Predictive Validity and Responsiveness of the Functional Ambulation Category in Hemiparetic Patients After Stroke

Article Outline

• Abstract
• Methods
  • Participants
  • Assessments
  • Intra- and Interrater Reliability
  • Examiners
  • Concurrent Validity
  • Predictive Validity
  • Responsiveness
• Results
  • Reliability
  • Concurrent Validity
  • Predictive Validity
  • Responsiveness
• Discussion
  • Reliability
  • Concurrent Validity
  • Predictive Validity
  • Responsiveness
  • Assessment of Walking
  • Study Limitations
• Conclusions
• Acknowledgment
• Appendix 1. The Functional Ambulation Categories10,11
• Appendix 2. Key Questions for Each FAC
• References
• Copyright

Abstract 

Mehrholz J, Wagner K, Rutte K, Meiβner D, Pohl M. Predictive validity and responsiveness of the Functional Ambulation Category in hemiparetic patients after stroke.

Objective

To determine the reliability, concurrent and predictive validity, and responsiveness of the Functional Ambulation Category (FAC) in hemiparetic patients after stroke.

Design

Prospective cohort.

Setting

An early rehabilitation center for patients with neurologic disorders.

Participants

Fifty-five nonambulatory patients after first-ever stroke, with duration of illness between 30 and 60 days, were included.

Interventions

Not applicable.

Main Outcome Measures

FAC, Rivermead Mobility Index (RMI), walking velocity, step length, and six-minute walking test (6MWT) were assessed at the beginning, after 2 and 4 weeks of rehabilitation, and again 6 months later. After 6 months, community ambulation was also assessed. Test-retest and interrater reliability, concurrent, discriminant, and predictive validity and responsiveness of the FAC were calculated.

Results

Based on video examinations, high test-retest reliability (Cohen κ=.950) and interrater reliability (κ=.905) were found. FAC scores at the beginning and after 2 weeks, 3 weeks, and 6 months correlated highly with the RMI (Spearman ρ=.686, ρ=.787, ρ=.825, ρ=.893, respectively), distance walked in the 6MWT (ρ=.949, ρ=.937, ρ=.931, ρ=.906, respectively), walking velocity (ρ=.952, ρ=.939, ρ=.902, ρ=.901, respectively), and step length (ρ=.952, ρ=.932, ρ=.896, ρ=.877, respectively) at the same time points (all P<.001). The RMI, walking velocity, step length, and distance walked in the 6MWT differed for each FAC category (P<.001). After 4 weeks of rehabilitation, an FAC score of 4 or higher predicted community ambulation at 6 months with 100% sensitivity and 78% specificity. FAC scores changed significantly between the first 2 and second 2 weeks (Wilcoxon z=8.7, z=7.9, respectively; both P<.001) of the inpatient rehabilitation program.

Conclusions

The FAC has excellent reliability, good concurrent and predictive validity, and good responsiveness in patients with hemiparesis after stroke.

Key Words: Gait, Outcome assessment (health care), Rehabilitation, Stroke

MANY PATIENTS REMAIN unable to walk or have difficulties with walking after stroke.¹ Gait rehabilitation that aims to improve walking ability is therefore of high importance to patients and their relatives.² Evidence³, ⁴, ⁵ suggests that gait training in specialized rehabilitation facilities improves walking function after stroke. However, there is still a need for specific walking assessment tools to measure the progress of patients (eg, during gait rehabilitation). Currently available gait-assessment tools range from complex and expensive laboratory techniques involving detailed analyses of kinematic and kinetic variables⁶ to simple measures used in the clinical setting for measuring such variables as walking speed.⁷ However, high costs, difficulties in interpretation and communication of the results, and problems with accessibility can sometimes make laboratory gait analysis impractical for everyday clinical use.

Clinically based assessments of walking ability such as the Timed Up & Go test and the timed walk test may therefore have advantages of immediate availability; ease of administration and interpretation; and, of course, low costs.⁸, ⁹ However, walking scales have to be validated with respect to reliability, responsiveness, and predictive and concurrent validity. The Functional Ambulation Category (FAC) is a common clinical gait assessment scale first described by Holden et al in 1984.¹⁰ The FAC distinguishes 6 levels of walking ability on the basis of the amount of physical support required. The FAC is a quick visual measurement of walking, is simple to use, easy to interpret, and cost-effective because only stairs and 15m of indoor floor are needed to administer the test.⁹ Research¹⁰, ¹¹, ¹² indicates that the FAC score correlates with walking velocity and step length.

Predictive validity and responsiveness to change (caused, eg, by recovery) have not before been investigated for the FAC, although they are very important in decision making regarding treatment and for intervention trials.

The aim of the present study was therefore to evaluate these essential but unknown psychometric properties of the FAC in a population of initially nonambulatory patients with hemiparesis after stroke.

Back to Article Outline

Methods 

Participants 

Patients immediately after admission to inpatient rehabilitation, who had a first-time supratentorial stroke, either ischemic or hemorrhagic, were aged 18 to 80 years and had a duration of illness shorter than 60 days were included in the study. Patients were all able to sit without holding on to any support (eg, on the edge of the bed and with feet unsupported by anyone and without contact with the floor), were either completely nonambulatory or required the assistance of 1 or 2 therapists to walk irrespective of the use of an ankle-foot orthosis or a walking aid (except rolling walker), were able to understand the meaning of the study and to follow instructions, and had signed an informed consent of participation approved by the local ethics committee.

All patients took part in a multicenter research project called the Deutsche Gangtrainer Studie (DEGAS).¹³ In the DEGAS, patients were randomized to receive either 20 minutes of locomotor training on an electromechanical gait trainer and 25 minutes of physiotherapy (PT) or only 45 minutes of PT without locomotor training every week day for 4 weeks during an inpatient rehabilitation program.¹³ Only the patients from the authors’ center who had complete data and video sets of all assessment time points were included in this study.

Assessments 

All patients were assessed at the beginning (baseline), after 2 and 4 weeks of therapy, and again after 6 months (follow-up). FAC, gait speed, stride length (measured with the ten-meter walk test [10MWT]), and walking distance (measured with the six-minute walking test [6MWT]) were assessed. The FAC discriminates 6 levels of walking abilities on the basis of the amount of physical support required (see appendix 1 for details).

The Barthel Index (score range, 0−100) is a valid and reliable index measuring activities of daily life.¹⁴, ¹⁵ Included are 10 items relating to the degree of independence from any help.¹⁵ The Barthel Index score was used to describe activities of daily life at baseline.¹⁴, ¹⁵

The Rivermead Mobility Index (RMI; score range, 0−15) is comprised of 15 mobility-related items, from turning over in bed to running, and is a reliable and valid measure of mobility restrictions and body functions.¹⁶

The 10MWT and the 6MWT were used to assess walking velocity, stride length, and walking distance.⁸, ¹⁷, ¹⁸ For walking-velocity evaluation, patients walked a distance of 15m twice at their maximum speed. The time was taken for middle 10m, and the mean velocity was calculated.¹³ During the 10MWT, we assessed step length according to prior publications.⁴, ¹⁹

To assess walking distance, patients walked 6 minutes without interruption (6MWT), and the maximum distance was noted. If patients had to stop during the 6MWT because of exhaustion, the distance covered to that point was considered.⁷

Intra- and Interrater Reliability 

One week after the first rating of each examiner, the second rating was performed to reduce the chance that the raters would remember the last rating, thus biasing the test-retest reliability. Every rater was kept blinded to any results of the other examiners. We determined interrater reliability and test-retest reliability of the FAC by using κ statistics. According to Altman,²⁰ the results of the κ statistics were considered as “excellent” if κ values were .80 or above.

To measure the reliability of the FAC, video sequences for every patient at every time point were recorded. The FAC video sequences were standardized for all 4 FAC assessment time points and included a video taken from the rear and from the side while the patients walked a 15-m distance. The videos gave information on the physical support needed by patients while walking, irrespective of the technical aids used. If the patients could climb stairs, an additional video helped to distinguish between FAC level 4 and 5.

Examiners 

All video sequences were assessed twice by each of the 4 examiners. All examiners were physiotherapists and had at least 5 years experience in stroke rehabilitation. Nonetheless, in addition to the described FAC, we used key questions for each of the 6 categories to identify the appropriate FAC level (appendix 2).

Concurrent Validity 

Concurrent validity was evaluated by correlating the FAC scores with RMI scores, walking velocity, stride length, and 6MWT of patients at baseline, after 2 and 4 weeks, and after 6 months. The RMI was chosen because it measures body function, (eg, sitting and standing balance), which is known to be associated with walking function.

Walking velocity, stride length, and distance walked in the 6MWT were used because these parameters represent basic gait tests broadly used in rehabilitation facilities and rehabilitation research.

The time points were chosen according to the DEGAS study protocol.¹³ These time points were rated to observe trends in the relationship across the time bands (increasing or decreasing over time).

The changes of FAC scores between study onset and 6 months later were correlated with changes in RMI scores, walking velocity, stride length, and 6MWT. Additionally, Spearman rank correlation coefficients were calculated for all comparisons.²¹

To examine whether the RMI, walking velocity, stride length, and 6MWT differ for each of the FAC categories, an analysis of variance approach was used.

Predictive Validity 

To measure predictive validity, functional community ambulation was used as a target outcome at 6 months after the study onset. The term “community ambulation” was used according to previous publications and was defined as the ability to walk faster than 73m/min, ability to walk longer than 332m, and ability to climb stairs and curbs as described by Lernier-Frankiel et al²² and Lord and Rochester.²³ If patients met all 3 predefined conditions, patients’ ability to walk was graded as “community ambulation.”

Receiver operating characteristics (ROCs) for all possible FAC cutoff points at 4 weeks were used. Sensitivity, specificity, negative and positive predictive values, and the area under the curve (AUC) were calculated to identify community ambulation at 6-month follow-up.

Responsiveness 

To measure responsiveness to change, standardized response means (SRMs) were calculated. This was done because SRMs may better reflect individual changes than effect sizes (SRM is equal to mean change divided by standard deviation [SD] of the difference). Differences were calculated between FAC scores at baseline compared with 2 weeks, 3 weeks compared with 4 weeks, and 4 weeks compared with 6 months. Wilcoxon signed-rank tests (a nonparametric alternative to the paired t test) were used to check for statistical significance of differences.²¹ The global α was set for all comparisons to .05. We used SAS/STAT software^a for all calculations.

Back to Article Outline

Results 

Between August 2002 and May 2003, a total of 55 patients were eligible and fulfilled the inclusion criteria. The patient characteristics are shown in table 1; the descriptive statistics of the study parameters are shown in table 2.

Table 1. Patient Characteristics at Baseline

Characteristic	Patients (N=55)
Age (y)⁎	62.8±10.2(65,40−78)
Sex (female/male)	15/40
Diagnosis
Stroke (ischemic/hemorrhagic)	41/14
Side of lesion
Hemisphere (left/right)	30/25
Duration of illness (d)⁎	30.6±15.5(28,9−60)
Barthel Index (score, 0−100)⁎	34.3±11.6(30,25−65)

Table 2. Descriptive Statistics of Study Parameters

Parameters	At Baseline	After 2 Weeks	After 4 Weeks	At 6-Month Follow-Up
FAC (score, 0−5)	0.44±0.69(0,0−3)	1.22±1.32(1,0−5)	1.98±1.50(2,0−5)	2.79±2.12(4,0−5)
RMI (score, 0−14)	2.51±1.62(2,1−10)	4.04±2.88(3,0−12)	5.76±3.93(5,1−14)	7.38±5.01(8,0−14)
6MWT (m)	15.9±34.3(0,0−175)	50.9±81.1(15,0−315)	83.9±107.8(40,0−460)	112.3±143.9(60,0−560)
Walking velocity (m/s)	0.07±0.14(0.01,0.01−0.82)	0.19±0.28(0.08,0.01−0.96)	0.33±0.46(0.15,0.01−1.96)	0.38±0.51(0.15,0.00−1.96)
Step length (m)	0.09±0.13(0,0.00−0.48)	0.18±0.19(0.18,0.00−0.61)	0.27±0.20(0.22,0.00−0.74)	0.28±0.26(0.26,0.00−1.06)

NOTE. Values are mean ± SD (median, range).

Reliability 

The test-retest and interrater agreement of ratings of the FAC were excellent, with kappas of .950 and .905, respectively.²⁰

Concurrent Validity 

The results of the correlations between the absolute values of the FAC, RMI, 6MWT, walking velocity, and step length are shown in table 3. Changes in FAC scores between baseline and 6 months after the end of the study correlated significantly with changes in the RMI (Spearman ρ=.841, P<.001), 6MWT (ρ=.795, P<.001), walking velocity (ρ=.767, P<.001), and step length (ρ=.805, P<.001). For each of the FAC categories, the RMI, walking distance, walking velocity, and step length differed significantly (all P<.001).

Table 3. Concurrent Validity

Parameter	FAC Scores at Baseline	FAC Scores After 2 Weeks	FAC Scores After 4 Weeks	FAC Scores at 6-Month Follow-Up
RMI	.686	.787	.825	.893
6MWT	.949	.937	.931	.906
Walkingvelocity	.952	.939	.902	.901
Steplength	.952	.932	.896	.877

NOTE. Values are Spearman ρ (all P<.001).

Predictive Validity 

The highest AUC was found for the cutoff FAC of 4 or higher (AUC=89%). The FAC cutoff of 4 or higher was more sensitive (100%) than specific (78%) in predicting community ambulation 6 month after study end. All results of the ROC are shown in table 4.

Table 4. Results of the ROC

Responsiveness 

The FAC scores changed over time; the median FAC score was 0 (mean ± SD, 0.44±0.69) at baseline, 2 (mean, 1.98±1.5) after 4 weeks, and 4 (mean, 2.79±2.12) 6 months after study end. The responsiveness of the FAC was good. FAC scores changed significantly between the first 2 study weeks (SRM=1.016, Wilcoxon z=−8.691, Bonferroni-adjusted P<.001), between the second 2 study weeks (SRM=.842, z=−7.900, Bonferroni-adjusted P<.001), and between study week 4 and 6 months after study end (SRM=.699, z=−6.368, Bonferroni-adjusted P<.001).

Back to Article Outline

Discussion 

The present study shows that the FAC has excellent test-retest and interrater reliable, has good concurrent and predictive validity, and is sensitive to change in the assessment of patients after stroke who cannot walk without personal assistance at the beginning of their inpatient rehabilitation.

Reliability 

Walking scales cannot be seen as valid without knowledge of reliability. The reliability of the FAC is only described in studies by Holden¹⁰ and Collen²⁴ and colleagues. In a first description of the FAC, Holden¹⁰ described good interrater reliability (κ=.72) when tested by 9 therapists on 5 patients. In contrast, Collen²⁴ achieved only fair interrater reliability (κ=.36) in 25 chronic patients after stroke. In both studies, test-retest reliability was not described. In our study, the interrater reliability was very good (κ=.91) and therefore slightly higher than previously reported.¹⁰, ²⁴ This could mainly be attributed to the use of key questions and videos and to the fact that our rater had at least 5 years experience in stroke rehabilitation.

Concurrent Validity 

Changes of FAC scores correlated significantly with changes in walking speed, step length, and the distance walked in the 6MWT. These walking variables are commonly regarded as indicators of progress in gait performance.⁸, ¹⁷, ²⁵, ²⁶ Therefore, the results of the present study suggest that improvements in FAC scores are associated with improvements in gait performance. These results are in line with the findings of Holden who showed a close relationship between the FAC and measures of gait performance such as walking velocity, cadence, step and stride length, and temporal-distance measures.¹⁰, ¹¹

On the one hand, we observed a slight decrease of the correlation between FAC and walking distance over time. On the other hand, we found a minor increase in the relation between FAC and RMI over time.

Kollen et al¹² recently described that walking velocities associated with the specific FAC vary in time from higher to lower speeds. In contrast in our study, we found only a 5% decrease in the relationship between FAC and walking velocity. Six months after stroke, however, 81% of the variance of the FAC scores was still explained by walking velocity.

The relationship between FAC scores and the RMI underscores an association between gait ability and some body functions. This could be interpreted in the sense that motor control of tasks such as rolling from one side to another, sit-to-stand maneuvers, and sitting and standing balance may be useful during walking.

Predictive Validity 

Our results suggest that the ability to ambulate after a 4-week rehabilitation program measured with the FAC predicts community ambulation at 6 months after stroke. We conclude, therefore, that a dichotomized FAC (in FAC ≥4 and FAC <4) may be useful in predicting a level of community ambulation with high sensitivity and specificity. This could be important for clinical rehabilitation goal planning and future studies investigating walking outcome.

Responsiveness 

The present study showed for the first time that the FAC is responsive to change between the first weeks of rehabilitation and after 6 months of rehabilitation because of the recovery process after stroke. Until now, the FAC may have been seen as too insensitive to change to be used as a reliable outcome measure for research purposes. However, our results indicate a good responsiveness of the FAC between the different investigations. Even from 4 weeks until 6 months after the study onset, the FAC scores proved responsive to change and could still detect improvements in walking function. Our results suggest, therefore, that the FAC can be used in clinical research to measure improvement and outcome in gait performance in primary nonambulatory patients after stroke.

Assessment of Walking 

There is no true criterion standard for the assessment of walking outcome after stroke. Despite gait analysis systems playing an important role in better understanding hemiparetic gait after stroke, there is a need for reliable walking scales for routine clinical use and additionally for a valid assessment of gait outcome in research projects.²⁷ An argument for the use of walking scales is that they are often easy to interpret. Walking scales such as the FAC measure functional aspects, which are often more important for patients after stroke and for their relatives (eg, how much personal assistance is required for walking). Additionally, the FAC allows a precise documentation of walking progress. Thus, the FAC is useful as a routine clinical assessment tool and also for research purposes to measure walking outcome.

Study Limitations 

The lack of a uniform time basis (the duration of illness of our cohort was between 9 and 60 days) limits the interpretation of our results with respect to nonlinear patterns of recovery after stroke.²⁸ However, for logistical reasons, all patients immediately after entering the inpatient rehabilitation program were included.

Furthermore, our results may be limited in the sense that only patients who were initially nonambulatory were included in the study. Therefore, the responsiveness and concurrent validity (eg, walking distance) of the scale may be overestimated to some extent. However, in planning clinical trials, it is desirable to include patients who are most responsive to change.²⁹

The revealed excellent reliability was at least partly influenced by the use of key questions to each FAC. The questionnaire was used to train the examiners but was, however, not tested for its reliability and validity.

Additionally, the video-based assessments are of limited practicability in everyday use. Therefore, the revealed reliability may be overestimated compared with the use of the FAC in clinical practice.

Recent studies²³, ³⁰ have questioned the common opinion that clinical walking scales are appropriate and valid measures for community ambulation. In this vein, the FAC and our definition for community ambulation²² are limited because crucial environmental and cognitive factors were not considered.³⁰ However, given the predictive validity of the FAC, the categories of the FAC could be used as a screening tool to identify patients who are at higher risk of not achieving community ambulation during rehabilitation. This might be an advantage in therapy planning and rehabilitation goal setting.

We have used the term “responsiveness” in our study. However, because “responsiveness” has different meanings in science, some authors would argue that the term “sensitivity to change” would be more appropriate. Different and sometimes contradictory recommendations exist in the literature for using these terms. In preparing this article, all authors decided to use the term “responsiveness,” which can be seen as the result of time but also could be the result of applied interventions.

Back to Article Outline

Conclusions 

In patients with hemiparesis after stroke, the FAC has very good reliability, good concurrent and predictive validity, and is responsive to change over time. For clinical practice and research purposes, the FAC may be an appropriate assessment tool in the measurement of walking ability.

Supplier

Back to Article Outline

Acknowledgment 

We thank Derek Barton for his assistance in preparing the manuscript.

Back to Article Outline

Appendix 1. The Functional Ambulation Categories¹⁰, ¹¹ 

Back to Article Outline

Appendix 2. Key Questions for Each FAC 

Key questions for a FAC level of 0:

“Is walking really only possible if two persons hold this patient?”

If answer is “yes” then FAC=0

Key questions for a FAC level of 1:

Two or more answers have to be “yes” for FAC=1. Two or more answers have to be “no” for FAC=2

Key questions for a FAC level of 2:

Answer to question 1 and 3 have to be “yes” and the answer to question 2 has to be “no” for FAC=2

Key questions for a FAC level of 3:

If one answer to the three questions is “no”, choose FAC=3

Key questions for a FAC level of 4:

Both answers have to be “yes” for FAC=5. If one answer is “no”, choose FAC=4

Key questions for a FAC level of 5:

“Would this patient be able to walk alone over a variety of surfaces, such as uneven ground, grass, stairs etc.?”

Answer with “yes” for FAC=5

Back to Article Outline

References 

1. Kollen B, van de Port I, Lindeman E, Twisk J, Kwakkel G. Predicting improvement in gait after stroke: a longitudinal prospective study. Stroke. 2005;36:2676–2680
   • View In Article
   • CrossRef
2. Bohannon R. Rehabilitation goals of patients with hemiplegia. Int J Rehabil Res. 1988;11:181–183
   • View In Article
   • CrossRef
3. Macko RF, Ivey FM, Forrester LW, et al. Treadmill exercise rehabilitation improves ambulatory function and cardiovascular fitness in patients with chronic stroke: a randomized, controlled trial. Stroke. 2005;36:2206–2211
   • View In Article
   • CrossRef
4. Pohl M, Mehrholz J, Ritschel C, Ruckriem S. Speed-dependent treadmill training in ambulatory hemiparetic stroke patients: a randomized controlled trial. Stroke. 2002;33:553–558
   • View In Article
   • CrossRef
5. Kwakkel G, Wagenaar RC, Koelman TW, Lankhorst GJ, Koetsier JC. Effects of intensity of rehabilitation after stroke (A research synthesis). Stroke. 1997;28:1550–1556
   • View In Article
   • MEDLINE
   • CrossRef
6. Bowden MG, Balasubramanian CK, Neptune RR, Kautz SA. Anterior-posterior ground reaction forces as a measure of paretic leg contribution in hemiparetic walking. Stroke. 2006;37:872–876
   • View In Article
   • CrossRef
7. Eng JJ, Chu KS, Dawson AS, Kim CM, Hepburn KE. Functional walk tests in individuals with stroke: relation to perceived exertion and myocardial exertion. Stroke. 2002;33:756–761
   • View In Article
   • CrossRef
8. Salbach NM, Mayo NE, Higgins J, Ahmed S, Finch LE, Richards CL. Responsiveness and predictability of gait speed and other disability measures in acute stroke. Arch Phys Med Rehabil. 2001;82:1204–1212
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (162 KB)
   • CrossRef
9. Wade DT. Measurement in neurological rehabilitation. Oxford: Oxford Medical Publications; 1992;p 25-30
   • View In Article
10. Holden MK, Gill KM, Magliozzi MR, Nathan J, Piehl-Baker L. Clinical gait assessment in the neurologically impaired (Reliability and meaningfulness). Phys Ther. 1984;64:35–40
    • View In Article
    • MEDLINE
11. Holden MK, Gill KM, Magliozzi MR. Gait assessment for neurologically impaired patients (Standards for outcome assessment). Phys Ther. 1986;66:1530–1539
    • View In Article
    • MEDLINE
12. Kollen B, Kwakkel G, Lindeman E. Time dependency of walking classification in stroke. Phys Ther. 2006;86:618–625
    • View In Article
    • MEDLINE
13. Pohl M, Werner C, Holzgraefe M, et al. Repetitive locomotor training and physiotherapy improve walking and basic activities of daily living in acute stroke patients: a single-blind, randomised multi-centre trial. Clin Rehabil. 2007;21:17–27
    • View In Article
    • MEDLINE
    • CrossRef
14. Laake K, Laake P, Ranhoff AH, Sveen U, Wyller TB, Bautz-Holter E. The Barthel ADL index: factor structure depends upon the category of patient. Age Ageing. 1995;24:393–397
    • View In Article
    • MEDLINE
15. Wade DT, Collin C. The Barthel ADL Index: a standard measure of physical disability?. Int Disabil Stud. 1988;10:64–67
    • View In Article
    • MEDLINE
16. Collen FM, Wade DT, Robb GF, Bradshaw CM. The Rivermead Mobility Index: a further development of the Rivermead Motor Assessment. Int Disabil Stud. 1991;13:50–54
    • View In Article
    • MEDLINE
17. Eng JJ, Dawson AS, Chu KS. Submaximal exercise in persons with stroke: test-retest reliability and concurrent validity with maximal oxygen consumption. Arch Phys Med Rehabil. 2004;85:113–118
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (70 KB)
    • CrossRef
18. van Loo MA, Moseley AM, Bosman JM, de Bie RA, Hassett L. Inter-rater reliability and concurrent validity of walking speed measurement after traumatic brain injury. Clin Rehabil. 2003;17:775–779
    • View In Article
    • MEDLINE
    • CrossRef
19. Pohl M, Rockstroh G, Ruckriem S, Mrass G, Mehrholz J. Immediate effects of speed-dependent treadmill training on gait parameters in early Parkinson’s disease. Arch Phys Med Rehabil. 2003;84:1760–1766
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (78 KB)
    • CrossRef
20. Altman D. Practical statistics for medical research. London: Chapman & Hall; 1991;
    • View In Article
21. Armitage P, Colton T. Encyclopedia of biostatistics. Chichester: Wiley; 1998;
    • View In Article
22. Lernier-Frankiel M, Vargas S, Brown M, Krusell L, Schoneberger W. Functional community ambulation: what are your criteria?. Clin Manag Phys Ther. 1986;6:12–15
    • View In Article
23. Lord SE, Rochester L. Measurement of community ambulation after stroke: current status and future developments. Stroke. 2005;36:1457–1461
    • View In Article
    • CrossRef
24. Collen FM, Wade DT, Bradshaw CM. Mobility after stroke: reliability of measures of impairment and disability. Int Disabil Stud. 1990;12:6–9
    • View In Article
    • MEDLINE
25. Olney SJ, Griffin MP, McBride ID. Temporal, kinematic, and kinetic variables related to gait speed in subjects with hemiplegia: a regression approach. Phys Ther. 1994;74:872–885
    • View In Article
    • MEDLINE
26. Kwakkel G, Wagenaar RC. Effect of duration of upper- and lower-extremity rehabilitation sessions and walking speed on recovery of interlimb coordination in hemiplegic gait. Phys Ther. 2002;82:432–448
    • View In Article
    • MEDLINE
27. Baer HR, Wolf SL. Modified Emory functional ambulation profile: an outcome measure for the rehabilitation of poststroke gait dysfunction. Stroke. 2001;32:973–979
    • View In Article
    • CrossRef
28. Kwakkel G, Kollen B, Twisk J. Impact of time on improvement of outcome after stroke. Stroke. 2006;37:2348–2353
    • View In Article
    • CrossRef
29. Haynes R, Sackett D, Guyatt G, Tugwell P. Clinical epidemiology: how to do clinical practice research. Philadelphia: Lippincott Williams & Wilkins; 2006;
    • View In Article
30. Shumway-Cook A, Patla AE, Stewart A, Ferrucci L, Ciol MA, Guralnik JM. Environmental demands associated with community mobility in older adults with and without mobility disabilities. Phys Ther. 2002;82:670–681
    • View In Article
    • MEDLINE

• a Version 9.1.3; SAS Institute Inc, 100 SAS Campus Dr, Cary, NC 27513.