VA Research and Development LOGO

Logo for the Journal of Rehab R&D
Vol. 40 No. 1, January/February 2003
Pages 19 — 28


Reliability of measures of gait performance and oxygen consumption with stroke survivors
Inácio Teixeira da Cunha-Filho PT, PhD; Helene Henson, MD; Sharmin Wankadia, PT, MS; Elizabeth J. Protas, PT, PhD, FACSM
Physical Therapy Department, UNIBH Minas Grerais, Brazil; Department of Physical Medicine and Rehabilitation, Houston Veterans Affairs Medical Center, Houston, TX; School of Physical Therapy, Texas Womens University, Houston, TX; Department of Physical Therapy, University of Texas Medical Center, Galveston, TX; Rehabilitation Research and Development Center of Excellence on Healthy Aging with Disabilities, Houston Veterans Affairs Medical Center, Houston, TX
Abstract — This study assessed the reliability of gait performance with concurrent measures of oxygen consumption (VO2) in stroke survivors (SS). Nine male SS (60.00 ± 15.08 yr) had a recent history of stroke (44.56 ± 51.35 days since the stroke) and were receiving rehabilitation. Four had a right cerebrovascular accident (CVA), and five had a left CVA. Subjects walked without assistance, although three used a single cane to complete the test. Within 30 minutes, subjects completed two trials of a 5 min walk while walking back and forth on a 5 m walkway wearing a portable gas analyzer to collect samples of gases. The intraclass correlation coefficient (ICC) was used to assess reliability. The ICC for gait energy expenditure, walk distance, gait speed, and gait energy cost were 0.64, 0.97, 0.95, and 0.97, respectively. Assessment of gait performance with concurrent measures of VO2 is a reliable procedure with SS.
Key words: gait performance, oxygen consumption, reliability, stroke.

Abbreviations: ANOVA = analysis of variance, CVA = cerebrovascular accident, D = distance, EE = energy expenditure, HR = heart rate, HVAMC = Houston Veterans Affairs Medical Center, ICC = intraclass correlation coefficient, S = speed, SDD = smallest detectable difference, SEM = standard error of the measurement, SS = stroke survivors.
This material was based on work supported partially by the Houston Veterans Affairs Medical Center: Rehabilitation Research and Development Center of Excellence on Healthy Aging with Disabilities and the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES), Brazil.
Address all correspondence and requests for reprints to Elizabeth J. Protas, PT, PhD, FACSM; Department of Physical Therapy, School of Allied Health Sciences, University of Texas Medical Branch, 301 University Boulevard, Galveston, TX 77555-1144; 409-772-3068; fax: 402-747-1613; email: ejprotas@utmb.edu.

INTRODUCTION

Mobility is often significantly affected by cerebrovascular accident (CVA). Hence, a principal goal of rehabilitation for patients who survive a CVA is to target mobility and independence and to recover function. Residual motor deficits, especially those linked to gait, impose further health-related risks and contribute to increased morbidity associated with this chronic condition. This population also has poor aerobic function, which may be further aggravated by a preexisting cardiovascular condition, such as hypertension, congestive heart failure, peripheral vascular disease, or pulmonary and metabolic diseases [1-3]. Energy expenditure (EE) required to perform routine ambulation is elevated approximately 1.5- to 2.0-fold in hemiparetic stroke patients compared to normal control subjects [4].

Increased energy demand reduces movement, favoring further decline in cardiopulmonary fitness and increasing disuse atrophy and weakness, which further impairs function [5]. Despite this, rehabilitative interventions usually fail to emphasize the need to improve or to prevent further deterioration of the ability to generate aerobic work, especially for patients in the acute stages of recovery from a stroke. The lack of aerobic training for these patients may be justified by the lack of strong evidence to suggest that aerobic exercise training can reduce the high EE or cardiovascular demands of walking in this population. Moreover, these patients are considered high risk for new cardiovascular events, and few studies define exercise parameters or the safety of this mode of rehabilitation for this population. Cardiovascular comorbidity is a significant health concern in approximately 75 percent of the patients who experience a stroke, and cardiac disease remains the leading cause of death in individuals who are long-term stroke survivors (SS) [6].

Few studies have actually assessed aerobic function, especially with subacute SS, because of the lack of protocols and guidelines [5-8]. Measurement of aerobic capacity by bike or treadmill protocols with actual measures of peak oxygen consumption (VO2) during maximal effort is expensive, requires specialized personnel, and demands time. This population of SS is considered to be high-risk for exercise stress testing given the frequent association between CVA and hypertension. Furthermore, the practicality of maximal aerobic stress tests in this population may be questionable, depending upon the magnitude of the motor and neurological impairment.

Indirect assessment of aerobic function with SS has been performed by measures of endurance obtained with timed walking tests, such as the 5 min and 6 min walks [8,9]. Timed walking tests have been developed to assess endurance with patients who have different disabling conditions, and the psychometric qualities of these tests, such as reliability and validity, have been reported [10,11].

More recently, the use of portable gas analyzers has allowed the assessment of VO2 during functional activities. Patients can carry these analyzers on their back while performing functional tasks. Measurement of VO2 during functional activities might indicate more appropriately the patient's capacity to generate aerobic work.

Neuromotor recovery is more pronounced during the first 3 months after a stroke and contributes to the variability of outcomes during this period. Any tool developed to evaluate outcomes with the neurologically impaired population must demonstrate stability over time so that the observed scores are as close as possible to the true scores. Therefore, the main objective of this paper was to assess the reliability of gait performance with concurrent assessment of VO2 with the use of a portable gas analyzer in a group of SS.

METHODS
Subjects

Ambulatory SS were recruited from a sample of convenience admitted to the Rehabilitation Service at the Houston Veterans Affairs Medical Center (HVAMC). Participants had a stroke less than 4 weeks before entering the study, and they provided written informed consent before participating. This study was approved by the Institutional Review Board for Human Subject Research for Baylor College of Medicine and Affiliated Hospitals.

Instrumentation

We obtained measurement of VO2 with an Aerosport portable gas analyzer, model KB1-C. This is a lightweight (1.25 kg) portable system that comprises a face mask for collecting expired air, a pneumotach, sensors for analyzing oxygen (O2) and carbon dioxide (CO2) content of expired air, a heart rate (HR) monitor with a chest strap transmitter worn around the chest, and a battery pack. The KB1-C system performs O2 and CO2 analysis continuously using a patented system called Electronic Variable Sampling. As air is exhaled through the pneumotach, a proportional microsample is drawn off into a mixing chamber inside the unit. The O2 concentration is measured with a galvanic fuel cell, and the CO2 concentration is measured by nondispersive infrared analysis. These are the same methods used in conventional metabolic measurement systems. Expired flow is measured with a flat-plate orifice pneumotach, and ventilatory volume is calculated by digital integration. VO2 and rate of carbon dioxide production (VCO2) are calculated according to standard equations for indirect calorimetry.

Before each assessment, the pneumotach was calibrated with a 3 L calibration syringe. The gas analyzers were calibrated according to the manufacturer's specifications with the use of a calibration gas of known O2 and CO2 composition [12 percent O2 and 5 percent CO2, and balance nitrogen (N2)]. The KB1-C system was set to sample data every 20 s for ventilation, fractions of inspired O2 and expired CO2, HR, respiratory exchange ratio, VO2, and VCO2. Participant information such as age, gender, height, and weight was entered in the portable gas analyzer before each measurement

Validation of the KB1-C system has been reported against a standard metabolic gas analyzer in which patients with paraplegia exercising to exhaustion during arm crank ergometry and with nondisabled subjects running to exhaustion on a treadmill (coefficient of determination of 0.94 for fractional expired oxygen) [12]. The KB1-C has also been validated against a criterion Doug-las bag system during cycle ergometry [13]. In this latter study, the validity of the system was acceptable for measuring O2 uptake in the 1.5 to 3.5 L/min range [13]. There are no reports on the validity of this system for the level of VO2 observed in this present study.

Protocol

Before beginning the test, each subject was instructed to walk back and forth at a usual gait speed, on a 5 m walkway for 2 min to become acquainted with the procedure. The subject was then allowed to rest for 6 min.

The subject was then fitted with the portable gas analyzer, a facemask, and a chest band HR transmitter. The participants were asked to walk on the 5 m walkway for a total time of 5 min. The floor was marked in 1 m intervals, and the subjects were asked to walk back and forth at their usual gait velocity throughout the test. We used a stopwatch to measure the walking time.

After this first trial, the mask was taken off and the information recorded was printed. The participant was allowed to rest for at least 10 min before performing the second trial. The portable gas analyzer was recalibrated, the facemask was repositioned, and the subject then completed the second trial.

The variables of interest were rate of VO2 during the 5 min walk, which was called EE, walk distance (D), gait speed (S), and gait energy cost (C). EE was calculated by averaging three samples of VO2 recorded every 20 s each minute during the 5 min walk. Data were recorded in milliliters of O2 per kilogram of body weight per minute walked (mLO2/kg/min). D was recorded in meters and was the total distance covered in 5 min. Since body weight contributes to the expression of VO2, VO2 is reported relative to body weight. Participants were weighed on a stand-on digital scale (Scale-Tronix 5005) immediately before the initiation of the protocol. Weight was assessed in kilograms with participants dressed and with shoes on, the same way they performed the tests.

Measurement of gait speed (m/s) was obtained by dividing the walk distance by the time walked in seconds for the entire 5 min. Cost was obtained by dividing the values of VO2 sampled during the 5 min walk by the speed values in meters per minute.

Although walk performance has been reported based on a 6 min walk test [2,9,11], we used the 5 min walk considering that the population studied is more acutely compromised and might be more limited compared to subjects with a chronic stroke. The 5 min walk as an indicator of gait endurance has been used with patients with different levels of neurological impairment [8,10].

The use of a 5 m walkway to assess gait performance was selected because it required different levels of neuromuscular integrity, such as the ability to change gait acceleration and to turn, which may impact walking performance. Moreover, because subacute SS may be discharged home or to skilled facilities, assessment of gait performance in such a short track might represent more of the requirements on discharge.

Statistical Analysis

Data are presented as means and standard deviations (SDs). The main item of interest in this study was the relationship between the paired measurements EE, D, C, and S during the 5 min walk for the same patient. We analyzed the data using repeated measures analysis of variance (ANOVA) with the trial as the independent variable and the patient as a random effect. The intraclass correlation coefficient (ICC), which is a measure of correlation that considers variance, describes the agreement between the repeated measures. The model ICC3,1 was chosen because we were looking for the stability of the measures based on one assessment. The ICC was calculated as described by Fleiss and Shrout and is defined mathematically as the ratio between the variance component obtained between subjects minus the residual error, divided by the between subjects variance [14].

We used the standard error of the measurement (SEM) to indicate absolute reliability and calculated it as the square root of the absolute error variance [15,16]. We also calculated the smallest detectable difference (SDD) for the measurements. This SDD is a clinically relevant measure because it suggests the change that might be expected because of an intervention rather than sampling error at a level of statistical significance of 0.05 level. The SDD is the 95 percent confidence interval (CI) of the SEM multiplied by the square root of 2; i.e., SDD = ± 1.96 x square root of 2 x SEM. Thus, we calculated the smallest change between two independent measurements for the outcomes of interest [17].

RESULTS

Nine male subjects were recruited to participate in this study. Table 1 depicts demographics, such as age, height, weight, number of days since the stroke, type and location of injury, comorbidity, and medication. Subjects had a mean age of 60.00 ± 15.18 years. The mean time since the CVA was 44.56 ± 51.35 days. Five had a right CVA, and four had a left-sided CVA. All participants were ambulatory and clinically in stable condition. Three participants used a single cane to complete the test, and one participant used a wheelchair for long distances.

Table 1.
Participants' characteristics.
Subject
Age
Height
Weight
 
No. Days Since Stroke
 
Type of Injury
 
Location
Comorbidity
 
Medication
1
69
185.90
101.60
 
7.00
 
H
 
R parieto-occiptal
CAD, Afib, MI, CHF
 
1,2,4,5,9
 
 
 
 
 
 
 
 
 
cortex
 
 
 
2
51
176.00
83.90
 
153.00
 
I
 
R subcortical
HTN, GI
 
3,4
3
73
108.10
71.70
 
6.00
 
I
 
L subcortical
CAD, PVD, CABG
 
4,9
4
69
173.00
79.80
 
2.00
 
I
 
R subcortical
HTN,
 
4,9
5
27
188.00
80.70
 
43.00
 
I
 
L hemisphere
-
 
9
6
57
175.00
98.00
 
28.00
 
I
 
L cortical
-
 
9
7
63
183.00
90.70
 
45.00
 
I
 
L subcortical
-
 
4,9
8
54
163.00
74.40
 
103.00
 
I
 
R frontal
-
 
8,9
9
77
177.80
83.50
 
14.00
 
I
 
L frontal
CAD, HTN, GI
 
1,4,8,9

Mean
60.00
169.99
84.92
 
44.56
 
 
 
SD
15.18
24.33
10.03
 
51.35
 
 
 

Afib = atrial fibrillation
CABG = coronary artery bypass grafting
CAD = coronary artery disease
CHF = congestive heart failure
GI = glucose intolerance
H = hemorrhagic
HTN = hypertension
I = ischemic
L = left
MI = myocardial infarction
PVD = peripheral vascular disease
R = right
 
TIA = transitory ischemic attack
SD = standard deviation
1 = beta-blocker
2 = diuretics
3 = ACE inhibitors
4 = calcium channel blocker
5 = nitrate
6 = antiarrhythmic
7 = digitalis
8 = alpha-blocker
9 = other

In Table 2, the values for each outcome variable are shown separately for each subject, across trials. The ANOVA indicated that the mean values for EE, D, C, and S were not different between trial 1 and trial 2: 6.75 ± 1.24 versus 6.93 ± 1.44 mLO2/kg/min (F1,8 = 0.23, p = 0.64), 152.94 ± 60.55 versus 158.06 ± 57.21 m (F1,8 = 2.38, p = 0.16), 0.25 ± 0.11 versus 0.24 ± 0.09 mLO2/kg/m, (F1,8 = 3.52, p = 0.10), and 0.51 ± 0.20 versus 0.53 ± 0.19 m/s (F1,8 = 2.46, p = 0.16), respectively. Table 3 depicts the reliability index obtained for each outcome variable (ICC), SEM, and SDD.

Table 2.
Individual scores for each variable evaluated in both trials.
Subject
 
EE
(mLO2/kg/min)
 
D (m)
 
C
(mLO2/kg/m)
 
S (m/s)
 
 
Trial 1
Trial 2
 
Trial 1
Trial 2
 
Trial 1
Trial 2
 
Trial 1
Trial 2
1
 
5.72
5.58
 
167.00
153.00
 
0.17
0.18
 
0.56
0.51
2
 
9.46
8.96
 
128.50
145.00
 
0.37
0.31
 
0.43
0.48
3
 
6.44
8.68
 
269.00
275.00
 
0.12
0.16
 
0.90
0.92
4
 
5.91
5.03
 
130.00
135.00
 
0.23
0.19
 
0.43
0.45
5
 
5.85
5.21
 
220.00
219.00
 
0.13
0.12
 
0.73
0.73
6
 
6.14
6.75
 
69.00
81.00
 
0.45
0.42
 
0.23
0.27
7
 
7.74
6.87
 
144.00
141.50
 
0.27
0.24
 
0.48
0.47
8
 
7.43
7.52
 
150.00
158.00
 
0.25
0.24
 
0.50
0.53
9
 
6.05
7.79
 
99.00
115.00
 
0.31
0.27
 
0.33
0.38

Mean
 
6.75
6.93
 
152.94
158.06
 
0.25
0.24
 
0.51
0.53
SD
 
1.24
1.44
 
60.55
57.21
 
0.11
0.09
 
0.20
0.19
p Value
 
0.64
-
 
0.16
-
 
0.10
-
 
0.16
-

EE = gait energy expenditure
D = walk distance
S = gait speed
 
C = gait energy cost
SD = standard deviation
 
 
 
 
 


Table 3.
Reliability indexes.
Measure
ICC3,1
SEM
SDD
EE (mLO2/kg/min)
0.64
0.80
2.22
 
 
 
 
D (m)
0.97
6.92
19.18
 
 
 
 
C (mLO2/kg/m)
0.95
0.02
0.06
 
 
 
 
S (m/s)
0.97
0.02
0.06

ICC3,1 = intraclass correlation coefficient (model)
SEM = standard error of measurement
SDD = smallest detectable difference (1.96 x squart root of 2 x SEM)
EE = energy expenditure (gait)
D = distance (walk)
C = cost (gait energy)
S = speed (gait)
 
DISCUSSION

The purpose of this study was to evaluate the reliability of gait performance with simultaneous measurement of VO2 with individuals who had had a recent CVA. The ICCs obtained with this study indicated that the reliability of VO2 during walking (EE) is modest, but can be excellent when its assessment is normalized by walking velocity, which is the gait energy cost.

For walking in normal individuals whose ages ranged from 20 to 80 years old, the reported gait EE is 12.01 mLO2/kg/min [18]. Previous studies involving subacute and chronic SS reported gait EE values of 8.9 mLO2/kg/min [19], 6.6 mLO2/kg/min for more impaired SS, and 8.15 mLO2/kg/min for less impaired SS [8]. In our study, the mean EE across trials was 6.84 mLO2/kg/min. This value suggests that the population evaluated in this study was comparable to the more impaired SS and demonstrated a level of VO2 during walking that was about half of the normal values.

In a previous generalizability study conducted in our laboratory involving 40 healthy individuals (ages ranged from 18 to 53 years), we observed that this portable gas analyzer generated a reliability coefficient of 0.93 for EE. In the present study, the reliability index obtained with SS was lower than the one observed for healthy individuals (ICC3,1 = 0.64). Measures of VO2 while subjects walked are prone to variability because they can be influenced by the walking speed and by the subject's size [21]. The variability in this measure might be even more evident in a population such as SS because of the impaired gait and different characteristics of the gait, such as asymmetry and reduced stride lengths.

When the VO2 (EE) was analyzed reflecting the efficiency of walking (gait energy cost), that is, normalizing the VO2 by the walking velocity, the reliability index increased to 0.95. This indicates, indeed, that gait speed affects the variability of VO2 and, therefore, should be included in the calculation of measures of VO2 during walking. The mean distance covered during the 5 min walk between trials was 155.5 = meters with preferred gait speed, and the reliability of this measure was high (ICC3,1 = 0.97). A walk distance of 144 m has been used to classify SS as moderately impaired on a 6 min walk [22]. If we extrapolate this value to a 5 min walk, a moderate impairment would be equivalent to a walk distance of 120 m. On admission, subacute SS walked a mean distance of 53.93 m ± 49.7 m, ranging from 1 m to 162 m, during a 5 min walk [23]. In this study, two out of nine participants walked less than 120 m in both trials. Therefore, our group ranged from moderate to less than moderate impairment while walking.

The gait speed of SS demonstrates substantial variability, depending on when the assessment was conducted during recovery, the amount of compromise, and the length of the track used [22,24,25]. Nevertheless, gait speed is a well-accepted indicator of neurological recovery in SS [22,24]. The mean speed between trials in this study was about 0.52 m/s, which, according to Wade et al., is a normal walk speed for SS [25]. Therefore, the use of a short walkway does not seem to underestimate the values for walking speed. The reliability of speed assessment in this study was excellent (ICC3,1 = 0.97).

Before test initiation, participants were instructed to walk back and forth on the 5 m walkway for 2 min to get used to the procedure and to decrease variability during the testing because of the lack of familiarity with the process. However, the participants tended to perform better during the second trial, although the instructions to perform the test were not altered between trials. The participants walked longer and faster, but tended to be more efficient (lower costs) in the second trial. This trend suggested that the participants walked more naturally and closer to their optimal performance at preferred walking velocity during the second trial.

We also used measures of SEM to indicate absolute reliability that can be expected in an individual's per-formance. The SEM values for each outcome variable were small enough to demonstrate that the measurements are stable over time.

We reported an analysis of the SDD for each outcome variable to illustrate that the measure of absolute reliability can also be used to indicate the size of change that is expected because of an intervention as compared to changes just related to sampling error. In our sample, a change between two consecutive measures of VO2 of more than 2.22 mLO2/kg/min could be considered a meaningful change. A meaningful change in walk distance should be higher than 19.18 m. For cost and speed, changes should be more than 0.06 mLO2/kg/m and 0.06 m/s, respectively.

The number of turns expected in such a short walkway (5 m) is known to affect both walking distance and velocity. Furthermore, more turns may have differential effects on patients with varying levels of gait impairment. For example, a more severely compromised stroke survivor may take longer to turn (more steps) and probably use more oxygen in the process. However, gait performance based on a short distance may represent a patient's daily life more, considering that they are more often required to overcome short distances during the day when discharged from acute rehabilitation and sent home. With this study, we did not find any indication that the 5 min walk completed on a 5 m walkway might have underestimated the participant's walk endurance or gait speed or that gait EE and gait cost were excessively scored, considering that the results are within the range for similar patients. Therefore, the use of a 5 m walkway to assess gait outcomes, such as endurance, EE, speed and energy cost, may be another possible choice to evaluate gait performance. Changes in acceleration accompanied by multiple turns may provide a more realistic estimate of walking required in daily life.

CONCLUSION

Since the sample size used was small, the generaliza-bility of the results is limited. In addition, the KB1-C system still needs further validation while measuring low levels of VO2, similar to the ones observed here. However, this study indicates that same-day measures of gait performance obtained during a 5 min walk completed on a 5 m walkway are reproducible with ambulatory subacute SS and were similar to other studies reported. The variability of VO2 during walking can be decreased by normalizing the oxygen values by the gait speed. One should examine the smallest detectable differences in clinical settings to sort out changes that may occur as a result of clinical interventions as opposed to changes secondary to natural recovery or the variability of the measure.

REFERENCES
1. Roth EJ. Heart disease in patients with stroke: incidence, impact, and implications for rehabilitation, part I: Classification and prevalence. Arch Phys Med Rehabil 1993; 74:752-60.
2. Olney SJ, Monga TN, Costigan PA. Mechanical energy of walking of stroke patients. Arch Phys Med Rehabil 1986; 67:92-98.
3. Potempa KL, Lopez M. Braun LT, Szidon P, Fogg L, Cicknell T. Physiological outcomes of aerobic exercise training in hemiparetic stroke patients. Stroke 1995;26: 101-5.
4. Gerson J, Orr W. External work of walking in hemiparetic patients. Scand J Rehabil Med 1971;3:85-88.
5. Macko RF, DeSouza CA, Tretter LD, Silver KH, Smith GV, Anderson PA, Tomoyasu N, Gorman P, Dengel DR. Treadmill aerobic exercise training reduces the energy expenditure and cardiovascular demands of hemiparetic gait in chronic stroke patients. Stroke 1997;28:326-30.
6. Sacco RL, Wold PA, Kannel EB, McNamara PM. Survival and recurrence following stroke: the Framingham Study. Stroke 1982;13:290-95.
7. Tsuji T, Liu M, Tsujiuchi K, Chino N. Bridging activity as a mode of stress testing for persons with hemiplegia. Arch Phys Med Rehabil 1999;80:1060-64.
8. Cunha I, Henson H, Qureshy H, Lim PAC, Monga TN, Protas EJ. Gait outcomes after acute stroke rehabilitation with supported treadmill ambulation training: a randomized controlled pilot study. Arch Phys Med Rehabil 2002; 83:1258-65.
9. Mayo NE, Wood-Dauphinee S, Ahmed S, Gordon C, Higgins J, McEwen S, Salbach N. Disablement following stroke. Disabil Rehabil 1999;21(5,6):258-68.
10. Protas EJ, Holmes SA, Qureshy H, Johnson A, Lee D, Sherwood AM. Supported treadmill ambulation training after spinal cord injury: a pilot study. Arch Phys Med Rehabil 2001;82:825-31.
11. King S, Wessel J, Bhambhani Y, Maikala R, Sholter D, Maksymowych W. Validity and reliability of the 6 minute walk in persons with fibromyalgia. J Rheumatol 1999; 26(10):2233-37.
12. Jacobs PL, Nash MS, Mintz D. Assessment of fractional expired gases and air flow by an ambulatory metabolic analyzer. JEP online 1999;2(4):20-28.
13. King GA, McLaughlin JE, Howley ET, Bassett DR, Ainsworth BE. Validation of Aerosport KB1-C portable metabolic system. Int J Sports Med 1999;20:304-8.
14. Fleiss JL, Shrout PE. Intraclass correlations: uses in assessing rater reliability. Psychol Bull 1979;86(2):420-28.
15. VanLeeuwen DM, Barnes MD, Pase M. Generalizability theory: a unified approach to assessing the dependability (reliability) of measurements in the health sciences. J Outcome Meas 1998;2:302-25.
16. Clark S, Rose DJ. Evaluation of dynamic balance among community-dwelling older adult fallers: a generalizability study of the limits of stability test. Arch Phys Med Rehabil 2001;82:468-74.
17. Roebroeck ME, Harlaar J, Lankhorst GJ. The application of generalizability theory to reliability assessment: an illustration using isometric force measurements. Phys Ther 1993;73:386-95.
18. Waters RL, Mulroy S. The energy expenditure of normal and pathologic gait. Gait Posture 1999;9:207-31.
19. Hash D. Energetics of wheelchair propulsion and walking in stroke patients. Orthop Clin North Am 1978;9:372-74.
20. Teixeira da Cunha Filho I, Henson H, Qureshy H, Lim PAC, Monga TN, Protas EJ. A comparison of regular rehabilitation and regular rehabilitation with supported treadmill ambulation training for acute stroke patients. J Rehabil Res Dev 2001;38(2):245-55.
21. Baker R, Hausch A, McDowell B. Reducing the variability of oxygen consumption measurements. Gait Posture 2001; 13:202-9.
22. Richards CL, Malouin F, Dean C. Gait in stroke. Stroke 1999;15(4):833-55.
23. Cunha-Filho IT, Henson H, Qureshy H, Lim PAC, Monga TN, Protas EJ. Performance-based gait tests for acute stroke patients. Am J Phys Med Rehabil 2002;81:848-56.
24. Perry J, Garrett M, Gromley J, Mulroy S. Classification of walking handicap in the stroke population. Stroke 1995; 26:982-89
25. Wade D, Wood V, Heller A, Maggs J, Hewer RL. Walking after stroke. Scand J Rehabil Med 1987;19:25-30.
Submitted for publication January 16, 2001. Accepted in revised form October 10, 2002.

Go to TOP

Go to the Contents of Vol. 40 No. 1