Intrasession reliability of center of pressure measures of postural steadiness in healthy elderly people1

896

Intrasession Reliability of Center of Pressure Measures of Postural Steadiness in Healthy Elderly People Danik Lafond, PhD, He´le`ne Corriveau, PT, PhD, Re´jean He´bert, MD, MPhil, Franc¸ois Prince, PhD ABSTRACT. Lafond D, Corriveau H, He´bert R, Prince F. Intrasession reliability of center of pressure measures of postural steadiness in healthy elderly people. Arch Phys Med Rehabil 2004;85:896-901. Objectives: To estimate the immediate test-retest reliability of a single measure of several center of pressure (COP) variables, to report the number of trials to be averaged to obtain a reliable measurement of postural steadiness, and to determine the minimal metrically detectable change (MMDC). Design: Cross-sectional study. Setting: University biomechanics laboratory. Participants: Seven community-living, healthy elderly people over the age of 60 years (range, 62–73y). Interventions: Not applicable. Main Outcome Measures: The COP was estimated from 2 force platforms and the following measures were calculated: (1) root mean square (RMS), (2) COP range, (3) COP mean velocity, (4) mean power frequency (MPF), (5) median power frequency (MedPF), and (6) sway area. Intraclass correlation coefficients (ICCs) were determined by using 9 successive quiet standing trials. Results: The ICCs obtained for 1 measure of 120 seconds were .58 and .58 for the RMS, .83 and .94 for the COP mean velocity, .52 and .62 for the COP range, .44 and .30 for the MPF, and .34 and .47 for the MedPF in anteroposterior (AP) and mediolateral (ML) directions, respectively. The ICC of the COP sway area obtained for 1 measure was .41. Only 2 trials had to be averaged to obtain an ICC over .90 for the COP mean velocity associated with an MMDC of ⫾1.2mm/s (AP) and ⫾0.6mm/s (ML). Conclusions: Mean velocity was the most reliable COP measure and using 2 repetitions allowed for reliable measurement of postural steadiness. For the other COP variables, 3 trials of 120 seconds were needed to obtain an ICC over .80. Key Words: Aged; Posture; Pressure; Rehabilitation; Reliability and validity. © 2004 by the American Congress of Rehabilitation Medicine and the American Academy of Physical Medicine and Rehabilitation HE INCIDENCE OF FALLS in the elderly as well as the social and medical impact caused by falls has led to much T research into postural stability. Postural instability has been

From the Departments of Kinesiology (Lafond, Prince) and Surgery (Prince), University of Montreal, Montreal; Gait and Posture Laboratory, Marie-Enfant Rehabilitation Center, Montreal (Lafond, Prince); and Ageing Research Center of Sherbrooke, Sherbrooke Geriatric University Institute, Sherbrooke (Corriveau, He´bert), QC, Canada. 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(s) or upon any organization with which the author(s) is/are associated. Reprint requests to Franc¸ois Prince, PhD, Dept of Kinesiology, Univ of Montreal, PO Box 6126, Downtown Station, Montreal, QC H3C 3J7, Canada, e-mail: [email protected]. 0003-9993/04/8506-7980$30.00/0 doi:10.1016/j.apmr.2003.08.089

Arch Phys Med Rehabil Vol 85, June 2004

associated with the incidence of falls.1-4 Several clinical and laboratory methods have been developed with which to assess different dimensions of the postural control system.5 In laboratory settings, the experimental protocol is generally designed to assess postural steadiness or postural balance. Postural balance refers to the ability to stay upright or to recover equilibrium after external dynamic perturbations, whereas postural steadiness refers to standing as still as possible on a force platform, sometimes under altered somatosensory conditions.6 Postural unsteadiness has been associated with a higher risk of falling.7-9 Many variables have been developed from force platform signals to quantify postural steadiness. Center of pressure (COP) is the most common and is defined as the point of application of the ground reaction forces under the feet.10 COP is the outcome of the inertial forces of the body and restoring equilibrium forces of the postural control system. COP displacement is used to make inferences about neurologic and biomechanic mechanisms of postural control. COP displacement can be characterized as 1- and 2-dimensional measures. These measures include but are not limited to (1) the rootmean-square error (RMS), (2) COP range, (3) mean COP position, (4) mean (MPF) or median (MedPF) power frequency, (5) fractal dimensions, (6) sway area, (7) mean COP velocity, and (8) COP path length. The RMS variance of the displacement and the velocity of the COP have also been proposed as measures.11 Like many biologic measurements, COP has an intrinsic variability that affects the reliability and the validity of postural control outcomes. Assuming that the measurement conditions are constant, the difference between 2 measures is attributable to the error variance, which is influenced by the variability of the phenomena measured and the precision of the instrumentation. Increasing the number of repetitions decreases the weight of the error variance compared with the true score.12 Averaging the results of many trials or repetitions improves the intrasession reliability—the immediate test-retest reliability. Intersession reliability is referred to as day-to-day reliability and introduces the concept of stability of the phenomenon measured over a given time period. To date, few metrologic studies have been performed to assess the reliability of COP measures in elderly subjects during quiet standing (table 1). To our knowledge, only 1 study has used an appropriate statistical design13 to assess the immediate test-retest reliability based on the analysis of variance (ANOVA) model with a specific version of the intraclass correlation coefficient14 (ICC). Corriveau et al14 estimated the intrasession reliability of a single measure of the COP– center of mass (COM) variable. They concluded that 4 trials must be averaged to obtain a reliable measure of postural control. By using the mean of 4 trials, they could perform an intersession and interrater reliability study of the COP-COM variable.15 Before considering the intersession reliability, it is important to determine how many trials must be averaged to obtain an immediate, reliable COP measure in 1 testing session. In addition, postural steadiness stabilogram measures should be used to quantify intervention effects or to document the evo-

897

RELIABILITY OF COP MEASURES, Lafond Table 1: Summary of Intrasession Reliability Studies of Different Postural Steadiness Variables Study

Geurts et al37 Samson and Crowe23 Benvenuti et al25

Corriveau et al14

Postural Stability Measure

RMS COP amplitude RMS COP velocity COP mean velocity COG area COG velocity COP velocity SD quadratic fit of COP RMS COP-COM

Sample Size

No. of Trial

Outcome

8

10

CV

15 36

10 1

7

9

CV ICC⫽.71 ICC⫽.76 ICC⫽.74 ICC⫽.76 (AP), .77 (ML) 1 trial ICC⫽.79 (AP); .69 (ML) 4 trials ICC⫽.94 (AP); 90 (ML)

Abbreviations: AP, anteroposterior; COG, center of gravity; CV, coefficient of variation; ICC, intraclass correlation coefficient; ML, mediolateral; SD, standard deviation.

lution of a particular neuromuscular condition that affects postural control. A follow-up evaluation of postural steadiness includes before and after measurements. The clinical difference of COP measures, that is, the change that could be considered clinically different between 2 measurements, should be greater than the noise induced by the intrinsic variability of the phenomena and instrumentation. The minimal metrically detectable change (MMDC) represents this difference between 2 measurements associated with the lack of perfect reliability or the measurement error.16 The objectives of this study, then, were to determine (1) how many trials must be averaged to obtain a reliable measure of several COP measures by determining the intrasession reliability and (2) the MMDC of these COP measures in healthy elderly subjects. METHODS Participants Seven healthy elderly subjects (4 women, 3 men) participated in this study. Their average age, mass, and height were, respectively, 67.9⫾4.3 years, 65.6⫾17.5kg, and 161⫾12cm. The recruitment procedures are detailed elsewhere14 and are briefly explained here. At the time of their evaluation, subjects reported having no neurologic or musculoskeletal impairments, were living independently in the community, and were more than 60 years of age. Subjects were excluded if they reported having had 1 fall in the past year. All subjects gave their informed consent to participate in the study, which had been previously approved by the university ethics board. Procedure Subjects were instructed to stand upright on 2 adjacent force platforms. Data were acquired with subjects in a double-leg stance with feet at pelvis width. They were instructed to look straight ahead, with their head erect and their arms at their sides in a comfortable position. Nine successive trials, with subjects’ eyes open for 120 seconds and with a rest between trials of approximately 5 minutes, were used for analysis. Outlines of the feet were traced to ensure that foot placement was constant across trials. Ground reaction forces and moments were acquired from 2 force platforms.a Analog signals were sampled at a frequency of 20Hz with an analog-to-digital converter and were recorded on a computer. The temperature of the force platforms was always stabilized for at least 60 minutes before any data were collected, to minimize any electronic drifts. Potentiometers of the amplifier were set manually to zero, and

a 1-second bias from each force platform channel was recorded at 20Hz before each experimental session. The mean bias was then removed from the experimental data to ensure that force platform data had zero drift after amplification. The data (in volts) collected were thereafter transformed (in newtons and newton meters) by multiplying the data array by the full calibration matrix—provided by the manufacturer—that is specific for each force platform. The force platform signals were filtered with a zero-lag sixth-order Butterworth low-pass filter at 10Hz. The COP from each force platform was first calculated and the netCOP, which is the weighted average position of COP from each force platform,17 was determined in both anteroposterior (AP) and mediolateral (ML) directions with respect to equation 1 below. The force platform data were processed by using Matlab, version 5.1.b netCOP⫽COP l ⫻

F z(l) F z(r) ⫹COP r ⫻ F z(l) ⫹F z(r) F z(l) ⫹F z(r)

(1)

where COPl and COPr are the center of pressure under the left and right foot, respectively; Fz(r) and Fz(l) are the vertical ground reaction force under the left and right foot, respectively. Data Analysis Several COP measures were calculated from the netCOP of each trial in both the AP and ML directions for: (1) the RMS amplitudes, (2) the COP range, (3) the COP mean velocity, (4) the MPF, and (5) the MedPF. Sway area, using the principal component analysis, was also calculated.18 Each of these COP measures was calculated in the first 30 seconds, 60 seconds, and 120 seconds of each quiet standing trial. The reliability of these COP variables was estimated by using the ICC described by Shrout and Fleiss.13 Derived from the ANOVA results, the ICC2,1 compares the within-subject variability with the between-subject variability and it considers random effects over time (equation 2). ICC 2,1 ⫽

MS B ⫺MS E MS B ⫹(n⫺1)MS E ⫹n(MS R ⫺MS E )/k

(2)

where MSB, MSR, and MSE are the mean squares of the 2-way ANOVA, n is the number of subjects, and k the number of trials. According to Donner and Eliasziw,19 7 subjects and 9 trials allows for significantly (P⬍.05) differentiating ICCs between .95 and .80 with 80% power. The Spearman-Brown formula for stepped-up reliability was used to estimate the number of trials (k) required to obtain an Arch Phys Med Rehabil Vol 85, June 2004

898

RELIABILITY OF COP MEASURES, Lafond Table 2: ICC Values of 1 Trial and Number of Trials (k) to be Averaged to Obtain an ICC >.90 of COP Variables for the 30-Second, 60-Second, and 120-Second Trials 30 Seconds COP Variables

60 Seconds

120 Seconds

Direction

ICC

k ICCⱖ.90

ICC

k ICCⱖ.90

ICC

k ICCⱖ.90

— AP ML AP ML AP ML AP ML AP ML

.22 .29 .44 .39 .35 .73 .87 .34 .35 .32 .41

31 22 12 14 16 3 2 18 17 20 13

.47 .38 .57 .52 .61 .77 .90 .09 .28 .02 .24

10 15 7 9 6 3 1 ⬎20 ⬎20 ⬎20 ⬎20

.41 .52 .62 .58 .58 .83 .94 .44 .30 .34 .47

13 8 6 7 8 2 2 11 ⬎20 18 10

Sway area COP range RMS COP velocity MPF MedPF

expected reliability and the 95% lower bound confidence interval (CI).12 The following equation allows an estimation of the reliability coefficient of the mean (Rk) by averaging k trials with a 1-measure reliability coefficient R: R k⫽

kR 1⫹(k⫺1)R

(3)

One can also estimate the number of trials (k) to be averaged to obtain a target coefficient of reliability (R*) by using the following formula: k⫽

R*(1⫺R) R(1⫺R*)

(4)

Postural steadiness stabilogram measures should be used to quantify the change over the time of postural control. To achieve that, it is important to distinguish the true value from the noise caused by the inherent variability of the measurement. The MMDC was defined as the 95% CI of the standard error (SE) of the measurement (⫾1.96Se). The SE of the measurement is defined by the equation Se⫽Sr公1⫺ICC where Sx is the standard deviation (SD) of the data. RESULTS The ICCs obtained for 1 measure are presented in table 2 for each trial duration. In general, the reliability increases with the duration of the trial. The ICCs obtained for 1 measure of 120 seconds were .58 (lower CI, .31) and .58 (lower CI, .30) for the

RMS, .83 (lower CI, .64) and .94 (lower CI, .85) the for COP mean velocity, .52 (lower CI, .25) and .62 (lower CI, .35) for COP range, .44 (lower CI, .18) and .30 (lower CI, .08) for MPF, and .34 (lower CI, .11) and .47 (lower CI, .21) for MedPF in the AP and ML directions, respectively. The ICC of the COP sway area obtained for 1 measure was .41 (lower CI, .16). Figures 1 to 3 show, for each trial duration, the ICCs estimated by averaging k trials for RMS, COP mean velocity, and COP range, respectively. For the 120-second trial duration, we found that for RMS, 7 trials had to be averaged to obtain an ICC of .90. Similar results were found for COP range, where 8 and 6 trials of 120 seconds had to be averaged to have an ICC of .90, respectively, in the AP and ML directions. COP mean velocity is the most reliable COP measure. Averaging 2 trials of 120 seconds allows an ICC of .90, whereas averaging 4 trials allows an ICC over .95 in both the AP and ML directions. MPF, MedPF, and COP sway area are the less reliable COP measures. MPF showed an ICC of .90 by averaging 12 and 21 trials of 120 seconds in the AP and ML directions, respectively. Averaging 10 and 12 trials gave a reliability coefficient over .90 for MedPF (in ML) and for sway area, respectively. MMDC was calculated by using the ICC and the SD of the data in both the AP and ML directions for sway area, COP range, COP mean velocity, and RMS. MMDC values presented in table 3 are for 120-second trial duration, with the appropriate number of trials averaged for a specific ICC value.

Fig 1. ICC values of the RMS of the COP in (A) the AP and (B) the ML directions.

Arch Phys Med Rehabil Vol 85, June 2004

RELIABILITY OF COP MEASURES, Lafond

899

Fig 2. ICC values of COP mean velocity in (A) the AP and (B) the ML directions.

DISCUSSION The intrasession reliability of several COP variables was estimated by using the ICC based on an ANOVA model. Our first objective was to determine the number of trials required to obtain a reliable measure of several COP variables. We found that COP mean velocity was the most reliable COP variable. Only 2 trials of 2 minutes in duration had to be averaged to obtain an ICC over .90 in both the AP and ML directions. Averaging more trials did not produce an important increase in reliability because of the curve plateau (fig 2). COP range and RMS COP displacement presented a good reliability, with an ICC over .80, by averaging at least 4 trials of 2 minutes (4 – 8 trials). Sway area and power spectral measures (MPF, MedPF) showed poor reliability where more than 5 trials had to be averaged to obtain an ICC over .80 (5–11 trials). Whether the reliability is acceptable is discussed elsewhere.14 An ICC over .80 represents good reliability20; a few authors have suggested that an excellent reliability coefficient should be over .85.21,22 The lack of variability of a single trial of COP variables in this study can be attributed to the inherent variability of the restoring equilibrium forces and not to the lack of precision of the apparatus. The resolution of the force platforms is .8␮V䡠v⫺1䡠N⫺1 and the accuracy of the COP position calculated in this study was .2mm. The reliability of postural steadiness outcome has been addressed only in a few studies. Samson and Crowe23 assessed the intrasession variability of COP path length and COP mean

velocity of ten 1-minute quiet standing trials. They recommended more than 10 trials to improve the consistency, but this recommendation was based on the coefficient of variation (CV), not on measurement error. Hufschmidt et al24 also used the CV to estimate the intraindividual variance of COP variables based on 10 trials. The sway area showed a high CV (58.9%). However, the authors proposed to increase the duration of trials to decrease the high variability of the COP measures that reduce the clinical significance of the stabilogram. Benvenuti et al25 used the ICC to estimate the reliability of several postural sway measures on a retest protocol with 4 hours between 2 sessions and found an ICC of .74 based on 2 consecutive trials for COP mean velocity. The authors concluded that this measure had poor reliability, which could be attributed to the variability of the subject’s postural balance strategy. We obtained, with an appropriate ICC, similar results with 1 measure of COP mean velocity (range, .80 –.84) and an excellent reliability by averaging 2 trials. Goldie et al26 used linear regression to investigate the retest reliability of 2 consecutive trials and concluded that the ground reaction forces are more reliable than COP displacement in both the AP and ML directions. Only 1 study14 was designed on an appropriate statistical model to determine the number of trials to be averaged to obtain a reliable measure of the postural steadiness. It was concluded that a reliable measure of the COP-COM necessitates an average of 4 trials (ICC⬎.90).

Fig 3. ICC values of COP range in (A) the AP and (B) the ML directions.

Arch Phys Med Rehabil Vol 85, June 2004

900

RELIABILITY OF COP MEASURES, Lafond Table 3: MMDC Values of COP Measures for the 120-Second Trial Duration

CONCLUSIONS

Direction

Mean ⫾ SD

Trial*

MMDC

Sway area (mm2)

—

135.8⫾46.3

COP range (mm)

AP

26.3⫾5.6

ML

14.1⫾4.3

AP

3.7⫾1.0

ML

2.3⫾0.9

AP ML

7.9⫾1.7 5.2⫾1.0

1 13 1 8 1 6 1 7 1 8 2 2

⫾68.4 ⫾27.2 ⫾7.4 ⫾3.5 ⫾5.1 ⫾2.7 ⫾1.2 ⫾0.9 ⫾1.2 ⫾0.6 ⫾1.0 ⫾0.6

COP Variables

RMS (mm)

COP velocity (mm/s)

*Number of trial averaged for ICC ⱖ.90.

The unreliability of postural steadiness measures decreases the power of a study to detect differences between groups because of the random measurement errors that cause an increase in the variance. Our results showed low reliability for several COP variables when measured once and may explain the results of the few studies intended to measure differences between groups (young, old, healthy, impaired subjects) or to detect an effect of treatment or experimental conditions. Crilly et al27 found no improvement of the postural steadiness measures (RMS, COP range, Rombert quotient) after a 12-week exercise program based on 2 trials. The RMS of COP displacement of one 40-second repetition did not differ between healthy elderly subjects and elderly subjects with diabetic peripheral neuropathy among several experimental conditions.28 It was shown that COP mean velocity was the most discriminating variable with which to assess the age-related changes of the postural steadiness and risk of falling.29-31 This could be explained by the high reliability of the COP mean velocity that we found and the poor reliability estimated for other COP variables that are also used to assess the age-related changes of postural steadiness. It may also explain why several studies found no intervention effects or differences between healthy subjects and patients using other COP variables.24,27,32 However, the effect of an exercise program was not significant when using COP velocity and sway area based on only 1 trial.33 Another study8 found no improvement in an exercise group compared with controls by averaging 2 to 4 trials of 10 seconds. The authors did not address the reliability of the COP measures of postural steadiness to explain their results. Our study showed an increase of reliability of COP measure by increasing the trial duration. We assessed the reliability of COP measures by using 3 different trial durations. In general, reliability increased the longer the duration of the trial. Recently, it was suggested that COP should be recorded for at least 60 seconds to increase the stationarity of the signal.34,35 However, signal stationarity cannot be addressed only on a time interval definition because of the long-range correlation of the COP.36 It is obvious that the duration of the trial in quiet standing is limited because of fatigue particularly for pathologic elderly patients. A trial duration of 120 seconds is a good compromise for signal stationarily and the reliability of the COP measures. It also permits good precision of the fundamental frequency when spectral analysis of the COP is preferred. Arch Phys Med Rehabil Vol 85, June 2004

This study established the MMDC of several COP measures. These values represent the lower bound of the clinically significant amount of difference that can be considered during a follow-up evaluation in a clinical setting. References 1. Tinetti ME, Speechley M, Ginter SF. Risk factors for falls among elderly persons living in the community. N Engl J Med 1988;319: 1701-7. 2. Lord SR, Clark RD, Webster IW. Postural stability and associated physiological factors in a population of aged persons. J Gerontol 1991;46:M69-76. 3. O’Loughlin JL, Robitaille Y, Boivin JF, Suissa S. Incidence of and risk factors for falls and injurious falls among the communitydwelling elderly. Am J Epidemiol 1993;137:342-54. 4. Shumway-Cook A, Baldwin M, Polissar NL, Gruber W. Predicting the probability of falls in community-dwelling older adults. Phys Ther 1997;77:812-9. 5. Horak FB. Clinical assessment of balance disorders. Gait Posture 1997;6:76-84. 6. Prieto TE, Myklebust JB, Myklebust BM. Characterization and modeling of postural steadiness in the elderly: a review. IEEE Trans Rehabil Eng 1993;1:26-34. 7. Brocklehurst JC, Robertson D, James-Groom P. Clinical correlated of sway in old age—sensory modalities. Age Ageing 1982; 11:1-10. 8. Lichtenstein MJ, Shields SL, Shiavi RG, Burger MC. Clinical determinants of biomechanics platform measures of balance in aged women. J Am Geriatr Soc 1988;36:996-1002. 9. Overstall W, Exton-Smith AN, Imms FJ, Johnson AL. Falls in the elderly related to postural balance. BMJ 1977;1:287-95. 10. Winter DA. A.B.C. (anatomy, biomechanics and control) of balance during standing and walking. Waterloo (ON): Waterloo Biomechanics; 1995. 11. Riley PO, Benda BJ, Gill-Body KM, Krebs DE. Phase plane analysis of stability in quiet standing. J Rehabil Res Dev 1995; 32:227-35. 12. Bravo G, Potvin L. Estimating the reliability of continuous measures with Cronbach’s alpha or the intraclass correlation coefficient: toward the integration of two traditions. J Clin Epidemiol 1991;44:381-90. 13. Shrout PE, Fleiss JL. Intraclass correlation: uses in assessing rater reliability. Psychol Bull 1979;86:420-8. 14. Corriveau H, He´ bert R, Prince F, Raıˆche M. Intrasession reliability of the “center of pressure minus center of mass” variable of postural control in the healthy elderly. Arch Phys Med Rehabil 2000;81:45-8. 15. Corriveau H, He´ bert R, Prince F, Raıˆche M. Postural control in the elderly: an analysis of test-retest and interrater reliability of the COP-COM variable. Arch Phys Med Rehabil 2001;82:80-5. 16. He´ bert R, Spiegelhalter DJ, Brayne C. Setting the minimal metrically detectable change on disability rating scales. Arch Phys Med Rehabil 1997;78:1305-8. 17. Winter DA, Prince F, Frank JS, Powell C, Zabjek KF. Unified theory regarding A/P and M/L balance in quiet standing. J Neurophysiol 1996;75:2334-43. 18. Oliveira LF, Simpson DM, Nadal J. Calculation of area of stabilimetric signals using principal component analysis. Physiol Meas 1996;17:305-12. 19. Donner A, Eliasziw M. Sample size requirements for reliability studies. Stat Med 1987;6:441-8. 20. Fleiss JL. The design and analysis of clinical experiments. New York: John Wiley & Sons; 1986. 21. Laurencelle L. The´ orie et techniques de la mesure instrumentale. Ste-Foy (QC): Presse de l’Universite´ du Que´ bec; 1998. 22. Weiner EA, Stewart BJ. Assessing individuals. Boston: Little Brown; 1984. 23. Samson M, Crowe A. Intra-subject inconsistencies in quantitative assessments of body sway. Gait Posture 1996;4:252-7.

RELIABILITY OF COP MEASURES, Lafond

24. Hufschmidt A, Dichgans J, Mauritz KH, Hufschmidt M. Some methods and parameters of body sway quantification and their neurological application. Arch Psychiatr Nervenkr 1980;228:135-50. 25. Benvenuti F, Mecacci R, Gineprari I, et al. Kinematic characteristics of standing disequilibrium: reliability and validity of a posturographic protocol. Arch Phys Med Rehabil 1999;80:278-87. 26. Goldie PA, Bach TM, Evans OM. Force platform measures for evaluating postural control: reliability and validity. Arch Phys Med Rehabil 1989;70:510-7. 27. Crilly RG, Willems DA, Trenholm KJ, Hayes KC, Delaquerrie`reRichardson LF. Effect of exercise on postural sway in the elderly. Gerontology 1989;35:137-43. 28. Dickstein R, Shupert CL, Horak FB. Fingertip touch improves postural stability in patients with peripheral neuropathy. Gait Posture 2001;14:238-47. 29. Prieto TE, Myklebust JB, Myklebust BM, Kreis DU. Intergroup sensitivity in measures of postural steadiness. In: Woollacott M, Horak F, editors. Posture and gait: control mechanisms. Vol 2. Portland (OR): Univ Oregon Books; 1992. p 122-5. 30. Era P, Heikkinen E. Postural sway during standing and unexpected disturbances of balance in random samples of men of different ages. J Gerontol 1985;40:287-95. 31. Maki BE, Holliday PJ, Fernie GR. Aging and postural control: a

32. 33. 34. 35. 36. 37.

901

comparison of spontaneous- and induced-sway balance tests. J Am Geriatr Soc 1990;38:1-9. Niam S, Cheung W, Sullivan PE, Kent S, Gu X. Balance and physical impairments after stroke. Arch Phys Med Rehabil 1999; 80:1227-33. Lichtenstein MJ, Shields SL, Shiavi RG, Burger MC. Exercise and balance in aged women: a pilot controlled clinical trial. Arch Phys Med Rehabil 1989;70:138-43. Carpenter MG, Frank JS, Winter DA, Peysar GW. Sampling duration effects on center of pressure summary measures. Gait Posture 2000;13:35-40. Carroll JP, Freedman W. Nonstationary properties of postural sway. J Biomech 1993;26:409-16. Duarte M, Zatsiorsky VM. Long-range correlations in human standing. Phys Lett A 2001;283:124-8. Geurts AC, Nienhaus B, Mulder TW. Intrasubject variability of selected force-platform parameters in the quantification of postural control. Arch Phys Med Rehabil 1993;74:1140-50.

Suppliers a. Model OR6-5; Advance Mechanical Technology Inc, 176 Waltham St, Watertown, MA 02472. b. The MathWorks Inc, 3 Apple Hill Dr, Natick, MA 01760-2098.

Arch Phys Med Rehabil Vol 85, June 2004