Application of a model to analyze shoulder biomechanics in adult patients with spinal cord injury when walking with crutches in two different gait patterns

Abstract

BACKGROUND: Specific biomechanical models have been developed to study gait using crutches. Clinical application of these models is needed in adult spinal cord injury (SCI) population walking with different patterns of gait with crutches to prevent overuse shoulder injuries.

OBJECTIVE: To apply a biomechanical model in a clinical environment to analyze shoulder in adult SCI patients walking with two different patterns of gait with crutches: two point reciprocal gait (RG) and swing-through gait (SG).

METHODS: Load cells were fixed to the distal ends and forearm cuffs of a pair of crutches. An active markers system was used for kinematics. Five cycles for each gait pattern were analyzed applying a biomechanical model of the upper limbs. Fifteen subjects with SCI were analyzed.

RESULTS: The flexo-extension range of motion was significantly greater when using SG (p < 0.01). Similarly, the superior, posterior and medial forces were significantly stronger for SG in all 3 directions. Flexion, adduction and internal rotation torques were also greater in SG (p < 0.01).

CONCLUSIONS: A biomechanical model was successfully applied to study shoulder biomechanics in adult patients with SCI walking with crutches in two different gait patterns. Greater loads exerted on the shoulder walking with SG were confirmed compared to RG.

Keywords

Rehabilitation spinal cord injury biomechanics shoulder gait crutches

1 Introduction

According to 2014 annual report of the National Spinal Cord Injury Statistical Center (NSCISC), 12,500 new cases of SCI are registered each year in the United States (University of Alabama at Birmingham, 2014b). Given a current population of 313 million people, this represents an incidence of around 40 cases per million population, and an associated prevalence of approximately 276,000, ranging between 240,000 and 337,000. The life expectancy of individuals that suffer SCI is currently quite similar to that of healthy individuals, with the exception of patients with high cervical injury. However, in many cases patients lose the ability to walk, although incomplete spinal cord injury (iSCI) patients are usually able to walk with the support of ortheses or technical aids. While the number of complete injuries was 1.5 times higher than incomplete ones in the 1970s, this ratio has since fallen to the point that nowadays, there is approximately the same number of complete and incomplete injuries (University of Alabama at Birmingham, 2014a). Accordingly, it has been estimated that 39.5% of SCI patients can walk, among which 51.3% use a technical aid for the upper limbs (ULs) (University of Alabama at Birmingham, 2014a).

In the case of walking with crutches, the individual’s weight must be largely unloaded onto the technical aid, which constitutes a considerable effort for the UL joints that are consequently predisposed to suffer injuries due to the overuse. Such injuries have a similar origin as those caused by manual wheelchair propulsion or weight transference where a force is applied from the carpus throughout the forearm to the glenohumeral joint, raising the humeral head and compressing the subacromial tendons (Dyson-Hudson et al., 2004). In fact, the load exerted on the UL during gait with crutches is considered as a potentially cause of shoulder pain (Sie et al., 1992). For this reason, it would be very useful to know the magnitude of the forces and loads to which the shoulder joints are subjected when patients use crutches in order to determine the potential sources of shoulder pain or injury. Indeed, some specific UL biomechanical models have been developed to study gait using crutches (Requejo et al., 2005; Slavens et al., 2011).

One of the immediate applications of these biomechanical models is to contribute data and criteria that can be applied to the design of crutches that diminish the impact of the load and forces on the UL joints. Furthermore this information can be employed in the process of gait rehabilitation, focusing patient treatment towards patterns that minimize the impact of the forces on the shoulder. In this sense, the gait patterns most commonly used by SCI patients with crutches during rehabilitation are swing-through gait (SG) and two-point reciprocal gait (RG). In the former, the patient walks with both crutches simultaneously, projecting their body and lower limbs in a pendular movement, and with the heels coming into contact with the ground ahead of the crutches. By contrast, the patient walks in an alternating manner in RG, moving the right crutch and the left foot at the same time, and subsequently the left crutch and the right foot (Faruqui & Jaeblon, 2010).

Most biomechanical studies into assisted gait have focused on the lower limbs (LL) (Gil-Agudo et al., 2009; Vankoski et al., 1997; Youdas et al., 2005), whereas ULs biomechanics have not been sufficiently characterized. ULs biomechanics studies reported in SCI patients that loads using RG are higher at the contralateral shoulder to the most affected lower limb than at the ipsilateral shoulder (Requejo et al., 2005)., and it appears that the effect of the crutch on the shoulder tends to elicit shoulder abduction and extension (Crosbie & Nicol, 1990; Opila et al., 1987). In addition, differences between SCI and healthy subjects have been found when walking with the SG pattern (Noreau etal., 1995). Most studies that focused on developing a method to assess the impact of technical aids on patients have involved a limited number of patients (Liggins et al., 2001; Requejo et al., 2005; Seireg et al., 1968; Slavens et al., 2011; Slavens et al., 2007; Slavens et al., 2009; Slavens et al., 2010). Alternatively, studies centered on the UL biomechanics of gait with crutches have been performed on different groups of patients or on those suffering from different conditions (e.g., comparing different gait patterns or technical aids involving the ULs) (Crosbie & Nicol, 1990; Haubert et al., 2006; Melis et al., 1999; Noreau et al., 1995; Opila et al., 1987; Requejo et al., 2005; Slavens et al., 2011; Waters et al., 1989). Up to now, UL biomechanical models have been applied in children population. The distinct loads on the shoulder while walking with RG and SG have been analyzed in a sample of children with myelomeningocele (Slavens et al., 2009), demonstrating that greater axial forces on the shoulder were associated with SG rather than RG. However, to the best of our knowledge no studies have applied biomechanical models to analyze the three-dimensional kinetics of the shoulder in adult SCI patients using these 2 different patterns of gait with crutches. It is interesting to check whether the data in adults are similar to those obtained in children.

Our hypothesis is that shoulder loads will be greater during SG than during RG in adult SCI patients as in children with myelomeningocele (Slavens et al., 2009). Thus, in this study the aim was to apply a biomechanical model to analyze shoulder kinematics and kinetics in adult SCI patients when walking with two different gait crutches patterns: RG and SG. We focused on the application of the methodology in a clinical environment to suggest future applications.

2 Materials and methods

2.1 Instrumentation

Two load cells (MCW-6-1000: AMTI, Watertown, MA, USA) were installed at the distal end of a pair of Guardian crutches (Sunrise Medical, Fresno, CA, USA) with 6 degrees of freedom (DOF). Another 2 load cells (model 5811: Burster, Gernsbach, Germany) with 1 DOF were placed at the joint between the crutches and the forearm cuffs. Each 6 DOF sensor incorporates strain gauges and it measures the force components and the torque, while the 1 DOF sensors measure the perpendicular force exerted by the forearm on the forearm cuff of the crutch. A motion capture system based on active markers was used to record the UL movement of the patients and the crutches (Charnwood Dynamics Limited, Leicestershire, England). This consisted of 2 scanners located on either side of a walking corridor that could detect the infrared light emitted by the markers. The signals generated by the load cells were sent through cables to the motion capture system, in order to synchronize their recordings with the markers’ position. Both the force sensors and the scanners were calibrated by the manufacturers and their functioning was verified in our laboratory.

2.2 Kinematic model

In accordance with the standards of the International Society of Biomechanics (ISB), a kinematic model has been developed to obtain the anatomical points of reference and the coordinate systems (CSs) of the UL segments from the three-dimensional positions of 28 active markers (Fig. 1A) (Pérez-Rizo et al., 2013; Wu et al., 2005). Following the ISB standards, the anteroposterior axis was defined as X-axis, with the anterior direction being positive, the longitudinal axis was defined as Y-axis, with the upward direction being positive, and the mediolateral was defined as Z-axis, with the lateral direction being positive (Wu et al., 2005). The center of the shoulder, elbow and wrist joints, and the CSs of the body segments corresponding to the thorax, arms, forearms and hands were defined (Fig. 1A). The lateral and medial epicondyles reference points were calculated from a rigid cluster of 4 markers attached to the upper arm by elastics straps (Fig. 1A). In the same way, the thorax reference points corresponding to the C7 vertebra, the incisura jugular, the T8 vertebra and the processus xyphoideus were calculated from a rigid cluster of 4 markers attached to the thorax by a harness (Fig. 1A). The clusters were oriented in such a way that the markers emitted the infrared light towards the scanners, and the thorax cluster was located on the back side of the thorax’s patient at such a distance that the light emitted by the marker did not interfere with the arms during the walk records (Fig. 2) (Cappozzo et al., 1995). With regards to the crutches, a CS was defined for each segment of the crutch from 3 markers used to record its movement (Fig. 1B) (Requejo et al., 2005).

In order to calculate the flexo-extension, adduction-abduction and internal-external rotation movements of the shoulder joint, Cardan-Euler algorithms were used according to the rotation sequence ZXY (Zatsiorsky, 1998). The spatiotemporal parameters were calculated from 2 reference points defined at the distal ends of the crutches (Fig. 1B).

2.3 Kinetic model

The kinetic model was developed based on the kinematic model and the forces given by the load cells. The segments corresponding to the thorax, arms, forearms and hands were defined, as well as one segment for each element of the crutches, i.e., tip, 6 DOF load cell, handle, upper handle bar, 1 DOF load cell, and forearm cuff (Fig. 1B). Each segment was considered as a rigid solid with a uniformly distributed mass. Forces and torques were determined between 2 consecutive rigid solids at each joint by inverse dynamics algorithms, according to the Newton-Euler methodology (Zatsiorsky, 1998): ${\vec{F}}_{proximal} = \sum_{i = 1}^{n} m_{i} ({\vec{a}}_{i} + \vec{g}) + \sum_{i = 1}^{q} {\vec{F}}_{j}$ (1)

$\begin{matrix} {\vec{T}}_{proximal} = \sum_{i = 1}^{n} {{\vec{M}}_{i} + {\vec{R}}_{i} \times [m_{i} ({\vec{a}}_{i} + \vec{g})]} \\ + \sum_{j = 1}^{q} ({\vec{P}}_{j} \times {\vec{F}}_{j}) + \sum_{k = 1}^{p} {\vec{τ}}_{k} \end{matrix}$ (2)

where ${\vec{F}}_{proximal}$ is the reaction force on the proximal analyzed joint in the Global Coordinate System, m_i is the mass of segment i, ${\vec{a}}_{i}$ is the acceleration of segment i, n is the number of distal segments connected in chain, q is the number of external forces, ${\vec{F}}_{j}$ is the applied external forces, ${\vec{T}}_{proximal}$ is the torque on the proximal analyzed joint in the Global Coordinate System, ${\vec{M}}_{i}$ is the inertial moment due to segment i, ${\vec{R}}_{i}$ is the distance from the center of mass of each distal segment i to the proximal joint, ${\vec{P}}_{j}$ is the vector from the application of the external force to the proximal joint, p is the number of external couples and ${\vec{τ}}_{k}$ is the applied external couples.

The point of application of the forces of the hand on the crutch handle was assumed to be at the projection of the center of the third metacarpal head on the longitudinal axis of the crutch handle. With regard to the kinetics between the cuffs of the crutches and the forearms, forces in the lateral and longitudinal directions and torques in the three directions were assumed to be negligible. Anteroposterior force measured by the corresponding cuff load cell was considered applied on the forearm, on the projection of the superior end of the load cell segment on the longitudinal axis of the forearm. The reaction forces and internal torques on the shoulder joint were calculated with respect to the thorax CS. Both the kinematic and the kinetic model were implemented using biomechanical analysis software (C-Motion Inc., Germantown, MB, USA).

2.4 Subjects

Fifteen adult subjects with SCI (12 males and 3 females) were recruited onto the study (mean age ± standard deviation: 38.5 ± 11.46 years), with a mean height of 1.76 m (±0.05) and weight of 80.33 kg (±9.06). See Table 1 for more detailed patient characteristics. All patients had suffered thoracic or lumbar iSCI, ASIA Impairment Scale D (Ditunno et al., 1994), and they were able to perform both RG and two-point SG continuously, without assistance, for at least 10 m. Fractures, luxations or orthopedic surgery of the UL were considered as exclusion criteria, as was UL pain impeding ambulation with crutches. Cardio-respiratory insufficiency was also considered an exclusion criterion. All patients trained the two different gait patterns studied during their rehabilitation.

2.5 Data recording, processing and analysis

After performing a clinical assessment to define the Motor Index of their lower limbs, the active markers of the motion capture system were placed on the patient. Each patient then walked along a 10 mwalking corridor trying out the 2 different patterns under study (SG and RG, in a randomized order) at their preferred speed, until 5 valid recordings were obtained in terms of signal synchronization and correct execution, discarding recordings with artifacts or patient instability. The frequency used by the motion capture system was 200 Hz, and the load cells recorded data at 1000 Hz. Low pass filters were applied to the signals from the position markers (6 Hz), and to the torques and forces (7 Hz).

For each valid recording, two consecutive cycles were selected (one from the right crutch and the other from the left one), both executed in the center of the walking corridor. A cycle is defined as the time period between consecutive contact between the crutch and the floor. The events when the crutches were lifted off the floor were used to distinguish between the support phase and the swing phase of the crutch.

In order to study the shoulder during the most demanding phase of the cycle in terms of load, only the data of the contralateral shoulder to the most affected lower limb were analyzed, since the loads were expected to be greater in this case (Requejo et al., 2005), and only the kinetic data from the support phase were analyzed. The biomechanical models were applied to the filtered signals, obtaining the flexo-extension, adduction-abduction and internal-external rotation joint trajectories and torques, as well as the vertical, anteroposterior and mediolateral forces for each register. Every joint variable was normalized with respect to the 100% of the cycle duration. Furthermore, the forces were normalized to the patient’s weight, whereas the torque was normalized to both the patient’s weight and height.

The maximum, minimum and range of motion (ROM) of the shoulder rotations, and the mean, maximum and minimum values of the kinetic variables were extracted, and the normalized time when these maximum and minimum values occurred were also calculated. Moreover, spatiotemporal parameters were calculated for each register. With the aim of characterizing SG and RG patterns, the mean and standard deviation of the joint variables were extracted for each 1% of the cycle. Graphical representations of the shoulder joint trajectories (Fig. 3), forces (Fig. 4) and torques (Fig. 5) are shown.

2.6 Statistical analysis

To obtain a descriptive analysis, the mean and standard deviation of the spatiotemporal parameters (Table 2), and the statistical parameters of the joint trajectories (Table 3), forces (Table 4) and torques (Table 5), were calculated for the SG and RG cycles. Given the sample size (15 subjects), the differences between the SG and two-point RG patterns were studied by applying a non-parametric test (Wilcoxon signed-rank) to both the spatiotemporal parameters, and the kinematic and kinetic statistics. P-values lower than 0.05 were considered as statistically significant. All statistical analyses were performed using SPSS software for Windows (version 12.0, SPSS Inc, Chicago, IL, USA).

3 Results

3.1 Spatiotemporal parameters

In this study, the cycle length, cycle duration, support and the swing phase durations, speed, cadence, % of the support phase duration and stride length was compared between the SG and two-point RG patterns of gait with crutches in iSCI patients. Most of these values were higher when the SG pattern was employed, yet not significantly (Table 2).

3.2 Shoulder kinematics

The most notable result in terms of shoulder kinematics was that the ROM in the sagittal plane (flexo-extension) was greater in the SG than in RG pattern (p < 0.05). However, the maximum and minimum values of abduction were higher in the RG than in the SG pattern and the maximum values were reached earlier in the two-point RG cycle than in the SG cycle (p < 0.05). The ROM of the internal-external shoulder rotation was also higher in the SG than in the RG (p < 0.05) (Table 3). A graphical representation of the mean cycle corresponding to each gait pattern can be found in Fig. 3.

3.3 Shoulder kinetics

The highest forces were fund in the SG pattern (Fig. 4). As expected, the maximum and the mean followed the upward direction and both were higher in the SG than in the RG (p < 0.01). The maximum force arose later in the SG than in the RG cycle (p < 0.01), and the mean and peak posterior force was greater in the SG than in the RG (p < 0.01). In the mediolateral axis, the strongest forces were the medial ones, for which the mean and peak forces were also greater in the SG than in the RG cycles (p < 0.01)(Table 4).

As was also expected, the torques were higher in the SG than in the RG cycles (Fig. 5), as seen for the mean and maximum flexor torque (p < 0.01). The mean and maximum adductor torques were also higher in the SG than in the RG (p < 0.01). The internal and external rotation torques were generally quite low, with higher internal rotation in the SG than in the RG (p < 0.01) (Table 5).

4 Discussion

We have applied a biomechanical model to analyze shoulder biomechanics when walking with crutches in clinical environment. This is the first study to compare three-dimensional kinematic and kinetic data obtained from adult SCI patients when walking with crutches using the RG and SG patterns. The data obtained confirmed our initial hypothesis that greater forces and torques were applied on shoulder when using the SG as opposed to the two-point RG in adults SCI patients walking with crutches.

The biomechanical model presented here is based on previous studies (Requejo et al., 2005; Slavens et al., 2011) and follows the ISB guidelines. We have adapted this model defining a segment for every part of the crutches instead of defining only two segments for the crutch (an upper segment and a lower segment) as in previous studies (Requejo et al., 2005; Slavens et al., 2011). The model is suitable for being used by a two scanners motion capture system of active markers recording both ULs simultaneously. Superior, posterior and medial forces were predominant in both the gait patterns studied and they were significantly higher in the SG than in the RG in all 3 directions. In terms of torques, the flexion, adductor and internal rotator torques were also predominant in both patterns, and again, significantly higher torques were produced in the SG than in the RG cycles in all 3 cases. Indeed, the flexor torques were the highest torques in both gait patterns.

Spatiotemporal parameters were not significantly different between the two patterns, although most of the parameters in the SG cycles were slightly higher than those in the RG cycles. The spatiotemporal parameters were calculated here with respect to the contact of the crutch with the floor, whereas elsewhere they were based on the contact of the foot with the floor (Crosbie & Nicol, 1990; Haubert et al., 2006; Noreau et al., 1995; Requejo et al., 2005; Slavens et al., 2007, 2009, 2010, 2011). However, no significant differences in speed were evident between the two patterns, which is in line with previous studies (Slavens et al., 2007, 2009). Conversely, we found significant differences in the shoulder flexo-extension ROM, as well as in the internal-external rotation, in contrast to the these previous studies (Slavens et al., 2007, 2009). These differences may be due to the high UL forces required to complete the SG cycle.

The forces exerted on the shoulder in this study were similar to those described elsewhere (Crosbie & Nicol, 1990; Haubert et al., 2006; Noreau et al., 1995; Requejo et al., 2005; Slavens et al., 2009), particularly in reference to two previous studies that presented highly relevant clinical information (Haubert et al., 2006; Slavens et al., 2009). In one of these, the loads exerted on the shoulder of 14 iSCI patients were analyzed while walking with crutches or with a walker, concluding that the forces are greater with crutches (Haubert et al., 2006). The second and more recent study is the only study that compared the differences between RG and SG, in this case in a sample of 9 children with myelomeningocele at the lumbar level (Slavens et al., 2009). The force values in the present study were similar to those values described earlier (Slavens et al., 2009) and it was concluded that the forces were greater in the SG than in the RG cycles, which is also in agreement with previous studies (Slavens, 2009).

Superior, posterior and medial forces were predominant in both gait patterns, which constitutes a threat to the integrity of the glenohumeral joint since it enables upward migration of the humeral head. Consequently, if this upward migration of the humeral head is not counteracted by an adequate response of the rotating shoulder cuff, subacromial impingement might occur. Moreover, the repetitive nature of these forces in iSCI patients walking with crutches may produce degeneration due to overuse. Although the posterior force was weaker than the superior one, the resultant superior-posterior force may compromise the posterior capsule, as well as the tendons of the supraspinatus, infraspinatus and teres minor muscles. While the forces exerted on the shoulder when walking with crutches and those produced by manual wheelchair propulsion are comparable (Haubert et al., 2006), the superior forces on the shoulder during RG are higher than during wheelchair propulsion on a flat floor and they only become comparable during propulsion up an 8% slope (Kulig et al., 1998). Thus, according to our data the forces during SG could be even greater than those produced during wheelchair propulsion.

With regard to the normalized gait cycle, the superior peak force took place earlier in the two-point RG cycle than in the SG cycle. This indicates that patients must achieve stability earlier in the RG than in the SG cycle to avoid falling, especially at slow speeds (Slavens et al., 2009). Although no differences were detected in spatiotemporal parameters, the fact that the peak force appears earlier in the RG cycle may be related to an attempt to decrease gait speed, since the ULs serve to control balance (Slavens et al., 2009).

In contrast to the pediatric population with myelomeningocele where no differences were evident between the SG and RG torques on the shoulder (Slavens et al., 2009), in the present study there were significant differences in all 3 axes. The magnitudes of the toques were found similar to previous studies (Noreau et al., 1995; Requejo et al., 2005; Slavens et al., 2009, 2010, 2011). Flexor, adductor and internal rotator torques were predominant, with higher values in the SG, particularly in the case of the flexor torque. Flexor torque is necessary to counteract the inertial forces generated during the pendular movement of the body while the crutches are on the floor supporting the patient’s weight. This torque also controls the speed of this movement and the precision of the initial contact of the feet with the floor (Noreau et al., 1995). The high mean shoulder flexor torques found during SG have a similar magnitude to the hip flexor torque associated with the gait of healthy subjects (Noreau et al., 1995; Slavens et al., 2009). Internal rotator torque could control the external rotation of the humerus, whereas the adductor torque would counteract the effect of the reaction force between the crutches and the floor, located laterally to the center of the glenohumeral joint.

The biomechanical model applied appears to be effective in detecting significant differences in the loads experienced on the shoulder when adults with iSCI walk with crutches using SG or two-point RG. The results presented here may help develop optimized protocols for the rehabilitation of gait with crutches and improve our understanding of the kinetic behavior of the shoulder during gait with crutches, possibly serving to design new crutches that minimize the impact of the forces exerted on theshoulder.

One limitation of this study is the variability in the analyzed sample. Another limitation is the fact that patients were more accustomed to walk using RG, although all of them used SG during theirrehabilitation. We consider that this methodology can be used in clinical settings to recommend a gait training protocol or a technical aid with less impact on shoulder. Another clear application would be to facilitate data to improve the design of more ergonomiccrutches.

5 Conclusions

A biomechanical model was successfully applied to study shoulder biomechanics in adult patients with SCI walking with crutches in two different gait patterns. The loads exerted on the shoulder of patients with iSCI were higher walking with crutches using the SG rather than the RG pattern. Superior, posterior and medial forces were predominant in both gait patterns, and the strongest forces were those in the superior direction for both types of cycles. The repetitiveness of these forces may cause damage in the capsular ligaments of the shoulder. Biomechanical analysis of gait with crutches can quantify these forces with the aim of preventing or minimizing such damage.

Conflict of interest

None of the authors have any financial or personal relationships, or affiliations, which inappropriately influence any decisions regarding the work or manuscript.

Ethical approval

We certify that all applicable institutional and governmental regulations concerning the ethical use of human volunteers were followed during the course of this research. The study was approved by the Local Ethics Committee and by the Research Commission of the National Hospital for Spinal Cord Injury, and it was conformed to the Helsinki Declaration. All patients gave their written informed consent to participate in the study.

Footnotes

Acknowledgments

This project was funded by Fundación Sociosanitaria de Castilla-La Mancha (No. PI-2010/050). We want to thank all the patients participating in this study for their effort and collaboration.

References

Cappozzo

, Catani

, Croce

U.D

, Leardini

(1995). Position and orientation in space of bones during movement: Anatomical frame definition and determination. Clinical Biomechanics (Bristol, Avon), 10(4), 171–178. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/11415549

Crosbie

W. J.

, & Nicol

A. C.

(1990). Reciprocal aided gait in paraplegia. Paraplegia, 28, 353–363. http://doi.org/10.1038/sc.1990.48

Ditunno

J. F.

, Young

, Donovan

W. H.

, & Creasey

(1994). The international standards booklet for neurological and functional classification of spinal cord injury. Paraplegia, 32(2), 70–80. http://doi.org/10.1038/sc.1994.13

Dyson-Hudson

T. A.

, & Kirshblum

S. C.

(2004). Shoulder pain in chronic spinal cord injury, Part I: Epidemiology, etiology, and pathomechanics. The Journal of Spinal Cord Medicine, 27(1), 4–17.

Faruqui

S. R.

, & Jaeblon

(2010). Ambulatory assistive devices in orthopaedics: Uses and modifications. The Journal of the American Academy of Orthopaedic Surgeons, 18(1), 41–50.

Gil-Agudo

, Pérez-Rizo

, Del Ama-Espinosa

, Crespo-Ruiz

, Pérez-Nombela

, & Sánchez-Ramos

(2009). Comparative biomechanical gait analysis of patients with central cord syndrome walking with one crutch and two crutches. Clinical Biomechanics, 24(7), 551–557. http://doi.org/10.1016/j.clinbiomech.2009.04.009

Haubert

L. L.

, Gutierrez

D. D.

, Newsam

C. J.

, Gronley

J. K.

, Mulroy

S. J.

, & Perry

(2006). A comparison of shoulder joint forces during ambulation with crutches versus a walker in persons with incomplete spinal cord injury. Archives of Physical Medicine and Rehabilitation, 87(1), 63–70. http://doi.org/10.1016/j.apmr.2005.07.311

Kulig

, Rao

S. S.

, Mulroy

S. J.

, Newsam

C. J.

, Gronley

J. K.

, Bontrager

E. L.

, & Perry

(1998). Shoulder joint kinetics during the push phase of wheelchair propulsion. Sep. 132–143.

Liggins

A. B.

, Coiro

, Smith

B. T.

, Mccarthy

J. J.

, & Forest

(2001). Motion, temporal and force profiles of lofstrand crutch usage. Gait & Posture, 13(Abstract), 295–296.

10.

Melis

E. H.

, Torres-Moreno

, Barbeau

, & Lemaire

E. D.

(1999). Analysis of assisted-gait characteristics in persons with incomplete spinal cord injury. Spinal Cord, 37(6), 430–439.

11.

Noreau

, Richards

C. L.

, Comeau

, & Tardif

(1995). Biomechanical analysis of swing-through gait in paraplegic and non-disabled individuals. Journal of Biomechanics, 28(6), 689–700. http://doi.org/10.1016/0021-9290(94)00118-N

12.

Opila

K. A.

, Nicol

A. C.

, & Paul

J. P.

(1987). Upper limb loadings of gait with crutches. Journal of Biomechanical Engineering, 109(4), 285–290. http://doi.org/10.1115/1.3138682

13.

Pérez-Rizo

, Solís-Mozos

, Belda-Lois

J. M.

, Page

Á.

, Taylor

, Pons

J. L.

, & Gil-Agudo

Á.

(2013). Instrumentation and biomechanical model for kinematic and kinetic analysis of upper limbs during gait with crutches. Journal of Accessibility and Design for All, 3(Cc), 127–148.

14.

Rab

, Petuskey

, & Bagley

(2002). A method for determination of upper extremity kinematics. Gait & Posture, 15(2), 113–119. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/11869904

15.

Requejo

P. S.

, Wahl

D. P.

, Bontrager

E. L.

, Newsam

C. J.

, Gronley

J. K.

, Mulroy

S. J.

, & Perry

(2005). Upper extremity kinetics during Lofstrand crutch-assisted gait. Medical Engineering & Physics, 27(1), 19–29. http://doi.org/10.1016/j.medengphy.2004.08.008

16.

Seireg

A. H.

, Murray

M. P.

, & Scholz

R. C.

(1968). Method for recording the time, magnitude and orientation of forces applied to walking sticks. American Journal of Physical Medicine, 47(6), 307–314.

17.

Sie

I. H.

, Waters

R. L.

, Adkins

R. H.

, & Gellman

(1992). Upper extremity pain in the postrehabilitation spinal cord injured patient. Archives of Physical Medicine and Rehabilitation, 73(1), 44–48.

18.

Slavens

B. A.

, Bhagchandani

, Wang

, Smith

P. A.

, & Harris

G. F.

(2011). An upper extremity inverse dynamics model for pediatric Lofstrand crutch-assisted gait. Journal of Biomechanics, 44(11), 2162–2167. http://doi.org/10.1016/j.jbiomech.2011.05.012

19.

Slavens

B. A.

, Frantz

, Sturm

P. F.

, & Harris

G. F.

(2007). Upper extremity dynamics during Lofstrand crutch-assisted gait in children with myelomeningocele. The Journal of Spinal Cord Medicine, 30(Suppl 1(6)), S165–S171. Retrieved from http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2031971&tool=pmcentrez&rendertype=abstract

20.

Slavens

B. A.

, Sturm

P. F.

, Bajournaite

, & Harris

G. F.

(2009). Upper extremity dynamics during Lofstrand crutch-assisted gait in children with myelomeningocele. Gait & Posture, 30(4), 511–517. http://doi.org/10.1016/j.gaitpost.2009.07.125

21.

Slavens

B. A.

, Sturm

P. F.

, & Harris

G. F.

(2010). Upper extremity inverse dynamics model for crutch-assisted gait assessment. Journal of Biomechanics, 43(10), 2026–2031. http://doi.org/10.1016/j.jbiomech.2010.03.026

22.

University of Alabama at Birmingham. (2014a). Complete public version of the 2014 annual statistical report for the Spinal Cord Injury Model System. Birmingham, Alabama: National Spinal Cord Injury Statistical Center. Retrieved from https://www.nscisc.uab.edu/

23.

University of Alabama at Birmingham. (2014b). Spinal Cord Injury Facts and Figures at a Glance. Birmingham, Alabama: National Spinal Cord Injury Statistical Center. Retrieved from https://www.nscisc.uab.edu/

24.

Vankoski

, Moore

, Statler

K. D.

, Sarwark

J. F.

, & Dias

(1997). The influence of forearm crutches on pelvic and hip kinematics in children with myelomeningocele: Don’t throw away the crutches. Developmental Medicine and Child Neurology. http://doi.org/10.1016/0966-6362(95)93501-3

25.

Waters

R. L.

, Yakura

J. S.

, Adkins

, & Barnes

(1989). Determinants of gait performance following spinal cord injury. Archives of Physical Medicine and Rehabilitation, 70(12), 811–818.

26.

, Vanderhelm

, Dirkjanveeger

, Makhsous

, Vanroy

, Anglin

,...& Wang

(2005). ISB recommendation on definitions of joint coordinate systems of various joints for the reporting of human joint motion?Part II: Shoulder, elbow, wrist and hand. Journal of Biomechanics, 38(5), 981–992. http://doi.org/10.1016/j.jbiomech.2004.05.042

27.

Youdas

J. W.

, Kotajarvi

B. J.

, Padgett

D. J.

, & Kaufman

K. R.

(2005). Partial weight-bearing gait using conventional assistive devices. Archives of Physical Medicine and Rehabilitation, 86(3), 394–398. http://doi.org/10.1016/j.apmr.2004.03.026

28.

Zatsiorsky

V. M.

(1998). Kinematics of human motion. Champaign, Illinois: Human kinetics.