Wheelchair Propulsion Training

Wheelchair training is one of the eight key phases for optimizing wheelchair service delivery outlined by the World Health Organization. For manual wheelchairs, there appears to be two distinct but related aspects of training in the literature; wheelchair skills training and manual wheelchair propulsion training. The former is covered in the Wheelchair Use section and relates to mastering management of the wheelchair in different situations and environments such as ramps, curbs, folding the manual wheelchair. The latter is reviewed in this subsection.

Manual wheelchair propulsion is studied using kinetics and kinematics such as contact angle, stroke frequency and mechanical efficiency, to evaluate how to optimize manual propulsion, thereby affecting the potential risk for chronic overuse injuries related to propulsion. Propulsion training focuses on how these optimized techniques are translated into everyday use. The studies reviewed explore the delivery of this training and the effect of this type of training over time.

Author Year

Research Design

Total Sample Size

Methods Outcome
Zwinkels et al. 2014


Review of published articles between inception to October 2013


Methods: Articles published in English focused on exercise training with at least one outcome measure for wheelchair propulsion (i.e., cardio-respiratory fitness, anaerobic capacity, muscular fitness, or mechanical efficiency).

Databases: PubMed and EMBASE.

Levels of EvidenceModerate quality: Low quality RCTS, prospective controlled trials; Very low quality: Case Series, case reports.

Questions/ Measures/ Hypothesis:

To review the literature on the effectiveness of training programs on improving hand-rim wheelchair propulsion capacity.

1.     There was a total sample of 249 (50% SCI).

2.     For all studies examining interval training (n=8), endurance wheelchair propulsion capacity was found to significantly improve in the experimental groups (ranging from 18-34% in individuals with disabilities).

3.     In studies that reported sprint wheelchair propulsion (strength studies, n=2), strength training was not found to be effective in improving sprint performance.

4.     Overall, Mixed Training (n=6) studies were shown to improve endurance wheelchair propulsion.

5.     For the endurance studies (n=5), three studies reported significant improvement in endurance outcomes, two in peak oxygen intake, and only one study (with an able-bodied sample) showed significant improvement in mechanical efficiency.

Rice et al. 2013





Population: Mean age :40.0 yr; Gender: males=24, females=3; Level of injury range: L3-C7; Mean time since injury: 18.0 yr.

Intervention: Compare 2 propulsion training methods (high and low tech) between experimental and control conditions to determine which system was more effective at teaching manual wheelchair users (MWUs) to increase contact angle (CA) and decrease stroke frequency (SF) during propulsion at two speeds (1.5 m/s or self-selected speed) on an overground course of 15m of level tile, of medium pile carpet and a 1.2° ramp. There were two experimental conditions: an instruction only (IO) group that received a multi-media presentation (MMP) over four sessions, and a MMP and real-time feedback (FB) group which received four sessions. The control group (CG) received no training but had three sessions where they propelled on the overground course and on the dynamometer without instruction. Participants used their own w/c throughout, with no changes in configuration. Data was collected pre-post the same day (n=27) and 3mo follow up (n=22)

Outcome Measures: CA (degrees), SF (strokes per second), peak resultant force [Fr; N/(m/s)], and rate of rise of Fr [rorFr (N/m)].

1.     In controlling for velocity, weight, time since injury and level of injury:

a) Both intervention groups showed increased CA and decreased SF in same day and 3 mo follow-up compared to the CG (p<0.05);

b) For SF, intervention groups decreased the identical amount but the IO group showed greater decrease at 3mo follow-up (p<0.05); FB group showed greater percent increase in CA compared to IO group, who showed a greater percent increase than CG at both time periods (p<0.05);

c) Both the FB and IO groups showed significant short-term increases in peak Fr at the handrim, with a larger percent increase for the FB(p<0.05), however long-term changes were not significantly larger than baseline; the CG showed a significant increase in long-term (3mo post intervention) peak Fr.

2.     The FB and IO groups showed significant short- and long-term reductions in peak rorFr compared to CG (p<0.05)

3.     There were no significant interactions for any of the three test groups for surface type suggesting the effects of training were not influenced by the surface type (carpet, ramp, tile).

4.     There were no significant interactions across test groups for propulsion speed.

5.     Results of the fixed effects analysis of CA, SF, peak force and rorF compared to demographics found: 1) older participants tend to use smaller CA (p,0.001), and more strokes (p=0.002) whereas lower level injured participants used fewer strokes (p=0.001); 2) older and heavier participants tended to use greater peak force (p=0.04) whereas lower level injured participants tended to use less peak force (p=0.001).

Effect Sizes: Forest plot of standardized mean differences (SMD ± 95%C.I.) as calculated from pre- and post-intervention data.

Rice et al. 2014





Population: Mean age: 38.3 yr; Gender: males=28, females=9; Level of injury: paraplegia=34, tetraplegia=3; Level of severity: AIS A=20, B=4, C=8, D=2, unknown=3; Mean time since injury: acute.

Intervention: Intervention group received education on wheeled mobility and upper limb clinical practice guidelines by a physical and occupational therapist (IG); control group received standard therapy services (SCG).

Outcome measures: Wheelchair setup, selection, propulsion biomechanics, pain, (numeric rating scale (NRS), Wheelchair Users Shoulder Pain Index (WUSPI) Satisfaction with Life Scale (SLS) and Craig Handicap Assessment and Reporting Technique scores. All measures completed at discharge, 6 mo and 1 yr.

1.      There were no significant between-group differences or within-subject differences for: 1) wheelchair setup (rear axle position in relation to acromium or elbow flexion position at the top of the push cycle); 2) wheelchair selection although at 6mo and 1 yr 100% of IG met the recommendation of an ultra-light wheelchair; 3) pain, immediate or long term (1 yr).

2.      In the SLS scores showed a trend for an increase in only the physical subsection between 6month and 1 yr (p=0.07) and the occupational subsection between 6mo and 1 yr (p=0.07).

3.      For propulsion biomechanics, compared to the SCG, the intervention group had significantly lower push frequency at discharge on tile (p=0.02) a trend effect on carpet (p=0.10) and used a significantly longer push length on ramps at all time points (p=0.03).

Effect Sizes: Forest plot of standardized mean differences (SMD ± 95%C.I.) as calculated from pre- and post-intervention data.

Morgan et al. 2017


Prospective controlled trial


Population: Mean age= 38±17.5 yr; Gender: males=4, females=2 ; Level of injury range: C6-L2.

Intervention: Manual wheelchairs (MWC) users participated in nine 90-min wheelchair training sessions 2-3 times per week, using motor learning principleswith a repetition-based approach; participants acted as their own control The aim of the training was to increase the push angle and efficiency, use a semicircular push pattern and, decrease push force Two baseline measures were taken three weeks apart , and the psot-test immediately after the intervention

Outcome Measures: Wheelchair push forces (WMS): Average force, Peak force, Slope of the force; Wheelchair Skills Test (WST), Kinematic Variables: Area of the push loop, hand-axle relationship, push angle; Wheelchair performance test (WPT): contact, recovery, speed, push effectiveness, push frequency.

1.     Area of the push loop significantly increased from pre to post test (p=0.05), as well as hand-axel relationship (p=0.03).

2.     A positive, but not statistically significant improvement was found for push angle pre- and post- intervention (p=0.07).

3.     No significant improvement was found for the WST

4.     Three items on the WPT improved significantly pre and post intervention: recovery (p<0.01), speed (p<0.01), push effectiveness (p=0.04).

5.     Slope of the force was the only factor that improved significantly on the WMS (p=0.03).

Blouin et al. 2015




Population: Mean age: 42.1 yr; Gender: males=16, females=2; Mean weight: 77.4 kg; Mean time since injury: 14.8 yr; Level of injury: C7 or LI; Severity of injury: AIS A, B or C.

Intervention: Patients participated in a training session in a standard manual wheelchair on a stimulator with haptic biofeedback (HB) in order to modify patient’s mechanical effective force (MEF) along push phase to achieve more effective MEF pattern. Two pre- and two post training trials were completed without hepatic feedback, each for 1 min. Training was in five 3-min blocks with a 2min rest between; heptic feedback was provided at five different, randomized levels. Visual feedback on the linear velocity was also provided.

Outcome Measures: Raw force measured using forces sensors on the wheels and simulator base and moment data measured using the SmartWheel, MEF (%push) patterns, mean wheelchair linear velocity, Mean biofeedback moments and mean power output.

1.     On average, participants increased mean MEF by up to 15.7% on right side and 12.4% on left side from pre-training to post-training.

2.     Power output was significantly higher during the training blocks compared to the pre-and post-training (p≤.007).

3.     Mean wheelchair velocities remained equivalent or slightly decreased during the training.

4.     No significant differences in ΔMEFrms scores were found neither between the pre-training and the training, nor between any pairs of training blocks (p>0.1).

5.     Biofeedback level had significant impact on mean MEF in both Q2 and Q3 quartiles and on both sides (p>0.02).

6.     Significant increases in mean MEF were found between the pre-training trial and training blocks BL3, BL4, and BL5 on the right side (p≤0.001).

7.     On the left side, mean MEF was significantly higher during training block BL5 in quartile Q2, and demonstrated a tendency to increase between the pre-training trial and training blocks BL3, BL4, and BL5 in quartile Q3 (p≤0.06).

8.     Mean MEF decreased slight during post-training compared to pre-training on left side, remained equivalent on right side, led to non-significant increase in ΔMEFrms.

DeGroot et al. 2009




Population: Mean age: 37 yr; Gender: males=6, females=3; Injury etiology: tetraplegia=2, paraplegia=4, cerebral palsy=1, spinal muscular atrophy=1, multiple sclerosis=1; Mean during of w/c use: 10 yr.

Intervention: Participants were trained on a wheelchair treadmill with verbal instruction (in-depth explanation of Boninger et al. propulsion principles – using a semicircular pattern, using long and smooth strokes and reducing push frequency) and visual instruction and feedback (1) video of an experienced wheelchair user demonstrating the four propulsion patterns – arc, single-loop-over, double-loop-over, and semicircular and 2) visual feedback of performance during propulsion)Training continued until trainer and trainee felt sufficient training and practice had occurred. 10 sec of data were collected immediately following training/practice.

Outcome Measures: push frequency, push length, peak push force, average push force, peak push force and average speed using a SMART wheel attached to the participants’ own MWC. Propulsion was on a wheelchair treadmill.

1.     Push length increased (p<0.05) pre-to post training.

2.     Push frequency decreased (p<0.01) pre-to post training.

3.     Peak (p<0.05) and average (p<0.01) forces increased pre-to post training.

4.     Average speed did not change.

5.     Graphic representations showed differences in propulsion characteristics between one participant with paraplegia and one participant with tetraplegia.

  • Tetraplegia participant propelled at slower speed than paraplegia participant.
  • Participant with tetraplegia had, on average, a lower push frequency than the participant with paraplegia.
  • Push force comparisons did not show clear patterns.


Morgan et al. (2017) explored the effectiveness of a repetition-based motor learning approach to improve propulsion techniques of longer strokes and changing the propulsion pattern to a semicircular pattern. The training program was based on the recommendations to reduce force and frequency of pushes from Clinical Practice Guidelines for Preservation of Upper Limb Function Following Spinal Cord Injury. Participants all made some improvement in propulsion across all the testd variables, suggesting that this type of approach to improve propulsion techniques is viable; however, the number of participants effects the strength of the study and therefore the ability to generalize to other people.

Rice et al. (2013) compared two propulsion training methods to determine the effectiveness of training in relation to contact angle (CA) (angle along the arc of the hand rim), stroke frequency (SF) (number of strokes per unit of time), peak resultant force (Fr) (the maximum forces experienced during the push phase of propulsion), and peak rate of rise of resultant forces (rorF) (how rapidly the forces are applied to the hand rim). Testing was completed using two speeds (1.5 m/s and a speed the participant selected) and three over-ground conditions (tile, medium pile carpet and 1.2° ramp) over three training sessions. The findings suggest that there are immediate benefits to propulsion training with carryover of benefits long term (three months) as compared to the provision of opportunities to practice propulsion but without instruction (control group) regardless of the speed of propulsion or the type of surface used.

It is worth noting however, the intervention groups also received weekly phone calls to remind them to continue to practice with the training techniques, the effect of which was not evaluated. Neither intervention required the presence of a health care professional; the multimedia presentation was a five-minute video and slide presentation emphasizing the importance of proper technique and defined the key parameters for monitoring such as CA. The second intervention group also received real-time feedback provided using a specialized wheel that collects data related to CA, SF, and velocity. This real-time feedback was projected onto a screen for the participant to view as they were propelling on the dynamometer. Variables were presented randomly and discontinuously in keeping with motor learning theory. It was noted that CA feedback was easier to react to than SF feedback and CA feedback had an inadvertent effect on SF as well.

Rice et al. (2014) examined if intervention that strictly adhered to the Clinical Practice Guidelines for Preservation of Upper Limb Function (Paralyzed Veterans of America) recommendations for wheelchair set up, selection and propulsion skills would decrease shoulder pain and improve satisfaction with life and participation as compared to standard care. Both the intervention group and the control group (standard care group) testing occurred on the same inpatient spinal cord injury rehabilitation program. Similar to Rice, et al. 2013, the intervention protocol also used a multimedia approach (printed material, DVD of propulsion skills and pictures) however different to the above study, therapists provided this training in addition to ongoing education and feedback related to the key concepts from the practice guidelines. All involved therapists received motor learning training in addition to training related to the practice guidelines. The significant findings from this study are fewer than the Rice et al. study; the findings of significance were a lower push frequency on tile at discharge from inpatients and a larger push length across all time points (discharge and six months and one year post discharge) for those who received the training compared to the control group who did not. However, the study did not find significance differences in pain with either pain scale used which differed from previous studies (identified within the article). Ongoing testing and follow-up over several years may be needed to determine if these findings persistent and if pain levels are impacted in the long term.

deGroot et al. (2009) examined the effects, both immediate and sustained, of a verbal and visual training intervention for manual wheelchair propulsion, comparing difference in effects on people with paraplegia versus people with tetraplegia. Researchers found that push length increased, and push frequency decreased immediately following training, however push forces increased which was not expected. The authors questioned if this latter finding was an inadvertent result of participants focusing on the propulsion pattern, or a result of efforts to decrease push frequency, participants felt they needed to push harder. Comparison of findings for participants who had a paraplegia and tetraplegia found that push frequency (pushes per second) showed that participants with tetraplegia had a lower push frequency and a higher number of pushes to complete the same ten-meter distance. It is important to note that these comparisons were based on one participant for each group and that the total number of participants with paraplegia versus tetraplegia was four and two respectively.

Blouin et al. (2015) explored the influence of haptic biofeedback on mechanical efficiency propulsion training. Training was completed in five, three-minute training blocks with two one minute pre-and post testing blocks without the feedback. This group developed a simulator which provides feedback by increasing or decreasing the rolling resistance and therefore mechanical effort, as propulsion patterns deviate from or approached the desired pattern respectively. The authors cite that unrestricted increases in propulsion mechanical effective force (MEF) which has been associated with an increase in load at the shoulders, and mechanical inefficiency in propulsion. The authors suggest that through training with haptic biofeedback the MEF can be moderated therefore more efficient with less negative effect on the shoulders. They found that participants could modify their MEF pattern to become more efficient but only during the middle portion of the push phase where the greatest push effort occurred. They also found that the effects of the training were not sustained at post-testing by all participants.


There is level 1b (from one blinded RCT study: Rice et al. 2013; one RCT study: Rice et al. 2013; one prospective controlled study: Morgan et al 2017; and two pre-post studies: deGroot et al. 2009; Blouin et al. 2015) evidence that wheelchair propulsion training result in improved biomechanics of propulsion which are sustained over time.

There is level 1b (from one blinded RCT: Rice et al. 2013; one RCT study: Rice et al. 2013; and one pre-post study: deGroot et al. 2009) evidence that using a multimedia approach results in improved wheelchair propulsion training outcomes.