UBC Faculty Research and Publications

Reliability and validity of using the Lokomat to assess lower limb joint position sense in people with… Domingo, Antoinette; Lam, Tania Dec 16, 2014

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


52383-12984_2014_Article_684.pdf [ 471.59kB ]
JSON: 52383-1.0223859.json
JSON-LD: 52383-1.0223859-ld.json
RDF/XML (Pretty): 52383-1.0223859-rdf.xml
RDF/JSON: 52383-1.0223859-rdf.json
Turtle: 52383-1.0223859-turtle.txt
N-Triples: 52383-1.0223859-rdf-ntriples.txt
Original Record: 52383-1.0223859-source.json
Full Text

Full Text

RESEARCHReliability and validity of uoSpinal cord injury (SCI) often results in complete or par- and increase participation and quality of life.J N E R JOURNAL OF NEUROENGINEERINGAND REHABILITATIONDomingo and Lam Journal of NeuroEngineering and Rehabilitation 2014, 11:167http://www.jneuroengrehab.com/content/11/1/167dinated movements, including walking, is proprioceptivesense - the sense of position and movement (kinesthesia) [8].2International Collaboration on Repair Discoveries (ICORD), Vancouver, BC,Canadatial paralysis, affecting the ability to walk and participatein physical activity. Because of this diminished mobility,people with SCI are at high risk of secondary complica-tions such as compromised cardiovascular health, pressuresores, and osteoporosis [1,2]. Therapeutic interventionsthat can improve walking ability are important becauseStrategies to improve ambulation in people with SCIhave largely focused on enhancing motor output [3].Intensive, task-specific gait retraining strategies thatprovide repeated practice of walking movements havebeen shown to improve walking function [4-6]. It isthought that the sensory information provided by the re-peated practice of movements promotes neural plasticitythrough use-dependent mechanisms [4,7].One key sensory modality critical for the control of coor-* Correspondence: antoinette.domingo@sdsu.edu1School of Kinesiology, University of British Columbia, Vancouver, BC, CanadaBackgroundAbstractBackground: Proprioceptive sense (knowing where the limbs are in space) is critical for motor control duringposture and walking, and is often compromised after spinal cord injury (SCI). The purpose of this study was toassess the reliability and validity of using the Lokomat, a robotic exoskeleton used for gait rehabilitation, toquantitatively measure static position sense of the legs in persons with incomplete SCI.Methods: We used the Lokomat and custom software to assess static position sense in 23 able-bodied (AB)subjects and 23 persons with incomplete SCI (American Spinal Injury Association Impairment Scale level B, C or D).The subject’s leg was placed into a target position (joint angle) at either the hip or knee and asked to memorizethat position. The Lokomat then moved the test joint to a “distractor” position. The subject then used a joystickcontroller to bring the joint back into the memorized target position. The final joint angle was compared to thetarget angle and the absolute difference was recorded as an error. All movements were passive. Known-groupsvalidity was determined by the ability of the measure to discriminate between able-bodied and SCI subjects. Toevaluate test-retest reliability, subjects were tested twice and intra-class correlation coefficients comparing errorsfrom the two sessions were calculated. We also performed a traditional clinical test of proprioception in subjectswith SCI and compared these scores to the robotic assessment.Results: The robot-based assessment test was reliable at the hip and knee in persons with SCI (P ≤ 0.001). Hip andknee angle errors in subjects with SCI were significantly greater (P ≤ 0.001) and more variable (P < 0.0001) than inAB subjects. Error scores were significantly correlated to clinical measure of joint position sense (r ≥ 0.507, P ≤ 0.013).Conclusions: This study shows that the Lokomat may be used as a reliable and valid clinical measurement tool forassessing joint position sense in persons with incomplete SCI. Quantitative assessments of proprioceptive deficitsafter neurological injury will help in understanding its role in the recovery of skilled walking and in thedevelopment of interventions to aid in the return to safe community ambulation.they can help to reduce these secondary complicationsassess lower limb joint pwith incomplete spinal coAntoinette Domingo1,2* and Tania Lam1,2© 2014 Domingo and Lam; licensee BioMed CCommons Attribution License (http://creativecreproduction in any medium, provided the orDedication waiver (http://creativecommons.orunless otherwise stated.Open Accesssing the Lokomat tosition sense in peoplerd injuryentral. This is an Open Access article distributed under the terms of the Creativeommons.org/licenses/by/4.0), which permits unrestricted use, distribution, andiginal work is properly credited. The Creative Commons Public Domaing/publicdomain/zero/1.0/) applies to the data made available in this article,Domingo and Lam Journal of NeuroEngineering and Rehabilitation 2014, 11:167 Page 2 of 10http://www.jneuroengrehab.com/content/11/1/167This is perhaps best illustrated by a case report of a personwho had lost all proprioceptive sensation below the neck [9].This person had to compensate for compromised balanceand stability by visually monitoring his steps and using a lar-ger base of support [9]. In cases of people with pyridoxine(vitamin B6) toxicity, which damages large-diameter affer-ents, there have been reports of ataxic gait, demonstratingthe important role of proprioception in inter-joint coordin-ation during walking [10]. Also, impairments in obstaclecrossing in people with peripheral neuropathy secondary todiabetes have been associated with impaired proprioception[11]. Given that a SCI could damage ascending proprio-ceptive tracts, it seems vital to understand how proprio-ceptive deficits in people with SCI impact functionalambulation, especially when performing skilled loco-motor tasks such as walking over uneven terrain, obs-tacle crossing or stair negotiation.Clinical assessments of proprioceptive sense used byclinicians are not quantitative and lack sensitivity [12].For example, one clinical measure of joint position senseinvolves the clinician grossly moving a limb and askingthe patient to simply indicate the direction that the limbwas moved [13]. Another measure of proprioception in-volves imitating a presented movement but quantificationof the response is usually only estimated [14]. In addition,when performing these manual test, the velocity of move-ment, points of contact, and force of contact applied tothe individual can vary, affecting the results of the examin-ation. The administration of these manual tests is verydifficult to standardize and quantify.Several groups have developed methods to quantita-tively measure proprioception in the upper extremities ofable-bodied subjects [15-19]. In addition, tools have beendeveloped to quantitatively measure joint position sensein the upper extremity in persons with stroke [20-22] andhemiplegic cerebral palsy [18]. There are a number ofstudies that quantitatively measure kinesthesia in thelower limb [23-26] and proprioception related to jointdysfunction [27-30]. However, there are no tools that aresuitable for quantitative testing of lower limb propriocep-tion in people with neurological injury. A reliable and pre-cise method to measure proprioceptive sense in the lowerlimbs, especially one that could be used for neurologicalpopulations, is needed. Precise clinical assessments of sen-sory function are necessary to evaluate the effectiveness oftreatments and understand the role of proprioceptivesense in locomotor recovery. This is an essential first stepthat precedes the development of treatments to improvesensory function and ultimately maximize skilled walkingand community ambulation.Thus, the purpose of our study was to test the reliabilityand validity of a new quantitative assessment tool of lowerlimb joint position sense using the Lokomat (HocomaAG, Volketswil, Switzerland), a robotic gait rehabilitationdevice. We hypothesized that the Lokomat-based assess-ment of proprioceptive sense would be a reliable and validmethod of measuring conscious proprioception in personswith incomplete spinal cord injury (iSCI).MethodsSubjectsIndividuals with iSCI were recruited to participate.Participants had to meet the following inclusion criteria: 1)at least 6 months post injury; 2) were in stable medicalcondition; 3) no history of musculoskeletal disease; 4)no cardiovascular condition where exercise was contrain-dicated; 5) weight was less than 300 lbs and height wasless than 6’1” due to the capacity limits of the roboticexoskeleton; 6) 19 years of age or older; 7) able to followdirections so they could complete the experiment.Able-bodied participants were included if they: 1) hadno neurological, cardiovascular, or musculoskeletal injur-ies interfering with their ability walk 2) weight was lessthan 300 lbs and height was less than 6’1” due to the capa-city limits of the robotic exoskeleton; 3) 19 years of age orolder.Subjects participated in this study with informed, writtenconsent. Experimental procedures were approved by theResearch Ethics Board at the University of British Columbiaand were conducted in accordance with the Declaration ofHelsinki.Robotic assessmentWe used the Lokomat, a robotic lower extremity exo-skeleton, and custom software to quantitatively assesslower extremity static joint position sense (Figure 1A).The Lokomat is a computer-controlled motorized gaitrehabilitation system consisting of a pair of robotic legsto which the thighs and lower legs are strapped. Thethigh and shank segments of the Lokomat only allowmovement in the sagittal plane and are moved by linearmotors housed within the exoskeletal structure. Encoderswithin the exoskeleton measure the hip and knee jointangles. The Lokomat was adjusted according to the lengthand size of the subject's legs. Subjects were secured to theLokomat by leg cuffs around the mid-thigh, upper shank,and lower shank, as well as a waist belt. Each robotic legattached to a central horizontal frame that was secured tothe subject around the pelvis.Subjects were suspended in an upright position abovethe ground using a body weight support system. Thishelped to ensure that the leg could move freely withouttouching the treadmill surface. The ankle of the test legwas fixed into a neutral position throughout the experi-ment with the use of passive foot lifter straps. Foampadding was placed in between the straps and the an-terior surface of the limb to decrease any sensory cuesfrom the straps as the leg was moved during testing.hectsSub aDomingo and Lam Journal of NeuroEngineering and Rehabilitation 2014, 11:167 Page 3 of 10http://www.jneuroengrehab.com/content/11/1/167Only one leg was tested per subject. In persons withspinal cord injury, we tested the leg with the leastamount of spasticity. In able-bodied subjects, we testedthe right leg. The foot of the untested leg was placedonto a platform so that the subject could bear someweight on that leg for comfort.AFigure 1 Experimental set-up and procedures. A. Subjects were attacweight support system with the ankle fixed in neutral position. All subjewas obscured with a curtain and only one joint was tested at a time. B.away from that angle, and the subject used the joystick to place the limactual angle were recorded as an error.The Lokomat moved the leg into predetermined posi-tions and speeds using custom software. When subjectswere asked to move their leg, they all used a joystick con-troller to change the hip or knee angles. This bypassed re-strictions due to variations in the extent of voluntarycontrol over the lower limb between individuals. We setthe Lokomat to move the leg at 7° per second, exceptwhen using the joystick controller to move the leg, wherethe speed was 3°or 6° per second, based on the joystickangle. If the subject moved the joystick slightly away fromcenter, the joint moved at 3°/sec, and when pushed fur-ther, the leg moved at 6°/sec. We chose to use a differentspeed when using the joystick controller so that subjectswould not use movement time as a cue to the location ofthe target position. Joint angle data from the encoderswere collected using custom software written in LabView(National Instruments, Austin, TX, USA).ProceduresTwo hip angles (10° extension, 30° flexion) and two kneeangles (10° flexion, 50° flexion) were used as target posi-tions for a total of 4 combinations of angles. These angleswere chosen because they spanned the range of motiontypically used during walking. Each target angle wasassessed 5 times. Only one joint was tested at a time, soeach combination was done twice, resulting in 40 trials persubject. The order of joint testing was randomized betweensubjects. Subjects were made aware of which joint wasgoing to be tested for each set of trials. The order of anglestested was randomized within each joint tested. Vision ofthe legs was obscured with a curtain. Subjects wereTarget Distractor ActualLokomat moves kneeSubject moves kneeMemorize knee angleBd to the Lokomat and suspended above the ground using a bodycontrolled passive leg movements with a joystick. Vision of the legsbjects were presented with a target angle, the leg was then movedt the remembered target angle. Difference in the target angle andinstructed to keep their leg passive throughout testing.Subjects were given breaks from the body weight supportas needed. Blood pressure was measured during the breaksto ensure it stayed near baseline values throughout thestudy. The assessment was approximately 1.5 hours in dur-ation for each subject, inclusive of all trials and rest breaks.The hip or knee was moved into a target position andheld there for 5 seconds. The subject was asked tomemorize the angular position of the joint being tested(the hip or knee). The Lokomat then moved the test jointinto another “distractor” position for 5 seconds while theother joint was maintained in the same position. Dis-tractor angles were either 15° or 30° away from the targetangle (see Tables 1 & 2). The subject was then asked tobring the test joint back into the memorized target pos-ition with the joystick controller. The final joint angle, or“actual” angle, was compared to the target angle and thedifference was recorded as an error (Figure 1B).These procedures were repeated on a second day atleast one week later to assess test-retest reliability.Clinical assessmentWe also performed a clinical test of proprioception [13] insubjects with SCI, where the leg was moved (either thehip or knee) ~10° by an experimenter from a randomTable 1 Tasks for hip proprioception assessmentTask Hip target angle (°) Distractor angle (°) Knee angle (°)1 30 F 0 10 F2 30 F 15 F 50 F3 10 E 5 F 10 FDomingo and Lam Journal of NeuroEngineering and Rehabilitation 2014, 11:167 Page 4 of 10http://www.jneuroengrehab.com/content/11/1/167starting position, and the subject had to indicate whetherthe movement was “up” or “down”. This was repeated 10times at each joint [14], and the number of incorrectresponses were compared to the error scores of the roboticassessment. The same experimenter performed the clinicalassessment for all subjects.EMGFor a subset of subjects (AB: N = 14, SCI: N = 16), EMGdata were recorded from the rectus femoris, medial ham-string, medial gastrocnemius and tibialis anterior duringthe robotic assessment. EMG signals were visually moni-tored on-line to ensure that the muscles remained quies-cent during the tests. If muscle activity was observed (dueto voluntary activation or spasticity), the trial was repeateduntil no muscle activity was observed.Data analysis and statisticsAll data analysis was performed using IBM SPSS Statistics21 (New York, NY). At each target angle, we took the ab-solute values of all the errors. Smaller errors are associatedwith more accurate joint position sense. Means were re-ported with ±1 standard error. Significance was evaluatedat α = 0.05 for all statistical tests.Discriminative validityTo test if there were differences between groups (SCI vs.AB), target angles, and repetitions, we performed a mixedfactorial analysis of variance (ANOVA) (within: 4 targetangles x 5 repetitions; between: 2 groups) comparingabsolute errors from the first day of testing, separately forthe hip and the knee. Post hoc analysis was performed asneeded to delineate specific differences between target an-gles (with a Sidak correction for multiple comparisons).4 10 E 20 F 50 FF: Flexion; E: Extension.These outcomes provided information on discriminativevalidity (between groups) of the robotic assessment.Table 2 Tasks for knee proprioception assessmentTask Knee target angle (°) Distractor angle (°) Hip angle (°)1 10 F 40 F 30 F2 10 F 25 F 10 E3 50 F 35 F 30 F4 50 F 20 F 10 EF: Flexion; E: Extension.We also performed a mixed factorial ANOVA on thestandard deviation of Day 1 error scores (within: 4 targetangles; between: 2 groups), separately for the hip and knee.Post hoc analysis was performed as needed to delineatespecific differences between target angles (with a Sidak cor-rection for multiple comparisons). Subjects with poorerproprioception and with larger angle errors would also beexpected to have greater variability in their responses.Test–retest reliabilityIn addition to the mixed factorial analysis, we alsocalculated intraclass correlation coefficients (ICC) [31]and Bland-Altman tests [32] to assess reliability. Theintraclass correlation coefficients were calculated onthe overall average scores of 20 trials for the hip andknee separately, for able-bodied subjects and subjectswith SCI. An ICC value of less than 0.40 indicates poorreproducibility, ICC values in the range 0.40 to 0.75indicate fair to good reproducibility, and an ICC value ofgreater than 0.75 shows excellent reproducibility [33].The Bland-Altman tests include (1) a graphical repre-sentation (Bland–Altman plot) of the difference be-tween test measures plotted against the mean of thetwo measures; (2) calculation of the mean of the differ-ence between test measures and 95% confidence inter-vals (CI); and (3) a measure of the limits of agreement(LOA) between the two measures, which is defined asd ± 1.96 x SDdiff, where d is the difference and SDdiff isthe standard deviation of the differences.Internal consistencyInternal consistency between the different trials was mea-sured using Cronbach’s alpha. We analyzed absolute errorscores from the first day of testing, separately for the hipand knee (20 trials at each joint) in both AB and SCI par-ticipants. We used this to determine if the number oftrials could be reduced for future experiments.An exploratory factor analysis (principal componentsmethod) was also performed to evaluate dimensionality ofthe assessment. Averaged errors at each target for the SCIdata were entered into separate analyses. The criteria forthe factors were based on Kaiser stopping criteria, whereselected factors had an eigenvalue above 1.0.Convergent validityRobotic assessment scores were compared to resultsfrom the clinical assessment of proprioception. We usedSpearman’s rank correlation to compare these results.ResultsParticipantsTwenty three able-bodied subjects (9 males, 14 females;age = 37.8 ± 14.1 years (mean ± SD)) and 23 subjects withincomplete SCI (19 males, 4 females; age = 40.5 ±14.0 years, American Spinal Injury Association Impair-ment Scale (AIS) = B-D, 6.3 ± 5.6 years post-injury) par-ticipated in the study. All participants with iSCI werecommunity dwelling but varied greatly with respect toinjury and walking ability (Table 3). The AB and SCIgroups were not significantly different in terms of age(P = 0.51), but the proportion of males and females weredifferent between groups. Most subjects were able tomaintain quiescent muscles throughout the testingperiod. For some subjects, if any increase in EMG wasobserved during a trial, subjects were able to reducetheir muscle activity when the trial was repeated.Comparison of absolute errors between groups, targetangles, and repetitionsAble bodied subjects had an overall average absoluteerror of 2.63° ± 0.17 (mean ± standard error) at the hipand 4.05° ± 0.28 at the knee, while participants with SCIhad an overall average of 6.64° ± 1.18 at the hip and13.31° ± 1.75 at the knee across 20 trials (5 repetitions of4 different target angles, Day 1) (Figure 2A). TheANOVA showed that there was a significant differencein error scores between the AB and SCI groups for thehip and knee (P ≤ 0.001 for both joints). At the hip, therewere no differences between target angles or repetitions(P = 0.199 and 0.426, respectively, Lower-bound correc-tion due to violation of sphericity). Similarly at the knee,there were no differences between target angles or repe-titions (P = 0.067 and 0.392, respectively, Lower-boundcorrection due to violation of sphericity). There were nointeraction effects.Comparison of standard deviation of error scoresSubjects with SCI had greater variability (larger standarddeviations) in their error scores than able-bodied subjects(hip and knee: P < 0.0001) (Figure 2B). There were no dif-ferences in variability between target angles (hip and knee:P > 0.129, Huynh-Feldt correction due to violation ofsphericity) or interaction effects (hip and knee: P > 0.207).Test-retest reliabilityThe calculated intraclass correlation coefficients show thatthere was fair to excellent agreement of absolute errorsbetween the 2 days of testing when comparing the overallTable 3 Subject demographics and clinical characteristicsSubject Sex Age (yr) AIS AIS Level Time post-injury (yr) 10 MWT (comf) (s) 10 MWT (max) (s) Assistive device1 F 55 D T3 4 25.8 19.8 FC, AFO & FES2 M 29 B C6-C7 8 97.7 NT FWW3 M 67 C C4-C5 4 24.8 19.1 FC4 M 61 B C5 7 128.3 NT FWW & Bilateral arm trough1Domingo and Lam Journal of NeuroEngineering and Rehabilitation 2014, 11:167 Page 5 of 10http://www.jneuroengrehab.com/content/11/1/1675 F 23 B/C T12 36 M 46 C T12 47 F 19 B T12 18 M 34 C C7 189 M 47 C C3-T1 610 M 25 C T10 211 M 65 D C4-C5 512 M 41 D C4-C5 313 M 30 C T5-T6 1014 M 46 D C3 215 M 33 B T4 416 M 28 C C5-C6 417 F 57 C T7 2318 M 36 C C4-5 719 M 49 B C5 220 M 38 B C7 1521 M 21 B C5-C6 422 M 42 C C1-C2 923 M 41 C/D C4-5 1AIS: American Spinal Injury Association Impairment Scale; 10MWT (comf): 10-meterat maximum speed, in seconds; FC: forearm crutches; AFO: ankle-foot orthosis; FES:SW: standard walker; R: right.19.1 17.2 FC47.8 NT FWWNT NT NA38.3 27.1 FWW11.1 9.9 NoneNT NT FC & braces20.7 14.5 FC7.9 7.3 none65.4 55.1 SW & Left AFO9.2 7.1 noneNT NT none91.7 75.4 FWW35.0 86.0 SW38.1 28.2 FC, swedish cage right knee, right AFONT NT NTNT NT NTNT NT NT16.2 9.6 4 wheeled walker & AFO26.0 15.9 FWWwalk test at comfortable speed, in seconds; 10MWT (max): 10-meter walk testfunctional electrical stimulation; FWW: front-wheeled walker; NT: not tested;Average absolute error (deg)Hip 30Knee 10Hip 30Knee 50Hip -10Knee 10Hip -10Knee 5001020304050 Hip 01020304050 Knee Knee 10Hip 30Knee 50Hip 30Knee 10Hip -10Knee 50Hip -10A810eg)Hip 15 KneeBABSCIMaxMinleDomingo and Lam Journal of NeuroEngineering and Rehabilitation 2014, 11:167 Page 6 of 10http://www.jneuroengrehab.com/content/11/1/1670246Hip 30Knee 10Hip 30Knee 50Hip -10Knee 10Hip -10Knee 50Hip target angle (deg)SD of absolute error (d0510Figure 2 Absolute angle errors and error variability. A. Absolute angaverage scores (Figure 3A & B). In participants with SCI,ICC = 0.55 for the hip assessment and ICC = 0.882 for theknee. In AB participants, ICC = 0.493 for the hip andICC = 0.656 at the knee (all P ≤ 0.008).The Bland Altman procedures showed that the 95%confidence interval for the mean difference betweentest sessions (d) included 0 (mean [95% CI] (hip): −0.37[−2.7, 1.9], mean [95% CI] (knee): 0.98 [−0.90,2.88]),subjects with SCI were greater than in AB subjects for both joints (P≤ 0.00target angle or repetitions for either group. Angle errors are labeled by kneblack and gray dots represent maximum and minimum absolute angle errotested at a time. B. Standard deviations of angle errors for the hip and knedeviations (were more variable) (P < 0.0001, repeated measures ANOVA) thor repetitions for either group. Angle error standard deviations are labeled0510152025ASDaDaICC = 0.493* ICC = 0.550*Hip average absolute error (deg)AFigure 3 Test-retest reliability: intraclass correlation coefficients (ICC)and B. knee joint. ICC analysis showed the test had fair to excellent reprodthe ICC correlation coefficients are indicated by asterisks (P < 0.05).Knee 10Hip 30Knee 50Hip 30Knee 10Hip -10Knee 50Hip -10Knee target angle (deg)errors for the hip and knee joints (mean ± SEM). Angle errors inshowing there was no systematic change in the errorsbetween days (Figure 4A). The plots do, however, indicatethe presence of heteroscedasticity, where the variability ofd is unequal across the range of mean errors. In this case,the variability of d is greater at greater error scores. Weconfirmed the presence of heteroscedasticity by calculatingthe correlation coefficient between the absolute differenceand the average of test sessions and found that it was1, repeated measures ANOVA). There were no significant differences ine or hip target angle with respective positions at the other joint. Thers, respectively, across subjects for each target. Only one joint wase joints (mean ± SEM). Subjects with SCI had higher standardan in AB subjects. There were no significant differences in target angleby knee or hip target angle with respective positions at the other joint.BCIy 1y 205101520253035ICC = 0.656* ICC = 0.882*Knee average absolute error (deg)B. Average absolute angle errors for Day 1 and Day 2 for the A. hip jointucibility for AB subjects and SCI subjects (all P ≤ 0.001). Significance ofsignificant (hip: r = 0.594, P = 0.003; knee: r = 0.451, P =0.031) (Figure 4B).Comparison of robotic assessment to clinical assessmentIn participants with SCI, clinical scores were significantlycorrelated to robotic assessment scores at both the hipand knee (hip: rho = 0.507, P = 0.013; knee: rho = 0.790,P < 0.0001; Spearman’s rank correlation) (Figure 5A & B).Internal consistencyCronbach’s alpha showed good internal consistencyamong the different test items in both groups of sub-jects, and reliability would be minimally affected if anitem was deleted from the assessment. In SCI partici-pants, overall α = 0.915 at the hip joint, and if an itemwas deleted, the range of Cronbach’s alpha was [0.905,0.919]. At the knee joint, α = 0.868, and if an item was de-leted, the range of Cronbach’s alpha was [0.853, 0.874]. InAB participants, overall α = 0.723 at the hip joint, and ifan item was deleted, the range was [0.675, 0.756]. At theknee joint, α = 0.764, and if an item was deleted, the rangeof Cronbach’s alpha was [0.698, 0.777].Factor analysisFactor analysis showed that a clear solution consistingof 1 factor resulted when using principal components,factoring for both the hip and knee in subjects withSCI, based on the scree plots and eigenvalues. For thehip, 43.4% of the variance observed was explained byone factor. For the knee, 44.4% of the variance observedwas explained by one factor. This analysis indicatedthat each target angle contributed to overall proprio-ception error scores.DiscussionWe showed that the Lokomat, used with custom soft-ware, is a valid and reliable tool to measureKnee Average of Day 1 & 2 (deg) AceBland-Altman Plots−15−10−5051015HipDay 1−Day 2 (deg) −15−10−5051015Kneeresf te95dDomingo and Lam Journal of NeuroEngineering and Rehabilitation 2014, 11:167 Page 7 of 10http://www.jneuroengrehab.com/content/11/1/1670 2 4 6 8 10 12 14 16 18024681012141618Hip Average of Day 1 & 2 (deg)|Day 1−Day 2| (deg) BHeteros0 2 4 6 8 10 12 14 16 18Average of Day 1 & 2 (deg) Figure 4 Bland-Altman plots and heteroscedasticity. Data points repagreement of mean absolute angle errors between two separate days osessions is plotted versus the mean of the test sessions across days. Theshown as dashed lines. The 95% limits of agreement are shown as dottewith SCI. There is a positive relationship between the absolute difference a(hip: P = 0.003; knee: P = 0.031).ent data from each participant with SCI. A. Bland-Altman plots forsting for the hip and knee joints. The difference between the two test% confidence intervals of the mean difference between test days arelines. B. Heteroscedasticity plots for the hip and knee joints in subjects0 5 10 15 20 25 30024681012dasticity0 5 10 15 20 25 30Average of Day 1 & 2 (deg) nd the average of the test session across the two days of testingdifficult to do in other planes of motion and at otherDalinrmDomingo and Lam Journal of NeuroEngineering and Rehabilitation 2014, 11:167 Page 8 of 10http://www.jneuroengrehab.com/content/11/1/167proprioceptive sense of the lower extremities in peoplewith incomplete spinal cord injury. This tool was sensi-tive enough to detect differences in proprioceptive sensebetween SCI and AB groups, with significantly greaterangle errors in the SCI group for both the hip and knee.There was also large variability observed in propriocep-tive sense among subjects with spinal cord injury. Thiswas expected, since the degree of injury to the structuresthat carry conscious proprioceptive sense may varygreatly between individuals. Conscious proprioception isprimarily relayed in the dorsal columns [13,34], and ithas been shown that proprioceptive signals are alsotransmitted to the cerebral cortex via the spinocervicalthalamic tract [35,36]. Proprioceptive sense is derivedfrom sensory information from muscle spindles, golgitendon organs and skin mechanoreceptors [13].0 1 2 3 4 5 6 70510152025Clinical score (number incorrect)Average robotic assessment score (deg) HipAFigure 5 Comparison of robotic assessment and clinical measures.robot-based assessment to clinical measures at the A. hip and B. knee. Cboth joints (knee: rho = 0.790, P < 0.0001; hip: rho = 0.507, P = 0.013, SpeaThere are several testing paradigms for assessing pro-prioceptive sense, but we chose our protocol on what wefelt would be most appropriate for the SCI population.Other robotic assessments of proprioception have usedcontralateral joint matching tasks, particularly in theupper extremities, where movement occurs in thecontralateral limb and then is copied in the test limb[21,37,38]. The advantage of this method is that there isno reliance on memory to complete the task, and thesubject can rely on reference information of the contra-lateral limb in real time. However, it is suggested thatcontralateral limb matching tasks may lead to greatermatching errors than in ipsilateral tasks because of therequired transfer of information between the hemi-spheres to the respective somatosensory cortices [39].Even more importantly, in the current study, a jointmatching paradigm would not be appropriate because itwould be impossible to tell which side the deficit exists(i.e. contralateral matching tasks assume an uninjuredside). Investigators have measured joint position sense ofjoints in the lower extremities. Another testing paradigmuses voluntary movement of the limb or body segmentto achieve the target angles [42]. Because of the highvariation in motor ability between subjects with SCI, wechose to keep all movements passive by using a joystickto move the leg to the target angle. Therefore, given theconstraints of the limbs and joints tested, our testingprotocol is likely the most appropriate for measuringjoint position sense of sagittal plane motion in the legs.Correlation to clinical test of proprioceptive senseClinical measures of proprioception usually involve mov-hip rotation in older adults [40] and children [41] withcerebral palsy using visual paradigms, but this would be0 1 2 3 4 5 6 7051015202530Clinical score (number incorrect)KneeBta points represent data from each participant with SCI. Correlation ofical scores were significantly correlated to robotic assessment scores atan’s rank correlation).ing a limb segment in one direction, and then having thepatient copy the movement with the opposite limb orverbalize the direction to the clinician [14]. In thepresent study, the Lokomat based assessment was highlycorrelated to a clinical measure of proprioception [13].However, it is likely that the Lokomat based assessmentwas more a sensitive assessment since we observed aceiling effect in the clinical score (Figure 5A & B). Insubjects where the clinical test showed completelyintact proprioception (zero incorrect responses), theLokomat-based test showed a wide range of angleerrors (Figure 5A & B).The clinical test we used likely contains elements ofboth static position sense as well as movement sense,but it is one commonly used by clinicians to assessjoint position sense [14,43,44]. We chose to use thisparticular clinical assessment because it only involvedone limb at a time and the movements were passive.This helped to maintain similar conditions as theLokomat-based assessment.been shown to have a major role in maintaining standingDomingo and Lam Journal of NeuroEngineering and Rehabilitation 2014, 11:167 Page 9 of 10http://www.jneuroengrehab.com/content/11/1/167balance [45]. In any case, this study shows a means forassessing proprioceptive sense in a quantitative and reli-able manner, and this approach could theoretically be im-plemented in a robotic device developed for testing theankle [46]. Discomfort in the some of the subjects hadoccurred during testing due to being suspended in theharness for an extended period of time with relatively littlemovement of the lower extremities. This was resolved bytaking frequent breaks, monitoring signs and symptoms ofautonomic dysreflexia (e.g., taking blood pressure inter-mittently), and encouraging movement during breaks tofacilitate blood flow. Reducing the number of total trialswould also help to resolve this issue, and is reasonable giventhe good internal consistency (Cronbach’s alpha of 0.868for the knee and 0.916 for the hip) of the Lokomat-basedassessment.Our sample population lacked any individuals withlower spinal cord injuries (e.g., cauda equina injuries),therefore the reliability and validity of the Lokomat-basedassessment for this group still needs to be evaluated. Inaddition, it may not be feasible to use this assessment inthose with severely limited joint range of motion sincemovement of the limb segment is needed to test eachAdvantages of robotic assessmentUsing the Lokomat with custom software provided a sys-tematic and reliable way to assess proprioceptive sense inpersons with SCI. When testing proprioceptive sense, it isvery important to be consistent between testing sessionswith tested angles, reference angles, time of movementand guessing time [39]. The computer controlled move-ment helped to keep the testing environment very consist-ent between trials and testing sessions, as long as thesubject was attached to the robot appropriately by the ex-perimenter. Although using a robotic device to test pro-prioception should be very reliable in theory, error scoresmay be inconsistent between testing sessions because ofintra-subject variability in individuals with spinal cordinjury, especially in those that have poor proprioceptivesense. As evidenced by the presence of heteroscedasti-city in persons with SCI, the difference in errors be-tween Day 1 and Day 2 of testing tended to be largerwhen the average error was larger (Figure 4B). This ideais also illustrated in Figure 3A & B, showing that sub-jects with SCI who had smaller angle errors tended tohave more similar scores between Day 1 and Day 2.Limitations of robotic assessmentOne limitation to using the Lokomat for assessment ofproprioception of the legs is that it unable to measure pro-prioceptive sense at the ankle. Ankle proprioception hasjoint. Individuals with cognitive impairments (e.g., due totraumatic brain injury) may not be able to participate inthis assessment due to the attention and memory require-ments of the task.ConclusionsThis work provides evidence that the Lokomat, when usedwith custom software, can provide a reliable and validmethod for quantifying joint position sense. This will bean essential tool when helping to understand the role ofproprioception in the recovery of functional tasks such asstanding and skilled walking function. Because we foundthat there were no differences in errors based on the an-gles tested and good internal consistency between testitems, it would be sensible to use only one combination oftarget and distractor angles for each joint in future proto-cols. Future studies should also quantitatively measuremovement sense. Ultimately these assessments will helpaid in the development of therapeutic interventions toimprove proprioceptive sense in people with neurologicalinjury, helping to maximize safe participation and qualityof life due to improved mobility.Competing interestsThe authors declare that they have no competing interests.Authors' contributionsAll authors contributed to the concept and project design. AD and TLprovided writing of the manuscript. AD conducted the data collection andanalysis. TL provided the facility and equipment. TL and AD completed thegrant application to fund the project. All authors read and approved the finalmanuscript.AcknowledgementsThe authors would like to thank Remco Benthem de Grave and LarsLünenburger for their technical assistance in the initial phases of this work,and Laurent Mingo and Jeswin Jeyasurya for their ongoing technical supportduring this study. In addition, we are grateful to the participants, Taha Qaiser,and Emily Kwee for their contributions to this study. This work wassupported by a grant from the International Foundation for Research inParaplegia. TL was supported by a Canadian Institutes of Health ResearchNew Investigator Award.Received: 7 October 2014 Accepted: 12 December 2014Published: 16 December 2014References1. McKinley WO, Jackson AB, Cardenas DD, DeVivo MJ: Long-term medicalcomplications after traumatic spinal cord injury: a regional modelsystems analysis. Arch Phys Med Rehabil 1999, 80:1402–1410.2. Groah SL, Stiens SA, Gittler MS, Kirshblum SC, McKinley WO: Spinal cordinjury medicine. 5. Preserving wellness and independence of the agingpatient with spinal cord injury: a primary care approach for therehabilitation medicine specialist. Arch Phys Med Rehabil 2002, 83:S82–S89.3. Ditunno JF Jr, Burns AS, Marino RJ: Neurological and functional capacityoutcome measures: essential to spinal cord injury clinical trials.J Rehabil Res Dev 2005, 42:35–41.4. Barbeau H: Locomotor training in neurorehabilitation: emergingrehabilitation concepts. Neurorehabil Neural Repair 2003, 17:3–11.5. Dietz V, Harkema SJ: Locomotor activity in spinal cord-injured persons.J Appl Physiol 2004, 96:1954–1960.6. Dobkin B, Apple D, Barbeau H, Basso M, Behrman A, Deforge D, Ditunno J,Dudley G, Elashoff R, Fugate L, Harkema S, Saulino M, Scott M: SpinalCord Injury Locomotor Trial Group: Weight-supported treadmill vsover-ground training for walking after acute incomplete SCI.Neurology 2006, 66:484–493.control study. J Neuroeng Rehabil 2013, 10:57.42. Goble DJ, Coxon JP, Wenderoth N, Van Impe A, Swinnen SP: Proprioceptivesensibility in the elderly: degeneration, functional consequences andplastic-adaptive processes. Neurosci Biobehav Rev 2009, 33:271–278.43. Fugl-Meyer AR, Jaasko L, Leyman I, Olsson S, Steglind S: The post-strokehemiplegic patient. 1. a method for evaluation of physical performance.Scand J Rehabil Med 1975, 7:13–31.44. Stillman BC: Making sense of proprioception: the meaning ofproprioception, kinaesthesia and related terms. Physiotherapy 2002,88:667–646.45. Fitzpatrick R, McCloskey DI: Proprioceptive, visual and vestibularthresholds for the perception of sway during standing in humans.J Physiol 1994, 478(Pt 1):173–186.46. Blanchette A, Lambert S, Richards CL, Bouyer LJ: Walking while resisting aperturbation: effects on ankle dorsiflexor activation during swing andpotential for rehabilitation. Gait Posture 2011, 34:358–363.doi:10.1186/1743-0003-11-167Cite this article as: Domingo and Lam: Reliability and validity of usingDomingo and Lam Journal of NeuroEngineering and Rehabilitation 2014, 11:167 Page 10 of 10http://www.jneuroengrehab.com/content/11/1/1677. Dietz V, Harkema SJ: Locomotor activity in spinal cord-injured persons.J Appl Physiol (1985) 2004, 96:1954–1960.8. Sherrington C: On the proprioceptive system, especially in its reflexaspect. Brain 1907, 29:467–482.9. Lajoie Y, Teasdale N, Cole JD, Burnett M, Bard C, Fleury M, Forget R, PaillardJ, Lamarre Y: Gait of a deafferented subject without large myelinatedsensory fibers below the neck. Neurology 1996, 47:109–115.10. Schaumburg H, Kaplan J, Windebank A, Vick N, Rasmus S, Pleasure D, BrownMJ: Sensory neuropathy from pyridoxine abuse. A new megavitaminsyndrome. N Engl J Med 1983, 309:445–448.11. Liu MW, Hsu WC, Lu TW, Chen HL, Liu HC: Patients with type II diabetesmellitus display reduced toe-obstacle clearance with altered gaitpatterns during obstacle-crossing. Gait Posture 2010, 31:93–99.12. Garraway WM, Akhtar AJ, Gore SM, Prescott RJ, Smith RG: Observervariation in the clinical assessment of stroke. Age Ageing 1976, 5:233–240.13. Gilman S: Joint position sense and vibration sense: anatomical organisationand assessment. J Neurol Neurosurg Psychiatry 2002, 73:473–477.14. Piriyaprasarth P, Morris ME, Delany C, Winter A, Finch S: Trials needed toassess knee proprioception following stroke. Physiother Res Int 2009, 14:6–16.15. van Beers RJ, Sittig AC, Denier van der Gon JJ: The precision ofproprioceptive position sense. Exp Brain Res 1998, 122:367–377.16. Wilson ET, Wong J, Gribble PL: Mapping proprioception across a 2Dhorizontal workspace. PLoS One 2010, 5:e11851.17. Gandevia SC, McCloskey DI: Joint sense, muscle sense, and theircombination as position sense, measured at the distal interphalangealjoint of the middle finger. J Physiol 1976, 260:387–407.18. Goble DJ, Brown SH: Upper limb asymmetries in the matching ofproprioceptive versus visual targets. J Neurophysiol 2008, 99:3063–3074.19. Fuentes CT, Bastian AJ: Where is your arm? Variations in proprioceptionacross space and tasks. J Neurophysiol 2010, 103:164–171.20. Carey LM, Oke LE, Matyas TA: Impaired limb position sense after stroke: aquantitative test for clinical use. Arch Phys Med Rehabil 1996, 77:1271–1278.21. Dukelow SP, Herter TM, Moore KD, Demers MJ, Glasgow JI, Bagg SD,Norman KE, Scott SH: Quantitative assessment of limb position sensefollowing stroke. Neurorehabil Neural Repair 2010, 24:178–187.22. Leibowitz N, Levy N, Weingarten S, Grinberg Y, Karniel A, Sacher Y, Serfaty C,Soroker N: Automated measurement of proprioception following stroke.Disabil Rehabil 2008, 30:1829–1836.23. Clark FJ, Burgess RC, Chapin JW, Lipscomb WT: Role of intramuscular receptorsin the awareness of limb position. J Neurophysiol 1985, 54:1529–1540.24. Cammarata ML, Dhaher YY: Proprioceptive acuity in the frontal andsagittal planes of the knee: a preliminary study. Eur J Appl Physiol 2011,111:1313–132025. Refshauge KM, Chan R, Taylor JL, McCloskey DI: Detection of movementsimposed on human hip, knee, ankle and toe joints. J Physiol 1995,488(Pt 1):231–241.26. Browne K, Lee J, Ring PA: The sensation of passive movement at themetatarso-phalangeal joint of the great toe in man. man. J Physiol 1954,126:448–458.27. Corrigan JP, Cashman WF, Brady MP: Proprioception in the cruciatedeficient knee. J Bone Joint Surg (Br) 1992, 74:247–250.28. Grigg P, Finerman GA, Riley LH: Joint-position sense after total hipreplacement. J Bone Joint Surg Am 1973, 55:1016–1025.29. Karanjia PN, Ferguson JH: Passive joint position sense after total hipreplacement surgery. Ann Neurol 1983, 13:654–657.30. Barrack RL, Skinner HB, Cook SD, Haddad RJ Jr: Effect of articular diseaseand total knee arthroplasty on knee joint-position sense. J Neurophysiol1983, 50:684–687.31. Shrout PE, Fleiss JL: Intraclass correlations: uses in assessing raterreliability. Psychol Bull 1979, 86:420–428.32. Bland JM, Altman DG: Statistical methods for assessing agreementbetween two methods of clinical measurement. Lancet 1986, 1:307–310.33. Rosner B: Fundamentals of biostatistics. Belmont, CA: Duxbury Press; 2005.34. Wall PD, Noordenbos W: Sensory functions which remain in man aftercomplete transection of dorsal columns. Brain 1977, 100:641–653.35. Afifi A, Bergman R: Functional neuroanatomy: text and atlas. New York, NY:McGraw-Hill; 2005.36. Johnson EO, Babis GC, Soultanis KC, Soucacos PN: Functionalneuroanatomy of proprioception. J Surg Orthop Adv 2008, 17:159–164.the Lokomat to assess lower limb joint position sense in people withincomplete spinal cord injury. Journal of NeuroEngineering andRehabilitation 2014 11:167.Submit your next manuscript to BioMed Centraland take full advantage of: • Convenient online submission• Thorough peer review• No space constraints or color figure charges• Immediate publication on acceptance• Inclusion in PubMed, CAS, Scopus and Google Scholar• Research which is freely available for redistribution37. Dukelow SP, Herter TM, Bagg SD, Scott SH: The independence of deficitsin position sense and visually guided reaching following stroke.J Neuroeng Rehabil 2012, 9:72.38. Kaplan FS, Nixon JE, Reitz M, Rindfleish L, Tucker J: Age-related changes inproprioception and sensation of joint position. Acta Orthop Scand 1985,56:72–74.39. Goble DJ: Proprioceptive acuity assessment via joint position matching:from basic science to general practice. Phys Ther 2010, 90:1176–1184.40. Wingert JR, Welder C, Foo P: Age-related hip proprioception declines:effects on postural sway and dynamic balance. Arch Phys Med Rehabil2014, 95:253–261.41. Damiano DL, Wingert JR, Stanley CJ, Curatalo L: Contribution of hip jointproprioception to static and dynamic balance in cerebral palsy: a caseSubmit your manuscript at www.biomedcentral.com/submit


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items