Comparison of Upper-Extremity Balance Tasks and Force Platform ...

Comparison of Upper-Extremity Balance Tasks and Force Platform Testing in Persons With Hemiparesis Background and Purpose. The purpose of this study was to investigate

the relationship between clinically accessible functional balance tools and sophisticated force platform measures in a standing position. Subjects. Twenty persons who had hemiparesis secondary to a stroke and were ambulatory (mean age=57.9 years, SD=13.6, range=35-79) were evaluated during a single testing session. Methods. Performances on self-generated upper-extremity balance tasks using the nonparetic side (Functional Reach Test [FRT], arm raise and arm reach tasks) were compared with responses to external perturbations on the Balance SystemTM(postural sway, symmetry of weight distribution). Results. No relationship was found between the upper-extremity balance tests and the force platform measures of postural sway. After suppressing the effect of age by means of partial correlation coefficients, the FRT was correlated with measures of postural symmetry in parallel stance on the Balance SystemTM(r=.66-.78). The FRT was only moderately correlated with the arm raise and arm reach tasks ( r =.43 and .44). Conclusion and Discussion. Postural sway in response to force platform perturbations may have little relation to the postural control necessary for self-generated upper-extremity balance tasks. In contrast, the FRT and the force platform measures of postural symmetry appear to be evaluating comparable standing-balance abilities in persons with hemiparesis. The modest relationship between the FRT and the arm raise and arm reach tasks limits the finding's clinical relevance. [Fishman MN, Colby LA, Sachs LA, Nichols DS. Comparison of upper-extremity balance tasks and force platform testing in persons with hemiparesis. Phys Ther. 1997;77:1052-1062.1

Key Words: Balance, Force platform, Function', Hemiparesis, Afeasurement. Marie N Fishman Lynn A Colby

Larry A Sachs Debmah S Nichols Physical Therapy. Volume 7 7 . Number 10 . October 1997

he manifestations of stroke are variable and can consist of motor deficits, sensory disturbances, perceptual impairments, functional limitations, and balance difficulties.' Normal standing balance has been defined as the ability to maintain the body's center of gravity over its base of support with limited postural Both anticipatory and compensatory postural adjustments are required for normal standing balance.%nticipatory postural movements that occur prior to self-generated perturbations of balance or in preparation for external perturbations of balance are controlled by feedforward mechanism^.^,^^ Compensatory adjustments t h a ~accompany the disturbances of equilibrium rely on visual, somatosensory, and vestibular feedback me~hanisms.l,~-7 In addition, maintenance of an upright posture is dependent on coordinated motor Communication between sensory system inputs and motor system outputs is believed to be required for normal standing b a l a n ~ e . ~ In an attempt LO measure standing balance, a large number of evaluation tools have been developed. Tools that assess an individual's ability to maintain balance during self-generated or external perturbations are

thought by some researchers9,10 to be more useful because control of the body during movement is the essence of practical standing balance. The quantification of standing balance in persons with stroke has taken several forms, including measuring performances on self-generated balance tasks".l0 and on force platform units capable of producing balance-threatening external perturbations.' The Functional Reach TestVFRT) and the arrn raise and arm reach tasks described by Goldie et all0 are examples of upper-extremity balance tasks involving self-generated perturbations of movement using the nonparetic limb. These upper-extremity tasks are based on current concepts of balance control. Numerous studies involving individuals without balance impairments have demonstrated that anticipatory postural adjustments occur prior to balance-disturbing vol~intary upper-extremity movements during standing.'?-(This anticipatory postural activity is hypothesized to keep the body's center of gravity within its base of support by counterbalancing upcoming perturbations generated by or the backvoluntary arm m o v e m e n t ~ . " ~ ~ - ' Fexample, ward and downward acceleration of the body's center of

MY F i s h r n ; ~ PT, ~ ~ . was a graduate st~tdent,Physical Therapy Division, School of Allied Medical Professions, T h e Ohio State Universitv, when this s t ~ ~ dwas y completed in pal-tial f~~lfillnicnt of the reqnirements for her h,Iaster of Sciencc degree.

1-1C:olby. PT, is .Ishistant Professor Enieritus, Physical Therapy Division, School of Allied Medical Professions, T h e Ohio State University. W Sachh, Phl), is Associate Professol. Emeritus, The Ohio State University. H e was .Issociate Director and Graduate Sti~dies(:hail-person, School of Allied Medical Protrssions, T h e Ohio Sratc University, when this study was completed. DS Nichols, PhD, PT, is Director and Associate Professor, Phvsical Therapy Division, School of Allied Mediral PI-ofessions. The Ohio Scate Uni~el-si~:y, 1583 Perry St, ( ; o l ~ ~ ~ n bOH u s . 43210 (LISA) (Nichols.SOposthox.acs.ohio-state-edu). Address all correspondence to Dr Nichols. This study was approved by T h e Ohio State University Biomedical Sciences H ~ u u a nSuhjects Review Committee Kcsults of thih study wel-e pl-esented at Physical Therapy '96: the AIII~I-icanPhysical Therapy Association's Scientific Meeting and Exposition; .Min~reapolis,Minn; June 14-18, 1996. 7'lilr

nrhrb roa5 J uhmztted Junp 3, 1996, ond ur(Ls (~((pptpd .%fin125, 1997

Physical Therapy . Volume 77 . Number 10 . October 1997

Fishman et a1 . 1053

gravity during forward flexion of the arm in standing is compensated for by anticipatory postural adjustments that precede the prime niover by 40 to 60 milliseconds and that produce an upward and forward a~celeration.~'~~2~ Horak et alw found that subjects with hemiparesis were unable to raise their nonparetic arm as rapidly as subjects without hemiparesis because of the greater latency of anticipatory postural activity on the contralateral hemiparetic side of the body relative to deltoid prime mover activity. Individuals with hemiparesis lacked vital postural stabilization on the paretic side." Voluntary upper-extremity movements in standing, therefore, appear to require normal anticipatory, as well as compensatory, postural adjustments." The FRT attempts to measure a person's ability to adapt to a single selfgenerated arm movement and to adjust to a self-imposed forward displacement of the center of mass.Vn contrast, the arm raise and arm reach tasks described by Goldie et all0 assess the ability to perform repeated self-paced movements using the nonparetic upper extremity during timed trials. Many commercial units are available for the evaluation of' standing balance. The Balance systemTM*is a force platform used to measure balance in response to externally imposed perturbations. According to the manufacturer," this device can be used to measure a person's center of balance, weight distribution, and postural sway in response to linear or angular platform displacements during a timed period. Many investigators have examined postural responses associated with force platform perturbations in subjects without known neurological impairment':'-" and in subjects with herniparesi~.H.'"'~ There is, however, still a lack of a documented relationship between self-generated upper-extremity balance tasks and force platform measures that depend on responses to external perturbations.

The objective of our study was to determine whether, in individuals with hemiparesis, performances of self-generated upper-extremity balance tasks are correlated with responses to external perturbations o n a force platform. Upper-extremity balance tasks are inexpensive, practical, and clinically accessible test^.^,^^) In contrast, the Balance SystemTMis a more complex and quantitative balance tool, which is not readily available in all clinical settings. In the event that a strong correlation is found, performance on one test may be used to reliably predict performance o n the other test. Because of the large number of standing balance tests available to physical therapists, we contend that relationships among these various balance tools must be investigated. Similar postural strategies are believed to be used for the maintenance of balance in expectation of self-generated upper-extremity movements and in response to externally imposed perturbation^."^"^ Better balance is associated with a lower amount of postural sway (improved stability),'.'H."-w greater postural ~ y m m e t r y , ' ~ ' ~and ~~~~" improved functional skill^.^^-^^ We therefore hypothesized that (1) a negative correlation would exist between self-generated upper-extremity balance tasks and postural sway in response to external perturbations on the Balance SystemTM,(2) a positive correlation would exist between self-generated upper-extremity balance tasks and a measure of symmetry of weight distribution in response to external perturbations o n the Balance SysternTh',and (3) a positive correlation would exist among various self-generated upper-extremity balance tasks.

Method Subjects

A sample of conveniencc, consisting of 20 subjects with

hemiparesis, was recruited from The Ohio State University Hospitals (Columbus, Ohio). Criteria for inclusion into the study were that the subjects (1) be within 12 months of their first and only unilateral cerebrovascular The influence of age on balance is well des~ribed."'~-:'~ accident, (2) be able to follow simple commands, Many r e s e a r c h e r ~ ' ~ ~ ~ ~ l -reported : ~ ~ a v e age-associated (3) report n o orthopedic, vestibular, or additional neuincreases in postural sway. In addition, slower arm rologic disease, (4) be able to stand for at least 60 n~ovements:~-',~%nd age-related declines in anticipatory seconds without assistance, ( 5 ) report n o diplopia or and reaction-time arm postural activity for ~elf-paced:~l visual-field defects, (6) have normal passive range of movementsm have been observed in elderly individuals motion of the nonparetic upper extremity and both without known neurological impairment. The FRT has lower extremities, as examined by the researcher, and also been shown to be age sensitive"'!1," as well as (7) be pain free in areas that may limit their ability to positively influenced by height.",30 We believe that reach with the nonparetic arm, stand, and balance. T o evaluate these criteria, each subject completed a quesresearchers investigating the relationships between such tionnaire and a single researcher physically examined measures as postural sway, functional reach, and each participant. The subjects ranged in age from 35 to balance-disturbing arm movements should consider the 79 years (R=57.9, SD=13.6), and all subjects were within influences of age and height. 28 weeks of their cerebrovascular accident at the time of testing (X=8.8, SD=6.8) (Tab. 1 ) . Nine subjects ambulated without the use of an assistive device, and 11 Chattanooga

C:r.ollr~ Inc,

1054 . Fishman et al

4717 Xdnrns Kd, PO Box 489, Hixsori, T N 57143.

Physical Therapy . Volume 77 . Number

10 . October 1997

Table 1. Subject Demographics No. of Subjects Hemiparesis Left Right

12 8

Gender Male Female

12 8

Age (Y) Height (m) No. of weeks after stroke

-

X

SD

Range

57.9 1.71

13.6 0.09

35-79 1.52-1.85

8.8

6.8

2-28

subjects required the use of a straight cane, quad cane, or walk.er. In this latter group, 5 subjects required the additional support of an ankle-foot orthosis (AFO) or an elastic wrap during amhulation to maintain ankle dorsiflexion on the paretic side. In accordance with university policy and to protect the rights of the subjects, informed consent was received from all subjects prior to participation in this study. lnstrurnentation The FFtT, developed by Duncan et al,%as one of the self-generated upper-extremity balance tasks used in this study. 'The subject assumes a comfortable stance and stands perpendicular to a wall where a yardstick is horizontally affixed at the level of the subject's acromion. With a fisted hand, the subject extends an arm along the yardstick. The position of the third metacarpal is recorded prior to and after a maximum forward reach along the yardstick. Scoring consists of the mean difference (in centimeters) between the two measurements for three successful trials of reaching9 This balance measure yields measurements that are both reliable (K Arnis, MTLockridge, unpublished research, 1995)"nd valid,q:!q,4" and it has been used extensively with various patient populations.47-5' In a sample of 16 persons with cerebrovascular accidents between the ages of 37 and 87 years, Amis and Lockridge (unpublished research, 1995) found same-day functional reach measures to have intrarater reliability with an intraclass correlation coefficient (ICC[l,l]) value of .92.

The arm raise and arm reach tasks described by Goldie et all0 require the use of a stopwatch, a tape measure, masking tape, and an adjustable standard walker. The stopwatch is needed to time the trials. The tape measure is needed to form a floor grid, demarcated by masking tape, for the alignment of both feet in the stance positions. In addition, the masking tape is used as targets o n a rail, wall, or standard walker. These self-generated upper-extremity balance tasks consist of the maximum Physical Therapy . Volume 7 7 . Number 10 . October 1997

number of arm raises o r arm reaches on the nonparetic side during 60-second periods. Each task is performed twice in both parallel and step stance positions. Scoring consists of the mean frequency of response of both trials for each stance position.10 These balance tests also yield reliablelO,wand valid" measurements of postural control in standing in individuals with hemiparesis. T h e force platform testing was performed using the Balance SystemTM. This device consists of a platform that is supposed to be capable of angular rotation and linear translation. Four independent electronic pressure transducers located in two footplates are used for measurement, and an adjustable safety harness and adaptable grab bars provide safety for this computer-driven device." The Balance SystemTM,according to its manufacturer,ll is capable of computing a person's center of balance, percentage of weight distribution, and amount (expressed as a percentage) of postural sway (sway index). The center of balance, or the location of the person's center of gravity in relation to the supporting area, represents the percentage of change in bodyweight distribution away from the geometric center of balance. This information is presented o n paper as x and y coordinates (COBx a n d COBy), and it is computed from the vertical distribution of force over the footplates. The system is also supposed to be able to measure the percentage of weight o n the toes and the heel of the person's left foot or right These data can be used to determine the symmetry of weight distribution for the duration of each trial. The sway index refers to the amount of postural sway relative to the mean center of balance during a timed trial. Postural sway is computed using the standard deviations of the coordinates during the timed trial, and it is calculated by the mean square method. A high sway index is supposed to be related to a large amount of sway and diminished balance, whereas a low sway index is said to occur when there is little postural sway and better balance,I1 but evidence for these claims has not been published in peer-reviewed form. At 100% of its capacity, the Balance SystemTMis supposed to be able to produce linear translations consisting of forward and backward displacements of a 1.9-cm (%-in) distance from the center position that occur at a speed of 2.54 cm (1 in)/0.8 s.I1 In contrast, the angular d cause displacements of 4 rotations are s ~ ~ p p o s eto degrees in the toes-up and toes-down positions at a speed of 2"/s. All descriptions of the unit's performance characteristics are based o n the manufacturer's information," as these performance characteristics were not verified in our study. The reliability and validity of measurements obtained with the Balance SystemTMhave been discussed p r e v i o ~ i s l y . ~ " - ~ ~

Fishman et al . 1055

There is moderate to high reliability for measurements of postural sway in individuals with hemiparesis in standing on stable and moving platforms (ICC[l,l]=.58 and .74, r e s p e ~ t i v e l y ) Similarly, .~ Levine et aP4 reported ICCs (3,l) of .75, .65, and .80, respectively, for postural sway for a static platform, linear translation, and angular rotation conditions in subjects with hemiplegia standing o n the Balance SystemTM.In addition, the reliability of measurements of the percentage of change in bodyweight distribution in a sitting position on this unit was very stroilg for subjects with no known balance impairment (ICC[2,1]=.86-.96) and moderate to very high for subjects with hemiparesis for stable and leaning conditions (ICC[2,1]=.30-.75 and .53-.95, respectively) .'" Although the Balance SystemTMis only poorly to moderately correlated with the Clinical Test of Sensory Interaction in Balance (CTSIB)" and the Functional Independence Measure (FIM)," it demonstrates greater discriminative validity than the CTSIB for postural sway in a standing position when comparing performances of subjects with no known balance impairment and subjects with h e m i p a r e s i ~ . ~ T hBalance e SystemTMhas been described elsewhere in great

Testing All measurements were obtained by one researcher during a single session. The tests were administered in random order. Testing was conducted in a quiet and nondistracting environment. For the purpose of our study, functional reach testing was performed with each subject in a standardized parallel stance position (feet 10 cm apart). This modification was made to ensure consistency of stance width with the tasks described by Goldie et al.1° The subject stood perpendicular to a wall where a yardstick was horizontally adjusted at the level of the acromioil on the nonparetic side." Good postural alignment, without excessive shoulder protraction or retraction, was ensured prior to each reach. The researcher observed the subject's scapula with the subject in a standing position and repositioned it into proper alignment if necessary. The subject was instructed to "reach as far foiward (as possible) without losing ... balance o r taking a step" and without touching the wall." The trial was otherwise thrown out and repeated." The reaching technique was not controlled, and, to ensure safety, the subject was supervised by the researcher.!) There were no practice trials. The mean filnctional reach measurement was calculated for three successful trials." Meas~rrementsfor the arm raise and arm reach tasks were obtained using the methods described by Goldie et al.1° Each task was performed twice, first in the parallel stance position (feet 10 cm apart) and then in the step stance position (paretic leg foward, heel to toe,

1056 . Fishmon et al

Figure. Testing protocol for the arm reach tosk. (Adopted and reprinted with permission of the American Physical Therapy Association from Goldie et a1.'0)

feet 10 cm apart). The subject stood perpendicular to a wall, with the medial border of the inner nonparetic foot 25 cm from the wall. A standardized floor grid, demarcated by masking tape, was used for all subjects. The researcher counted aloud during the 60-second trials and monitored the subject for safety. The paretic foot was stabilized at the heel by the researcher's own foot during step stance testing. In the event that the stance position was not maintained, the trial was eliminated and repeated immediately.1° There were no practice trials. Targets made of masking tape delineated the starting and end positions of each movement used in the tasks described by Goldie et al.1° During the arm raise task, the subject lifted and lowered the nonparetic arm between two targets on the wall as frequently as possible during 60 seconds. The lower target was identified by asking the subject to raise the arm forward, without leaning at the trunk, to a point on the wall at hip level. The upper target was placed on the wall at the subject's shoulder level directly above the lower target. The mean frequency of response for both trials was calculated for each stance position. The arm reach task required the use of the testing arrangement displayed in the Figure. The subject touched a side target on the wall and reached for a front target on a standard walker with the nonparetic upper extremity as many times as possible during a timed 60-second period. The subject was instructed to visually follow the reaching maneuver during testing. The mean frequency of response for both trials was also computed for each stance position for the arm reach task. l o

Physical Therapy

. Volume 77 . Number 10 . October

1997

Table 2. Means and Standard Deviations for the Upper-Extremity Tasks [N=20)

FRTa (cml

Arm Raise (repetitionslmin) PSb

ssc

Arm Reach (repetitionslmin)

ss

PS

I

FRT=Functional Reach Test." "PS=parallel stance: feet 10 cm apart. 'SS=step stance: paretic leg fowal-d. heel to toe, feet 10 cm apart "

Each si~bjectalso underwent testing with eyes open on the Balance SystemTM.The unit was calibrated prior to each testing session. The same standardized grid was reproduced on the system's platform, and the subject was tested first in the parallel stance position and then in the step stance position. The subject was tested under three different conditions for each stance position: stable platform, linear translation, and angular perturbation. To become familiar with the Balance SystemTM, the subject was always tested first on the stable condition. This condition consisted of standing on the stationary platform for the timed duration. Randomization of the perturbations was conducted to prevent any potential effects of order. Testing was carried out at 50% of the unit's capacity because full-capacity testing was thought to be too challenging for our subjects. The linear translations, therefore, involved forward and backward displacements of a 1.9-cm (%-in) distance from the center position, which occurred at a speed of 1.27 cm ( % in)/O.8 s.ll Angular rotations consisted of displacements of 4 degrees in the toes-up and toes-down positions, which occurred at a speed of l"/s.l1 Prior to each trial, the subject was warned of the upcoming condition yet was unaware of the initial platform direction (eg, forward versus backward). The subject was told that the platform would move forward and backward or tilt up and down. A single trial lasting 10 seconds was performed for each condition. In the event that the subject was unable to complete the first trial, a repeat trial was performed. The subject was eliminated from the study, however, if the second attempt was unsuccessful. The Balance SystemTMcomputed the subject's percentage of weight distribution on the paretic side and amount of postur;al sway (expressed as a percentage) for the duration of each testing condition.ll Data Analysis

Pearson product-moment correlation coefficients were used to investigate the relationships between the numerous vai~iables(maximum functional reach, frequency of arm raise and arm reach response, postural sway, and symmetry of weight distribution). Scores for symmetry of weight distribution were calculated by taking the absolute values of the percentage of weight on the paretic side as computed by the Balance SystemTMminus 50%. Physical Therapy . Volume 7 7 . Number 1 0 . October 1 9 9 7

We considered these absolute scores to be indicative of postural symmetry and not representative of whether too much or too little weight was shifted onto the paretic limb for various platform conditions. The lower the score, the more symmetrically the subject stood on the Balance SystemTM.Height and age were taken into account as possible confounding variables using partial correlation coefficients. The level of significance was set a priori at an alpha level of .05. Correlations were deemed to be low if below .40, moderate if between .40 and .65, and strong if above .65.

Results A total of 21 subjects agreed to participate in this study. Twenty subjects were able to successfully perform the tests in the study's protocol. Only one subject, a woman, had to be eliminated from the study because she was unable to complete a testing condition on the Balance SystemTMon her second attempt. Means and standard deviations for all variables obtained are shown in Tables 2 and 3. Relationship Between Upper-Extremity Tasks and Postural Sway

We first examined the relationship between the selfgenerated upper-extremity tasks and postural sway by condition on the Balance SystemTM. No relationship was found, and the correlations ranged from a negligible value of .OO to a low value of -.29. Height and age also were not correlated with postural sway for the six tested conditions on the Balance SystemTM. Relationship Between Upper-Extremity Tasks and Postural Symmetry

We next investigated the relationship between the selfgenerated upper-extremity tasks and postural symmetry by condition on the Balance SystemTM.The FRT was moderately correlated with measures of postural symmetry in the parallel stance position on the Balance SystemTMfor the stable platform ( r =.66, Ped a pel-centagr. ' Expressed as a prrcentage :rnd r.;~lc~~lated using tlrc [ormula: I.~bsolutevalu~.(L%, wcight on parclic si(ir-3(l%,)1

Table 4. Correlation Coefficients Between the Upper-Extremity Tasks and the Balance SystemTMby Condition for Symmetry of Weight Distribution (N=20)

Arm Raise FRTa PS Platform stable Angular rotation Linear translation

.66*" .49* .54"*

SS Platform stable Angular rotation Linear translation

.24 .05 .05

PSb .06 .02 .06

-.41d ~ . 4 2 ~ -.52"

FRT=F~unrtional Reach Test." Single asterisk (*) indicates P