JRRD logo
Vol. 37 No. 3, May/June 2000
Pages 283 - 296

Influence of trunk flexion on biomechanics of wheelchair propulsion

Mary M. Rodgers, PhD, PT; Randall E. Keyser, PhD; Elizabeth R. Gardner, MS, PT; Pamela J. Russell, PhD; Peter H. Gorman, MD

University of Maryland School of Medicine, Departments of Physical Therapy and Neurology, Baltimore, MD; Veterans Affairs Maryland Health Care System, Research & Development and Physical Medicine and Rehabilitation Services, Baltimore, MD

Abstract — Propulsion styles that are characterized by high stresses may influence the susceptibility of manual wheelchair users (MWCU) to upper limb injury. An experimental, cross-sectional study was designed to compare physiological and biomechanical characteristics of wheelchair propulsion in two groups of MWCU--a trunk flexion propulsion style group (FG) and a non-trunk flexion propulsion style group (NFG)--across fresh and fatigued states. Data on joint kinetics and kinematics, handrim kinetics, propulsion temporal characteristics, and electromyography were collected at the fresh and fatigued states, and oxygen uptake was collected continuously, to characterize wheelchair propulsion performance of 19 MWCU during a submaximal exercise test to exhaustion. The FG was characterized by a more flexed trunk position accompanied by greater shoulder flexion and elbow extension (p<0.05), which was accentuated with fatigue when compared to the NFG. When fatigued, marked decreases (p<0.05) in key propulsion muscle activity were observed in the FG, but not in the NFG. Temporally, the FG decreased contact time on the handrim by 1 percent of the propulsion cycle when fatigued, in contrast to the NFG who increased contact time by 7 percent (p<0.05). Results suggest that a trunk flexion style of wheelchair propulsion may lead to potentially debilitating upper limb injury, since these individuals appear to be compensating for peripheral muscle fatigue.

Key Words: biomechanics, fatigue, kinematics, kinetics, propulsion style, wheelchair propulsion.

This material is based on work supported by the Department of Veterans Affairs, VA Rehabilitation Research and Development Service, Washington, DC 20420
Presented in part at the Third North American Congress on Biomechanics, Waterloo, Ontario, Canada, 1998. Current Address: Pamela J. Russell, PhD, Dept. of Movement Arts, Health Promotion, & Leisure Studies, Bridgewater State College, Bridgewater, MA.
Address all correspondence and requests for reprints to: Mary M. Rodgers, PhD, PT; University of Maryland, School of Medicine, Department of Physical Therapy, 100 Penn St., Baltimore, MD 21201-1082; email:


  Chronic health problems due to long-term use of manual wheelchairs significantly impair the independence of individuals who depend upon wheelchairs for locomotion. Manual wheelchair users (MWCU) are likely to experience upper limb pain (1-4) which limits physical performance (4,5) and quality of life (6-8). In different study samples, sixty-four to 73 percent of MWCU with spinal cord injuries (SCI) have reported shoulder or wrist joint pain (1,2,4,9) and 67 percent of MWCU have upper limb mononeuropathies defined by strict electrodiagnostic criteria (3). Gellman and coworkers (9) and Pentland et al. (5) reported increases in the prevalence of upper limb pain as a function of time since sustaining SCI. Rodgers, et al. (10) suggested the frequency and duration of arm use may compound the risk for overuse injury in MWCU. Others have reported mechanical inefficiency during wheelchair propulsion (11-13) and activities of daily living (14,15), suggesting propulsion mechanics may be a pathogenic mechanism for these injuries.

  Rapid application of propulsion force (16,17), loading of joints while in extreme positions (18-21), high strength requirements (21-23), and metabolic demands (14,15,24,25) appear to be important contributors to upper limb injury in MWCU. Propulsion style, therefore, seems implicated as a possible determinant of the condition. Individual variants in propulsion style have been noted (26,27), with the amount of trunk flexion being widely variable among MWCU (27,28) and increasing with propulsion speed (29,30). Important differences in upper limb kinematics (28), propulsion temporal characteristics (28), and energy expenditure (30) have been identified based on varying degrees of trunk flexion.

  We tested the hypothesis that a propulsion style incorporating excessive or exaggerated trunk flexion would result in biomechanical and physiological characteristics thought to incur painful upper limb injury, as compared to a style that does not incorporate trunk flexion. We predicted that a trunk flexion style of propulsion, specifically, would result in increased oxygen uptake (VO2), and potentially harmful joint kinematics, kinetics, muscle activation patterns, and propulsion temporal characteristics. The aim of this study was to determine the degree to which theorized mechanisms of upper limb injury differ as a result of trunk flexion and non-trunk flexion propulsion styles in the MWCU.


  Nineteen non-athlete MWCUs participated in this study (Table 1). Three were female and the majority of participants had spinal cord injuries. Ages ranged from 21-68 years and durations of wheelchair use ranged from 3-38 years. Self-perceived fitness ratings were in the medium range, and none were actively participating on a sport team or in recreational sports. Before testing, a medical examination was completed by a physician familiar with the requirements for study participation. Inclusion criteria included use of a manual wheelchair for at least one year prior to the study, wheelchair use for the majority of home and community mobility, and absence of upper limb involvement, ventilatory involvement, or systemic diseases that would preclude or limit exercise testing. Before testing, written, informed consents were obtained in accordance with the procedures approved by the Institutional Review Board.

Table 1.
Subject characteristics.

Sex Age Injury Control Time Hgt Wgt Arm Trunk

M 21 T5 absent  7 180.5 72.4 47.0 47.0
M 30 T3 incomplete absent 12 185.4 66.4 77.0 51.0
M 32 spina bifida present  6 157.5 56.8 73.0 51.0
M 33 bilat TTS* present 10 188.0 97.3 80.0 57.0
M 36 T12 present 10 154.9 102.3  66.0 43.2
F 37 T7 absent 34 121.9 52.3 60.0 32.0
F 39 multi-trauma present 12 177.8 75.0 70.5 40.0
M 40 T2 absent 10 167.5 100.0  73.0 47.0
M 42 T12 incomplete present 11 179.1 95.5 82.0 45.0
M 44 T3 absent 24 178.8 71.2 68.5 46.5
F 45 T7 absent 24 162.5 44.9 43.0 43.0
M 49 L5 incomplete present  3 187.0 127.0  50.5 48.0
M 50 T5 absent 28 186.0 90.9 45.0 48.5
M 52 T6 absent 12 180.3 66.8 74.0 42.0
M 52 T5 absent 26 180.3 78.5 75.0 49.0
M 54 multi-trauma present 26 182.9 77.3 73.0 51.0
M 54 T5 incomplete present  5 180.3 72.7 71.1 45.7
M 58 T8 absent 38 178.0 76.3 46.0 42.0
M 68 T4 incomplete present 21 187.5 79.5 45.0 51.0
Mean 44.0     16.8  174.5 79.1 64.2 46.3
SD 11.3     10.4  16.1 19.6 13.5  5.4

Age in years; Control=trunk control; Time=time in wheelchair, in years; Hgt=height in cm; Wgt=weight in kg; Arm=arm length in cm; Trunk=trunk length in cm; *TTS=tarsal tunnel syndrome.

  The wheelchair measurement system included the instrumented wheelchair ergometer, three Peak three-dimensional (3-D) Charged Coupled Device (CCD) cameras and video acquisition system (Peak Performance Technologies, Colorado Springs, CO), an eight channel surface Electromyographic (EMG) telemetry system (Noraxon USA Inc, Scottsdale, AZ), and a metabolic cart (Cardio 2, Medical Graphics Corp, St. Paul, MN). All exercise tests were conducted on the wheelchair ergometer (Figure 1) with a 22-in diameter handrim. Components of a stationary bicycle ergometer were used to provide frictional propulsion resistance through a chain and sprocket system connected to the wheelchair axle at one end and to a flywheel at the other end (Figure 2). Torque applied by the subject to the handrims of the wheels propelled the flywheel. A nylon belt was used to create a pulley system to which known weights could be applied, enabling precise control of resistance to the flywheel that was then transmitted to the wheels.

Figure 1. A manual wheelchair user prepared for data acquisition. Note instrumented wheelchair ergometer and locations of reflective markers used for joint kinematic data acquisition.


Figure 1.
A manual wheelchair user prepared for data acquisition. Note instrumented wheelchair ergometer and locations of reflective markers used for joint kinematic data acquisition.

Figure 2. Schematic diagram of the wheelchair ergometer showing the method used to apply resistance to the handrim.

Figure 2.
Schematic diagram of the wheelchair ergometer showing the method used to apply resistance to the handrim.

  The wheels of the wheelchair ergometer had no camber, and the backrest was positioned perpendicular to the floor. Subjects used their own wheelchair cushions in the wheelchair ergometer seat. Wheelchair ergometer seat width and height were adjusted to match as closely as possible the wheelchair seat of each individual. Subjects were positioned in the wheelchair with their feet on a stepstool the height of which corresponded to that required to keep the thighs parallel to the floor.

  The wheelchair ergometer was instrumented with a PY6-4 six-component force/torque transducer (Bertec Corp., Worthington, OH) in its wheel hub. The Bertec transducer used bonded strain gauges to measure handrim forces and moments in 3-D (six channels). It had a maximum torque (Mz) capacity of 150 N-m and a maximum plane-of-wheel force (Fx and Fy) capacity of 3,500 N. These capacities corresponded to a gain setting of unity, and the maximum usable range was set by user-selected gains. The full-scale output of the transducer was ±10 V. Since the transducer rotated with the wheel, connections were enabled using a slip ring. A potentiometer monitored the angular positions of the wheel, transducer, and handrim assembly. The amplified electrical signals from the force transducer and potentiometer were collected using an analog-to-digital converter and acquisition software (Peak Performance Technologies, Colorado Springs, CO). Handrim kinetic, temporal, and potentiometer data were collected at 360 Hz. A bicycle speedometer with a digital display was attached to the right wheel of the chair and placed in view of the participant to provide visual feedback of propulsion velocity. The speedometer (Cateye Cyclocomputer, Model CC-CD 100, Osaka, Japan) had an accuracy of ±0.2 km/hr (±1rpm) and range of 0-100km/hr. Wheelchair propulsion velocity was maintained at 32 rpm (3 km/hr) for all testing. The relatively slow propulsion velocity of 3 km/hr was selected because of the inclusion of elderly individuals in the sample and because of the difficulty of maintaining higher speeds at high resistance loads.

  Kinematic data were collected at 60 frames/s using the video cameras and data acquisition system. Calibration was performed by videotaping a precisely calibrated cube with 24 known coordinate points and digitizing each point location. Direct linear transformation computations were performed using the Cartesian (X-Y-Z) coordinate system, with camera lens distortion corrected using the Peak software package. System accuracy of the Peak Performance Measurement System has been reported by the company to exhibit a 0.5 mm average error when calibrating in static conditions. Our own dynamic evaluations have determined the angular calculations to be within 1° at speeds up to 300°/s. Cameras were positioned so that all markers could be viewed in at least two cameras at any time during the experiment. One camera was located directly in front of the wheelchair ergometer at a distance of 25 feet and the other two cameras were approximately 45° to each side of the center camera. Spherical retroreflective markers were placed on the dorsal surface of the fifth metacarpal head, medial styloid process, lateral styloid process, radial head, acromion, and greater trochanter. Joint marker displacements were recorded using the camera, and joint angles (trunk, shoulder, elbow and wrist flexion/extension, shoulder abduction/adduction, and wrist radial/ulnar deviation), velocities and accelerations were calculated using the Peak Performance software.

  Joint kinetics were calculated using a 3-D, linked segment model (31). This model used an inverse dynamics approach, employed the Newton-Euler method (based on body coordinate systems), and assumed the arm to be three rigid segments (hand, forearm and upper arm) connected by the wrist, elbow and shoulder joints. Definitions for the global and local (handrim, hand, forearm, upper arm, and trunk) coordinate systems have been described in detail (Figure 3). Displacement data collected using the video cameras were differentiated to form velocity and acceleration vectors. The linear acceleration of each marker was used to interpolate the linear acceleration of the center of mass for each limb segment. These estimates were then transformed into respective body coordinates. The forces and moments measured from the hub transducer were transformed from handrim to hand coordinates, using the radius of the handrim of the wheelchair, and were used as the input to the hand equations. Raw data were smoothed using a Butterworth low-pass filter with a cutoff frequency of 6 Hz. Contact phase was defined as the entire time of handrim loading. These motion vectors, forces, and torques, together with anthropometric data, were the input variables to a program that computes the forces and moments of the wrist using an inverse dynamics process. The recursive program then determines the joint forces and moments of the elbow and shoulder.

Figure 3. Global and local coordinate systems for the wheelchair propulsion model.

Figure 3.
Global and local coordinate systems for the wheelchair propulsion model. Body segments (T, trunk; u, upper arm; L, lower arm or forearm; H, hand) and segmental motions as a function of moments about local axes (flexion/extension, abduction/adduction, ulnar/radial deviation) are indicated. Note: x'y'z' are coordinates for the hub transducer; x'w-y'w-z'w coordinates are parallel to xw-yw-zw coordinates; xyz coordinates with a w subscript refer to wheel coordinates. (Reprinted by permission from Mary M. Rodgers, Srinivas Tummarakota, and Junghsen Lieh, 1998, "Three dimensional dynamic analysis of wheelchair propulsion," Journal of Applied Biomechanics 14(1):80-92.)

  EMG data were collected from an eight-channel telemetry system at 960 Hz (Noraxon USA Inc, Scottsdale, AZ). Pairs of surface electrodes were placed on the right side of each subject, over motor point locations for the pectoralis major, anterior and posterior deltoids, middle trapezius, biceps and triceps brachii, flexor carpi ulnaris and extensor carpi radialis muscles. Signals were processed by computing the mathematical root-mean square (RMS) value of the EMG signal (32). The signal was then normalized to the EMG level recorded during maximal isometric voluntary contraction (MVC) for each muscle using standardized manual muscle test positions (33).

  Heart rate was monitored by a telemetered pulse rate monitor that included a transmitter, attached by a belt to the thorax, and a receiver with a digital display. Cardiorespiratory measurements were made using breath-by-breath open circuit spirometry. The metabolic cart system included rapid response oxygen (zirconium cell) and carbon dioxide (infrared cell) analyzers, and a pneumotachometer, all interfaced with a microcomputer. Cardiorespiratory measurements were averaged over 30 s for reporting purposes.

Exercise Tests
  A maximal graded exercise test (GXT) on the wheelchair ergometer was used to establish resistance load for the fatigue test. For the GXT, subjects rested (6 min), propelled the wheelchair at a velocity of 3 km/hr (32 rpm) without a load (3 min), then continued while weight was incrementally added at a rate of 0.3 kg every 3 min, to increase power output. The test was terminated at volitional exhaustion, defined as the self-reported inability to maintain the target velocity. Subjects were monitored and encouraged to maintain the designated velocity. Heart rate was recorded at the end of each 3-min stage and cardiorespiratory information was continuously recorded as 30-s averages.

  Two to seven days following the GXT, subjects completed the fatigue test. Load corresponded to 75 percent of the peak VO2 that occurred during the GXT. Since resistance load used for the fatigue test was based on the individual's performance on the GXT, resistance loads varied from individual to individual. For the fatigue test, subjects rested (6 min), propelled the wheelchair at 3 km/hr without a load (3 min), then continued propelling at the sub-maximal load until volitional exhaustion. Subjects were monitored and encouraged to maintain the designated velocity. Heart rate was recorded at the end of each 3-min stage and cardiorespiratory information was continuously recorded as 30-s averages. Propulsion mechanics, including handrim kinetics, joint kinematics, EMG, and temporal characteristics, were collected for 6 s (3 propulsion cycles) under three conditions: during the last 30 s of wheeling without a load, after 2.5 min of wheeling with a load (fresh), and just before exhaustion (fatigued).

Strength Tests
  Strength of the dominant upper limb was tested using two techniques. Handgrip strength was tested in all subjects using a Baseline hydraulic hand dynamometer. Three maximal grips were averaged to represent grip strength. Eleven of the subjects were also tested using a Kincom isokinetic dynamometer (60°/s) to test other upper limb joint motions which included shoulder flexion/extension (0-55°), shoulder internal/external rotation (0-55°), elbow flexion/extension (5-70°), wrist flexion/extension (0-45°), and wrist radial/ulnar deviation (0-45°). The dynamometer speed of 60°/s corresponded to the joint angle velocities occurring during wheelchair propulsion at a velocity of 3 km/hr. Three maximal voluntary contractions were collected concentrically and eccentrically. Mean peak moment was used to characterize strength for each motion.

  Subjects were divided into two groups based on trunk position (upright sitting position=90° relative to the horizontal plane). Trunk angles greater than 90° were defined as trunk extension and those less than 90° were defined as trunk flexion. Subjects who demonstrated more than 10° of trunk flexion (in the 0-80° range) during contact when fresh or fatigued, and those whose trunk flexion increased more than 10° from the fresh to the fatigued state were included in the flexion group (FG). All others were included in the non-flexion group (NFG). Based on the group assignment criteria, 9 subjects were assigned to the FG and 10 to the NFG. The FG included four who had increased trunk flexion angle, one who had increased flexion range from fresh to fatigued state, and four who met both criteria. Anthropometric characteristics of subjects are shown in Table 1. Wheelchair seat widths ranged from 15-19 in (38-48 cm) with a mean of 16.4±1.8 in (41.7±4.6 cm).

Data Analysis
  EMG peak amplitude was determined from the average of three propulsion cycles for each trial. Timing of muscle activity and integrated EMG (total activity or area under the curve) were determined from a representative cycle for each trial and related to propulsion and recovery phases. Propulsion phase (contact time) was defined as the entire time of handrim loading, with recovery phase being defined as the time when the hand was not in contact with the handrim. All EMG data were normalized to the maximum isometric voluntary contraction of each muscle (percent MVC). Kinetic and kinematic data were averaged over three cycles (contact to contact) for each condition (fresh and fatigued). The decision to average three cycles was based on the low cycle-to-cycle variability achieved, attributed to the fact that wheelchair propulsion is a constrained activity which was also performed at a set velocity. Any cycles that were aberrant for any reason (i.e., miscontact or slippage on the wheel) were not included in the analysis.

  Joint kinetics and kinematics, handrim kinetics, propulsion temporal characteristics, and EMG data were compared between groups during the fresh and fatigued states. Oxygen uptake (VO2) during the fatigue test was characterized by absolute VO2 (ml/min), oxygen uptake divided by maximal oxygen uptake (VO2/VO2max), and oxygen uptake divided by power output (VO2/resistance load). These variables were compared at times corresponding to one-third, two-thirds, and completion of the test for both groups. Comparisons were performed using Analysis of Variance (ANOVA), with first effect repeated (fresh and fatigued states) and second effect by group (FG and NFG). Interaction effects were further examined through post-hoc analysis. Type-I error threshold was held at or below 5 percent (pless than or equal to0.05).

  Subject characteristics, including age, height, weight, length of time in a wheelchair, and upper limb strength were compared between groups using a two-sample t-test. Trunk control was classified as present or absent for all subjects. Those with motor-complete spinal cord injuries above the twelfth thoracic vertebra (T12) were classified with absent trunk control. Those with motor incomplete spinal cord injuries, complete injuries below T12, or other types of injuries were classified with trunk control. Gender and trunk control were compared between groups using a two-way frequency analysis. Statistical significance was established when the type-I error probability was not observed above five percent (pless than or equal to0.05).


Trunk Flexion Versus Non-trunk Flexion Groups
  No significant inter-group differences in age, height, weight, length of time as a wheelchair user, gender, or trunk control were observed. Upper limb strength (grip and isokinetic) was not significantly different between groups, with the exception of the concentric shoulder extension moment, which was greater (p<0.04) in the NFG (67±4 N-m) than the FG (41±18 N-m). Failure of differences in the means to reach the 0.05 threshold for significance across all muscle groups appeared to be the result of the small number of subjects completing isokinetic dynamometer tests (NFG=3 subjects; FG=8 subjects).

  Kinematic differences are depicted for representative subjects from each group in Figure 4. Significant inter-group differences in trunk position were found (Table 2). Inter-group kinematic differences at the shoulder and elbow were also observed (Tables 3 and 4, respectively). During contact (p<0.006) and at release (p<0.004), the FG demonstrated greater shoulder flexion than the NFG. Elbow extension was greater for the FG during contact (p<0.013) and at release (p<0.031). No significant differences were found in the wrist kinematics (Table 5).

Figure 4. Representative subjects from the flexion group (FG) and non-flexion group (NFG) depicting joint kinematic differences.

Figure 4. Representative subjects from the flexion group (FG) and non-flexion group (NFG) depicting joint kinematic differences. Greater trunk flexion, along with associated differences in shoulder and elbow angles, can be seen in the FG compared to the NFG at contact, during contact and at release during the fresh and fatigued states.

Table 2.
Trunk flexion kinematics.

    FG NFG  

At contact       State
Fresh   91±11 97±7 94±9 
Fatigued   83±12  99±9* 91±11
  Group 87±12  98±8  
During contact       State
Fresh   81±11  92±6* 87±9 
Fatigued   75±10  93±8* 84±9 
  Group 78±11  93±7  
At release       State
Fresh   85±11  94±6* 90±9 
Fatigued   78±9   94±8* 86±9 
  Group 82±10  94±7  
ROM (during contact)     State
Fresh   12±6    7±5 10±6 
Fatigued   13±8    7±4 10±6 
  Group 13±7     7±5#  

Joint angles in degrees; FG=trunk flexion; NFG=non-trunk flexion; ROM=range of motion; *=significantly different interaction (p<0.05); #=significantly different main effect (p<0.05); scores are means±SD.

Table 3.
Shoulder kinematics.

  Flexion     Abduction  
  FG NFG     FG NFG  

At contact   State       State
Fresh -42±12 -47±10 -45±11   41±8 44±11 43±10
Fatigued -36±15 -48±9* -42±12   38±8 44±11 41±10
Group -39±14  -48±10   Group 40±8 44±11  
During contact   State       State
Fresh 23±11   8±12  16±12   43±9 47±9  45±9
Fatigued 25±8   13±10 19±9   41±8 46±8   44±8 
Group  24±10    11±11#   Group 42±9 47±9   
At release   State       State
Fresh 20±11   5±12 13±12   22±9 28±7  25±8
Fatigued 23±7   10±9  17±8   24±7 29±10  27±9 
Group 22±9     8±11#   Group 23±8 29±9   
ROM (during contact)   State       State
Fresh 67±9  58±21  63±15    23±10 20±7  22±9
Fatigued 63±14 63±15  63±15   19±9 21±7   20±8 
Group 65±12 61±18   Group  21±10 21±7   

Joint angles in degrees; FG=trunk flexion; NFG=non-trunk flexion; ROM=range of motion; *=significantly different interaction (p<0.05); #=significantly different main effect (p<0.05); scores are means±SD.

Table 4.
Elbow extension kinematics.

    FG NFG  

At contact     State
Fresh   121±8 112±7  117±8 
Fatigued    115±13 111±5 113±9
  Group  118±11 112±6  
During contact     State
Fresh   153±8 142±9 148±9
Fatigued   152±6 145±9 149±8
  Group 153±7  144±9#  
At release     State
Fresh   150±9  139±10  145±10
Fatigued    149±7   142±10  146±9 
  Group 150±8   141±10#  
ROM (during contact)     State
Fresh     48±11   40±13  44±12
Fatigued     48±10   45±10  47±10
  Group   48±11   43±12  

Joint angles in degrees; FG=trunk flexion; NFG=non-trunk flexion; ROM=range of motion; # significantly different main effect (p<0.05); scores are means±SD.

Table 5.
Wrist kinematics.

  Extension     Deviation  
  FG NFG     FG NFG  

At contact   State       State
Fresh -24±16 -23±19 -24±18    5±12 7±15  6±14
Fatigued -28±14 -20±20 -24±17   6±7 3±16  5±12
Group -26±15 -22±20   Group  6±10 5±16  
During contact   State       State
Fresh -37±12 -28±21 -33±17   37±12 29±22  33±17
Fatigued -35±10 -26±19 -31±15   34±13 27±23  31±18
Group -36±11 -27±20   Group 36±13 28±23  

At release   State       State
Fresh   6±12  10±16   8±14   24±8 15±19  20±14
Fatigued   5±13  10±19   8±16   24±7 14±17  19±12
Group   6±13  10±18   Group 24±8 15±18  

ROM (during contact)   State       State
Fresh  66±16  49±22   58±19    38±17 31±16  35±17
Fatigued  59±20  51±21   55±21    31±14 30±22  31±18
Group  63±18  50±22   Group  35±16 31±19  

Joint angles in degrees; FG=trunk flexion; NFG=non-trunk flexion; Extension=wrist extension; Deviation=wrist ulnar deviation; ROM=range of motion; scores are means±SD.

  Resistance applied to the wheelchair (FG=1,500±1,071 g, NFG=1,350±852 g) and the length of time the wheelchair was propelled (FG=24.8±15.9 min, NFG=29.9±26.4 min) were similar between groups. VO2max during the fatigue test was also similar between groups (FG=1087±456 ml/min, NFG=1,080±319 ml/min). There were no significant inter-group differences in VO2, VO2/VO2max, and VO2/load values for each group at the beginning third, middle third, and end of the fatigue test (Table 6). Only two inter-group differences in joint kinetics were apparent. The FG had 7 N less (p<0.022) posterior force (Fx) and 29 N more (p<0.046) medial force (Fz) at the elbow during contact compared to the NFG.

Table 6.
Oxygen uptake.

  VO2 (ml/min) VO2/VO2max VO2/load

1/3 of test 1035±431  846±459 96±12  78±30 83±32  72±36
2/3 of test 1027±440  964±276 94±5   91±15 82±32  91±44
At exhaustion 1015±419  979±295 94±5   91±8 83±36  93±45

FG=trunk flexion; NFG=non-trunk flexion; scores are means±SD.

  Group differences in muscle activity patterns were seen in three muscles. The FG demonstrated 20 percent earlier cessation of flexor carpi ulnaris (p<0.001) and pectoralis major (p<0.031) activity. The total biceps activity was greater (p<0.034) for FG compared to NFG (FG=18 percent, NFG=10 percent).

Fresh Versus Fatigued State
  Fatigue main effects were seen in some of the variables. Both groups demonstrated more shoulder flexion (p<0.047 during contact, p<0.018 at release) when fatigued. Both groups applied more inward force (Fz) to the handrim (p<0.03) when fatigued than when fresh (Table 7). Both groups demonstrated less (p<0.024) wrist flexion moment when fresh (-5±5 N-m) than when fatigued (-8±7 N-m), and less (p<0.022) radioulnar shear force (Fx) when fatigued (45±25 N fresh, 37±21 N fatigued). Fatigue altered muscle activity patterns in three muscles. With fatigue, both groups demonstrated significant decreases in peak amplitude of the biceps (p<0.006) and pectoralis major muscles (p< 0.025). Finally, fatigue-induced changes in both groups included decreased total biceps (p<0.006) and pectoralis major (P<0.04) activity, and earlier onset (p<0.02) and peak activity of the triceps (p<0.01).

Table 7.
Handrim kinetics.

  Peak Forces (N)

  Fx (tangential) Fy (radial) Fz (medial)
      State     State     State
Fresh 70±32 67±14 69±23 -67±21 -65±19 -66±20 11±14 7±6   9±10 
Fatigued 67±25 68±18 68±22 -75±37 -74±26 -75±32 15±15 13±9 14±12
Group 69±29 68±16   -71±29 -70±23   13±15 10±8  
  Peak Moments (N-m)
  Mx My Mz*
Fresh -4±4 -3±1 -4±3 -3±2 -3±2 -3±2 -22±9 -20±5 -21±7
Fatigued -3±4 -3±3 -3±4 -2±2 -2±2 -2±2 -21±8 -21±5 -21±7
Group -4±4 -3±2   -3±2 -3±2   -22±9 -21±5  

FG=trunk flexion; NFG=non-trunk flexion; # significantly different main effect (p<0.05); scores are means±SD; *=propulsive moment.

  Interaction effects were seen for several variables (p<0.05). In the FG, trunk flexion increased 7-10° more with fatigue than in the NFG. When fatigued, shoulder flexion increased at contact by 6° in the FG, but not in the NFG. Propulsion temporal characteristics are shown in Table 8. The FG decreased contact time on the handrim by 1 percent of the propulsion cycle when fatigued (0.04 s decrease), in contrast to NFG who increased (p<0.033) contact time by 7 percent of the cycle (0.04 s increase) as they fatigued (Figure 5).

Table 8.
Propulsion temporal characteristics.

  Stroke Frequency Contact (sec) Contact (% cycle)


Stroke frequency in cycles per second; FG=trunk flexion; NFG=non-trunk flexion; *=significantly different interaction (p<0.05); scores are means±SD.

Figure 5. Significant inter-group differences (mean±sd) in handrim contact time with fatigue. In contrast to the non-flexion group (NFG), the flexion group (FG) demonstrated decreased contact time with fatigue.

Figure 5. Significant inter-group differences (mean±sd) in handrim contact time with fatigue. In contrast to the non-flexion group (NFG), the flexion group (FG) demonstrated decreased contact time with fatigue.

  Interaction effects were also observed in muscle activity patterns. When fatigued, FG demonstrated significant decreases in peak amplitude of the flexor carpi ulnaris and triceps muscles, which did not occur in NFG (Figure 6). Although only significant for the flexor carpi ulnaris and triceps, the FG demonstrated declines in peak amplitude with fatigue across all muscle groups tested, while the NFG demonstrated increases with fatigue in all but three muscle groups.

Figure 6.Mean (±sd) EMG peak magnitude of the flexor carpi ulnaris (FCU) and triceps (TRI) muscles at the fresh and fatigued states for both groups. EMG peak magnitude is normalized to maximum isometric voluntary contraction (percent MVC). Note decreases in EMG activity with fatigue in the FG not evident in the NFG that were statistically significant.

Figure 6.Mean (±sd) EMG peak magnitude of the flexor carpi ulnaris (FCU) and triceps (TRI) muscles at the fresh and fatigued states for both groups. EMG peak magnitude is normalized to maximum isometric voluntary contraction (percent MVC). Note decreases in EMG activity with fatigue in the FG not evident in the NFG that were statistically significant.


Trunk Flexion Versus Non-trunk Flexion Groups
  Inter-group propulsion style differences were characterized by variations in shoulder flexion and elbow extension. A pattern of greater shoulder flexion and elbow extension that accompanied trunk flexion was seen in the FG. Consistent with the observations of others (28,29), this pattern of movement allowed the FG to rely on trunk excursion to generate the translational forces necessary for wheelchair propulsion. In contrast, the NFG made contact with the wheel in a position of greater shoulder extension than the FG. This pattern did not change with fatigue. Trunk excursion was not characteristic of the NFG style, which relied on force generation from sources other than those translated through the trunk, such as muscular power (34), for propulsion. The FG also had larger medial elbow joint forces, which may be related to the greater elbow extension.

  Inter-group propulsion style differences did not appear to be attributed to differences in the degree of volitional trunk control. Subjects with inadequate volitional trunk control would not be expected to adopt a trunk flexion propulsion style. Subjects in this study with the most impaired trunk control had spinal cord injuries between T2-T4 (n=4), with the majority at the T5 level and below or with injuries not involving the trunk (n=15). Most of the subjects in the current study appeared to have sufficient trunk control to allow some degree of forward lean. Other investigators have noted propulsion similarities for other characteristics of MWCU with high and low paraplegia (26,35), suggesting that the difference in trunk control did not affect wheelchair propulsion ability. The degree of volitional trunk control was not measured for this study. Trunk control was classified by the level and type of injury, and from the medical screening examination. Information regarding the extent to which subjects are able to lean forward in the wheelchair and return to the starting position would be valuable in understanding the relationship between trunk control and propulsion style.

Fresh Versus Fatigued State
  The trunk flexion movement pattern appeared to have been compensatory for muscle fatigue, since it was more pronounced in the fatigued state. The inter-group similarity in load, test duration, and VO2 supported the premise that the trunk flexion style of propulsion was adopted to compensate for muscle fatigue rather than for aerobic economy. The increase in inward handrim force (Fz) may be a compensation for the fatiguing of the muscles used in grasping the handrim. Dallmeijer et al. (26) found that subjects with cervical level spinal cord injuries (C4-C8) who lacked grasping ability used the inward handrim force for additional friction during propulsion.

  The fatigue model was used in this study in an attempt to simulate the conditions that have been suggested to contribute to overuse injuries. It is clear that repetitive activities are associated with increased risk of joint degenerative changes. These changes may result from damage to the static and dynamic controlling mechanisms of joints (ligaments, joint capsules, menisci, and muscles). Work by Pugh (36) has shown that muscular control of joint movement is directly related to shock absorption. Based on this work, the fatigued muscle would appear to be an inadequate shock absorber, contributing to the probability of joint degradation. The changes in muscle activity patterns and biomechanical variables with fatigue that we have observed in our studies seem to support the use of the fatigue model to identify those MWCU most susceptible to overuse injuries. Since subjects had to be pain-free to participate in this study, experience in the wheelchair could not be correlated with current overuse injury. The only way to directly relate our measurements to overuse injury would be to follow our subjects over time to see what characteristics were consistent in those who developed overuse injuries. Longitudinal studies are very difficult, but are definitely warranted.

  Differences in the magnitude and timing of muscle activity with fatigue were observed according to propulsion style. During local muscle fatigue, EMG amplitude is known to increase due to increased motor unit recruitment, among other factors (37). In the current study, EMG peak amplitude of muscles critical to propulsion increased in the NFG from the fresh to fatigued state. However, there was a decline in peak amplitude of all muscles tested in FG, particularly in the flexor carpi ulnaris and triceps muscles. This decline may be interpreted as the recruitment of fewer motor units in upper limb muscles tested at the point of exhaustion. A fatigue-mediated reduction in motor unit recruitment appears to have been enabled by the employment of a compensatory movement strategy in the FG (i.e., trunk flexion, to generate propulsion moment). Such a strategy may not have been necessary for the NFG, who did not demonstrate the degree of fatigue shown by FG, suggesting better fitness of key propulsion muscles in the NFG. Extremely large skeletal muscle fiber areas in the triceps muscles of wheelchair athletes have been reported and attributed to hypertrophy from constant use during wheeling and other daily activities (38). Dynamic muscular strength of the triceps has been identified as an important determinant of handrim impulse generation (39), highlighting the important role of triceps activity during wheelchair propulsion and supporting the notion that triceps fatigue may trigger compensatory propulsion strategies.

  The muscular fatigue demonstrated by the FG may place them at higher risk for injury. Muscle damage, as evidenced by elevated plasma creatine kinase, lactate dehydrogenase, and myoglobin content, has been noted in able-bodied individuals following progressive, submaximal exercise on a wheelchair treadmill (40). The reduction in contact time that accompanies fatigue may also place those with the trunk flexion style of propulsion at greater risk for injury. Faster application of handrim forces, and therefore joint forces, would be expected with decreases in contact time. Rapid loading of the arms during propulsion has been associated with certain overuse injuries among MWCU, such as carpal tunnel syndrome (17). Conversely, longer contact times and lower peak forces applied to the handrim are characteristic of experienced MWCU, versus inexperienced non-wheelchair users, thereby lessening potentially harmful joint forces (16). The longer contact times occurring with fatigue in the NFG may be a benefit of their propulsion style in terms of injury risk reduction. Further investigation of propulsion pathomechanics may identify additional factors contributing to injury. Further research is necessary to determine if overuse injuries can be prevented by strength and endurance training of muscles critical to propulsion (e.g., triceps) in those who demonstrate the trunk flexion propulsion style, since they appear to be compensating for peripheral muscle fatigue.


  A trunk flexion style of wheelchair propulsion did not result in increased VO2 compared to a non-trunk flexion style of propulsion. However, the differences observed for kinematics, kinetics, muscle activation patterns, and propulsion temporal characteristics supported our hypothesis that a trunk flexion style of wheelchair propulsion may, in fact, lead to painful upper limb injury which limits physical performance and quality of life. Definitive studies on pathomechanical mechanisms of upper limb injuries in MWCU may support these findings and facilitate interventions for the resulting impairment and disability. Implications from the current study are that strength and endurance training of the muscles critical to propulsion (e.g., triceps) may be important in MWCU who demonstrate the trunk flexion style, since they appear to be compensating for peripheral muscle fatigue.


The authors express sincere thanks to James Parker for technical assistance, Shawn Moore for the production of figures, and Dan Hsu for original artwork. We also thank our research participants for their contributions of time and effort.

  1. Pentland WE, Twomey LT. The weight-bearing upper extremity in women with long term paraplegia. Paraplegia 1991;29:521-30.
  2. Sie IH, Waters RL, Adkins RH, Gellman H. Upper extremity pain in the postrehabilitation spinal cord injured patient. Arch Phys Med Rehabil 1992;73:44-8.
  3. Davidoff G, Werner R, Waring W. Compressive mononeuropathies of the upper extremity in chronic paraplegia. Paraplegia 1991;29:17-24.
  4. Subbarao JV, Klopfstein J, Turpin R. Prevalence and impact of wrist and shoulder pain in patients with spinal cord injury. J Spinal Cord Med 1994;18:9-13.
  5. Pentland WE, Twomey LT. Upper limb function in persons with long term paraplegia and implications for independence: part I. Paraplegia 1994;32:211-8.
  6. Lundqvist C, Siosteen A, Blomstrand C, Lind B, Sullivan M. Spinal cord injuries: clinical, functional and emotional status. Spine 1991;16:78-83.
  7. Post MWM, deWitte LP, van Asbeck FWA, van Dijk AJ, Schrijvers AJP. Predictors of health status and life satisfaction in spinal cord injury. Arch Phys Med Rehabil 1998;79:395-402.
  8. Wagner Anke AG, Stenehjem AE, Kvalvik Stanghelle J. Pain and life quality within 2 years of spinal cord injury. Paraplegia 1995;33:555-9.
  9. Gellman H, Sie I, Waters RL. Late complications of the weight-bearing upper extremity in the paraplegic patient. Clin Orthop 1988;233:132-5.
  10. Rodgers M, Gayle W, Figoni SF, Kobayashi M, Lieh J, Glaser RM. Biomechanics of wheelchair propulsion during fatigue. Arch Phys Med Rehabil 1994;75:85-93.
  11. van der Woude LHV, Hendrich KMM, Veeger HEJ, van Ingen Schenau GJ, Rozendal RH, deGroot G, Hollander AP. Manual wheelchair propulsion: effects of power output on physiology and technique. Med Sci Sports Exerc 1988;20:70-8.
  12. van der Linden ML, Valent L, Veeger HEJ, van der Woude LHV. The effect of wheelchair handrim tube diameter on propulsion efficiency and force application (tube diameter and efficiency in wheelchairs). IEEE Trans Rehabil Eng 1996;4:123-32.
  13. Brown DD, Knowlton RG, Hamill J, Schneider TL, Hetzler RK. Physiological and biomechanical differences between wheelchair-dependent and able bodied subjects during wheelchair ergometry. J Appl Physiol 1990;60:179-82.
  14. Janssen TWJ, van Oers CAJM, van der Woude LHV, Hollander AP. Physical strain in daily life of wheelchair users with spinal cord injuries. Med Sci Sports Exerc 1994;26:661-70.
  15. Dallmeijer AJ, Hopman MTE, van As HHJ, van der Woude LHV. Physical capacity and physical strain in persons with tetraplegia; the role of sport activity. Spinal Cord 1996;34:729-35.
  16. Robertson RN, Boninger ML, Cooper RA, Shimada SD. Pushrim forces and joint kinetics during wheelchair propulsion. Arch Phys Med Rehabil 1996;77:856-64.
  17. Baldwin MA, Boninger ML, Cooper RA, Shimada SD, Connor TJ. Wheelchair propulsion biomechanics and median nerve neuropathy. Proceedings of the 1st National Meeting of the VA Rehabilitation Research & Development Service; 1998 Oct 1-3, Washington, DC. Washington, D.C.: VA R&D, 1998; p. 170.
  18. Gellman H, Chandler DR, Petrasek J, Sie I, Adkins R, Waters RL. Carpal tunnel syndrome in paraplegic patients. J Bone Joint Surg 1988;70-A:517-9.
  19. Blankstein A, Shmueli R, Weingarten I, Engel J, Ohry A. Hand problems due to prolonged use of crutches and wheelchairs. Orthopaedic Rev 1985;XIV(12):735-40.
  20. Bayley JC, Cochran TP, Sledge CB. The weight-bearing shoulder. J Bone Joint Surg 1987;69-A:676-8.
  21. Veeger HEJ, Meershoek LS, van der Woude LHV, Langenhoff JM. Wrist motion in handrim wheelchair propulsion. J Rehabil Res Dev 1998;3:305-13.
  22. van der Helm FCT, Veeger HEJ. Quasi-static analysis of muscle forces in the shoulder mechanism during wheelchair propulsion. J Biomech 1996;29:39-52.
  23. Mulroy SJ, Gronley JK, Newsam CJ, Perry J. Electromyographic activity of shoulder muscles during wheelchair propulsion by paraplegic persons. Arch Phys Med Rehabil 1996;77:187-93.
  24. Janssen TWJ, van Oers CAJM, Veeger HEJ, Hollander AP, van der Woude LHV, Rozendal RH. Relationship between physical strain during standardised ADL tasks and physical capacity in men with spinal cord injuries. Paraplegia 1994;32:844-59.
  25. Hjeltes N, Vokac Z. Circulatory strain in everyday life of paraplegics. Scand J Rehabil Med 1979;2:67-73.
  26. Dallmeijer AJ, Kappe YJ, Veeger HEJ, Janssen TWJ, van der Woude LHV. Anaerobic power output and propulsion technique in spinal cord injured subjects during wheelchair ergometry. J Rehabil Res Dev 1994;2:120-8.
  27. Rao SS, Bontrager EL, Gronley JK, Newsam CJ, Perry J. Three-dimensional kinematics of wheelchair propulsion. IEEE Trans Rehabil Eng 1996;3:152-60.
  28. Sanderson DJ, Sommer III HJ. Kinematic features of wheelchair propulsion. J Biomech 1985;6:423-29.
  29. Veeger HEJ, van der Woude LHV, Rozendal RH. Wheelchair propulsion technique at different speeds. Scand J Rehabil Med 1989;21:197-203.
  30. Vanlandewijck YC, Spaepen AJ, Lysens RJ. Wheelchair propulsion efficiency: movement pattern adaptations to speed changes. Med Sci Sports Exerc 994;26:1373-81.
  31. Rodgers MM, Tummarakota S, Lieh J. Three-dimensional dynamic analysis of wheelchair propulsion. J Appl Biomech 1998;14:80-92.
  32. LeVeau BL, Andersson G. Chapter 5. Output forms: data analysis and applications. In: Selected topics in surface electromyography for use in the occupational setting: expert perspectives. US Department of Health and Human Services, Public Health Service, Centers for Disease Control, National Institute for Occupational Safety and Health, DHHD (NIOSH); 1992. Publication No. 91-100.
  33. Kendall FP, McCreary EK, Provance PG. Muscles, testing and function. 4th ed. Baltimore (MD): Lippincott Williams & Wilkins; 1993.
  34. Veeger HEJ, van der Woude LHV, Rozendal RH. Load on the upper extremity in manual wheelchair propulsion. J Electr Kinesiol 1992;1:270-80
  35. Newsam CJ, Mulroy SJ, Gronley JK, Bontrager EL, Perry J. Temporal-spatial characteristics of wheelchair propulsion. Am J Phys Med Rehabil 1996;4:292-9.
  36. Pugh J. Biomechanical aspects of osteoarthritic joints: mechanisms and noninvasive detection. In: DN Ghista, ed. Osteoarthromechanics. Washington: Hemisphere Pub. Corp.; 1982. p. 161-91.
  37. McArdle WD, Katch FI, Katch VL. Exercise physiology; energy, nutrition, and human performance. 3rd ed. Philadelphia: Lea and Febiger; 1991.
  38. Taylor AW, McDonnell E, Royer D, Loiselle R, Lush N, Steadward R. Skeletal muscle analysis of wheelchair athletes. Paraplegia 1979;17:456-60.
  39. Tupling SJ, Davis GM, Pierrynowski MR, Shephard RJ. Arm strength and impulse generation: initiation of wheelchair movement by the physically disabled. Ergonomics 1986;2:303-11.
  40. Ide M, Ogata H, Kobayashi M, Wada F. Muscle damage occurring in wheelchair sports people. Spinal Cord 1997;35:234-37.

  Submitted for publication March 30, 1999. Accepted in revised form June 15, 1999.

Go to TOP

Contents Vol. 37, No. 3

  Last Updated Wednesday, November 5, 2008 11:08 AM