of up to 14.1 ± 12.1 deg in internal rotation and 1.3 ± 1.4 mm in translation can be expected between limbs in an asymptomatic cohort. The SSD_As found in the present study are larger than those previously reported in this cohort for treadmill walking (up to 9.0 ± 7.4 deg in rotation and 0.6 ± 0.6 mm in translation) and smaller than those reported for bodyweight squatting (up to 18.6 ± 6.7 deg in rotation and 1.0 ± 0.4 mm in translation) [20]. Additionally, the natural asymmetry evident in our asymptomatic cohort is larger than previously reported “significant” differences in hip kinematics measured using dual fluoroscopy and marker-based motion capture. In a surgical population, Dimitriou et al. identified differences of 3.5 deg in internal external rotation and 0.8 mm of anterior-posterior translation between the operated hip and the native contralateral hip following total hip arthroplasty (THA) when performing a step-up task onto a low step [15]. Similarly, Shrader et al. demonstrated that individuals with THA exhibited a greater minimum hip flexion angle during a stair ascent task than controls (15.0 deg for the THA group compared to 7.6 deg for the control group) using marker-based motion capture techniques [31]. The current study results suggest that those previously reported differences in kinematics between THA and contralateral hips may be well within the natural side-to-side differences that exist in asymptomatic native hips. However, it should be noted a relatively small number of hips were analyzed in the present study (Fig. 3), and future studies that include more participants and a wider age range is needed to assess the generalizability of this result.

Although all participants in the current analysis had no current or prior history of hip pathology, injury or surgery, the presence of asymptomatic cam FAI or DDH morphologies may influence the study results. Previous research has shown that individuals with symptomatic cam FAI have decreased hip flexion while performing a squatting motion than their asymptomatic counterparts; however, hip flexion angles did not differ between the asymptomatic cam FAI and control groups [5], suggesting that morphologic FAI is not the sole determinant of aberrant hip kinematics. The presence of symptomatic dysplastic hip anatomy has been associated with shorter abduction moment arms during gait compared to healthy hips [32], which could manifest as altered kinematics. Analyzing a more demanding task, such as unilateral ascension of a step or a single-leg squat, may exacerbate these kinematic differences as a greater force is required to stabilize the stance limb. However, the effects of asymptomatic DDH on hip kinematics are currently unknown. Since all asymptomatic cam FAI in the present study were found in men, this may partially explain the greater movement variability in the male cohort, particularly in abduction-adduction and anterior-posterior translation. Studies with a larger number of hips are needed to explore the effects of asymptomatic morphology on variability in kinematics.

This was a cross-sectional observational study of healthy young adults, and the results should not be extrapolated to older populations, populations with higher body mass index, populations with symptomatic pathology, or hip kinematics exhibited in free-living environments. Other important limitations include the fact that foot placement and angulation during the motion was not controlled, which could vary between trials and participants and in turn affect the abduction and internal rotation of the hip throughout step ascent movement. Participants were instructed to stand with their feet in their natural position and no coaching was made with respect to feet turning in/out or stance width. Participants were allowed to perform the step ascent motion with their natural biomechanical pattern since one goal of the present study was to quantify the normative asymmetry inherent in unilateral extension from deep hip flexion. Observed kinematic differences between sexes and between individuals could have been due, in part, to differences in neuromuscular function or additional bony morphologies that were not measured during this study. Strength and neuromuscular control could impact hip kinematics and symmetry during deep extension movements; however strength of the muscles surrounding the hip, knee, and ankle was not quantified in the present study. Similarly, limb dominance was not included as a covariate in the present analysis, but research suggests that limb dominance does not affect hip muscle strength in healthy individuals [33]. Future work should further consider the interplay between limb dominance and hip kinematics during unilateral and bilateral-load tasks.

The results of this study are only applicable to the deep flexion portion of the step ascent motion where participants remained within the field of view of the biplane radiography system. Radiographic imaging of the hip during deep flexion–extension is challenging due to soft tissue occlusion from the thigh and abdomen and the limited field of view. Our imaging approach, shown in Fig. 2, minimizes soft tissue occlusion during deep flexion by directing X-rays downward at an angle of 10 deg. Regardless, we only obtained high-quality bilateral data during the step-up motion from 12 of the original 24 participants in this larger healthy adult cohort [20,21]. Due to the difficulty in collecting high-quality biplane radiography data during this motion, our required sample size to detect medium differences in kinematics was not achieved, and therefore the study was powered only to detect large differences in step ascent hip kinematics. Similarly, kinematics were collected separately for the right and left hip, and natural trial-to-trial variability within a participant further limited the number of participants included in our analysis of kinematic asymmetry since the SSD_A calculation requires an overlapping range of motion from the right and left hips in order to calculate SSD_A for an individual participant (Fig. 3). Collecting multiple trials during different phases of the step up motion (i.e., initial extension, midextension, and end-range extension) would increase our knowledge of how the hip moves dynamically over a larger range of motion, but would increase radiation exposure to the participant.

The results of this study indicate that bilateral asymmetry of up to 14.1 ± 12.1 deg in rotation and 1.3 ± 1.4 mm in translation can be expected in a young, asymptomatic cohort performing step ascent tasks and that sex-based differences in step ascent kinematics exist in abduction and medial translation (approximately 10.2 deg and 0.2 mm). However, it should be noted that the current small sample size limits the generalizability of these results. This information provides valuable normative data to clinicians and researchers regarding asymmetry that may be inherent during movements involving unilateral extension from deep hip flexion and provides a reference dataset for future, larger studies that evaluate side-to-side differences in kinematics following surgery or rehabilitation.

Funding Data

• National Center for Advancing Translational Sciences (Award No. UL1 TR001857; Funder ID: 10.13039/100006108).

Conflicts of Interest

MM has previously received royalties from Elizur LLC for a device unrelated to the present study. All other authors have no professional or financial affiliations that may be perceived to bias this presentation.

Data Availability Statement

The datasets generated and supporting the findings of this article are obtainable from the corresponding author upon reasonable request.

Nomenclature

CT =

computed tomography

FAI =

femoroacetabular impingement

DDH =

developmental dysplasia of the hip

OA =

osteoarthritis

IRB =

Institutional review board

BMI =

body mass index

DOF =

degrees of freedom

SSD_A =

absolute side-to-side differences

REML =

restricted maximum likelihood

THA =

total hip arthroplasty

LCEA =

lateral center edge angle