Kinetics and Kinematics of Wheelchair Propulsion on Level Surfaces

This subsection focuses on research articles which examined the trunk and upper extremity kinetics (forces, mechanical loads, moments (torque) and kinematics (movement at joints or between body segments) of manual wheelchair propulsion on level surfaces. Level surfaces included surface such as stationary treadmills, ergometers, and/or dynamometers and, smooth floor surfaces. The studies that follow examine propulsion more in-depth but each from a slightly different perspective, looking at slightly different parameters and environments.

Summarized Level 5 Evidence Studies

The following level 5 evidence studies have been reviewed, and the overarching findings from the studies are highlighted in this section. As noted at the start of this chapter, these types of studies are not included in the discussion or in the conclusions.

Mulroy et al. (1996) examined 12 deep and superficial muscles of the shoulder of 17 men with complete paraplegia (T10-L3) during wheelchair propulsion, to identify which muscles might be at risk for fatigue and overuse. Using electromyography, the activity in each muscle was recorded and determined as having mainly push or mainly recovery activity. The authors identify “two synergies of shoulder muscle function during wheelchair propulsion”; push phase was primarily shoulder flexion and scapular protraction, the recovery phase extension, abduction and scapular retraction. The authors’ noted that the cadence for wheelchair propulsion averaged 67 cycles per minute, thus is a repetitive activity. The authors identified that the supraspinatus and pectoralis major muscles may be at higher risk of fatigue due to their involvement in both cycles of propulsion and their high peak intensities during the push phase.

Desroches et al. (2010) described the upper limb joint dynamics during propulsion of a manual wheelchair, particularly the contribution of joint moment to joint stabilization. Their findings indicated that stabilization during propulsion and recovery phases were a large component of the joint forces and moments. Findings indicate that during propulsion, wrist and elbow joints were in the stabilization configuration of wrist extension, ulnar deviation and elbow adduction (angles close to 90°) while the shoulder flexion was in a propulsion configuration but approached the stabilization configuration of flexion and internal rotation (angles primarily greater than 60°).The authors conclude that these results confirm their hypothesis that an important part of joint moment is the contribution to stabilizing joints, in addition to contributing to the force to create propulsion. The authors further discuss how from a mechanical point of view this could be perceived as inefficient however; from an anatomical point of view stabilization is essential to support movement as well as maintaining the integrity of the joint during force application such as during wheelchair propulsion. The authors question if this partially explains the low mechanical efficiency of manual propulsion, and the potential for injury at these joints.


Upper Extremity Kinetics and Kinematics During Propulsion

VanLandewijick et al. (1994) studied the movement and muscular activity of the upper extremity during the push and the recovery phases of propulsion at three different speeds on a treadmill. The participants were 40 “highly trained” athletes with diagnoses of T3-L5 spinal cord injury (n=22), polio myelitis (n=13), spinal bifida (n=2) and lower limb amputation (n=3). The results were analyzed separately as push phase and recovery phase. The results indicate that the amount of elbow movement is dependent on the velocity of the push, with the amount of elbow extension decreasing as velocity increases. The results also indicated that the shoulder was in near maximum abduction at the point where the hand contacts the push rim and that as velocity increased the range of shoulder motion in the first half of the push cycle increased but decreased in the second half. Trunk inclination range did not change however the amount of time in the forward range increased at higher push velocities. Results related to the recovery phase suggest that positive mechanical work exists during this phase at velocities higher than 1.67 ms, approaching one-third of the entire mechanical work of the full propulsion cycle.

Kim et al. (2015) compared the neck and upper limb muscle activity of eight participants with T1-T12 paraplegia with eight able-bodied participants using electromyography. All participants received wheelchair propulsion training such that they could propel 200 meters in 1.5 to three minutes. Test conditions were propelling 200 m three times. The only difference in muscle activity of significance was that of the sternocleidomastoid muscle being more active in the test group. The authors note that the latissimus dorsi muscle was also more active in the test group than the control group, but it did not reach significance. The authors reported that these findings suggest that training and therapy should include education and treatment for the sternocleidomastoid muscle to reduce overuse and possible symptoms similar to visual display terminal syndrome.

Mercer et al. (2006) examined shoulder forces and moments during propulsion at two speeds to determine if biomechanics were related to shoulder injury pathology as identified from MRI and physical exam results. Findings suggest that body mass is associated with higher forces (posterior and lateral) and moments (internal rotation and extension) during the push phase of propulsion therefore higher mass is associated with increased risk for shoulder pathology especially acromioclavicular joint edema or coracoacromial ligament thickening. Findings also suggest that increased speed results in increased stoke frequency and use of larger shoulder forces and moments. Participants’ who used higher posterior force, lateral force or extension moment during propulsion were more likely to have CA ligament edema noted on the MRI; those who used larger lateral forces or abduction moments were more likely to have CA ligament thickening noted. Participants who used higher superior forces and internal rotation moments during propulsion showed signs of shoulder pathology in the physical exam. The authors suggest the necessity for interventions to reduce the forces and moments such as the use of lightweight wheelchairs to reduce rolling resistance and the forces required to propel, as well as proper set-up, body weight maintenance, training in propulsion techniques or alternative methods of propulsion.

Similarly, Gil-Agudo et al. (2014) examined the acute changes of the shoulder cuff soft tissue pre and post wheelchair propulsion at two different speeds but used ultrasound technology. Results indicated that joint forces were stronger in all directions and most moments in the higher intensity propulsion protocol, but the ultrasound parameters were not different before and after each test. Relating kinetic and ultrasound results indicated that high intensity propulsion increased long biceps tendon thickening when medial and inferior forces increased, and that the subacrominal space decreased with increased medial shoulder forces. The authors suggest that the shoulder forces and moments increase as the propulsion intensity increases which may contribute to the development of shoulder pain.

Bregman et al. (2009) compared total propulsion force to tangential propulsion force in 16 participants (five non-disabled, three with tetraplegic level of SCI, eight with paraplegic level of SCI) to determine if the tangential propulsion force results in a greater physiological cost that the total propulsion force (experimental condition). Participants propelled a study wheelchair on a level treadmill for 30 seconds; data from 10 consecutive propulsion cycles was used which was resampled to 100 samples for comparison and averaging. The kinetic and kinematic data was then inputted into the Deflt Shoulder and Elbow model to determine the physiological cost of the two conditions. The results of the kinetic and kinematic data indicate that: 1) the average propulsion cycle was 1.34(0.27) seconds for all three participant groups; 2) the push phase was 51.7(6.33) % of the full cycle; 3) mean force at exerted on the handrim was 18.8(4.7) N with no significant differences between groups; 4) the tangential component of the propulsion force was 11.7(2.8) N resulting in an fraction effective force (FEF) of 63.2(12.6)%, but no significant differences between groups were found. The authors report that based on the output from the DSEM, that the efficiency in manual wheelchair propulsion is related to the co-contraction around the elbow and the higher energy requirements of the shoulder during tangential propulsion compared to the experimental condition. Generally, the results indicate that the forces and moments in tangential propulsion are higher, often significantly higher compared to the total propulsion forces. The authors suggest that propulsion training should therefore not be focused on optimizing force but more so on finding the balance between the direction of force application on the hand rim and the musculoskeletal constraints of the person propelling.

Ambrosio et al. (2005) sought to determine if a correlation existed between shoulder strength and hand rim kinetics and between muscle imbalance and hand rim kinetics. The authors support that based on the finding of a positive correlation between strength and total resultant force (FR) at the hand rim, and that there was no correlation with decreased cadence, that strategies for both stretching and strengthening of the shoulder muscles as well as proper propulsion techniques are essential for rehabilitation.

Soltau et al. (2015) evaluated the symmetry of bilateral propulsion of 80 participants with paraplegia (injury levels not provided) that did not have shoulder pain. The findings suggest there is some asymmetry in propulsion from left to right, which increases with increasing demand on the upper extremity as was found on the 8% grade. The significant differences in joint range of motion (ROM) while statistically significant, were thought not be clinically significant as the differences were almost all less than 5°. The authors conclude that asymmetries in bilateral propulsion are minimal, and that the assumption that propulsion is symmetrical is reasonable for people without shoulder pain or injury that affects strength or ROM.

Jayaraman et al. (2015) examined propulsion kinetics and kinematics of 22 participants who used either the DLOP or semicircular (SC) stroke pattern to determine the influence of an ergonomic metric termed jerk, on shoulder pain. Jerk was measured at the change in direction during the recovery phase. Participants were divided into two groups based on their stroke pattern, and then sub-divided based on presence or absence of shoulder pain. The push phase was identified as being the point when the moment applied to the hand rim was greater than (start) or less than (end) one Nm for a minimum of 10 seconds. The findings suggest that the DLOP results in higher jerk forces than the SC likely due to the increased number of sharp directional changes coupled with increased acceleration/deceleration in the former pattern. The results also identified presence of shoulder pain influenced the jerk forces in that they were lower than in participants without reported shoulder pain. The authors suggested that, based on other non-wheelchair related research on jerk forces the participants with shoulder pain developed a smoother stroke pattern to minimize the impact of pain on propulsion, but they did not negatively affect propulsion effectiveness. The authors also suggested that it would be beneficial to incorporate jerk based metrics into propulsion training/practice in clinical settings.

Russell et al. (2015) observed the kinetics and kinematics differences between two propulsion conditions; self-selected free propulsion and self-selected fast propulsion. The results indicate that there is variability in the effect of increased reaction force magnitude on shoulder net joint moment (NJM) and net joint force (NJF), associated with increased speed of propulsion. The authors suggest that the “magnitude of the shoulder NJM depends on the proximal distal moments created by the NJFs about the centre of mass of the forearm and upper arm segments as well as the adjacent NJM at the elbow.” These results suggest that the positon of the upper extremity in relation to the rear wheel has significant effect on the forces influencing the shoulder during fast propulsion. Additionally, the results suggest that many participants use positional strategies to affect the load at the shoulders during fast propulsion. The authors suggest that comparing these two propulsion conditions in clinical practice may prove useful in propulsion training.

Koontz et al. (2012) compared kinetic and temporal propulsion variables between a level smooth tile surface and a wheelchair dynamometer to determine if differences existed. Force data was collected from the push phase of the propulsion cycle only. Their findings suggest that people who push with higher forces and moments and larger push angles can do so on both the dynamometer and the tile surface. However, there were changes noted in the propulsion curve (moment about the wheel hub), with a shift from predominantly bimodal or flat curves on the dynamometer to predominantly unimodal curves on the tile. The authors also conclude that the correlation between propulsion forces on the dynamometer and body weight can provide a means to estimate the peak propulsion forces on the tile surface (83% of variability accounted for by these two variables)., The authors did not comment on the amount of force they used to define higher forces, larger angles, etc.; it is assumed that those participants who would propel with low forces or smaller angles may not be as well correlated between the two surfaces. Since the participants in this study were experienced with wheelchair use (between six and 28 years of experience), it is not clear if the results apply to people with less experience. The authors identify the use of self-selected speeds as another limitation of the study as they differed across conditions. Since there was not a constant speed condition across subject’s performances it is questioned if the forces could be different at different speeds, however the authors identify numerous issues with obtaining a constant speed condition especially on the tile floor.

Gil-Agudo et al. (2010) examined differences in shoulder kinetics and kinematics of propelling on a treadmill at 3 km/hr compared to four km/hr. Overall, increasing speed increased shoulder net joint forces and moments, as well as cadence and propulsion angle. Analysis revealed that the predominant force on the shoulder during the push phase was posterior which increased in magnitude as propulsion speed increased and the prominent moment was shoulder flexion. This study also found that during the recovery phase the predominant force was anterior and was greater than the posterior force during the push phase. The authors suggest that study of propulsion should therefore include both the push and the recovery phases; the current tendency is to study only the push phases. It is worth noting that the authors indicated that movements of the trunk, scapula or clavicle were not included in their analysis.

Dallmeijer et al. (1998) explored the effectiveness of force application at the hand rim through the energy output and energy expenditure as an indication of propulsion mechanical efficiency, comparing differences for paraplegic and tetraplegic levels of spinal cord injury. They found that mechanical efficiency was lower in the tetraplegic participant group than the paraplegic participant group. Specifically, differences were noted in the force application to the hand rim which resulted in a significantly lower mechanical efficiency in the participants with tetraplegia. The main differences were a larger lateromedially and reduced frontal plane force application at the hand rim which is consistent with the typical muscular movements available for this group. The authors also found that increasing the intensity (speed) of propulsion resulted in an increased stroke angle for participants in the paraplegic group but a decreased stroke angle in participants in the tetraplegic group. The authors suggest that the effectiveness of force application at the hand rim plays a large role in propulsion mechanical efficiency therefore should be part of propulsion training programs in clinical settings.

Yang and colleagues (2012) investigated the effect of back rest height on propulsion patterns on a level surface and 3° slope. The study suggests that the low backrest (defined as ½ the trunk height as measured from seat base to acrominon) allows for greater shoulder ROM, lower cadence and greater length of stroke as evidenced by differences in start and end hand positions on the rim. Propulsion patterns changed with increased slope, independent of the backrest height. During the 3° slope cadence increased and ROM decreased as did the length of the stroke. Although the kinetic force impulse on the pushrim was the same for both back support heights, the authors propose that because the hand remained on the rim longer with the low back testing, the force was distributed over a longer time period therefore the effective force was lower. This in combination with lower cadence suggests a lower overall force applied to the pushrim thereby having potential to reduce propulsion injuries. Authors indicate that frequency of pushrim contact has been associated with median nerve injury therefore the height of the back support was important to consider in optimizing propulsion. The authors do note that their participants all had a low-level paraplegia for which a low back support may be appropriate, and that clinical reasoning is required when generalizing these study results to clinical practice. The authors also identified that the use of sling backrests in their study may have influenced the results in relation to propulsion forces due to postural differences between sling and rigid backrests.

Raina et al. (2012a) quantified and compared the scapular kinematics under two different load conditions during wheelchair propulsion on an ergometer. Load conditions were equated to the propulsive resistive forces that would be experienced on flat smooth surface such as tile (no load condition) and on an incline (8% grade for participants with paraplegia and 4% grade for those with tetraplegia). Participants who needed trunk control assistance were strapped to the back support during testing which was not accounted for in the analysis. The findings in this study suggest that on average there are similarities in scapular movement (anterior tilt, downward rotation and protraction) during the push phase of wheelchair propulsion for people with paraplegia and tetraplegia, with a greater ROM used when propelling up an incline. Participants with tetraplegia demonstrated a significantly higher rate of anterior scapular tilting compared to participants with paraplegia. This group also demonstrated a higher rate of change in scapular motion during the push phase of incline propulsion. The authors propose that the significant differences in downward rotation and protraction of the scapula during incline propulsion are associated with higher risk of shoulder impingement due to the reduction of acromial space in this position. While the differences in forces affecting propulsion were accounted for in the two load conditions, there was not an actual change in the level of the surface therefore there was not a change in the body position in relation to the wheelchair as there is in ascending an actual incline. For this reason, it is questioned if the results are fully representative of propulsion up an actual incline.

Qi et al. (2018) assessed the effect of propulsion speed on manual wheelchair user’s shoulder muscle coordination. Propulsion at higher speeds required significantly more propulsive muscle activity and energy expenditure. Specifically, the findings showed more muscle activity in the early push phase and in the transition between push and recovery phases at higher speed. The authors suggest that this provides further evidence that faster propulsion places higher demands on muscles to provide joint stabilization during transitions. Therefore, strength training and propulsion techniques that improve transitions may reduce UE demands and improve rehabilitation outcomes. This study was performed using a wheelchair ergometer, which may limit the applicability of results to everyday propulsion.

Cloud et al. (2017) examined the impact of seat dump angle on shoulder and scapular motion during propulsion on a set of custom rollers in individuals whose SCI level ranged from C6-L2. Scapulothoracic internal rotation and downward rotation both increased with increased dump. The implication of these differences towards shoulder health is not clear at this time. Glenohumeral kinematics were also measured but no significant difference was found. The authors suggest that risk of subacromial impingement may therefore be similar regardless of seating condition. Long-term effects were not examined in this study, neither was motion of the pelvis, forearm, hand, etc. Spine motion was captured and is commented on below in the trunk movement section.

Goins et al. (2011) described the horizontal and vertical translation of the elbow and elbow angle during three different speeds of propulsion (participants’ own normal, 20% less than normal and 20% more than normal) on two different randomly chosen surfaces (tile & low pile carpet) for people with tetraplegia. Three distinctive elbow movement patterns as well as three distinct elbow angle patterns were noted amongst the seven participants. With this limited number of participants is it difficult to surmise if these are typical patterns or if with an increase in number of participants if the number of patterns would also increase. The primary finding from this study was that with increased speed elbow translation changes, but the range of elbow flexion remained consistently within a mid-range.

Gil-Agudo et al. (2016) studied shoulder kinetics and ultrasonography before and after a high intensity wheelchair propulsion test in both SCI subjects and healthy controls. Peak shoulder forces and moments increased after the test in almost all directions for both SCI and control groups. Ultrasound parameters did not change before and after the test for individuals with SCI. The control group showed changes in Girometti Index and decreasing long-axis biceps tendon thickness. Tendon thickness did not increase as expected; the authors suggest that the test protocol may have been too short to provoke such changes. The authors also note that some of the differences between groups may indicate beneficial adaptations by manual wheelchair users to generate a longer and smoother stroke, reducing upward shoulder peak force and potentially decreasing risk of shoulder pathology. Worth noting, there were no female subjects tested, which may reduce the applicability of the results to the general population.

Gagnon et al. (2016) examined the association between performance-based manual wheelchair tests (MWPT) and upper extremity strength, trunk strength, and postural stability. Shoulder adductor strength on the weakest side was found to best predict performance during the 20 m maximal velocity test. The authors suggest that the complementary role of shoulder adductors in trunk stability may help to optimize the force applied at the handrim during propulsion. Similarly, shoulder adductor strength and anterior seated reaching are two key predictors of performance on the slalom test, explaining 71.3% of variance. This result emphasizes the high demands on dynamic postural control that are imposed by the numerous trajectory changes of this test. In contrast, handgrip strength best predicted performance on the 6-minute propulsion test.The authors note two reasons that handgrip strength may be principal: it has previously been found to characterize overall UE strength and that handgrip strength is key for the frequent stops and start at high velocity that are incorporated in the 6 min. propulsion test. In summary, MWPT performance is explained by a combination of factors; these results support the relevance of UE and trunk strengthening and dynamic sitting balance training in rehabilitation. However, the authors note that assumptions should not be made regarding causative factors based on their results. Small sample size is also identified as a limitation to the study.

Trunk Movement During Propulsion

Julien et al. (2014) completed analysis of kinematic data from a previous study for seven people with C5-7 spinal level injuries to describe the trunk and neck movements associated with manual wheelchair propulsion, in relation to speed of propulsion. The study found that forward flexion at the trunk and neck significantly increased during the push phase of propulsion but not during the recovery phase. Increased speed resulted in greater neck and trunk forward flexion. Lateral flexion and axial rotation were variable among participants with no identifiable patterns and did not change significantly with speed. The study concluded that trunk and neck forward flexion play a part in manual wheelchair propulsion for people with tetraplegia, and as such the neck, trunk and core musculature should be considered in conjunction with the upper extremity in future studies of manual wheelchair propulsion particularly around pain and overuse injuries. It is worth noting that there was variability in the identified AIS level with two of seven participants being an AIS C/D.

Rodgers et al. (2000) completed a prospective controlled trial to determine the impact of trunk flexion during propulsion compared to non-flexed trunk propulsion has on the biomechanical and physiological characteristics considered to be precursors to shoulder pain and/or injury. Participants were assigned to the flexion group (FG) based on trunk flexion past 90° from upright and/or trunk flexion more than 10° during propulsion. All others were assigned to the non-flexion group (NFG). Results indicate that the FG experienced greater shoulder flexion and elbow extension during propulsion than the NFG. The authors suggest this pattern allows greater reliance on trunk excursion to “generate translational forces necessary for wheelchair propulsion.” This reliance increased in the fatigued test for both groups but VO2 max did not increase suggesting trunk flexion is used to compensate for muscle fatigue and not to increase aerobic capacity during high demand prolusion.

Triolo et al. (2013) explored the effect of trunk and pelvis stabilization using electrical stimulation on the trunk and hip extensor muscles on the kinetics and kinematics of propulsion. Five of the six participants completed all three propulsion tasks (self-selected walking speed, sprint and incline) with and without muscle stimulation, the results of which were compared as a series of case studies with each participant being their own control. The results were variable, with stimulation significantly decreasing peak resultant handrim forces, improving efficiency and the ability to lean forward in same three of five participants but only during the level self-selected walking speed propulsion; the effect on the other participants were not changed with stimulation. The small number of participants and the effects of stimulation being seen primarily with the same participants and no changes noted in the other participants, suggests that further research is needed to determine if the benefit noted in this study has clinical application.

In addition to shoulder measurements described above, Cloud et al. (2017), measured thoracolumbar spine curvature with respect to seat dump angle. Contrary to their hypothesis, they discovered that participants had significantly less lordosis with increased seat dump angle of 14°. The authors discuss that this may be a result of more hip flexion causing the pelvis to tilt posteriorly, which in turn flattens the lumbar spine. Kyphosis was also measured but was not significantly affected by seat dump angle.

Gagnon et al. (2016) also comment on the role of trunk strength and postural stability in wheelchair propulsion; these results are discussed in the UE section above.


There is level 1b evidence (from two RCTs: Qi et al. 2018 and Cloud et al. 2017) that seat dump angle affects spinal curvature and scapulothoracc kinematics during wheelchair propulsion; however, the glenohumeral joint may not be affected.

There is level 1b evidence (from one RCT: Triolo et al. 2013) to suggest that electrical stimulation of the hip flexors and trunk muscles during manual wheelchair propulsion on a level surface may reduce the impact on the upper extremity at the handrim.

There is level 4 (from one post-test study: Koontz et al. 2012) evidence to suggest that when propulsion force and body weight are correlated, propulsion force on a wheelchair dynamometer correlates to propulsion force on a smooth level surface such as a tile floor.

There is level 1b evidence (from one RCT crossover: Goins et al. 2011, one prospective controlled study by Gil-Agudo et al. 2016, three post-test studies by Gil-Agudo et al. 2010, Mercer et al. 2006 and VanLandewijck et al. 1994, and one pre-post study by Gil-Agudo et al. 2014) that increasing speed/intensity of manual wheelchair propulsion results in an increase in cadence, increases in shoulder forces primarily in a posterior direction and, changes in elbow translation all of which may contribute to the development of upper extremity pain. However, no differences in shoulder ultrasound parameters were observed (Gil-Agudo et al. 2016).

There is level 1b evidence (from one RCT: Qi et al. 2018) that faster propulsion requires significantly higher propulsive muscle activity and energy expenditure and that faster propulsion requires more muscle activity in the early push phase and in the transitions between push and recovery.

There is level 4 evidence (from one post-test study: Bregman et al. 2009) to suggest that tangential propulsion forces are higher compared to total propulsion forces for people with paraplegic and tetraplegic levels of spinal cord injury as well as for people without a disability.

There is level 4 evidence (from one pre-post study: Russell et al. 2015) that suggests that the forces at the shoulder during fast propulsion are dependent on the forces around the centre of mass at the forearm and upper arm and therefore the position of the upper extremity during the propulsion cycle has a significant effect on shoulder forces.

There is level 4 evidence (from one post-test study: Dallmeijer et al. 1998) to suggest that there are differences in the efficiency of force application at the hand rim between participants with paraplegia and tetraplegia which are a result of differences in available muscle movement/function; force application at the hand rim contributes to a large degree to overall propulsion mechanical efficiency.

There is level 4 evidence (from one post-test study: Mercer et al. 2006) that higher body mass increases shoulder forces and moments, therefore may be associated with a higher risk of propulsion related injuries.

There is level 4 evidence (from one post-test study: Yang et al. 2012) that back rest height influences range of motion used for propulsion, cadence and length of stroke used during propulsion.

There is level 4 evidence (from two post-test studies: Yang et al. 2012 and Raina et al. 2012a) that to propel up a slope cadence increases and a greater range of motion is used at the shoulder and scapula.

There is level 1b evidence (from one RCT: Julien et al. 2013 and one prospective control study by Rodgers et al. 2000) to suggest that trunk and neck flexion during propulsion significantly changes propulsion forces at the handrim and shoulder for people with paraplegia or tetraplegia.

There is level 2 evidence (one prospective controlled trial: Kim et al. 2015) that indicates the sternocleidomastoid muscle is more active during propulsion in people with thoracic level paraplegia than in non-disabled people.

There is level 4 evidence (from one post-test study: VanLandewijck et al. 1994) to suggest that different muscles are primarily active in the push phase than in the recovery phase and that the onset of the different muscle activity does not coincide with the start of each phase.

There is level 2 evidence (from one cohort study: Jayaraman et al. 2015) to suggest that the change in directions during the recovery phase of propulsion result in high forces at the shoulder, (termed jerk) and varies by the type of stroke pattern used and the presence of shoulder pain.

There is level 4 evidence (from one post-test study: Gil-Agudo et al. 2010) that the predominant shoulder force during the recovery phase is anterior and is greater than the posterior force exhibited in the push phase of propulsion.

There is level 1b evidence (from one RCT: Gil-Agudo et al. 2014 and one post-test study by Ambrosia et al. 2005) to suggest that both stretching and strengthening of the shoulder muscles and training for optimal wheelchair propulsion techniques are needed as part of rehabilitation.

There is level 4 evidence (from one post-test study: Gagnon et al. 2016) that anterior and lateral flexion trunk strength, anterior seated reaching distance, and shoulder, elbow, and handgrip strength are moderately or strongly correlated with results of performance-based manual wheelchair propulsion tests.

There is level 4 evidence (from one post-test study: Soltau et al. 2015) to suggest that there are minimal kinetic and kinematic differences between left and right upper extremity propulsion, therefore propulsion effort can be considered symmetrical.