Hip and Trunk Muscle Activity and Mechanics During Walking With and Without Unilateral Weight

Abstract

Pelvic drop is caused by decreased hip abductor muscle activity and is associated with lower extremity injury. Hip abductor strengthening exercises are well-established; however, no standard method exists to increase hip abductor activity during functional activities. The purpose of this research was to study the effects of walking with a unilateral weight. Twenty-six healthy adults walked on an instrumented treadmill with and without hand-held weight (15–20% body weight). Muscle activity, kinematic, and kinetic data were collected using surface electromyography, motion capture, and force plates, respectively. Average hip and trunk muscle activity, hip, pelvic, and trunk angles, and peak internal hip moments during stance were compared for each side (contralateral/ipsilateral to the weight) between conditions (unweighted/weighted) using a generalized linear model with generalized estimating equation correction. Interactions between condition and side were observed for muscle activity, frontal plane pelvic and trunk angles, and frontal plane hip moments (p≤.003). Compared to the unweighted condition, the weighted condition had higher hip abductor activity contralateral to the weight (p<.001), while no change was found ipsilateral to the weight (p≥.790). Similar changes were found for kinematic and kinetic variables. Walking with a unilateral weight may be a therapeutic option to increase functional hip abductor activity.

Introduction

Pelvic drop is a movement pattern observed in the frontal plane during single-limb stance that is characterized by the non-stance side of the pelvis dropping inferiorly (Figure 1). Pelvic drop may place excessive stress on hip and knee joints due to its influence on lower extremity kinematics, joint moments, and joint and soft tissue loading^1–3. Increased pelvic drop has been demonstrated in individuals with gluteal tendinopathy⁴, early stage hip osteoarthritis⁵, and patellofemoral pain⁶ compared to healthy individuals. Pelvic drop has also been shown to contribute to the progression of knee osteoarthritis^1,3.

Figure 1 —

The two models demonstrate right single-limb stance with (A) Neutral Pelvis and (B) Pelvic Drop. Graphic created in OpenSim.

Pelvic drop is commonly thought to result from decreased hip abductor strength. However, various studies have found no relationship between hip weakness and pelvic and hip kinematics^7–11. In addition, hip strengthening interventions alone do not always modify lower extremity kinematics^11,12 or kinetics^13,14. It is therefore argued that a motor control adaptation may be responsible for this movement pattern^15,16 or a combination of motor control and strength deficits^2,12 as several studies have revealed hip weakness in populations with hip and knee injuries^6,17–24. These findings indicate the need for a patient-specific approach when addressing pelvic drop.

Current interventions for pelvic drop include gluteus medius strengthening exercises^25–34 and motor control exercises focused on lower extremity alignment during single leg activities. Many gluteus medius strengthening exercises are performed in positions that are not used during functional activities. Furthermore, most focus on concentric activation of the gluteus medius while this muscle is regularly challenged eccentrically. Thus, the motive for this research was to investigate a functional activity that may have the potential to promote eccentric hip abductor strengthening and neuromuscular re-education for improved gluteus medius muscle recruitment. Walking with a unilateral weight, which is also known as a suitcase carry or unilateral farmer’s carry exercise³⁵, is functional, clinically applicable, and has the ability to challenge the hip abductor musculature eccentrically based on biomechanical principles^36,37. The purpose of this study was to explore how walking while carrying a unilateral hand-held weight affects hip and trunk muscle activity, hip, pelvic, and trunk kinematics, and hip joint moments. We hypothesized that walking with a unilateral weight would result in increased muscle activity contralateral to, but not ipsilateral to, the weight. We also hypothesized that carrying a unilateral weight would promote improved neutral pelvic alignment compared to normal walking.

Methods

Participants:

A convenience sample of twenty-six healthy adults (14 females; 12 males; age: 25.3 ± 5.4 years (mean ± standard deviation); height: 1.73 ± 0.10 m; mass: 68.1 ± 13.8 kg; BMI: 22.6 ± 2.9 kg/m²) participated in this study between March, 2018 and November, 2018. Upon arrival, participants reviewed the data collection protocol with a researcher and provided written informed consent. Participants were eligible if they were between the ages of 18 and 50 years old, had experience walking on a treadmill, and reported feeling comfortable walking on a treadmill with and without a kettlebell weight in one hand. Participants were required to weigh 104.3 kg or less as the study involved walking with a kettlebell weight of 15% to 20% body weight and the kettlebell weights available in the laboratory were between 4.5 and 15.9 kg in 2.3 kg increments. Participants were excluded if they reported any history of the following: neurological disorder; cardiac or respiratory problems related to exercise; neck, shoulder, back, or lower extremity orthopedic surgery; or neck, shoulder, back, or lower extremity pain over the last month that caused a change in their exercise routine. This research was approved by the Institutional Review Board of Boston University.

Instrumentation:

Muscle activity data were collected using a surface electromyography (EMG) system (Trigno Wireless EMG System, Delsys Inc., Natick, MA) at a sampling rate of 2000 Hz. The skin was prepared for electrode placement by scrubbing the area with a cotton ball and rubbing alcohol. Surface electrodes were placed over the muscle bellies of gluteus maximus (GMAX), gluteus medius (GMED), tensor fascia lata (TFL), and lumbar erector spinae (lumbar ES) bilaterally according to guidelines for surface electrode placement³⁸. To confirm proper electrode placement, EMG signal was visually inspected as the participant performed the action of each muscle against manual resistance.

Kinematic data were collected using a 10-camera motion capture system (Vicon Motion Systems Ltd, Centennial, CO) at a sampling rate of 100 Hz. The cameras recorded the location of reflective markers over bony landmarks on the participant’s trunk, pelvis, and lower extremities bilaterally. Specifically, markers were placed over the participant’s seventh cervical vertebra, acromion processes, xiphoid process, anterior superior iliac spines (ASIS), iliac crests, sacrum at the height of the posterior superior iliac spines, greater trochanters, medial femoral epicondyles, lateral femoral epicondyles, medial malleoli, lateral malleoli, first metatarsal heads, fifth metatarsal heads, and calcanei. Additional markers were placed on the pelvis between the ASIS and iliac crest, and between the iliac crest and sacrum, bilaterally to improve tracking. Cluster marker sets consisting of four markers on a semi-rigid surface were placed on the participant’s thighs and shanks and secured with felt wrap and pre-wrap. Ground reaction force data were collected using force plates embedded in an instrumented split-belt treadmill (Bertec Corp., Columbus, OH) at a sampling rate of 2000 Hz.

Procedures:

Following arrival and informed consent processes, participants’ weight was measured and height was self-reported. Participants were then asked to perform three single leg squats on each leg with their non-stance leg in back, keeping their non-stance thigh perpendicular to the ground³⁹. The lead investigator (KAG) observed their pelvic control in the frontal plane during each squat. The stance limb that demonstrated greater contralateral pelvic drop was chosen as the focus limb for the study; participants were asked to walk with a kettlebell weight in the hand contralateral to this limb. If no side-to-side difference in pelvic drop was observed, a limb side was chosen using a random number generator. The kettlebell weight used during the data collection was 15–20% of the participant’s body weight. Based on pilot work, 15–20% of body weight was heavy enough to elicit increased muscle activity while still being light enough to carry. If two kettlebell weights fell within the predetermined range, participants were given the opportunity to hold each weight and choose which weight they felt most comfortable carrying. Participants also had the option of using a weight lifting strap to assist with grip. The weight lifting strap attached around the kettlebell weight and the participant’s wrist. After the weight and strap options were determined, final eligibility was confirmed by asking the participant to hold the weight in the hand contralateral to the focus limb and report if they felt comfortable holding and walking with the weight for the data collection.

After placement of the EMG electrodes, muscle activity was collected during maximal voluntary isometric contraction (MVIC) trials for each muscle group. MVIC trials consisted of maximal effort for an estimated three-second hold (muscle contraction) while the participant pushed against a strap that resisted movement. Standard manual muscle testing positioning was used for GMAX and lumbar ES⁴⁰, while GMED and TFL were tested together as participants performed standing hip abduction with a resistance strap placed around bilateral ankles. Participants performed two repetitions for each muscle group with a thirty-second rest period between each repetition. Verbal encouragement was provided with testing.

After placement of the reflective markers, a static standing trial was recorded to create a model of the participant in Visual3D (C-Motion Inc., Germantown, MD). During this trial, participants were asked to stand on a split-belt treadmill with feet shoulder-width apart and arms out to the side. Participants then walked on the split-belt treadmill for one minute at 1.25 m/s to gather baseline data of each participant’s muscle activity, kinematics, and kinetics during walking. Data were collected during the last 30 seconds of this trial, resulting in an average of 26.3 ± 2.3 strides for analysis. Participants then walked with the predetermined kettlebell weight in the hand contralateral to the focus limb for four minutes at 1.25 m/s. Muscle activity, kinematic, and kinetic data were recorded during the first 20 seconds of each minute of this trial, beginning when the treadmill reached full speed, resulting in an average of 65.8 ± 12.7 strides for analysis.

Data Processing:

Raw signals were bandpass filtered (20–450 Hz) by the Delsys hardware (Trigno Wireless EMG System; Delsys Inc, Natick, MA). We further processed EMG signals using Visual3D. All signals were high-pass filtered with a fourth-order Butterworth filter with a cut-off frequency of 20 Hz, full-wave rectified, and low-pass filtered with a fourth-order Butterworth filter with a cutoff frequency of 6 Hz to create a linear envelope⁴¹. Muscle activity data were normalized to each participant’s MVIC data for each respective muscle group, and these values were used to calculate average muscle activity during stance.

Hip, knee, and ankle joint angles were defined as the angle between proximal and distal segments. The pelvis and hip joint centers were defined using the CODA model⁴². Pelvis and trunk segment angles were determined with respect to the global coordinate system. Marker trajectories were low-pass filtered using a fourth-order Butterworth filter with a cutoff frequency of 6 Hz⁴³. Joint kinematics were calculated from the marker data using an 8-segment hybrid model with a Cardan x-y-z (mediolateral, anteroposterior, vertical) rotation sequence⁴⁴ in Visual3D. Ground reaction forces were low-pass filtered using a fourth-order Butterworth filter with a cutoff frequency of 10 Hz⁴³. Kinematic data and ground reaction force data were used to calculate internal joint moments in Visual3D.

Data Analysis:

All recorded strides with valid ground reaction force and EMG data were included in the analysis. Values for all variables of interest were calculated during the stance phase of the gait cycle for each side for the unweighted and weighted conditions. Side was defined as contralateral to or ipsilateral to the weight during the weighted condition, and this was maintained for the unweighted condition. Stance phase consisted of the time from heel strike to toe off on the same limb. These two time points were determined from the ground reaction force data, and they were verified through visual inspection in Visual3D. Normalized average muscle activity was calculated for each muscle (GMAX, GMED, TFL, and lumbar ES). Peak frontal plane angles were calculated for the hip, pelvis, and trunk, and peak internal hip moments were calculated for hip abduction, hip adduction, and hip extension.

Statistical Analysis:

To determine differences in muscle activity, kinematics, and kinetics, linear regressions were run with generalized estimating equation (GEE) corrections. For muscle activity, linear regressions were created for each muscle separately (GMAX, GMED, TFL, and lumbar ES). The kinematic and kinetic variables of interest were peak hip, pelvic, and trunk angles, and peak hip joint moments, respectively. All linear regressions used two within-subjects factors: condition (unweighted versus weighted) and side (contralateral versus ipsilateral to the weight). Significant findings were followed by pairwise comparisons with Least Significant Difference (LSD) adjustment. All analyses were conducted in IBM SPSS Statistics Version 25 (IBM Corporation, Armonk, NY) with an alpha level of .05.

Results

During the unweighted condition, no differences were observed between sides for muscle activity (p≥.088), kinematic variables (p≥.405), or kinetic variables (p≥.613). Comparisons between the unweighted and weighted conditions are reported below.

For muscle activity, GEE models revealed main effects of condition (p<.001) and side (p≤.004), and an interaction between the two (p<.001) for GMAX, GMED, TFL, and lumbar ES. Pairwise comparisons were used to investigate the effect of condition on muscle activity for each side separately.

In the weighted condition compared to the unweighted condition, GMED activation was 58% higher (22% vs 14% MVIC, mean difference: 7.9% [95% Wald Confidence Interval (CI) for Difference: 6.1%, 9.5%]), TFL activation was 65% higher (11% vs. 6.9%, mean difference: 4.5% [95% CI: 3.7%, 5.3%]), and lumbar ES activation was 90% higher (16% vs 8.6%, mean difference: 7.8% [95% CI: 6.1%, 9.4%]) during stance phase of the limb contralateral to the weight (p<.001). GMED and TFL activation during stance phase of the limb ipsilateral to the weight were not affected by condition (14% vs. 13%, mean difference: 0.1% [95% CI: −1.3%, 1.4%]; 6.7% vs. 6.7%, mean difference: 0.1% [95% CI: −0.5%, 0.7%], respectively) (p≥.790), while lumbar ES activation was 11% lower (6.8% vs. 7.6%, mean difference: 0.8% [95% CI: 0.1%, 1.6%]) (p=.034). GMAX activation was higher during stance phase of both the limb contralateral (64% higher; 10% vs. 6.4%, mean difference: 4.1% [95% CI: 2.9%, 5.2%]) and ipsilateral (22% higher; 7.0% vs. 5.8%, mean difference: 1.2% [95% CI: 0.6%, 1.9%]) to the weight (p<.001) (Figure 2).

Figure 2 —

Average muscle activity during stance. Contralateral to the weight, hip abductor and trunk muscle activity increased in the weighted condition compared to the unweighted condition. Ipsilateral to the weight, hip abductor activity did not change between conditions, but trunk activity decreased. Gluteus maximus activity increased bilaterally in the weighted condition compared to the unweighted condition. Abbreviations: GMED, gluteus medius; TFL, tensor fascia lata; Lumbar ES, lumbar erector spinae; GMAX, gluteus maximus; UW, unweighted condition; W, weighted condition; Contra, contralateral to the weight; Ipsi, ipsilateral to the weight. *Significant difference between the unweighted and weighted conditions (P<.05)

For joint kinematics, GEE models indicated no main effects of condition (p=.927) nor side (p=.462), nor an interaction between the two (p=.944) for peak hip adduction angle. A main effect of condition (p<.001), but not side (p=.068), was found for peak pelvic drop angle, while main effects of both condition (p<.001) and side (p=.023) were found for peak ipsilateral trunk lean. An interaction between condition and side was found for pelvic drop and ipsilateral trunk lean (p≤.003). Pairwise comparisons were used to investigate the effect of condition on peak pelvic drop and peak ipsilateral trunk lean for each side separately.

In the weighted condition compared to the unweighted condition, pelvic drop was 2° less (1.2° vs. 3.3°, mean difference: 2.1° [95% CI: 1.6°, 2.7°]), and ipsilateral trunk lean was 2° greater (3.6° vs. 1.4°, mean difference: 2.2° [95% CI: 1.1°, 3.3°]) during stance phase of the limb contralateral to the weight (p<.001). During stance phase of the limb ipsilateral to the weight, pelvic drop and ipsilateral trunk lean were not affected by condition (3.2° vs. 3.2°, mean difference: 0.01° [95% CI: −0.5°, 0.5°]; 0.9° vs. 1.7°, mean difference: 0.8° [95% CI: −0.1°, 1.7°], respectively) (p≥.076) (Figure 3).

Figure 3 —

Peak joint angles during stance. Peak hip adduction angle did not change between conditions for either side. Contralateral to the weight, peak pelvic drop decreased, while peak ipsilateral trunk lean increased, in the weighted condition compared to the unweighted condition. Ipsilateral to the weight, peak pelvic drop and peak ipsilateral trunk lean did not change between conditions. Abbreviations: UW, unweighted condition; W, weighted condition; Contra, contralateral to the weight; Ipsi, ipsilateral to the weight. *Significant difference between the unweighted and weighted conditions (P<.05)

For joint kinetics, GEE models revealed main effects of condition (p<.001) and side (p<.001) for peak internal hip abduction and adduction moments. A main effect of condition (p<.001), but not side (p=.101), was found for peak internal hip extension moment. An interaction between condition and side was found for each joint moment studied (p≤.005). Pairwise comparisons were used to investigate the effect of condition on each peak joint moment for each side separately.

In the weighted condition compared to the unweighted condition, the hip abductor moment was 43% higher (79 N·m vs. 56 N·m, mean difference: 24 N·m [95% CI: 20 N·m, 28 N·m]), and the hip adductor moment was 29% lower (6.0 N·m vs. 8.4 N·m, mean difference: 2.4 N·m [95% CI: 1.6 N·m, 3.3 N·m]) during stance phase of the limb contralateral to the weight (p<.001). During stance phase of the limb ipsilateral to the weight, the hip abductor moment was 19% lower (45 N·m vs. 56 N·m, mean difference: 10 N·m [95% CI: 8.3 N·m, 13 N·m]), and the hip adductor moment was 75% higher (15 N·m vs. 8.7 N·m, mean difference: 6.5 N·m [95% CI: 5.0 N·m, 8.1 N·m]) (p<.001). The hip extension moment was 10% higher during stance phase of the limb contralateral to the weight (52 N·m vs. 48 N·m, mean difference: 4.6 N·m [95% CI: 2.8 N·m, 6.4 N·m]) and 15% higher during stance phase of the limb ipsilateral to the weight (56 N·m vs. 48 N·m, mean difference: 7.3 N·m [95% CI: 5.5 N·m, 9.1 N·m]) (p<.001) (Figure 4).

Figure 4 —

Peak internal hip moments during stance. Contralateral to the weight, peak hip abductor moment increased, while peak hip adductor moment decreased, in the weighted condition compared to the unweighted condition. Ipsilateral to the weight, peak hip abductor moment decreased, while peak hip adductor moment increased. Peak hip extension moment increased bilaterally in the weighted condition compared to the unweighted condition. Abbreviations: UW, unweighted condition; W, weighted condition; Contra, contralateral to the weight; Ipsi, ipsilateral to the weight. *Significant difference between the unweighted and weighted conditions (P<.05)

Discussion

Our study demonstrated that walking with a unilateral weight increased hip abductor activity contralateral to the weight and caused no change in hip abductor activity ipsilateral to the weight compared to walking without the weight. Our results also revealed that the hip abductor moment increased contralateral to the weight and decreased ipsilateral to the weight compared to the unweighted condition, in agreement with biomechanical predictions and models^36,37. As demonstrated by Neumann, a model of static equilibrium of the pelvis in the frontal plane allows us to estimate the hip abductor force required to maintain neutral pelvic alignment during single limb stance^36,37. If viewed from the front, the body mass of an individual standing on their right leg creates a clockwise torque about the right hip joint. To counteract this torque and maintain neutral pelvic alignment, the individual’s right hip abductors must generate a counterclockwise torque of similar magnitude. Holding a weight in one hand adds another torque, influencing what is required of the hip abductors. If a weight is held opposite the stance limb, the hip abductors must generate substantially more torque to counteract not only the torque from body mass but also from the hand-held weight, which has a larger moment arm than body mass (Figure 5A). Conversely, if a weight is held on the same side as the stance limb, the torque created by the weight assists the right hip abductors in counteracting the torque created by body mass, decreasing the amount of hip abductor force required (Figure 5B).

Figure 5 —

The two models demonstrate static equilibrium of the pelvis in the frontal plane during right single limb stance. In both models, the right hip is the axis about which torque is generated. The force generated by the hip abductors (Hip Abd.) times its moment arm (dotted line) generate a counterclockwise torque. Body weight (BW) times its moment arm (dashed line) generates a clockwise torque. The external load times its moment arm (solid line) generates a torque which is clockwise in direction and larger in magnitude when held in the left hand compared to the counterclockwise torque it creates when held in the right hand. In static equilibrium, the sum of the counterclockwise torques equals the sum of the clockwise torques. (A) Contralateral Load: The counterclockwise torque is from the Hip Abd. The clockwise torques include the torque due to BW and the torque due to the load. (B) Ipsilateral Load: The Hip Abd. and BW continue to generate counterclockwise and clockwise torques, respectively, while the load now generates a counterclockwise torque. In model A, the hip abductors must generate more force to counteract the clockwise torques generated by both BW and the load (with its large moment arm), whereas in model B, the hip abductors are assisted by the load to counteract the torque of BW.

These changes in hip abductor activity are supported by walking studies by Neumann et al.^36,37,45,46 as well as a study comparing load position during walking lunges⁴⁷. Interestingly, Neumann et al. reported that changes in muscle activity during stance phase of the limb ipsilateral to the weight depend on the magnitude of the unilateral weight carried³⁷. They found that walking with a unilateral weight between 3% and 23% body weight led to a significant decrease in hip abductor activity during stance phase of the limb ipsilateral to the weight, while walking with a weight outside of this range led to no change in muscle activity on this side³⁷. Our study involved walking with weight of 15–20% body weight, yet we did not find significant differences in hip abductor activity ipsilateral to the weight even though we found a significant decrease in hip abductor moment on this side. The lack of change in muscle activity despite the decreased moment reveal there is not a direct relationship between joint moments and muscle activity, highlighting the complexity of measuring human movement.

To our knowledge, this study is one of few^48,49 that examine changes in lumbar ES activity during walking with and without unilateral weight. Our research expands on previous studies by using a larger population and by comparing weighted walking to unweighted walking as well as exploring side-to-side differences. In the weighted condition compared to the unweighted condition, we found that lumbar ES activity on the side contralateral to the weight increased during stance phase of that limb and lumbar ES activity on the side ipsilateral to the weight decreased during stance phase of that limb, consistent with our expectations based on biomechanical principles.

This study advances our understanding of movement by investigating kinematics in conjunction with muscle activity and joint moment data. Our kinematic results demonstrated no change in peak hip adduction angle between the two conditions on either extremity. In the weighted condition compared to the unweighted condition, pelvic drop was 2° less during stance phase of the limb contralateral to the weight, while no changes were found during stance phase of the limb ipsilateral to the weight. Although the change in pelvic drop on the contralateral side was small, it is important to recognize that our participants did not present with substantial pelvic drop at baseline. We believe the increase in hip abductor activity contralateral to the weight decreased pelvic drop during the weighted condition. However, the increase in ipsilateral trunk lean on this side may have also contributed to decreased pelvic drop, similar to a compensated Trendelenburg gait pattern. Although we found an increase in ipsilateral trunk lean, it is important to note that the increase in trunk angle was small. In addition, our participants demonstrated significant increases in hip abductor activity during the weighted condition on this side, revealing that the trunk lean did not fully compensate for the increased hip abductor moment. Because our kinematic changes were small, these findings should be considered with caution.

Our results reveal that walking with a unilaterally hand-held weight is a therapeutic option to increase hip abductor activity on a targeted side. A unilateral weight of 15–20% body weight may be an appropriate intensity to improve hip abductor endurance and motor control. Clinicians should monitor trunk position during this exercise to avoid excessive trunk lean.

This exercise may be more effective than exercises typically used in the clinic due to its ability to provide specific eccentric hip abductor training, which is required to prevent excessive pelvic drop during functional activities (e.g. walking, stair descent). This intervention is also specific to the task of walking and may be more successful at increasing hip abductor activity during walking. As demonstrated in neuro-rehabilitation and motor learning literature, task-specific training translates to greater transfer of learned behavior to a trained task than an untrained task^12,50. In addition, walking with a unilateral weight requires minimal equipment, costs little, and can easily be incorporated into daily activities, which may improve patients’ adherence to exercise within a home program.

This study has limitations. Our participant population consisted of healthy asymptomatic individuals, which limits generalizability to a symptomatic population. We also assumed pelvic control was transferrable between tasks of single leg squat and walking as we used single leg squat pelvic drop to determine which side to hold the kettlebell weight during our weighted walking trial. We were able to observe pelvic alignment in a slower and more distinguishable manner during the single leg squat. Furthermore, we did not provide our participants with cues regarding trunk position during the walking trials as we wanted to capture natural performance. In addition, we did not provide our participants with an extensive treadmill acclimation period prior to data collection procedures, which may impact our findings⁵¹. Lastly, this was a single session study, so we were unable to study the effects of using this exercise as a training tool.

Walking with a unilateral weight increases contralateral hip abductor moment and hip abductor muscle activity. Therefore, this intervention may be a viable option for task-specific training to increase hip abductor muscle activity during walking as well as improve hip abductor eccentric strength to a targeted side.