shoulder biomechanics during the push phase of wheelchair propulsion: a multisite study of persons...

O

SPJS

PcmR

sptws

i

g

dk

bcsco

ii7d

allBCMiWUN

(mPN

soTHbf

LP

667

RIGINAL ARTICLE

houlder Biomechanics During the Push Phase of Wheelchairropulsion: A Multisite Study of Persons With Paraplegia

ennifer L. Collinger, BS, Michael L. Boninger, MD, Alicia M. Koontz, PhD, RET, Robert Price, MSME,
ue Ann Sisto, PT, MA, PhD, Michelle L. Tolerico, MS, Rory A. Cooper, PhD
pom

iws

n

cR

Pmftwsbum7

wctptmsdpb

sinSor4CegjAjisv

o

ABSTRACT. Collinger JL, Boninger ML, Koontz AM,rice R, Sisto SA, Tolerico ML, Cooper RA. Shoulder biome-hanics during the push phase of wheelchair propulsion: aultisite study of persons with paraplegia. Arch Phys Medehabil 2008;89:667-76.

Objectives: To present a descriptive analysis and compari-on of shoulder kinetics and kinematics during wheelchairropulsion at multiple speeds (self-selected and steady-statearget speeds) for a large group of manual wheelchair usersith paraplegia while also investigating the effect of pain and

ubject demographics on propulsion.Design: Case series.Setting: Three biomechanics laboratories at research

nstitutions.Participants: Volunteer sample of 61 persons with paraple-

ia who use a manual wheelchair for mobility.Intervention: Subjects propelled their own wheelchairs on a

ynamometer at 3 speeds (self-selected, 0.9m/s, 1.8m/s) whileinetic and kinematic data were recorded.Main Outcome Measures: Differences in demographics

etween sites, correlations between subject characteristics,omparison of demographics and biomechanics between per-ons with and without pain, linear regression using subjectharacteristics to predict shoulder biomechanics, comparisonf biomechanics between speed conditions.Results: Significant increases in shoulder joint loading with

ncreased propulsion velocity were observed. Resultant forcencreased from 54.4�13.5N during the 0.9m/s trial to5.7�20.7N at 1.8m/s (P�.001). Body weight was the primaryemographic variable that affected shoulder forces, whereas

From the Human Engineering Research Laboratories, VA Rehabilitation Researchnd Development Center, VA Pittsburgh Healthcare Systems, Pittsburgh, PA (Col-inger, Boninger, Koontz, Tolerico, Cooper); Departments of Bioengineering (Col-inger, Boninger, Koontz, Cooper), Physical Medicine and Rehabilitation (Collinger,oninger, Tolerico), and Rehabilitation Science and Technology (Boninger, Koontz,ooper), University of Pittsburgh, Pittsburgh, PA; Department of Rehabilitationedicine, University of Washington, Seattle, WA (Price); Kessler Medical Rehabil-

tation Research and Education Corp, Spinal Cord Injury Rehabilitation Research,est Orange, NJ (Sisto); and Department of Physical Medicine and Rehabilitation,niversity of Medicine and Dentistry of New Jersey-New Jersey Medical School,ewark, NJ (Sisto).Supported by the National Institute on Disability and Rehabilitation Research

grant no. H133A011107), Veterans Affairs Rehabilitation Research and Develop-ent Service, U.S. Department of VA Affairs (grant no. B3057R), University ofittsburgh Model Center on Spinal Cord Injury (grant no. H133N000019), and aational Science Foundation Graduate Research Fellowship.A commercial party having a direct financial interest in the results of the research

upporting this article has conferred or will confer a financial benefit upon the authorr 1 or more of the authors. Boninger and Cooper have a nonfinancial affiliation withhree Rivers Holdings in the form of subcontracted grants. In addition, Three Riversoldings licenses patents unrelated to this publication from the University of Pitts-urgh. Boninger and Cooper receive royalties through the University of Pittsburghrom the sales of these licensed inventions.

Reprint requests to Michael L. Boninger, MD, Human Engineering Researchaboratories, VA Pittsburgh Health Care System, 7180 Highland Dr, 151R1-H,ittsburgh, PA 15206, e-mail: [email protected].

s0003-9993/08/8904-00308$34.00/0doi:10.1016/j.apmr.2007.09.052

ain did not affect biomechanics. Peak shoulder joint loadingccurs when the arm is extended and internally rotated, whichay leave the shoulder at risk for injury.Conclusions: Body-weight maintenance, as well as other

nterventions designed to reduce the force required to propel aheelchair, should be implemented to reduce the prevalence of

houlder pain and injury among manual wheelchair users.Key Words: Biomechanics; Rehabilitation; Shoulder; Spi-

al cord injuries; Wheelchairs.© 2008 by the American Congress of Rehabilitation Medi-

ine and the American Academy of Physical Medicine andehabilitation

EOPLE WITH SPINAL CORD injury (SCI) often rely ontheir ability to propel a manual wheelchair for independent

obility. Wheelchair propulsion requires a person to impart aorce to the wheelchair pushrim to move forward. As a result,he joints of the upper limb are loaded repeatedly as the manualheelchair user performs activities of daily living.1-5 The

houlder joint in particular is designed for mobility, not loadearing. This may be the reason that many manual wheelchairsers report shoulder pain. Estimates of shoulder pain amonganual wheelchair users with paraplegia range from 30%6 to

3%.7

Many investigators believe that repetitive loading duringheelchair propulsion, termed overuse syndrome, is a potential

ause for pain.8-10 Our most recent investigation11 supportedhis idea, because joint kinetics resulting from wheelchair pro-ulsion were linked to shoulder pathology. Mercer et al11 foundhat people who experienced larger forces and moments wereore likely to have coracoacromial pathology or to exhibit

igns of pathology on physical examination. It has been wellocumented that manual wheelchair users experience shoulderain; however, it is not known how pain affects shoulderiomechanics during wheelchair propulsion.A few studies1-5,12-14 have described 3-dimensional (3D)

houlder biomechanics during propulsion. Most of these stud-es have been conducted at a single site with a relatively smallumber of subjects, usually fewer than 20 participants.1,3,4,12,14

ome studies focused solely on shoulder kinetics3,4 and othersn shoulder kinematics.12,14 The largest study we are aware ofeported kinetics and kinematics of wheelchair propulsion for7 manual wheelchairs users with varying medical conditions. 2

omparisons between studies are difficult because of differ-nces in testing conditions. An instrumented wheelchair er-ometer was used in some studies2-5,14; others1,13 tested sub-ects in their own wheelchairs but on a dynamometer setup.lso, different coordinate systems are used when reporting

oint kinetics and kinematics.2,4,12-14 Inconsistency also existsn the propulsion speeds, with some studies focusing on steady-tate speeds1,2,5,12,13 and others examining only self-selectedelocities.3,4

Our goal is to present a descriptive analysis and comparisonf shoulder kinetics and kinematics during wheelchair propul-
ion at multiple speeds (self-selected and steady-state target
Arch Phys Med Rehabil Vol 89, April 2008

mailto:[email protected]

spdscetvsIEwdpcatpbaisscFpsp

nK((i

P

2bwythicdwclwFrhwqpy

I

sep

taSmfds1tdssm

skastwtmpwcfifisoidmoifsbcw(st

umtHuQsmmsmebdmkwB

D

a

668 MULTISITE WHEELCHAIR BIOMECHANICS, Collinger

A

peeds) for a large group of manual wheelchair users witharaplegia and to investigate the effect of pain and subjectemographics on propulsion. It is important to study self-elected velocity because the way a person propels the wheel-hair on an everyday basis may be linked to pathology. How-ver, because biomechanics vary with propulsion speed,3,4,13

arget speeds are valuable for directly comparing biomechanicariables between subjects. We hope to move toward a moretandardized description of shoulder kinematics by using thenternational Society of Biomechanics (ISB)–recommendeduler angle sequence.15 Glenohumeral forces and momentsill be referenced to local coordinate systems with 3 rotationalegrees of freedom. Despite differences in testing conditions,revious studies3,4,13 with different setups have observed in-reased joint loading at faster different speeds of propulsion,nd we expected the current study to confirm those results inhe largest subject population to date. We also believe thatersons with shoulder pain will experience less joint loadingecause of a modified propulsion style. Using a multisitepproach to recruit a large group of subjects also allows us tonvestigate the influence of subject demographic characteristicsuch as age, years since injury, injury level, height, weight, andex on propulsion biomechanics. We hypothesized that in-reased subject weight would result in increased joint loading.inally, we describe the relationship between the timing ofeak shoulder kinetics and arm posture, because loading thehoulder in vulnerable positions may contribute to shoulderathology.

METHODSThree sites participated in data collection: the Human Engi-

eering Research Laboratories (HERL) in Pittsburgh, PA;essler Medical Rehabilitation Research and Education Corp

KMRREC) in Orange, NJ; and the University of WashingtonUW) in Seattle, WA. This study was approved by each site’snstitutional review board.

articipantsA total of 61 subjects (21 from HERL, 20 from KMRREC,

0 from UW) volunteered and provided informed consentefore participation in this study. All subjects used a manualheelchair as their primary means of mobility, were over 18ears old, and had an SCI below T1 that had occurred morehan 1 year before participation in the study. Each subject alsoad a wheelchair with quick-release wheels to ensure compat-bility with the kinetic measurement device. People were ex-luded from this study if they had a history of fractures orislocations in the arms including the shoulder, elbow, andrist; upper-limb dysthetic pain as a result of a syrinx or

omplex regional pain syndrome type II; or if they had upper-imb pain that prohibited them from propelling a manualheelchair. Nondominant-side data were used for all analyses.ive of the 61 subjects were left-handed; all others wereight-hand dominant. Demographic information includingeight, weight, age, years since injury, injury level, and sexas collected from all subjects. Subjects were also asked 2uestions about shoulder pain: (1) Have you had any shoulderain in the last month? (2) Does your shoulder hurt you whileou are propelling your wheelchair?

nstrumentation and Data CollectionWheelchair dynamometer. Each subject’s wheelchair was

ecured to a dynamometer that had 2 independent rollers, 1 forach wheel. The resistance of the rollers is comparable to
ropelling over a tile surface.16 All 3 dynamometers used in u
rch Phys Med Rehabil Vol 89, April 2008

his study were fabricated and assembled at HERL, and theyre checked and maintained every 6 months at each site.ubjects were instructed to acclimate themselves to the dyna-ometer setup before testing. Real-time speed and direction

eedback were displayed on a monitor in front of subjectsuring the trials. Subjects participated in 3 speed trials: aelf-selected comfortable pace, 0.9m/s (3.2km/h [2mph]), and.8m/s (6.4km/h [4mph]). Subjects performed the self-selectedrial first so that they would not be influenced by the speedisplay, followed by the 0.9m/s and 1.8m/s trials. After aubject reached a steady-state speed, data were collected for 20econds. Subjects were allowed to rest as needed, approxi-ately 1 minute, between trials.Kinetic data. The SmartWheel,a a 3D force- and torque-

ensing device, was used at each site to measure propulsioninetics at the pushrim.17 Each SmartWheel was fitted bilater-lly at HERL and KMRREC, and unilaterally at UW, to eachubject’s own wheelchair. At UW, the SmartWheel was fixedo each subject’s nondominant side while an inertia-matchedheel was fitted to the other side. Because UW only had access

o 1 SmartWheel, the nondominant side was chosen because itay be less affected by pathology not related to wheelchair

ropulsion. Attaching the SmartWheel to a subject’s ownheelchair does not change the wheel placement, alignment, or

amber. Kinetic data were collected at 240Hz and digitallyltered with an eighth-order, zero-phase, low-pass Butterworthlter with a 20-Hz cutoff frequency. Kinetic data were down-ampled to 60Hz for comparison with kinematic data. Previ-usly, investigators of this multisite study identified differencesn pushrim kinetics between sites, presumably because of smallifferences in rolling resistance of the dynamometers. Aethod based on deceleration on the dynamometer was devel-

ped to correct for the kinetic differences to combine the datan future analyses.18 This method calculates a coefficient ofriction for each dynamometer system based on rolling re-istance and normal force (a percentage of the subject’sody weight distributed by the rear wheel). Differences inoefficients of friction between sites and individual bodyeights were used to adjust data from the collaborating sites

KMRREC, UW) so that data were comparable with the leadite (HERL). Adjusted pushrim forces were used as input tohe inverse dynamics model.

Kinematic data. Different motion-capture systems weresed at each of the 3 sites, but all were capable of outputting 3Darker position data relative to a global origin (located be-

ween the 2 rollers of the wheelchair dynamometer). TheERL site used 2 Optotrak 3020 systems,b the KMRREC sitesed a Vicon 612 Workstation,c and the UW site used aualisys MCU240 system.d The resolution of the Optotrak

ystem is .01mm at a camera distance of 2.25m, and theaximum residual marker error of both the Vicon and Qualisysotion capture systems is less than 1.5mm. The same marker

et was used at all 3 sites and included markers at the thirdetacarpophalangeal joint, radial styloid, ulnar styloid, lateral

picondyle, acromion, sternal notch, C7 vertebrae, T3 verte-rae, and greater trochanter. Each site was responsible foretermining the optimal sampling frequency and interpolationethods for their motion-capture system. For final analysis, all

inematic data were down-sampled to 60Hz. Kinematic dataere digitally filtered with a fourth-order, zero-phase, low-passutterworth filter with a 7-Hz cutoff frequency.

ata AnalysisInverse dynamics. Cooper et al1 previously described the

nthropometric model used for this study. Segment lengths and
pper-extremity circumferences of all subjects were measured

aiwdvdsa(mslnotflKttseadcflst

rsao

csotpdltspdfv

gtnbthvsefl0vsca

S

Wvtstdhvwwfarreruiitdvasapru

Fs

669MULTISITE WHEELCHAIR BIOMECHANICS, Collinger

s input to Hanavan’s mathematic model, which calculates thenertial properties of each body segment.19 Pushrim forcesere transformed to the glenohumeral joint using a previouslyescribed inverse dynamics model.1 Calculations for the in-erse dynamics model were performed using Matlab.e Shoul-er joint forces were transformed to the anatomic coordinateystem of the proximal segment of the shoulder joint, the trunk,s follows: anterior (x), posterior (�x), superior (y), inferior�y), medial (z), and lateral (�z) (fig 1). Shoulder joint mo-ents were calculated relative to the humeral local coordinate

ystem described in previous work.1 The humeral and trunkocal coordinate systems are coincident when the arm is in aeutral posture. Abduction (�) and adduction (�) momentsccurred about the x axis, external (�) and internal (�) rota-ion produced moments about the y axis, and extension (�) andexion (�) moments occurred about the z axis. HERL andMRREC collected bilateral kinematic data, and therefore the

runk local coordinate system was calculated using markers athe acromions and greater trochanters as previously de-cribed.11 UW collected unilateral kinematic data, so a differ-nt method, using markers at the sternal notch, C7 vertebra,nd T3 vertebra, was used to determine the trunk local coor-inate system; however, both methods approximate the localoordinate systems recommended by ISB.15 To minimize dif-erences between sites, a kinematic calibration trial was col-ected before testing. For this trial, subjects were instructed toit in their wheelchairs such that the trunk was perpendicular to

ig 1. Trunk local coordinate system. Reprinted with permis-ion.11

he ground aligned with the global coordinate system. A cor- a

ective transformation matrix was calculated for each subject toatisfy this condition. This transformation matrix was thenpplied to the trunk local coordinate system as calculated in allther trials.Forces and moments are presented with left shoulder sign

onvention to allow for averaging of all data regardless of eachubject’s nondominant side. Peak values during the push phasef propulsion were calculated and averaged over all strokes ofhe 20-second trial. Push phase was defined as a deviation ofushrim force and moment data from baseline (0) and wasetermined through visual inspection kinetic data using Mat-ab. This represents the period in which the hand is contactinghe pushrim. Resultant force at the shoulder was calculated byumming the directional force components. The timing of theeak loading, expressed as percentage of push cycle, wasetermined for the directional force and moments, as well asor resultant force. In addition, stroke frequency and meanelocity were calculated using data from the SmartWheel.Kinematics. Shoulder position was described as Euler an-

le rotations derived from the transformation matrix describinghe position of the humerus relative to the trunk local coordi-ate system using the rotation order (y, x=, y==) recommendedy ISB.15 Rotation about the y axis represents plane of eleva-ion. A positive value indicates that the humerus is behind aorizontal line connecting the 2 acromions, and a negativealue indicates that the humerus is in front of this line. Forimplicity, a positive rotation about y will be referred to asxtension, and a negative rotation about y will be referred to asexion. Rotation about the x= axis represents elevation where° corresponds to the arm at the subject’s side and a positivealue indicates abduction. Rotation about the y== axis repre-ents external (�) or internal (�) rotation. Euler angles werealculated during the push phase of propulsion, and maximumnd minimum values were averaged over all strokes.

tatistical AnalysisAll statistical analyses were performed using SPSSf forindows. Descriptive analysis including means, standard de-

iation, and frequencies for categoric variables was performedo summarize demographic (for each site individually and allites combined) and biomechanic data. Demographic informa-ion was compared for differences between sites. Normallyistributed, continuous variables (age, years since injury,eight, weight) were compared using a 1-way analysis ofariance (ANOVA). Median injury level, an ordinal variable,as compared using a Kruskal-Wallis test, and sex differencesere examined using a Fisher exact test. Although not a major

ocus of this study, correlations between demographic vari-bles were identified before creating linear regression modelselating demographics to biomechanic variables. Pearson cor-elations were calculated between all demographic variables,xcluding sex, which was examined using the Spearman �. Theelationship between pain and demographics was investigatedsing the Fisher exact test for sex, Kruskal-Wallis H test fornjury level, and independent samples t test for age, years sincenjury, height, and weight. Demographics were compared be-ween subjects with shoulder pain and those without. Indepen-ent samples t tests were used to compare peak biomechanicariables (including stroke frequency, velocity, shoulder forcesnd moments, and shoulder kinematics) between subjects withhoulder pain and those without. An exploratory correlationalnalysis between demographics and biomechanic variables waserformed to identify demographic variables most stronglyelated to propulsion biomechanics. These variables were thensed in linear regression models to predict biomechanic vari-
bles (peak shoulder kinetics and kinematics) at each propul-

sgbasputfcvt

D

fs(attrK

(wPw(s

I

rsuprhirUpv

sa

lsTbpwt

D

pswsipscas

wwosssnge(1c0sAtlttms

NAc*†

‡ com§ paris


A

ion speed while controlling for the interrelation of demo-raphic variables. Normalized group means for shoulderiomechanic variables including shoulder forces, moments,nd Euler angles were calculated using 5 strokes from eachubject to visually illustrate changes in biomechanics due toropulsion speed. A 1-way repeated-measures ANOVA wassed to compare biomechanic variables between speed condi-ions. Bonferroni-adjusted pairwise comparisons were per-ormed where appropriate. Nonparametric repeated-measuresomparisons (Friedman test) were made between peak timingariables. The level of statistical significance was set at P lesshan .05 for all statistical analyses.

RESULTS

emographicsTable 1 summarizes subject characteristics by site, as well as

or the overall subject group. Significant differences betweenites were noted for years since injury (P�.008), heightP�.004), and sex (�2�6.785, P�.034). Specifically, subjectst KMRREC were closer to the time of injury (P�.008) thanhose at HERL. Subjects from UW were shorter in stature thanhose from HERL (P�.006) and KMRREC (P�.025). UWecruited significantly more women than HERL (P�.023) orMRREC (P�.028).Correlation analysis showed that subjects who were older

r�.296, P�.021) and taller (r�.364, P�.004) tended toeigh more. Women tended to weigh less than men (��.253,�.049) and were also shorter (��.559, P�.001). Peopleho were further removed from injury also tended to be older

r�.604, P�.001). No other statistically significant relation-hips were identified between the variables listed in table 1.

nfluence of Shoulder PainForty-six percent of wheelchair users (n�28) in this study

eported experiencing shoulder pain in the last month. Sixteenubjects reported bilateral shoulder pain, and 12 experiencednilateral pain. Twenty-three percent of subjects (n�14) re-orted shoulder pain while propelling the wheelchair. Of thoseeporting shoulder pain in the last month, over 60% (n�17)ad seen a doctor for treatment. Sex, injury level, years sincenjury, height, and weight did not affect whether subjectseported shoulder pain in the last month or during propulsion.sing a 1-way ANOVA, we found that participants who re-orted pain during propulsion tended to be younger (36.6�8.8y

Table 1: Subje

Characteristics HERL

n 21Age (y) 46.6�9.9Years since injury 19.4�9.0‡

Height (m) 1.79�0.07§

Weight (kg) 77.5�11.0Injury level (median) 7.75 (T7–8)Sex (male/female) 19/2§

OTE. Age, years since injury, height, and weight values are meanNOVA; median injury level is presented and compared for differeompared between sites using a Fisher exact test.Significant difference between sites (P�.05).Significantly different from HERL (Bonferroni-adjusted pairwise comSignificantly different from KMRREC (Bonferroni-adjusted pairwiseSignificantly different from UW (Bonferroni-adjusted pairwise com

s 45.1�12.2y) than those who did not. The presence of b


houlder pain in the last month was not influenced by subjectge.

When comparing subjects who reported shoulder pain in theast month with those who reported no shoulder pain, noignificant differences in propulsion biomechanics were found.he same was true when comparing propulsion biomechanicsetween persons who did and did not experience pain whileropelling. Because subjects with and without shoulder painere so similar in terms of demographics and biomechanics,

he data were combined for all further analyses.

emographics and BiomechanicsA preliminary correlational analysis showed that men, taller

eople, and heavier people tended to use larger forces. Someignificant correlations between kinematics and sex, height,eight, and age were noted. However, this was not seen at all

peed conditions or in all directions of shoulder motion. Thendependent variables used in the linear regression models toredict shoulder kinetics were sex, height, and weight. Propul-ion velocity influences biomechanics, and therefore this wasontrolled for in the regression analysis. Sex, height, weight,ge, and velocity were used in linear regression models withhoulder kinematics as the dependent variables.

Sex, height, and weight were forced into regression modelshile controlling for velocity to predict shoulder kinetics. Sexas not a significant predictor of peak shoulder kinetics in anyf the linear regression models. Subject height was only aignificant predictor of shoulder extension moment at the self-elected and 0.9m/s speed conditions. However, subject weighteemed to be the primary factor contributing to shoulder ki-etics, because it was the single significant (P�.05) demo-raphic predictor of shoulder kinetics in many directions. As anxample, for a regression model predicting resultant forceself-selected: r2�.462, P�.001; 0.9m/s: r2�.560, P�.001;.8m/s: r2�.501, P�.001), body weight was the only signifi-ant independent variable (self-selected: ��.646, P�.001;.9m/s: ��.67, P�.001; 1.8m/s: ��1.002, P�.001). Heavierubjects experienced a higher resultant force during propulsion.t all speed conditions, increased subject weight was predic-

ive of higher anterior force, posterior force, inferior force, andateral force. Prediction of shoulder moments was less consis-ent among speed conditions. At the self-selected speed condi-ion, body weight was not a significant predictor of shoulderoments when entered into a regression model with velocity,

ex, and subject height. At the 0.9m/s speed condition, higher

haracteristics

RREC UW Overall

20 20 61�14.0 44.1�11.0 43.1�12.0�11.0† 14.7�9.4 14.6�10.5*�0.07§ 1.71�0.10†‡ 1.76�0.09*�16.2 70.4�13.4 75.9�14.0(T7–8) 8.5 (T8–9) 8.0 (T8)8/2§ 12/8†‡ 49/12*

andard deviation (SD) and compared between sites using a 1-waybetween sites using a Kruskal-Wallis test; sex, reported as n, was

isons).parisons).ons).

ct C

KM

38.69.5

1.7879.87.5

1

� stnces

par

ody weight predicted larger internal and external rotation

mip

pseatl

B

tijwthwf1psmap

svpotccitastTanbfnkadi

apf

NA*†

‡ usted§ usted

NA*†

‡

§


oments. At the 1.8m/s speed condition, higher abduction,nternal rotation, external rotation, and flexion moments wereredicted by body weight.Shoulder kinematics during propulsion were largely inde-

endent of subject characteristics when age, height, weight,ex, and velocity were entered into a regression model. How-ver, some statistically significant relationships were noted: atll speeds, greater subject weight predicted less shoulder ex-ension, and during the 1.8m/s condition, older subjects usedess external rotation and shoulder flexion during propulsion.

iomechanics and Propulsion SpeedFifty subjects had data from all 3 speed conditions, and

herefore this subset of subjects was used to analyze differencesn shoulder biomechanics due to propulsion speed. Two sub-ects were unable to reach the 1.8m/s condition. One subjectas almost double the target speed for the 0.9m/s speed, and

herefore these data were excluded. The remaining 8 subjectsad erroneous and/or noisy data for at least 1 speed condition,hich affected the peak values. Table 2 presents temporal data

or each of the speed conditions, which were compared using a-way repeated-measures ANOVA. Mean velocity, as ex-ected, was significantly different between all speeds. Meanelf-selected speed was 1.09m/s, which fell in between theean velocities of the 2 target speed conditions. Subjects usedhigher stroke frequency to propel at the 1.8m/s speed com-

ared with the self-selected or 0.9m/s condition.

Table 2: Temporal Data for 3

Temporal Data SS Speed (n�50) 0.9

Mean velocity (m/s) 1.09�0.31‡§ 0.9Stroke frequency (1/s) 0.95�0.25§ 0.9

OTE: Values are mean � SD.bbreviation: SS, self-selected.Significant difference between speed conditions (P�.05) detectedSignificantly different from self-selected speed condition (BonferroSignificantly different from 0.9m/s speed condition (Bonferroni-adjSignificantly different from 1.8m/s speed condition (Bonferroni-adj

Table 3: Peak Shoulder Kinetics f

Kinetics SS Speed (n�50) 0

Forces (N)Anterior force 17.0�10.6§

Posterior force 40.9�14.0‡§

Superior force 6.7�20.6‡§ �

Inferior force 47.6�12.5Medial force 9.2�7.5§

Lateral force 17.6�9.7§

Max resultant force 59.1�15.8‡§

Moments (Nm)Abduction moment 3.2�1.7Adduction moment 7.1�4.2‡§

External rotation moment 5.1�2.8§

Internal rotation moment 15.3�6.1‡§

Flexion moment 5.4�3.2§

Extension moment 10.8�6.0‡§

OTE: Values are mean � SD.bbreviation: NS, not significant.Significant difference between speed conditions (P�.05) detectedSignificantly different from self-selected speed condition (Bonferro
Significantly different from 0.9m/s speed condition (Bonferroni-adjustedSignificantly different from 1.8m/s speed condition (Bonferroni-adjusted
Table 3 summarizes peak shoulder kinetics for the 3 propul-ion speeds. Joint kinetics are strongly affected by propulsionelocity. All shoulder forces and moments followed the sameattern observed for the mean velocities, with the lowest valuesccurring for the 0.9m/s and the highest values resulting duringhe 1.8m/s trial. The kinetic values for the self-selected speedondition fell between these extreme values and tended to belosest to the quantities measured for the 0.9m/s trial. Signif-cant differences due to speed, evaluated using a 1-way repea-ed-measures ANOVA, were noted for all shoulder joint forcesnd moments except the abduction moment. Specifically, thehoulder kinetics for the 1.8m/s speed were significantly higherhan those for both the self-selected speed and the 0.9m/s trial.he only exception to this was inferior force; only the 0.9m/snd 1.8m/s values were statistically different. Statistically sig-ificant but small differences were noted for some variablesetween the self-selected and 0.9m/s conditions: posteriororce, superior force, resultant force, adduction moment, inter-al rotation moment, and extension moment. All of theseinetic variables are active joint forces, meaning they occur asresult of the force applied to the wheelchair. The largest

irectional force component was the posterior force, and thenternal rotation moment was the largest directional moment.

Figures 2 and 3 depict the group means for shoulder forcesnd moments normalized to a single propulsion cycle. Bothush and recovery phases are shown in the figures, although weocused on the push phase in this study. The transition between

ds of Wheelchair Propulsion

�50) 1.8m/s (n�50) Significance of Speed (P )

.14†§ 1.74�0.18†‡ �.001*

.24§ 1.29�0.30†‡ �.001*

a 1-way repeated-measures ANOVA.justed pairwise comparisons).pairwise comparisons).pairwise comparisons).

peeds of Wheelchair Propulsion

(n�50) 1.8m/s (n�50) Significance of Speed (P )

�10.4§ 29.5�15.9†‡ �.001*�10.4†§ 55.2�17.7†‡ �.001*�13.0†§ 21.2�25.8†‡ �.001*�13.5§ 52.4�19.3‡ .003*�6.8§ 14.2�11.3†‡ �.001*�8.4§ 24.6�10.3†‡ �.001*�13.5†§ 75.7�20.7†‡ �.001*

�1.7 3.4�2.3 NS�3.4†§ 10.6�5.3†‡ �.001*�3.3§ 7.4�5.1†‡ �.001*�5.2†§ 20.6�8.0†‡ �.001*�2.7§ 7.3�3.9†‡ �.001*�4.7†§ 14.3�6.0†‡ �.001*

a 1-way repeated-measures ANOVA.justed pairwise comparisons).

Spee

m/s (n

5�09�0

usingni-ad

or 3 S

.9m/s

15.935.60.4446.47.2

16.154.4

2.95.45.1

13.54.89.5

usingni-ad

pairwise comparisons).pairwise comparisons).


pbc

ttdamplitagfldna

3s4(rmprwpiERa

K

mcd

Fffts

Fsamb

F(ts


A

hases is represented by a band, rather than a single line,ecause the mean time of transition varied between speedonditions. The transition occurred at 42%, 46%, and 49% of

ig 2. Group mean shoulder forces—(A) Fx, anterior and posteriororce; (B) Fy, superior and inferior force; (C) Fz, medial and lateralorce; and (D) resultant force—during 3 speeds of propulsion. Theransition from push phase to recovery is shaded because it differslightly between speed conditions.

ig 3. Group mean shoulder moments during 3 speeds of propul-ion. (A) Mx, abduction and adduction moment, (B) My, externalnd internal rotation moment, and (C) Mz, flexion and extension

doment. The transition from push phase to recovery is shaded

ecause it differs slightly between speed conditions.


he propulsion cycle for the 0.9m/s, self-selected, and 1.8m/srials, respectively. Maximum joint loading tends to occururing the push phase in most directions; however, figures 2nd 3 show some exceptions, such as inferior (see fig 2B) andedial force (see fig 2C). As the hand applies force to the

ushrim to propel the wheelchair forward, the shoulder isoaded in the posterior direction (see fig 2A) and the humeruss pushed in the superior direction (see fig 2B), counteractinghe weight of the arm. Through the push phase, the humeruslso applies a more medially directed force (see fig 2C) to thelenoid. The application of force to the pushrim, along with theexion, adduction, and external rotation of the humerus (fig 4)uring the push phase, generates adduction (see fig 3A), inter-al rotation (see fig 3B), and extension (see fig 3C) momentsbout the shoulder.

Table 4 summarizes peak kinematics of the shoulder for thepropulsion speeds. Near the beginning of the push phase, the

houlder is maximally extended (see fig 4A), abducted (see figB), and internally rotated (see fig 4C). Only small differences�2°) were noted between speeds at this end of the shoulderange of motion (ROM). The shoulder is maximally flexed andinimally abducted and internally rotated near the end of the

ush phase. The shoulder had greater flexion and less internalotation (closer to a neutral posture) at the end of the strokeith increasing speed. Minimum abduction was unaffected byropulsion speed. The lack of speed dependence is furtherllustrated in figure 4, which depicts group means for shoulderuler angles normalized over a single push phase. Although theOM is slightly larger for the fastest speed condition, thectual angle difference is minimal (only a few degrees).

inetic and Kinematic TimingTable 5 summarizes the timing of peak shoulder forces andoments during the push phase of propulsion. For all speed

onditions, the peak resultant force at the shoulder occurreduring the first half of the push phase. The same is true for most

ig 4. Group mean shoulder Euler angles—(A) plane of elevation,B) elevation, and (C) rotation—during 3 speeds of propulsion. Theransition from push phase to recovery is shaded because it differslightly between speed conditions.

irectional forces and moments. Figure 4 shows that the hu-

mttids

stipapCa

rcecd

icoatba

tapsswpstdpmpcraetpiwcpetpatfplpi

*u†

r‡

a§

a

N*†

‡ usted§ usted


erus is extended, abducted, and internally rotated throughouthe push phase, with the most extreme postures occurring at theime of contact. Peak resultant force occurred later as speedncreased. As a result, the humerus was less extended, ab-ucted, and internally rotated (closer to neutral) at the time thehoulder experienced maximal joint force.

DISCUSSIONTo our knowledge, this article describes the largest study of

houlder biomechanics for a group of manual wheelchair userso date. Recruiting from multiple sites offered the advantage ofncreasing the number, and potentially diversity, of subjectsarticipating in the study. Although a multisite study has manydvantages, it also presents some difficulties. Data collectionrocedures and instrumentation must be uniform across sites.onsistency in protocol execution was maintained throughnnual meetings of all sites involved, calibration protocols,

Table 5: Timing of Shoulder Kinetics for 3 Speeds ofWheelchair Propulsion

BiomechanicsVariables

Median Timing of Peak(% Push Phase)

�22

Significanceof Speed

(P )SS

Speed 0.9m/s 1.8m/s

Anterior force 0.4 0.3 1.4 4.093 .129Posterior force 50.3§ 43.7§ 66.3†‡ 23.510 �.001*Superior force 55.4‡ 36.5†§ 45.5‡ 9.918 .007*Inferior force 17.2 10.8§ 22.2‡ 7.260 .027*Medial force 70.6§ 70.8§ 66.6†‡ 10.082 .006*Lateral force 38.8‡ 34.5†§ 41.6‡ 11.878 .003*Resultant force 38.7 25.9§ 42.5‡ 16.449 �.001*Abduction moment 9.8 12.7 11.9 2.358 .308Adduction moment 78.1 77.5 74.9 3.435 .180External rotation

moment 0.8 0.6 1.3 2.667 .264Internal rotation

moment 43.8‡ 39.1† 37.1 9.702 .008*Flexion moment 6.8 4.5 6.4 5.144 .076Extension moment 36.9 30.3 37.9 4.667 .097

Significant difference between speed conditions (P�.05) detectedsing the Friedman test.Significantly different from self-selected speed condition (Bonfer-oni-adjusted pairwise comparisons).Significantly different from 0.9m/s speed condition (Bonferroni-djusted pairwise comparisons).

Table 4: Peak Shoulder Kinematics

Kinematics SS Speed (n�50)

Shoulder anglesMax extension (deg) 47.1�10.5Max flexion (deg) 23.7�18.6‡§

Max abduction (deg) 52.4�7.5Min abduction (deg) 30.5�6.2Max internal rotation (deg) 83.9�11.5Min internal rotation (deg) 9.8�23.0‡

OTE: Values are mean � SD.Significant difference between speed conditions (P�.05) detectedSignificantly different from self-selected speed condition (BonferroSignificantly different from 0.9m/s speed condition (Bonferroni-adjSignificantly different from 1.8m/s speed condition (Bonferroni-adj

ASignificantly different from 1.8m/s speed condition (Bonferroni-djusted pairwise comparisons).

outine equipment maintenance, and bimonthly conferencealls. In addition, we previously determined that slight differ-nces in dynamometer rolling resistance existed, but this wasorrected using a deceleration trial for all subjects, making theata comparable between all sites.18

Some differences were noted in the demographic character-stics of subjects from each site. Specifically, this study in-luded more women (from UW) and persons closer to the timef injury (from KMRREC) compared with the sample recruitedt HERL. Ultimately a larger, more diverse subject populationranslates to stronger evidence linking subject characteristics,iomechanics, and the development of upper-limb injuriesmong manual wheelchair users with paraplegia.

To our knowledge, our study is the first to directly examinehe relationship between shoulder pain, both in the last monthnd during propulsion, and biomechanics during wheelchairropulsion. Our study is unique because we describe propul-ion at a self-selected speed in addition to 2 steady-state targetpeeds. We found no significant differences in the way subjectsith shoulder pain propelled their wheelchairs compared witharticipants without pain. We previously reported11 that per-ons who experience higher posterior forces and internal rota-ion moments during propulsion are more likely to have shoul-er pathology compared with manual wheelchair users withoutathology. Although it seems likely that higher forces andoments lead to pathology, rather than pathology causing

ersons to use higher forces and moments, we cannot draw thisonclusion with a cross-sectional study design. However, theesults of the current study imply that shoulder pain does notlter the way a person propels their wheelchair. This strength-ns our previous suggestion that propulsion biomechanics con-ribute to pathology, rather than pain or pathology influencingropulsion style. Persons were excluded from the current studyf they had pain severe enough to limit their ability to propel theheelchair, which likely contributed to the lack of biome-

hanic differences between subjects with and without shoulderain. This does not diminish our findings, because the samexclusion criterion was used in our study linking biomechanicso shoulder pathology. Certainly more severe pain may limit aerson’s ability to propel the wheelchair or require them tolter their propulsion style; however, this does not seem to behe case for most manual wheelchair users. In addition, weound that people who experienced shoulder pain during pro-ulsion tended to be younger, with a mean age difference ofess than 10 years. Although this is difficult to explain, it isossible that older people have modified their propulsive stroken such a way that they experience less pain during propulsion.

Speeds of Wheelchair Propulsion

.9m/sn�50)

1.8m/s(n�50) Significance of Speed (P )

�10.2§ 47.1�9.3‡ .030*�17.8†§ 27.6�17.3†‡ �.001*�8.2 53.3�7.8 .042*�6.6 31.4�6.2 NS�12.0 83.7�10.4 NS�20.0†§ 6.9�21.0‡ �.001*

a 1-way repeated-measures ANOVA.justed pairwise comparisons).pairwise comparisons).pairwise comparisons).

for 3

0(

45.915.652.630.883.115.6

usingni-ad

lternatively, some people with shoulder pain during manual


wbat

idtskdsrsa

ptcbwfhtreTisdmdpwsalrw

iistatadpdaIsmrirtrficdAdoe

ip

iwiawiptspdbtesOaTrrctc

obsdtarthatn

hoofmfidIifUOptcarsOfe

pm


A

heelchair propulsion may have transitioned to powered mo-ility and would not have been included in our study, becausell participants were manual wheelchair users. This is a selec-ion bias of our study.

Other studies have investigated shoulder biomechanics dur-ng wheelchair propulsion, but differences in test conditions,ata collection, biomechanic models, and subject characteris-ics make direct comparisons of results difficult. In particular,tudies use different coordinate systems when reporting jointinetics and kinematics. We have made an effort to reduce thisifficulty in the future by reporting kinematics using ISBtandards.15 We are unaware of any standards in terms ofeporting glenohumeral joint loading, although we used theame local coordinate systems for both kinetic and kinematicnalysis.

In the current study, results are presented only for the pushhase of propulsion, although mean shoulder biomechanics forhis group of subjects are presented for the entire propulsionycle in figures 2 to 4. We chose to focus on the push phaseecause this is when forces are being applied by the hand,hich directly leads to loading at the shoulder. It is obvious

rom the figures that glenohumeral loading occurs while theand is off the pushrim, and these forces are sometimes largerhan forces during the push phase. Shoulder kinetics duringecovery are due to the weight of the arm, as well as upper-xtremity motion resulting from the chosen recovery pattern.his is much more variable between subjects, because the hand

s not constrained to the path of the pushrim. Future workhould examine how different recovery patterns20 affect shoul-er biomechanics. We presented both shoulder forces andoments, because although related, they can affect the joint

ifferently. Shoulder forces generated during wheelchair pro-ulsion can drive the humerus upward toward the acromion,hich affects the soft tissue surrounding the joint. Muscles

urrounding the glenohumeral joint generate force to counter-ct the shoulder moments generated during propulsion. Unevenoading on surrounding muscles during propulsion may be aeason for muscle imbalance frequently observed in manualheelchair users.21

Despite many differences, trends exist throughout most stud-es. Shoulder joint forces and moments have been shown toncrease at faster speeds, as was confirmed in the currenttudy.3,4,11,13 Compared with our most recent study, whichested an entirely different group of persons with paraplegia on

similar dynamometer system, our results are fairly consis-ent.11 The force magnitudes are slightly lower, but this can bettributed to differences in the rolling resistance of the 2ynamometer systems. The 2 largest shoulder joint forces wereosterior and inferior force. Joint loading in the posteriorirection results from forces actively applied to the pushrim,nd inferior force is due primarily to the weight of the arm.ncreased posterior force has previously been associated withhoulder joint pathology, and as such, it is important to mini-ize shoulder joint loading in this direction.11 The internal

otation moment was the largest directional moment observedn this study. Previously, the extension moment was the largesteported moment, but these studies all reported moments rela-ive to the trunk or a global coordinate system, rather thaneferenced to the humerus.3,13 This further highlights the dif-culty in making comparisons between different studies. In-reased internal rotation moments have been related to shoul-er pathology as measured on magnetic resonance imaging.11

lso of note, the active propulsion moments (extension, ad-uction, internal rotation) were at least twice as large as thepposing moments about the same axes (flexion, abduction,
xternal rotation). This may contribute to rotator cuff muscle d

mbalance, which has been identified as a risk factor for im-ingement.21

Shoulder kinematics were less affected by propulsion veloc-ty than glenohumeral joint loading; however, some differencesere noted at the end of the push phase. Shoulder flexion

ncreased and the humerus was less internally rotated (closer toneutral posture) as propulsion speed increased. A larger ROMas used to propel at faster speeds, even though cadence

ncreased. Because the hand was moving faster along theushrim at the faster speed, it is likely that momentum carriedhe arm further forward during the recovery phase before thehoulder was actively extended in preparation for the nextropulsive stroke. As previously reported,13,14 maximal shoul-er extension, abduction, and internal rotation occur near theeginning of the push phase, and then the humerus movesoward a more neutral posture at the end of push phase. Koontzt al13 presented a summary of shoulder angles during propul-ion for 5 studies, which compare favorably with our results.ne study14 used Euler angles to describe shoulder kinematics

nd reported a slightly larger ROM than our current study.hese differences can be explained because the previous work

eferenced humeral kinematics to a global coordinate system,ather than the trunk, and also studied the entire propulsionycle as opposed to focusing on the push phase. Trunk ROM inhe sagittal plane can be as high as 15°, and therefore should beonsidered when comparing results from different studies.22

We were also interested in the relationship between the timef maximal shoulder joint loading and arm posture, because weelieve that loading in a vulnerable position may contribute tohoulder injuries. Peak resultant force and most of the peakirectional forces and moments occurred during the first half ofhe propulsion cycle. The humerus was extended (18°–33°),bducted (45°–49°), and internally rotated (54°–68°). Internalotation combined with abduction or shoulder flexion leaveshe supraspinatus tendon vulnerable to impingement.23 Theumerus was closer to the end ROM at the time of peak loadingt the lower speed conditions. This is important because evenhough forces are lower at the slow speeds, the arm is in a lesseutral posture.We found that many demographic variables (sex, age,

eight) correlated to biomechanic variables, but we were curi-us whether these correlations were simply a secondary effectf body weight differences among sexes, ages, and heights. Inact, when all of these factors were forced into a regressionodel, body weight was the sole predictor of shoulder joint

orce. At all propulsion speeds, heavier subjects experiencedncreased loading in the anterior, posterior, inferior, and lateralirections. They also experienced a greater resultant force.ncreased joint loading can be detrimental when applied repet-tively as in wheelchair propulsion. This emphasizes the needor manual wheelchair users to maintain a healthy body weight.nfortunately, this is not always easy to achieve or enforce.ther interventions may be able to reduce the force required toropel a wheelchair independent of body weight. For example,he lightest-weight wheelchair possible should be prescribed toomplement weight management strategies. Adjusting the axles far forward as possible, without compromising stability,educes the rolling resistance of the wheelchair and has beenhown to reduce superior force experienced at the shoulder.4,24

ther benefits of a forward axle position include reduced strokerequency and less activity in prime movers of the upperxtremity.25,26

A long, smooth stroke that maximizes contact with theushrim and avoids abrupt changes in direction is recom-ended to reduce stroke frequency and peak force.27,28 Shoul-

er extension, abduction, and internal rotation cannot be

aloiiIprm

S

csevAistsovfdIcdirpw

tmbitotiirdchw

bpsjjPpgfeabrrs

sp

1

1

1

1

1

1

1

1

1

1


voided during propulsion. However, instead of applying aarge peak force over a short time period, distributing the forcever the length of a large contact angle may reduce the risk ofnjury. This study found that subject characteristics did notnfluence or limit the ROM used during wheelchair propulsion.nstead, shoulder kinematics are restricted during the pushhase because the hand must remain in contact with the push-im. Other factors such as shoulder position relative to the axleay influence shoulder kinematics more strongly.

tudy LimitationsOne limitation of studies of wheelchair biomechanics, in-

luding the current study, is that testing is completed in aimulated environment, either on a dynamometer or wheelchairrgometer. Testing on a dynamometer allowed us to controlelocity while testing each subject in his/her own wheelchair.lso, kinematic data can be collected for an extended (unlim-

ted) period of time, and variability in propulsion surfaces andlopes is eliminated, which is particularly important in a mul-isite study design. However, the self-selected velocity ob-erved in the current study is slower than would be expected forverground propulsion.29 Although these simulated testing en-ironments afford many advantages for conducting research,uture efforts should incorporate overground testing to fullyescribe wheelchair propulsion in real-world environments.nstrumentation can also affect the results in a study of wheel-hair biomechanics. The SmartWheel is a widely used, vali-ated device that allows researchers to monitor pushrim kinet-cs wirelessly. However, the SmartWheel may increase rollingesistance because it has a solid insert, but this ensures that tireressure does not vary between subjects or sites, as is possibleith pneumatic tires.Although this study is descriptive in nature, future work will

ake further advantage of the large sample size afforded by aultisite recruitment effort. The relationship between shoulder

iomechanics and injury, measured by magnetic resonancemaging, radiography, and physical examination, will be inves-igated. The effect of propulsion pattern and wheelchair setupn shoulder biomechanics may also provide insight into poten-ial interventions to reduce shoulder loading during this repet-tive task. The ultimate goal of this large, multisite study is todentify risk factors for upper-limb repetitive strain injuriesesulting from wheelchair propulsion. This will lead to theevelopment of interventions, such as propulsion training orhanges to wheelchair equipment, that we hope will reduce theigh incidence of shoulder pain and injury among manualheelchair users.

CONCLUSIONSA multisite study design enabled us to investigate shoulder

iomechanics during wheelchair propulsion in a large group ofersons with paraplegia. This study describes the impact ofubject characteristics, pain, and propulsion speed on shoulderoint loading and kinematics. Significant increases in shoulderoint loading were observed as propulsion velocity increased.ain, however, did not affect shoulder biomechanics duringropulsion. In addition, body weight was the primary demo-raphic variable related to glenohumeral joint loading. Weound that peak shoulder joint loading occurs when the arm isxtended and internally rotated, which may leave the shouldert risk for injury. Previous work28,30-32 in both wheelchairiomechanics and ergonomics has identified a link betweenepetitive loading tasks and the development of overuse inju-ies. Body weight maintenance and other interventions de-
igned to reduce the force required to propel a wheelchair,
hould be implemented to reduce the prevalence of shoulderain and injury among manual wheelchair users.

References1. Cooper R, Boninger M, Shimada S, Lawrence B. Glenohumeral

joint kinematics and kinetics for three coordinate system repre-sentations during wheelchair propulsion. Am J Phys Med Rehabil1999;78:435-46.

2. Finley M, Rasch E, Keyser R, Rodgers M. The biomechanics ofwheelchair propulsion in individuals with and without upper-limbimpairment. J Rehabil Res Dev 2004;41:395-402.

3. Kulig K, Rao S, Mulroy S, et al. Shoulder joint kinetics during thepush phase of wheelchair propulsion. Clin Orthop Relat Res1998;Sep(354):132-43.

4. Mulroy S, Newsam CJ, Gutierrez D, et al. Effect of fore-aft seatposition on shoulder demands during wheelchair propulsion: part1. A kinetic analysis. J Spinal Cord Med 2005;28:214-21.

5. Rodgers M, Gayle GW, Figoni S, Kobayashi M, Lieh J, Glaser R.Biomechanics of wheelchair propulsion during fatigue. Arch PhysMed Rehabil 1994;75:85-93.

6. Ballinger D, Rintala D, Hart K. The relation of shoulder pain andrange-of-motion problems to functional limitations, disability, andperceived health of men with spinal cord injury: a multifacetedlongitudinal study. Arch Phys Med Rehabil 2000;81:1575-81.

7. Pentland WE, Twomey LT. The weight-bearing upper extremityin women with long term paraplegia. Paraplegia 1991;29:521-30.

8. Bayley JC, Cochran TP, Sledge CB. The weight-bearing shoulder.The impingement syndrome in paraplegics. J Bone Joint Surg Am1987;69:676-8.

9. Nichols PJ, Norman PA, Ennis JR. Wheelchair user’s shoulder?Shoulder pain in patients with spinal cord lesions. Scand J RehabilMed 1979;11:29-32.

0. Subbarao J, Klopfstein J, Turpin R. Prevalence and impact ofwrist and shoulder pain in patients with spinal cord injury. J SpinalCord Med 1995;18:9-13.

1. Mercer JL, Boninger M, Koontz A, Ren D, Dyson-Hudson T,Cooper R. Shoulder joint kinetics and pathology in manual wheel-chair users. Clin Biomech (Bristol, Avon) 2006;21:781-9.

2. Boninger ML, Cooper RA, Shimada SD, Rudy TE. Shoulder andelbow motion during two speeds of wheelchair propulsion: adescription using a local coordinate system. Spinal Cord 1998;36:418-26.

3. Koontz AM, Cooper RA, Boninger ML, Souza AL, Fay BT.Shoulder kinematics and kinetics during two speeds of wheelchairpropulsion. J Rehabil Res Dev 2002;39:635-50.

4. Rao S, Bontrager EL, Gronley J, Newsam CJ, Perry J. Three-dimensional kinematics of wheelchair propulsion. IEEE TransRehabil Eng 1996;4:152-60.

5. Wu G, van der Helm F, Veeger HE, et al. ISB recommendation ondefinitions of joint coordinate systems of various joints for thereporting of human joint motion—part II: shoulder, elbow, wristand hand. J Biomech 2005;38:981-92.

6. DiGiovine CP, Cooper RA, Boninger ML. Dynamic calibration ofa wheelchair dynamometer. J Rehabil Res Dev 2001;38:41-55.

7. Cooper R, Robertson R, VanSickle D, Boninger M, Shimada S.Methods for determining three-dimensional wheelchair pushrimforces and moments: a technical note. J Rehabil Res Dev 1997;34:162-70.

8. Koontz A, Yang Y, Price R, et al. Multi-site comparison ofwheelchair propulsion kinetics in persons with paraplegia. J Re-habil Res Dev 2007;44:449-58.

9. Hanavan E. A mathematical model of the human body. Wright-Patterson Air Force Base: Aerospace Medical Research Labora-
tories; Oct 1964. Report no. AMRL-TR-64-102.

2

2

2

2

2

2

2

2

2

2

3

3

3

a

b

cdef


A

0. Boninger ML, Souza AL, Cooper RA, Fitzgerald SG, Koontz AM,Fay BT. Propulsion patterns and pushrim biomechanics in manualwheelchair propulsion. Arch Phys Med Rehabil 2002;83:718-23.

1. McMaster W, Long S, Caiozzo V. Isokinetic torque imbalances inthe rotator cuff of the elite water polo player. Am J Sports Med1991;19:72-5.

2. Bednarczyk JH, Sanderson DJ. Kinematics of wheelchair propul-sion in adults and children with spinal cord injury. Arch Phys MedRehabil 1994;75:1327-34.

3. Neer C. Impingement lesions. Clin Orthop Relat Res 1983;Mar(173):70-7.

4. Brubaker CE. Wheelchair prescription: an analysis of factors thataffect mobility and performance. J Rehabil Res Dev 1986;23:19-26.

5. Boninger M, Baldwin MA, Cooper R, Koontz A, Chan L. Manualwheelchair pushrim biomechanics and axle position. Arch PhysMed Rehabil 2000;81:608-13.

6. Gutierrez D, Mulroy S, Newsam CJ, Gronley JK, Perry J. Effectof fore-aft seat position on shoulder demands during wheelchairpropulsion: part 2. An electromyographic analysis. J Spinal CordMed 2005;28:222-9.

7. Paralyzed Veterans of America Consortium for Spinal Cord Med-icine. Preservation of upper limb function following spinal cordinjury: a clinical practice guideline for health-care professionals.J Spinal Cord Med 2005;28:434-70.

8. Boninger M, Koontz A, Sisto S, et al. Pushrim biomechanics andinjury prevention in spinal cord injury: recommendations based onCULP-SCI investigations. J Rehabil Res Dev 2005;42(3 Suppl1):9-20.

9. Newsam CJ, Mulroy S, Gronley J, Bontrager EL, Perry J. Tem-
poral-spatial characteristics of wheelchair propulsion: effects of

level of spinal cord injury, terrain, and propulsion rate. Am J PhysMed Rehabil 1996;75:292-9.

0. NIOSH. Shoulder musculoskeletal disorders: evidence for work-relatedness. Musculoskeletal disorders and workplace factors: acritical review of epidemiology for work related musculoskeletaldisorders of the neck, upper extremity, and low back. Cincinnati:National Institute for Occupational Safety and Health, Publica-tions Dissemination; 1997. p 122-94.

1. National Research Council, Institute of Medicine. Musculoskele-tal disorders and the workplace: low back and upper extremities.Panel on Musculoskeletal Disorders and the Workplace. Commis-sion on Behavioral and Social Sciences and Education. Washing-ton (DC): Natl Acad Pr; 2001.

2. Steering Committee for the Workshop on Work-Related Muscu-loskeletal Injuries: The Research Base, Committee on HumanFactors, National Research Council. Work-related musculoskele-tal disorders: a review of the evidence. Washington (DC): Na-tional Academy Pr; 1999.

Suppliers. Three Rivers Holdings, 1826 W Broadway Rd, Ste 43, Mesa, AZ

85202.. Northern Digital Inc, 103 Randall Dr, Waterloo, ON N2V 1C5,

Canada.. Vicon, 9 Spectrum Pointe Dr, Lake Forest, CA 92630.. Qualisys AB, Packhusgatan 6, S-411 13 Gothenburg, Sweden.. The MathWorks Inc, 3 Apple Hill Dr, Natick, MA 01760-2098.. Version 11.0; SPSS Inc, 233 S Wacker Dr, 11th Fl, Chicago, IL

60606-6307.

shoulder biomechanics during the push phase of wheelchair propulsion: a multisite study of persons...

Documents