Dynamic balance control in elders: Gait initiation assessment as a screening tool

490

Dynamic Balance Control in Elders: Gait Initiation as a Screening Tool Hwa-am

Chang, MS, PF David E. Krebs, PhD, PT

ABSTRACT. Chang H, Krebs DE. Dynamic balance control in elders: gait initiation assessment as a screening tool. Arch Phys Med Rehabil 1999;80:490-494. Objective: To determine whether measurements of center of gravity-center of pressure separation (CG-CP moment arm) during gait initiation can differentiate healthy from disabled subjects with sufficient specificity and sensitivity to be useful as a screening test for dynamic balance in elderly patients. Subjects: Three groups of elderly subjects (age, 74.97 -C 6.56 yrs): healthy elders (HE, n = 21), disabled elders (DE, rr = 20), and elders with vestibular hypofunction (VH, IZ = 18). Design: Cross-sectional, intact-groups research design. Peak CG-CP moment arm measures how far the subject will tolerate the whole-body CG to deviate from the ground reaction force’s CP; it represents dynamic balance control. Screening test cutoff points at 16 to 18cm peak CG-CP moment arm predicted group membership. Results: The magnitude of peak CG-CP moment arm was significantly greater in HE than in DE and VH subjects (p < .Ol) and was not different between the DE and VH groups. The peak CG-CP moment arm occurred at the end of single stance phase in all groups. As a screening test, the peak moment arm has greater than 50% sensitivity and specificity to discriminate the HE group from the DE and VH groups with peak CG-CP moment arm cutoff points between 16 and 18cm. Conclusions: Examining dynamic balance through the use of the CG-CP moment arm during single stance in gait initiation discriminates between nondisabled and disabled older persons and warrants further investigation as a potential tool to identify people with balance dysfunction. 0 1999 by the American Congress of Rehabilitation Medicine and the American Academy of Physical Medicine and Rehabilitation AIT INITIATION is prerequisite to locomotion, although it is unknown how sensitive an indicator it is to dynamic balance dysfunction. l-5 How elders initiate gait is important because elderly people may have decreased dynamic balance control and, thus, are at risk for falling.4,6 The gait initiation biomechanics literature for the geriatric population is scant and incomplete, however. Gait initiation is a dynamic balance control challenge, because it is the transitional phase between

G

From the Graduate Program of Postprofessional Physical Therapy at Massachusetts General Hospital Institute of Health Professions, Boston, MA. Submitted for publication June 22, 1998. Accepted in revised form October 29, 1998. Supported by the National Space Biomedical Research Institute, National Institute for Disability and Rehabilitation Research (H133G60045), and National Institutes of Health (ROlAG11255 and ROlAG12561). No commercial party having a direct financial interest in the results of the research supporting this article has or will confer a benefit upon the authors or upon any organization with which the authors are associated. Reprint requests to David E. Krebs, PhD, PT, Professor and Director, Massachusetts General Hospital Biomotion Laboratory, MGH Institutes of Health Professions, 101 Mmimac Street, Boston, MA02114-4719. 0 1999 by the American Congress of Rehabilitation Medicine and the American Academy of Physical Medicine and Rehabilitation 0003.9993/99/8005-5090$3.00/O

Arch

Phys

Assessment

Med

Rehabil

Vol80,

May

1999

static balance in upright position and the start of steady-state walking.1,3-5Understanding the control of balance during gait initiation should improve the ability of gerontologists to discriminate between stable and unstable patients. Center-of-gravity (CG) measurements reflect body position, whereas those for center of pressure (CP) reflect weight shifts and muscular control during dynamic posture changes.2 Because gait initiation begins with the separation of CP and CG,’ studying the interaction between the two could clarify how postural and intentional movement components are coordinated during locomotion. Separation of CP and CG, described as CG-CP moment arm, is proportional to the CG horizontal acceleration in an ideal inverted pendulum model of the body pivoting around the ankle. 1,4Thus, the CG-CP moment arm is the primary variable by which horizontal CC acceleration may be predicted. Moreover, the peak CG-CP moment arm during activities indicates the subject’s tolerance of dynamic unsteadiness2 Some articles on gait initiation have described the details of the CGKP trajectory in normal young subjects,1,3-5but quantitative measurement of the peak CG-CP moment arm during gait initiation as a dynamic balance indicator among elders has not been described. This study was undertaken to assessand compare dynamic balance control associated with gait initiation by analyzing CG-CP moment arm among three elderly groups: healthy elders (the HE group), disabled elders (the DE group), and elders with known balance disorders from vestibular hypofunction (the VH group.) We also sought to determine how much CG-CP moment arm would differentiate HE from DE and VH groups. We hypothesized that DE and VH would have significantly less dynamic balance control at gait initiation than HE and that CG-CP moment arm measurement might be an effective screening test. METHODS Subjects Fifty-nine subjects (21 HE, 20 DE, and 18 VH) participated in this study. They were volunteers in a research project and participated in various tests of balance. Inclusion criteria for healthy elders: age, ~65 yrs, cognitively intact, no functional limitation, no assistive device during functional activities, negative neurologic and orthopedic examination, and no medication that could influence balance. Inclusion criteria for disabled elders: age ~6.5 yrs, at least two functional limitations on the SF36 physical subscale (excluding the vigorous activity item),7 and cognitively intact. Inclusion criteria for elders with vestibular hypofunction: age 265 yrs, in good health except for unilateral or bilateral vestibular hypofunction documented by sinusoidal vertical axis rotation, electronystagmography and otoneurologist’s examination.8 All VH subjects had reduced vestibular function, not distorted function such as occurs in benign paroxysmal positional vertigo or Meniere’s disease. Because the VH group had requested rehabilitation for balance and gait instability, we defined them as “gold standard” balance impaired, whereas the balance status of the DE group was unknown.

BALANCE

100 1

0

DURING

ELDERS’

GAIT

INITIATION,

491

Chang

Procedure Subjects were barefooted and were asked to stand as usual, with one foot on each force platform. The subjects were then asked to start walking at a self-selected speed. Data collection began about 750 milliseconds prior to gait initiation and lasted 5 seconds. Subjects were allowed to practice two times and data from one trial were collected and tabulated.

5 L\___

80' F

I

Data Collection Superplot software developed at our institution, Massachusetts General Hospital, in the MGH Biomotion Laboratory was

_____ --Ii 2cl-

0 1

-2OL 0.0

.

/

I,

I

0.5 second

.O

Fig 1. Representative example of a DE subject’s vertical ground reaction force normalized as percentage of body weight (%BW) on the right leg (dashed line) and the left leg (solid line) during gait initiation, showing that the subject did not step out of the force platform in the first step. Phases: 0, the period before onset of gait initiation; 1, onset of gait initiation to heel-off of the swing leg; 2, heel-off to toe-off of the swing leg; 3, toe-off to initial contact of the swing leg (single stance phase); and 4, initial contact of the swing leg to toe-off of the stance leg (double stance phase).

6

1

0 I-

1 I

Exclusion criteria for the three groups were: uncontrolled hypertension or an exercise blood pressure response exceeding 210/l IOmmHg, a recent malignancy treated with active chemotherapy or radiation, significant coronary heart disease, chronic hepatitis, renal disease, diabetes, hyperthyroid or parathyroid dysfunction, congestive heart failure, significant cardiac arrhythmias, and rheumatoid arthritis, or history of central neurologic disease (such as stroke, dementia, and Parkinson disease). All subjects had a brief physical examination (eg, gait analysis), and had normal cognitive ability to follow multiple commands by the laboratory physicians or physical therapists. Instrumentation Four Selspot II optoelectric camerasaand two force platforms were used to collect data as described in detail elsewhere.2,sThe Selspot system detects active infrared light emitting diodes (LED) in rigid plastic arrays anchored to 11 body segments (head, trunk, upper arms, pelvis, thighs, shank, feet). Each body segment was modeled with 6 degrees of freedom (3 translations and 3 rotations). The kinematic data were determined using TRACKb software at a data-collection rate of 150Hz. The software determined the position of the LED arrays with the accuracy of < lmm and < 1o.9 Using this kinematic data acquisition system, Krebs and associatesto found high trial-totrial reliability (r 2 X8) for the upper-body angular kinematic measurement in healthy subjects. Kinetic data were determined using force platforms, which assessedanterior-posterior, vertical, and lateral ground reaction forces as well as the center of pressure.A software package (Newtonb) applied inverse Newtonian dynamics to the force plate data and kinematic data to determine net forces and torques on joints.

I

I

I

I

I

2

I

I

I

1

4

5

i

I

’

’

1 I

20

i

2

c

0

Fig 2. Representatives of examples of the CG-CP moment arm (cm) during gait initiation in (A) an HE subject and (B) a DE subject showing that the peak CG-CP moment arm occurred at the end of phase 3 (single stance phase). Phases 0, 1,2,3, and 4 are as defined for figure 1.

Arch

Phys

Med

Rehabil

Vol

80, May

1999

492

BALANCE

DURING

ELDERS

GAIT

INITIATION,

Chang

Fig 3. Whole-body schematic 3D representations of an HE subject (left), a DE subject (middle), and a VH subject (right) showing their differences in peak CG-CP moment arm (the lines under each left foot) at the end of singlestance phase before the first swing leg initial contact. Peak CG-CP moment arm (cm): HE, 26.757; DE, 15.839; and VH, 15.052.

used to display, analyze and calculate the anatomic segment positions and orientations to estimate body CG kinematics, ground reaction forces and center of pressure. Data Analysis Gait initiation was defined as the interval between the start of weight shifting toward the first swing leg to heel-off of the first stance leg (fig 1). The dynamic balance measure was the magnitude of CP-CG separation (CG-CP moment arm) (fig 2), as described previously 2,4during gait. The timing and magnitude of the peak CG-CP moment arm was used for data analysis. The CG-CP moment arm data were excluded after the first swing leg’s initial contact because some of the subjects stepped off the force platform in the first step and, therefore, the CP data were missing. However, according to the data from the subjects across all groups who did not step out of the force platform in the first step, the peak CG-CP moment arm still occurred before initial contact of the first swing leg (fig 3). Analysis of variance (ANOVA) was used to compare differences among groups, and post hoc testing at the .Ol level was used to determine where the difference lay. Descriptive statistics were used to characterize the subjects in each group. The association between the peak CG-CP moment arm and subject characteristics (age, height, weight, body mass index and steady-state gait speed) were tested using Pearson correlation analysis in each group. The value of Pearson correlation (r) was statistically significant whenp < .Ol.” The sensitivity and specificity of the peak CG-CP moment arm as an indicator of dynamic balance control (healthy or balance-impaired) were investigated by screening test with cross-tabulation (table 1). The cutoff points for classifying the peak CG-CP moment arm data were selected at each l-cm interval. The screening test validly discriminates healthy subjects from balance-impaired subjects whenever both the sensitivity and specificity were over 50%. RESULTS Table 2 describes the subject characteristics in each group. There were no differences in age, height, weight, and body mass index among the HE, DE, and VH groups. The steady-state gait Arch

Phys

Med

Rehabil

Vol 80, May

1999

speed was 30% greater in the HE group than the DE group (p = .OOl) and 24% greater in the HE group than the VH group (p = .007), but was not different between the DE and VH group (p = .57). For all subjects’ GI, the peak CG-CP moment arm ranged from 10.37 to 41.77cm (table 3), with a mean value of 17.75 ? 5.8cm. The mean and standard deviation of the peak CG-CP moment arm among groups were: HE, 21.49 5 7.53; DE, 15.42 5 2.87; and VH, 15.98 rt 3.45. The peak CG-CP moment arm was 40% greater in the HE group than the DE group (p = .OOl) and was 35% greater in the HE group than the VH group (p = .004), but was not different between the DE and VH groups (p = .94). Therefore, the DE subjects were as balanceimpaired as the “gold standard” balance-impaired VH subjects. The peak CG-CP moment arm in all the groups occurred at the end of single stance phase and just before initial contact of the first swing leg (fig 2). The correlations between the peak CG-CP moment arm and subject characteristics were poor in each group (table 4). Therefore, dynamic balance control as measured by the peak CG-CP moment arm during gait initiation, was not related to subject characteristics, with the exception of height (r = .64, p = .004) in VH subjects. Moment arm in gait initiation was significantly correlated with steady-state gait speed (Y = X62, p < .OOl)in VH subjects. To develop a dynamic balance screening test, the DE and VH groups were combined and compared with the HE subjects (table 5). The cutoff points for classifying the peak CG-CP Table

1: Typical

Screening

Test

Cross-Tabulation

Healthy

CG-CP

moment

arm

> X

True

positive

a CG-CP

moment

arm

5 X

False C

negative

Balance-Impaired

False b

positive

True cl

negative

The screening test presented here in a 2 X 2 matrix is classified according to the known balance impairment status, compared with the predictor variable, peak CG-CP moment arm value; X is the cut point value. Sensitivity = a/(a + b); specificity = d/(c + d).

BALANCE

DURING

ELDERS’

Table Group

2: Subject

Height(m)

Age (yr)

HE DE

73.730 78.083

I 1.312 2 1.695

1.667 1.602

VH pValue

74.132

2 1.350 ,135

1.672

Values reported as mean i standard deviation. Abbreviations: HE, healthy elders; DE, disabled average forward gait velocity. * pvalue shows significant difference among

t i

MA Range (cm) lO
s 14

14 < MA 15 < MA 16 < MA

5 15 5 16 5 17

17
subjects

Frequency Moment HE (4

105.573 81.174

k 21.552 i 18.108

ir ,022 ,023

72.364

2 2.742 ,359

25.833

i ,802 ,038

85.346

I 27.831 ,002"

elders

with

vestibular

hypofunction;

BMI,

body

mass

index;

Gait

Speed,

steady-state

HE, DE, and VH groups.

of Peak

VH (n)

CG-CP

DE + VH (n) 1

0 1

3 2

2 1

5 3

1 1 2

2 4 2

1 2 3

3 6 5

1 3

1 2 2

0 3 2

1 5

0 1

0 2

1 0

0 1

3 1 1

0 20

0 18

0 38

Abbreviations: MA, peak CG-CP moment subjects combined as the “balance-impaired”

Gait Speed (cm&c)

!I .667 t ,868

1

7 21

BMI (kg/m*) 24.859 27.058

0

0 1 1

(kg) + 2.714 2 2.291

0

3 0

Characteristics

69.455 69.306

elders;VH,

Distribution Arm Data DE In)

493

Chang

,021 ,017

DISCUSSION The CG-CP moment arm during gait initiation clearly discriminated unsteady elders from healthy elders. The results supported the hypothesis that DE and VH subjects have significantly less dynamic balance control at gait initiation than those in the HE group. The data indicate that the most challenging point in gait initiation for stability is the instant prior to the initial contact of the first swing leg. According to indirect data obtained from Burleigh and colleagues5 and Jian and associates,r the peak CG-CP moment in a normal young subject is about 23cm during gait initiation (n = 2). These results are close to our results from healthy elders (HE mean: 21.49cm). Therefore, our HE subjects were, as expected, “healthy” people and did not have more dynamic 3: Grouped

INITIATION,

Weight

moment arm data were selected at each l-cm interval according to the grouped frequency distribution shown in table 3. We found that the screening tests with cutoff points of the peak CG-CP moment arm at the levels of 16, 17, and 18cm were about equally valid, because the sensitivity and specificity in these screen tests were all greater than 50% (table 5). Thus, there is more than a 50% chance that an elder is balanceimpaired (specificity) if his or her peak CG-CP moment arm is less than 16 to 18cm and more than a 50% chance of an elder being healthy (sensitivity) if his or her peak CG-CP moment arm is greater than 16 to 18cm. When the cutoff point decreased from 18cm to 16cm (table 5), sensitivity increased from 57% to 76%, whereas the specificity decreased from 80% to 65% in the HE versus DE comparison and from 72% to 56% in the HE versus VH comparison. Specificity was greater in the HE versus DE comparison than in the HE versus VH comparison.

Table

GAIT

arm;

DE + VH, group.

Table 4: Pearson With Age, Height,

4 0

DE and

balance impairments than healthy young subjects. With similar subject characteristics (age, weight, height, and body mass index) across groups (table 2), we found that the peak CG-CP moment arm in the HE group was 40% greater than the DE group and was 34% greater than the VH group during gait initiation. The DE and VH subjects did not tolerate as much CG-CP separation as the HE subjects did; thus, DE and VH subjects have poorer dynamic balance than HE subjects. As a screening tool, the cutoff points of the gait initiation peak CG-CP moment at 16, 17, or 18cm differentiated stable elders (HE) from unstable elders (DE and VH) (table 5). Moreover, we found that the test specificity for determining the DE subjects as “balance impaired” was higher than that for determining the VH subjects. The difference may be due to the lower variability among the DE subjects (GI peak CG-CP moment arm standard devitations: DE, 2.87cm; VH, 3.45cm). The difference in variability may have resulted from various degrees of vestibular insufficiency and idiosyncratic individual recovery rates.* The largest standard deviation (7.53) of the GI peak CG-CP moment arm was found in the HE group, a finding that may be explained by variations of motivation, personality, and each subject’s interpretation of “natural” walking.12-1s Dynamic balance requires the central nervous system to integrate multiple sensory and motor pathways so that the body can coordinate both postural and intentional movement components during locomotion. 8,12*13 Either singly or in combinations physiologic impairments in the sensory and motor systems may alter balance function.8,12%13 These impairments include insufficient visual acuity, vestibular function disabilities, and proprioceptive disorders, muscle weakness, and reduced general mobility. In addition, fear of falling may result in changes of balance contro1.6~8~12-14 We found that steady-state gait speed and peak CG-CP moment arm during gait initiation were both greater in the HE group than in the DE and VH groups. Decreased gait speed is one sign of balance impairment.6~8~13-i5 However, the correlation between the steady-state gait speed and the GI peak CG-CP moment arm was significant only in the VH group (table 4). This implies that decreased gait speed was a sign of balance impairment in the VH group. In the DE group, however, factors caused by balance impairment during gait initiation do not Moment Arm Gait Speed,

Group

Age

Height

Weight

BMI

HE

,297 ,259 p.154

-.247 ,395

-.293 -.270

~.I81 p.523

DE VH

VH

Correlation (r) of Peak CG-CP Weight, BMI, and Steady-State by Subject Group

,640"

,156

Phys

Med

,376 ,144

-.383

Abbreviations: HE, healthy elderly; DE, disabled with vestibular hypofunction; BMI, body mass steady-state average forward gait velocity, * p<.Ol.

Arch

Gait Speed

Rehabil

,862"

elderly; index;

VH, elderly Gait Speed,

Vol 80, May

1999

494

BALANCE

Table

5: Sensitivity

and Specificity of Peak at 16,17, and 18cm HE vs DE

MA = 18cm Sensitivity Specificity MA = 17cm Sensitivity Specificity MA = 16cm Sensitivity Specificity

HE vs VH

CG-CP

DURING Moment

ELDERS’ Arm

HE “s (DE + VHI

(%) (%)

57.14 80.00

57.14 72.22

57.14 76.32

(%) (%)

71.43 70.00

71.43 55.56

71.43 63.16

(%) (%)

76.19 65.00

76.19 55.56

76.19 60.53

Abbreviations: MA, peak CG-CP moment arm; HE, healthy disabled elders; VH, elders with vestibular hypofunction; subjects combined as the “balance-impaired” group.

elders; DE, DE + VH,

appear to contribute to a steady-state gait speed change. Also, the peak CG-CP moment arm was not related to the steady-state gait speed in the HE group. Therefore, we suggest that results of different variables from different tasks such as gait initiation versus steady-steady gait,2*s might reflect different levels of

dynamic balance control and neural adaptation and should be interpreted carefully. In an ideal inverted pendulum model of the body pivoting around the ankle, the CG-CP moment arm is proportional

to the

CG height during gait initiation. 2,4,5A taller person has a higher CG. However, human movement is more complex than an inverted pendulum since there are multiple segments in the human body. This may explain why the CG-CP moment arm did not correlate with CG height in our HE and DE groups during gait initiation. In contrast, the VH subjects tend to maintain the verticality of the head and trunk to enhance gaze and postural stability during gait initiation to compensate their vestibular insufficiency. 8~13This rigid body segment motion is similar to the inverted pendulum of the body pivoting around the ankle, and may explain why we found the peak CG-CP moment arm correlated highly with body height only in the VH subjects (Y = .64,p = .004). Gait initiation is a functionally-related task.16Analyzing gait initiation is less time-consuming and energy-consuming than conventional tests and will benefit those who cannot tolerate walking more than few steps and those who fatigue during the experimental tests. Performing gait initiation also requires a relatively small area and, therefore, is less costly and more convenient for the laboratory. However, the clinical application of the techniques used in this study is difficult because of the complex procedures and expensive equipment required. Further refinement of CG-CP moment arm measurement may be necessarybefore gait initiation moment arm screening is widely adapted in clinics. In summary, peak CG-CP moment arm measurement during gait initiation is a valid tool to identify people with balance dysfunction and warrants further investigation in the elderly and in patient populations.

Arch

Phys

Med

Rehabil

Vol 80, May

1999

GAIT

INITIATION,

Chang

Acknowledgments: We thank Massachusetts General Hospital Biomotion Laboratory for facilities and conduct of this study. The helpful advice of Professor Patricia Sullivan, MGH Institute of Health Professions, and English Editor Ann Butman are gratefully acknowledged. References 1. Jian Y, Winter DA, Ishac MG, Gilchrist L. Trajectory of the body COG and COP during initiation and termination of gait. Gait Posture 1992;1:9-22. 2. Riley PO, Mann RW, Hodge WA. Modelling of the biomechanics of posture and balance. J Biomech 1990;23:503-6. 3. Mann RA, Hagy JL, White V, Liddell D. The initiation of gait. J Bone Joint Surg 1979;61A:232-9. 4. Winter DA. A.B.C. (anatomy, biomechanics and control) of balance during standing and walking. Waterloo, Canada: Waterloo Biomechanics; 1995. 5. Burleigh AL, Horak FB, Malouin F. Modification of postural responses and step initiation: evidence for goal-directed postural reactions. J Neuronhvsiol 1994:72:2892-902. 6. Gehlsen GM, Whale; MH. Falls in the elderly: part I, gait. Part II, balance, strength, and flexibility. Arch Phys Med Rehabil 1990;71: 735-69. I. Ware JE, Snow KK, Kosinski M, Gandek B. SF 36 health survey: manual and interpretation guide. Boston: the Health Institute, New England Medical Center; 1993. 8. Krebs DE, Gill-body KM, Riley PO, Parker SW. Double-blind, placebo-controlled trial of rehabilitation for bilateral vestibular hypofunction: preliminary report. Otolaryngol Head Neck Surg 1991;109:735-40.

9. Antonsson DK, Mann RW. Automatic 6-DOF kinematic trajectory acquisition and analysis. J Dynamic Systems Measurement Control 1989;111:31-9. 10. Krebs DE, Wong D, Jevesevar DS, Riley PO, Hodge WA. Trunk kinematics during locomotor activities. Phys Ther 1992;72: 505-14. 11. Fisher RA, Yates F. Statistical tables for biological, agricultural and medical research. Harlow, United Kingdom: Longman Group; 1974. 12. Chiam R, Laxman N, Bernard I. Balance function in elderly people who have and who have not fallen. Arch Phys Med Rehabil 1988;69:261-4. 13. Chandler JM, Duncan PW. Balance and falls in the elderly: issues in evaluation and treatment. In: Guccione AA, editor. Geriatric physical therapy. St. Louis (MO): Mosby-Year Book, Inc; 1993. p. 237-51. 14. Imms FJ, Edholm OG. Studies of gait and mobility in the elderly. AgeAging 1981;10:147-56. 15. Podsiadlo D, Richardson S. The timed “up & go”: a test of basic functional mobility for frail elderly persons. J Am Geriatr Sot 1991;39:142-8.

16. Berg KO, Maki BE, Williams JI, Holliday PJ, Wood-Damphinee SL. Clinical and laboratory measures of postural balance in an elderly population. Arch Phys Med Rehabil 1992;73:1073-80. Suppliers a. jelspot AB, Flojelbergsgatan 14, S-431 37 Molndal, jelspot System Ltd, Troy, MI 48093. b. tiassachusetts Institute of Technology, Cambridge, MA.

Sweden;