Kinetics and Kinematics of Wheelchair Propulsion on Non-Level Surfaces

The physical environment influences how and where a manual wheelchair is used. Richter et al. (2007b) define cross slope as the slope of a surface perpendicular to one’s path of travel. Sidewalks, pathways and roads have some degree of cross slope to drain water.

This subsection reviews research articles that examined specifically the kinetics and/or kinematic properties of propulsion on non-level surfaces. Several of the articles used the test items from some of the formal wheelchair skills programs to frame the study but did not report on outcomes of those programs therefore were included in this section as opposed to the wheelchair skills subsection. The non-level surfaces explored in these studies included wheelies, curb ascent, ramps, soft surfaces such as carpet and grass, and cross slopes.

Author Year

Research Design

Total Sample Size

Methods Outcome
Martin-Lemoyne et al. 2016





Population: Mean age= 36.0 yr; Gender: males=6, females=4; Level of injury range: C6-T11; Mean time since injury: 11.9 yr.

Intervention: Mechanical load and muscular demands were measured for manual wheelchair (MWC) users using a SMARTWheel installe don participant’s own w/c, an Optotrack motion analysis system, and surface electromyography on the shoulrder muscles. Participants propelled up a ramp with and without a mobility assistance dog (ADMob).  The course had a 4 metre long, 8.5o ramp covered with a thin layer of asphalt. Each intervention was completed 3 times by each participant with rest periods between as needed

Outcome Measures: Spatiotemporal parameters: push phase, recovery phase, propulsion cycle, contact angle, speed; Pushrim kinetic: total force (Ftot),  tangential force (Ftan), mechanical effective force (MEF); Shoulder moments: flexion(flex)-extension(ext), adduction (add)-abduction  (abd), internal rotation (IR)-external rotation (ER); Muscular utilization ratio (MUR); Perception of upper limb effort as measured on a 10 point visual analog scale

1.     The use of an ADMob allows manual wheelchair users to ascend the ramp significantly faster while requiring significantly less upper limb efforts.

2.     Traction significantly increased (p=0.037) wheelchair speed with the ADMob compared with the same task without the ADMob.

3.     A significantly shorter (p=0.013) push phase and significantly longer (p=0.028) recovery phase when using the ADMob compared to without.

4.     Ftot and Ftan were significantly reduced with the use of the ADMob compared to without (p=0.005, and p=0.002, respectively).

5.     The maximum shoulder flexion (p=0.047), add-abd (p=0.017), and IR-ER (p=0.028) net joint moments were significantly reduced with the traction provided by an ADMob.

6.     MUR was significantly reduced for all tested muscles (p<0.022).

7.     The perception of upper limb effort was significantly reduced (p=0.005) when performing the experimental task with traction provided by the ADMob.

Gagnon et al. 2015




Population: Mean age: 40.8 yr; Gender: males=17, females=1; Level of injury: cervical=1, thoracic=17; Level of severity: AIS A=12, AIS B=3, AIS C=2, AIS D=1; Mean time since injury: 8.2 yr.

Intervention: Participants propelled their manual wheelchair (MWC) at a self-selected natural speed on a treadmill at different slopes (0, 2.7, 3.6, 4.8, and 7.1 degrees) which reflected an increase from one unit in height to 20, 16, 12 and 8 units of length respectively. Each angle had two trials lasting 1 min with a 2 min rest between tests.

Outcome Measures: The last 10 consecutive complete propulsion cycles were used to calculate outcomes Temporal parameters (push phase duration, Recovery phase duration, Total cycle duration, Trunk and shoulder movement kinematics (minimum, maximum, excursion movement amplitudes), Shoulder kinetics (flexion/extension, adduction/abduction, internal/external rotation moments), Peak and mean muscular utilization ratio (MUR) and the indicator of muscle work (IMW) for the anterior deltoid, Posterior deltoid, Pectoralis major clavicular fibers, Sternal fibers. Significance was inferred at p≤0.0125.

1.     The average durations of the push phase were similar for all tested slopes (p=0.267), whereas the average duration of the recovery phase declined as the slope become steeper (p=0.043).

2.     The total duration significantly decreased as the slope became steeper, except for during the 2.7° to 3.6° where the slope increment remained similar (p≤0.001).

3.     At the trunk, all minimum, maximum, and excursion movement amplitudes significantly increased as the slope became stepper (p<0.0125), except for minimum and maximum values during the 2.7° to 3.6° slope increment that remained similar (p>0.0125). At the 7.1° slope the greatest maximum forward trunk flexion (60.9°) and the greatest forward trunk excursion (22.4°) was reached.

4.     The mean and maximum shoulder flexion moments significantly improved as the slope increased (p<0.0125), except for the 3.6° to 4.8° and 4.8° to 7.1° slope increments.

5.     The mean adduction moments only significantly improved as the slope increased between 0° and 2.7° (p<0.001), whereas the peak mean value only significantly improved as the slope increased between 0° to 2.7° (p<0.001), 3.6° to 4.8° (p=0.002), and 4.8° to 7.1° (p=0.002) slope increments.

6.     The mean and maximum internal rotation moments significantly increased as the slope became steeper (p<0.0125), except for the 3.6° to 4.8° slope increment.

7.     The mean and maximum MURs and their indicator of muscle work value, significantly increased (ANOVA p<0.001) as the slope became steeper except for the posterior deltoid and that remained comparable between 2.7° to 3.6° slope increment.

Gagnon et al. 2014




Population: Mean age: 40.8 yr; Gender: males=17, females=1; Level of injury: cervical=1, thoracic=17; Injury severity: AIS A=12, AIS B=3, AIS C=2, AIS D=1; Mean time since injury: 8.2 yr.

Intervention: Participants propelled their manual wheelchair (MWC) at a self-selected natural speed on a level treadmill and then at randomly assigned slopes (0°, 2. 7°, 3.6°, 4.8°, and 7°) Each angle had two trials lasting 1min with a 2min rest between trials. Self-selected speeds were determined by timing propulsion over a 20m tile floor three times with a 2min rest between trials

Outcome Measures: Data was divided into the push phase (hand in contact with rim) and the recovery phase (hand not in contact with rim). Data was collected using the SMARTWheel on the non-dominant side. The last 10 consecutive complete propulsion cycles for each trial were used to calculate means for: 1) duration of push and recovery phases and propulsion cycle (both push and recovery phases), 2) contact angles, 3) total force, 4) tangential force, 5) mechanical effective force (MEF), 6) perceived effort. Significance was inferred at p<0.001.

1.     The recovery phase at 0° was 54 to 70% longer than for the other different slopes (recovery phase at: 0°=0.59±0.22, 2.7°=0.27±0.10, 3.6°=0.26±0.09, 4.8°=0.22±.0.08, 7.1°=0.18±0.05; p<0.001).

2.     The final contact angle was similar across all slopes except for the 0° slope, which was significantly lower than all other slopes (final contact angle at: 0°=45.97±9.04, (p≤.001) 2.7°=52.04±9.20, 3.6°=53.46±10.36, 4.8°=57.92±11.82, 7.1°=65.54±9.82,).

3.     Total contact angle remained greater during the level surface than all other slopes ((p≤.005) with the slopes presenting similar total contact angles (p=0.14, p=0.24).

4.     The greatest mean difference of total force and tangential force was found between 0° and 2.7° slopes compared with the differences observed between the other consecutive slopes (mean total force at: 0°=39.56±11.15, 2.7°=76.25±19.55, 3.6°=81.49±18.86, 4.8°=95.49±21.16, 7.1°=119.21±18.42; p<0.001. mean tangential force at: 0°=24.52±8.84, 2.7°=48.04±13.08, 3.6°=52.25±14.27, 4.8°=58.00±14.69, 7.1°=68.05±16.61; p<0.001). 5.     The MEF values were similar across all slopes located at approximately 80% of the propulsion phase (MEF values at: 0°=0.43±0.09, 2.7°=0.44±0.06, 3.6°=0.45±0.10, 4.8°=0.42±0.06, 7.1°=0.38±0.10; p>0.05).

6.     The perceived effort increased as slope angle increased, with the 0° slope having the lowest perceived effort and the 7.1° slope showing the greatest perceived effort (perceived effort at: 0°=1.18±1.10, 2.7°=3.78±2.83, 3.6°=4.06±2.69, 4.8°=5.27±2.80, 7.1°=6.86±2.68; no p-value provided).

Pierret et al. 2014




Population: Mean age: 38.9 yr; Gender: males=25, females=0; Level of injury: T3-L4; Mean time since injury: 10.6 yr.

Intervention: Participants performed two tests: 1) a test involving sub-maximal exertion on an arm ergocycle on the first day to estimate peak oxygen uptake up to 85% maximum heart rate, and 2) eight laps of a 50 m propulsion track with a cross slope (Cs) of 0, 2, 8, and 12 % each at two different velocities (one self-selected, one imposed rate). The intersession interval between tests was at least 2 days.

Outcome Measures: Heart rate (HR), absolute cardiac cost (ACC), relative cardiac cost (RCC), peak oxygen uptake (VO2), energetic cost per meter travelled and per kg weight (ECmkg), relative energetic cost (REC), Rating of Perceived Exertion (RPE) scale.

1.     5 participants were unable to complete the last 50 m lap under all test conditions.

2.     No significant differences were noted in HR or VOfor the 0% and 2% Cs.

3.     The HR, ACC, and RCC are all significantly altered by the velocity conditions (F>95; p<0.001) and, for each velocity, by the three different Cs (p<0.001).ACC also increased by user weight (p<0.001), age (p<0.001), injury level (p<0.001) and VO2 max decrease (p<0.001).

4.     The VO2, ECmkg and the REC values (energetic strain) are all significantly altered by the velocity conditions (p<0.005) and by the Cs for each velocity (p<0.001). The energetic strain increases when age (p<0.001) or body mass index (p<0.001) increase or when physical activity (p<0.001), injury level (p<0.001) or VO2Max (p<0.001) decrease. 5.     The RPE results remain unaltered by the velocity (p>0.04), but the Cs increase significantly the RPE (p<0.001).

Marchiori et al. 2014


Post Test


Population: Mean age: 31.8yr; Gender: males=9, females=2.

Intervention: Participants were instructed to approach an obstacle 8 cm high at a comfortable speed, then lift the caster wheels off then ground just before it, without stopping, and ascend it, using their own wheelchair. The ascent was divided into three phases based on the angle formed between the wheelchair frame and the ground: caster pop (P1), rear-wheel ascent (P2), and post ascent (P3). Participants used their own manual wheelchair.

Outcome Measures: SMARTWheel and eight camera video system to capture 3D joint power, 3D angle between the wrist, shoulder and elbow joint moments and angular joint velocity (moment).

1.     The highest moment and peak net moment of the three joints (i.e., shoulder, elbow, and wrist) was found during P2 in flexion.

2.     Forward trunk flexion started early in the caster pop phase

3.     According to the 3D angle:

·       The wrist was more in a stabilizing configuration during P1 and P2, and generated energy during P1.

·       The shoulder joint was in a stabilizing configuration during obstacle ascent and generated energy during P3.

·       The elbow was in a stabilizing configuration during P3, absorbing energy during P1 and P2.

Lalumiere et al. 2013a




Population: Mean age: 38.0 yr; Gender: males=14, females=1; Level of Injury: T2=1, T4=1, T5=1, T6=2, T7=2, T8=8, T10=3, T11=2, T12=2; Level of severity” AIS A=13, AIS B=1, AIS C=1; Mean time since injury: 9.5 yr. All MWU >4 hr/day, and self-reported independence with curb ascents of ≤12 cm with no shoulder pain.
Intervention: Participants were asked to complete three curb ascend tasks (curb height=4cm, 8cm, and 12cm) at a self-selected speed in their own w/c with 3m approach.Outcome Measures: Trunk and upper extremity kinematics and shoulder, elbow and wrist net joint moments using: a motion analysis system (Optotrak) with 23 skin-fixed markers and four markers attached to w/c frame; two instrumented rear wheels (SMART wheels) and; surface electromyography. Measures compared at caster pop, rear-wheel ascent and post ascent phases to determine related effect of curb height.
1.       All participants ascended 4 and 8 cm curbs; 80% (n=15) were able to ascend the 12 cm curb.

2.       Curb approach speeds differed significantly (p<0.0001) with speeds progressively increasing as the curb height increased. 3.       Curb height did not affect total duration (p=0.7), the duration of the caster pop phase (p=0.849) or the rear wheel ascent (p=0.077). 4.       In the sagittal plane of motion most movement differences were noted. maximum trunk flexion along with the total excursion of trunk flexion, maximum shoulder flexion, and greater flexion, extension and movement excursion in the plane of motion at the elbow, all progressively increased as the height of the curb was increased from 4cm to 8 cm (p≤0.001, p≤0.0001, p≤0.004 respectively), and then from 8cm to 12cm (p≤.0001, p=0.008, p≤0.004 respectively). However, the excursion of shoulder movement in the sagittal plane only improved significantly when the curb height was increased from 4 cm to 8 cm (p≤0.0001). No movement difference was confirmed at the wrist across the various curb heights (p>0.05).

5.       Compared to the 4 cm curb, all mean and peak total net moments produced at the shoulder, elbow and wrist significantly increased when ascending the 8 cm (p≤0.0001) or 12 cm curb (p≤0.01).

6.       Compared to the 8cm high curb, only the mean shoulder (p=0.001) as well as the peak and mean elbow total net joint moments (p≤0.009) further increased to a significant extent when ascending the 12cm high curb.

7.       Compared to the height of 4 cm, the peak rate of rise (ROR) values of the total shoulder net joint moment and of the shoulder flexion net joint moment were found to be significantly greater when ascending a height of 8 or 12 cm (p≤0.005). However, these values were similar when ascending an 8cm or 12 cm curb (p≥0.299).

8.       All mean (p≤0.031) and peak (p≤0.039) muscular utilization ratio (MUR) values for the upper extremity muscles assessed differed significantly across all heights.

Lalumiere et al. 2013b




Population: Mean age: 38.1yr; Gender: males=15, females=1; Level of injury: T=15 (T2-12), C=1 (C7); Mean time since injury: 9.2 yr.

Intervention: Compare the effects of four distinct rolling resistances (RRs) on the intensity of handrim kinetic measures on the non-dominant upper-limb (U/L) as well as symmetry (i.e., dominant versus on-dominant) of forces during the execution of wheelies among manual wheelchair users with SCI. Four wheelies per four randomized RRs including: (1) natural surface of painted high-grade smooth composite board (NAT), (2) 5-cm thick urethane soft yellow foam (LOW), (3) 5-cm medium viscoelastic pink memory foam (MOD), and (4) two 5-cm high wooden blocks with rear wheels completely blocked (HIGH).

Outcome Measures: Handrim kinetics: resultant force (Ftot), medial force (Fz) and tangential component of the resultant force (Ftg) measured using two instrumented wheels (Smart Wheels) during four phases of the wheelie: preparation, take-off, balance, and landing as measured by the angle between the w/c frame and ground surface. Motion analysis system used to synchronize data from instrumented wheels; symmetry index intensity measured to verify if forces were similar bilaterally.

1.     No significant differences in duration of each phase of the wheelie, except for the wheels blocked (High) for take-off and landing which were longer than all other surfaces.

2.     The mean and maximal Ftot were greater (p=0.001-.009) during the HIGH RR compared to the other RRs. During the preparation phase, Ftg patterns showed a forward force application compared to a quick backward force with all other RRs.

3.     The maximal Fz was similar across all RRs.

4.     The mean and max Ftot were greater during the take-off phase of performing a wheelie, compared with the other phases (preparation, balance, and landing phases) for all RRs. The mean and max Ftg were greater also during the take-off phase compared with all other phase regardless of RR. The mean Fz was similar during the balance and landing phases, however was significantly greater during the take-off phase compared to the preparation phase.

Nagy et al. 2012




Population: Mean age: 38 yr; Gender: males=20, females=3; Level of injury: tetraplegia=5 (C6-T1), paraplegia=19 (T4-L3); Mean time since injury: 14.8 yr.

Intervention: All participants used their own ultra-lightweight manual wheelchair and seating. Each had one practice and then one test trial of a series of eight of the following skills from the Wheelchair Skills Test: 10m tile surface, 10m of carpet surface, soft surface, 5° and 10° ramps, 2 cm, 5 cm and 15 cm curbs.

Outcome Measures: SmartWheel used to analyze push rim forces exerted during propulsion. Peak force for the first four skills was calculated from the entire performance; peak for the remaining skill were taken from the pushes that allowed successful completion. Mean peak force comparisons were completed using paired t-test for each skill to the 10 m tile skill.

1.     The mean peak pushrim forces were as follows for the skills: 10 m tile=101 N, 10 m carpet=103 N, soft surface=148 N, 5° ramp=138 N, 10° ramp=157 N, 2cm curb=119 N, 5 cm curb=155 N, 15 cm curb=232 N. **Only 6 szzubjects completed the 15cm curb).

2.     Comparison between mean peak forces of each skill compared to 10 m tile were all statistically significant (p=0.0001-.267) except the 10m carpet.

Morrow et al. 2010




Population: Mean age: 43 yr; Gender: males=11, females=1; Injury etiology: SCI=11, spinda bifida=1; Duration of manual w/c use: 18 yr.

Intervention: Five trials, with rest between, propelling at a self-selected speed for each condition in the following order: 1) push phase of level propulsion, 2) push phase of ramp propulsion (1:12 incline), 3) push phase of start, 4) negative acceleration phase of stop, 5) weight relief maneuver (push up and hold for 3 sec).

Outcome Measures: Two instrumented rear wheels (SmartWheels) on participants on manual wheelchair to capture force data at handrim; Motion analysis system (Real-time Eagle) with 15 markers on the trunk and right upper extremity and three each on the rear wheels to capture moments; Force of direction was defined as anterior (+) and posterior (-) of the x axis, medial (+) and lateral (-) of the y axis and superior (+) and inferior (-) of the z axis. Moment direction was defined as flexion or extension about the trunk z axis, elevation abduction and elevation adduction about the humerus x axis and internal and external rotation about the humerus z axis.

1.     There was a significant main effect of condition for the shoulder intersegmental forces in 4 of 6 force directions: anterior (p=0.001), posterior (p<0.001), medial (p=0.003), and superior (p<0.001).

2.     Post hoc analysis of the intersegmental shoulder forces indicated that: 1) in ramp condition the anterior force was significantly higher than level propulsion, weight relief , start and stop conditions, 2) posterior force of the ramp and weight relief conditions were significantly higher than level, start and stop conditions, 3) weight relief medial force was significantly higher than level, start and stop conditions, 4) the level, start and stop conditions were all statistically equivalent for all force conditions.

3.     There was a significant main effect for the shoulder intersegmental moments for three of six moment directions: extension (p<0.001), adduction (p=0.009), and external rotation (p=0.004).

4.     Post hoc analysis of the intersegmental shoulder moments indicated that: 1) extension moment for weight relief was equal to start but significantly greater than level, ramp and stop conditions, 2) Adduction moment for ramp was significantly higher that level condition, 3) external rotation moment of ramp and start were significantly greater that in the level condition, 4) abduction (p=0.092) or internal rotation (p=0.102). There was no main effect of condition for flexion.

Hurd et al. 2008




Population: Mean age: 43.6 yr; Gender: males=11, females=1; Injury etiology: SCI=11, spina bifida=1; Level of injury range: T4-L10; Duration of w/c use: 18 yr.

Intervention: Evaluated U/L symmetry during self-selected propulsion rates across eight different terrain conditions consisting of propelling straight forward in laboratory, outdoor community and indoor community. The outdoor community was a single continuous 500m concrete sidewalk that progressed across four conditions in this order; 1) 2° right side lower cross slope; 2) smooth level surface; 3) level aggregate (textured) surface; 4) 3° ramp (1:19 rise to run) smooth surface. Indoor community =1) 10 m level, low pile carpet and 2) 4.8° ramp (1:12 rise to run) with low pile carpet. Laboratory= 1) 10 m smooth level tile surface and 2) dynamometer with level surface. 1 trial completed for outdoor community items; three trial of indoor community and 1st laboratory items and; 1 30 trial on dynamometer.

Outcome Measures: Three push cycles using two instrumented rear wheels (Smart Wheels) were averaged to capture propulsion timing, effort and force using variables of moment, total force, tangential force, fractional effective force, time-to-peak propulsion moment, average work in joules, contact (length of push cycle) and instantaneous power. Symmetry index was used to determine symmetry of U/L propulsion (perfect symmetry=0 used for comparison).

1.     Symmetry indexes were significantly different within each condition across all variables.

2.     Between conditions, symmetry variables were also significantly different (propulsion moment, p<0.001; total force, p=0.004; tangential force p<0.001; fractional effective force, p<0.001; time-to-peak propulsion moment, p=0.001; work, p<0.001; contact, p<0.001; power, p<0.001)

3.     Comparing the within lab conditions (tile floor versus dynamometer) indicated no differences in symmetry indices for any variable

4.     Comparing the lab versus indoor conditions indicated no significant differences in symmetry indices for any variable.

5.     Comparing lab versus outdoor conditions resulted in significant differences in symmetry indices for all variables with outdoor being greater than lab except for time-to-peak moment (p=0.188)

6.     No patterns of dominant versus non-dominant upper limb contribution to propulsion were noted.

Richter et al. 2007b




Population: Mean age: 36 yr; Gender: males=19, females=7; Level of injury: paraplegia=24, spina bifida=2; Chronicity: chronic.

Intervention: Propulsion of personal wheelchair on a treadmill set at level, 3° and 6° inclines.

Outcome Measures: Speed, Force, Torque and loading rate, Cadence, Push angle, Power output, Push distance.

1.    All kinematic factors increased significantly when the incline increased from level to 6°: peak handrim force, 1.4 increase; loading rate, 1.3 increase; axial moment, 1.8 increase (p=0.00). Push angle and cadence were not affected.

2.    As the incline increased, distance traveled forward per push dropped (3°, p=0.034; 6°, p=0.00). Subjects utilized approximately 80 and 100 more pushes/km for the 3° and 6° inclines.

3.    Coast time decreased from 0.43 sec (level) to 0.35 sec (6° incline).

4.    Power output for the downhill wheel increased 1.6 and 2.3 times more than level for 3° and 6° (p=0.00).


Richter et al. (2007b) investigated the effect of cross slope on wheelchair handrim biomechanics. The data from this study indicates that more pushes are required to cover the same distance when on a cross slope and that the power required increased by a factor of 2.3 on a six-degree cross slope. Users must push harder on the downhill handrim and this increased loading may result in overuse injuries.

LaLumiere et al. (2013b) compared the effects of different rolling resistances on hand rim kinetics during manual wheelchair wheelies performed by people with a spinal cord injury (T12-C7) who had no history of shoulder pain. The rolling resistance (RR) was created by the surface on which the wheelie was performed; painted, high grade smooth composition board (NAT); five cm thick urethane soft yellow foam (LOW), 5-cm medium viscoelastic pink memory foam (MOD), and two five-cm high wooden blocks with rear wheels completely blocked (HIGH). The wheelie was analyzed in four phases; preparation, take-off, balance and, landing. Findings indicate that the HIGH RR was the least desirable surface for performing wheelies. The HIGH RR produced the greatest mean and total hand rim forces at all phases, showed a forward force application to lift the casters off the ground whereas all others used a quick backward force. The authors also found that the take-off phase mean and maximum resultant forces and mean and maximum of the tangential components of the resultant forces were greater than all other phases regardless of the RR.

The authors conclude that completing wheelies with the rear wheels blocked requires different motor learning strategies than on the other surfaces. Symmetry between dominant and non-dominant upper extremities was also evaluated in this study, with the findings suggesting that exertion forces are symmetrical in each phase. However, during the balance phase, the direction of the exerted forces differed on the NAT and LOW surfaces with the different direction oscillating between the dominant and non-dominant upper extremities to maintain balance. The authors reported looking to another study (Boninger et al 1999) in which the same propulsion forces were used to compare propulsion forces to the forces required to complete a wheelie on the NAT surface. From this comparison, they conclude that the forces are similar between these two skills, and that given the frequency of propulsion compared to performing wheelies, wheelies may represent a decreased risk to UL’s versus propulsion. However, the authors did not expand on this comparison so it is questioned that if the intensity of these two skills are the same, would an increase in the frequency of wheelies result in a similar risk exposure as propulsion.

LaLumiere et al. (2012a) compared movement strategies (kinematics), mechanical loads (kinetics) and relative muscle demands on the non-dominant side while 15 people with paraplegia ascended curbs of four, eight and 12 cm heights; participants propelled a three-metre approach at a self-selected speed. The authors hypothesized that the mechanical loads and muscular demands, especially at the shoulder would increase as curb height increased. The curb ascent was divided into the phases of caster pop-up, rear wheel ascent and post ascent phases. The authors report that the greatest net joint moment for all curb heights was shoulder flexion, closely followed by shoulder internal rotation and elbow flexion, which were corroborated by their EMG results. This study found limited elbow extension effort with this skill of curb ascension; in fact, the elbow flexors (long head of biceps) were used to succeed with ascending curbs. The muscle utilization ratio (MUR) at the pectoralis major, anterior deltoid and biceps brachii indicate these muscles contribute highly to these primary moments involved in ascending curbs. The moment demands placed on the shoulder and elbow joints progressively increased from a four to 12 cm curb, specifically 2.2 times for shoulder flexion and internal rotation, 2.8 times for shoulder adduction and 1.8 times for elbow flexion.

Similarly, the muscle demands as measured by EMG, increased as the curb height progressively increased. Considering the substantial shoulder and elbow demands with this task found in this study, the authors suggest that it is plausible that a decreased strength-generating capability at the shoulder flexors/adductors or at the elbow flexors could increase the mechanical demand and increase risk of musculoskeletal injury. The authors also found that forward trunk flexion increased as the curb height increased, suggesting that the forward momentum created by flexing the trunk and head in the direction of movement assisted in the second phase of rear wheels ascending the curb. The authors do report that the possible contributions of using forward trunk flexion were not fully examined in this study but they do propose there is benefit to include trunk flexion strategies in curb ascent training to augment the increasing demands on the shoulders and elbows as the curb height increases. Based on this study’s findings, the authors highlight clinical implications for injury prevention focused on 1) the individual and optimizing strength at shoulder flexors, shoulder adductors, and elbow flexor muscles, and determining the ability to use forward trunk flexion and 2) the environment by continuing to advocate for barrier free environments to decrease upper extremity risk exposure.

Marchiori et al. (2014) examined the joint angle and velocity during obstacle ascent in a manual wheelchair by 11 people up an 8 cm curb. Findings suggest increases in peak moments in the wrist, elbow and shoulders compared to propulsion, although their study did not measure level propulsion. Forward trunk flexion during the caster pop phase was stated to be supported by other study results, suggesting forward trunk flexion during this phase may reduce upper extremity strain, but this study did not provide supporting data.

Nagy et al. (2012) examined the pushrim forces during various advanced manual wheelchair skills compared to forces exerted during propulsion over a 10-metre tile surface. Advanced skills tested were from the Wheelchair Skills Test developed at Dalhousie University, which included; 10 meters of carpet, a soft surface, 5° and 10° ramps and 2 cm, 5 cm, and 15 cm curbs. The primary finding that the more advanced the skill the more force required. The authors note an increase in forces ranging from 18 to 130% but do not provide details of calculations. Discussion in this article focuses on the need to consider the forces being exerted during advanced wheelchair skills and the need to preserve upper extremity integrity through minimizing repetitive forces. However, the authors did not note if the participants were experienced with basic or advanced wheelchair skills nor the potential influence of skill experience on the forces exerted during the skills measured. The authors also did not discuss the implications or the need to balance minimizing the impact of pushrim forces with maintaining an active lifestyle or to the impact of wheelchair set-up/technique on the force.

Hurd et al. (2008) examined the symmetry of propulsion across a variety of terrains, for people with paraplegia. Findings indicated that propulsion asymmetries exist for all conditions with the magnitude of the difference being affected by the environment/terrain. Outdoor condition had the greatest magnitude of propulsion asymmetry. No differences were found in the magnitude between laboratory (tile floor and dynamometer) and indoor community conditions. The authors note that their results could not explain these differences, but they question the effect of fatigue on the results as the outdoor conditions were completed in one continuous pathway to simulate actual outdoor conditions, whereas the others were single testing conditions with rests in between. Dominance did not appear to have a role as no patterns of dominant versus non-dominant upper limb use during propulsion was detected for any condition. For these reasons the authors caution the use of single or averaged bilateral data for propulsion-based studies. The authors also highlight that these results, despite the limitations, underscore the need to complete propulsion evaluations and training in the person’s own natural environments to fully understand propulsion kinetics and kinematics.

Morrow et al. (2010) examined intersegmental shoulder forces and moments during everyday propulsion activities for daily life and mobility. Findings indicated that forces and moments vary significantly across the conditions used to simulate daily life and mobility activities. Not surprisingly, the condition of pushing up and hold for three seconds produced significantly higher shoulder forces than the level, start and stop propulsion conditions. The magnitude of forces at the shoulder was highest for the push-up condition followed by the ramp condition in most directions of force. The push-up maneuver resulted in a peak superior direct force two times greater than the magnitude of ramp propulsion and three times the magnitude of level propulsion. The authors suggest that the findings indicate that the push-up maneuver and propelling up a ramp are very high loading activities compared to level propulsion and as such the frequency of these high loading activities needs to be considered as part of maintaining shoulder health. Regarding shoulder moments, most shoulder moments during ramp propulsion and start conditions were equivalent but higher than level propulsion. Extension and abduction moments were higher in ramp propulsion, weight relief and start conditions compared to level propulsion. The authors suggest these findings are indicative of push-up condition, ramp propulsion and start conditions placing the largest estimated loads on the shoulder during propulsion.

Gagnon et al. (2014) examined the spatiotemporal propulsion cycle and push rim kinetics of the non-dominant hand during manual wheelchair propulsion in 18 people with spinal cord injury on a level surface and up four different slopes on a wheelchair treadmill. The slopes chosen corresponds to a 1:20, 1:16, 1:12 and 1:8 ratio of vertical height to horizontal length of the slope, similar to standards for ramps. Overall, they found that the push phase remained relatively the same on all slopes however, the recovery phase became shorter as the slope increased, with the recovery phase at the level surface being significantly longer than the slopes (54% – 70%). Therefore, as the authors suggest, the pushing frequency increases to offset the gravitational effect of the slope on the wheelchair. The initial contact on the rim moved forward with increasing slope and contact angle remained similar on the slopes equal to or greater than 3.6⁰. The authors question if the contact angle results are related to the forward flexion of the trunk during propulsion on slopes which was explored in their 2015 study (see below). Forces applied to the push rim increased as the slope increased, 200% at the greatest slope, however no similarities between the slopes were found which, the authors suggest, indicates the relationship between slope and push force is not linear. The authors suggest that these findings support the need for ramps with smaller slopes (2.7⁰ or 3.6⁰ which correspond to 1:20 and 1:16 respectively) as these slopes require similar effort and the greater slopes of 1:12 (4.8⁰) require greater effort, use greater forces, require more frequent push phases therefore have greater implications for shoulder integrity maintenance.

Gagnon et al. (2015) also examined the kinematic changes of the trunk and non-dominant shoulder in 18 people with spinal cord injury during manual propulsion up five different slopes (0°, 2.7°, 3.6°, 4.8° & 7.1) at a self-selected speed. All participants could maintain their self-selected propulsion speed of 1.17±0.18 m/s on the level surface and the 2.7° slope but only 88.9%, 77.8% and 55.6% were able to maintain it on the 3.6°, 4.8° and 7.1° slopes respectively. Forward trunk flexion, peak shoulder flexion, and shoulder mechanical and muscular efforts all increased as the slope increased. The authors suggest that the forward trunk flexion in conjunction with the forward trunk excursion may assist in moving the centre of mass anteriorly to prevent backward tipping as the slope increased. The authors also suggest that the increase in shoulder flexion but comparable flexion excursion across all slopes may be related to the need to accommodate at the shoulder for the forward trunk flexion. Additionally, the muscular and mechanical demands of the shoulder, particularly of the posterior deltoid muscle at the end of the push phase, also increased as the slope increased. The authors suggest that these finding support the clinical practice of high-intensity, short duration strength training for the upper extremity, especially the shoulders, to reduce the risk of shoulder integrity issues.

Martin-Lemoyne et al. (2016) also examined the mechanical load and muscular demands of the shoulder when ascending a 4-metre-long, 8.5° ramp; however the focus was on comparing ramp ascent with the assistance of a mobility assistance dog (ADMob) and without assistance. They found that with the ADMob the ascent was 38.3%faster, the push phase was 45.4% faster and the recovery phase was 38.6% longer. Participants also demonstrated significantly lower shoulder net movements (flexion, adduction and internal rotation) and lower upper limb exertion at the lower deltoid, biceps, triceps and pectoralis major muscles. Overall the authors noted reduced mechanical and muscular demands and participant perceived upper limb effort was 62.8% lower when using the ADMob. The authors suggest that more research is required to explore the effects of this type of intervention on the dog, as well as to explore potential challenges that may arise in daily life with a dog.

Pierret et al. (2014) examined the cardiorespiratory effect and perceived strain experienced by 25 men who sustained a thoracic or lumbar spinal cord injury during manual wheelchair propulsion on cross slopes of zero, two, eight and 12%. They found that cross slopes of zero percent and two percent did not differ in cardiorespiratory and subjective strains but that the 8% cross slope was found to have significant effects on cardiorespiratory strain and perceived strain but all participants were able to manage; not all participants were able to manage the 12% cross slope.


There is level 4 evidence (from one post-test study: Richter et al. 2007b) that wheeling cross slope results in increased loading on users’ arms and may lead to overuse injuries.

There is level 4 evidence (from one post-test study: Nagy et al. 2012) that advanced wheelchair skills require greater peak forces at the hand rim, however there is level 4 (from one post-test study by LaLumiere et al. 2013b) evidence that wheelies require a mean peak hand rim force similar to that of wheelchair propulsion.

There is level 4 evidence (from one post-test study: LaLumiere et al. 2013a) that ascending curbs of increasing height increases the mechanical and muscular demands at the shoulder and elbow joints placing these joints at risk of injury especially if adequate strength in the associated muscles is not present.

There is level 4 evidence (from one post-test study: Hurd et al. 2008) upper limb asymmetries exist in manual wheelchair propulsion with greater asymmetry in outdoor versus laboratory (tile floor and dynamometer) conditions.

There is level 4 evidence (one post-test study: Morrow et al. 2010) that the daily life and mobility activities of a push-up, ramp propulsion and the start phase of propulsion place the larger estimated loads on the shoulder and use greater shoulder abduction and extension moments compared to level propulsion.

There is level 2 evidence (from one lower RCT study: Martin-Lemoyne et al. 2017) that mechanical and muscular demands as well as perceived upper limb effort are significantly reduced when ascending a steep ramp with the assistance of a mobility assistance dog compared to without.

There is level 4 evidence (from one pre-post study: Pierret et al. 2014) that suggests the physiological demands of propulsion increase with increasing cross slopes beyond 2%, and that slopes greater than 8% significantly pose significant challenges both physiologically and physically.