Development of a Position-Specific Index of Muscle Strength to be Used in Stroke Evaluation

Article Outline

• Abstract
• Methods
  • Participants
  • Main Outcome Measure
  • Data Analysis
• Results
• Discussion
• Conclusions
• References
• Copyright

Abstract 

Moriello C, Mayo NE. Development of a position-specific index of muscle strength to be used in stroke evaluation.

Objective

To develop a position-specific index of muscle strength for individuals with stroke.

Design

Cross-sectional design.

Setting

A major teaching hospital in a Canadian urban city.

Participants

Sixty-three patients with poststroke onset between 3 and 12 months.

Interventions

Not applicable.

Main Outcome Measure

The muscle strength of the lower-extremity muscles was tested bilaterally in multiple positions using hand-held dynamometry.

Results

A principal components analysis resulted in grouping the muscles of the affected and unaffected sides of the gravity related and gravity eliminated positions into 5 indices. The 5 indices were moderately to highly correlated (r² range, .59−.81) with each other and so were combined into 1 global index. The gravity related muscle strength on the affected side was, on average, 85% of the unaffected side (range, 37%–157%); the gravity eliminated muscle strength of the affected side was, on average, 92% of the unaffected side (range, 53%–121%).

Conclusions

This study resolves the methodologic issue of how to summarize multiple data points that relate to one construct, namely, strength of different muscle groups assessed in several positions.

Key Words: Lower extremity, Muscles, Rehabilitation, Stroke

MUSCLE WEAKNESS ARISING both from disuse and from upper motoneuron inhibition¹ is a common consequence of stroke², ³ and is manifested by a complete or partial inability to generate force. The side of the body contralateral to the brain lesion (affected side) is predominantly weaker than the ipsilateral side (less affected side).⁴ Commonly, muscle strength of the affected lower extremity is reduced from 23% to 90% of the strength on the less affected side.⁵ However, recently evidence is emerging that muscle weakness also occurs on the less affected side.⁴ Muscle strength is an important construct to study in stroke because strength is required for the performance of basic activities of daily living, such as walking and stair climbing.⁶, ⁷, ⁸ Methods to assess muscle strength vary, such as isokinetic dynamometry, manual muscle testing, and hand-held dynamometry. With isokinetic dynamometry, the limb is maintained at a constant speed throughout the testing maneuvers and muscular torque is measured at different speeds (0°/s to 300°/s). There is a standard protocol to follow for each muscle group and, although accurate, this method of testing requires specialized expensive equipment and an extensive amount of training on the equipment.

Manual muscle testing is the most widely used technique for assessing muscle strength in a clinical setting. There is a standard protocol for assessment of muscle strength. It is inexpensive, because equipment is not required; however, an extensive amount of training is required to obtain a reproducible assessment of muscle strength across time and across evaluators. A disadvantage of using this method is that muscle strength is rated on an ordinal scale rather than measured directly.

Hand-held dynamometry is another method of manually assessing muscle strength. It provides a measure in pounds or kilograms or newtons. The equipment is much more affordable than the isokinetic dynamometer. Training is required for use of this equipment; however, the training is more practice than technical. The disadvantage of using hand-held dynamometry in a stroke population is that standard protocols, such as described by Bohannon,⁹ are not always applied and many clinicians revert to assessing patients in positions that are most convenient. Despite the disadvantages of using hand-held dynamometry, it seems to be the most appropriate method of assessing muscle strength in a clinical setting.

When hand-held dynamometry is used clinically, the strength of several muscle groups may be measured in several different positions. The amount of data accumulated as a result of muscle testing is often mystifying to clinicians because it is difficult to summarize strength for an individual when different muscles have been assessed in different positions. To condense these measures, a method to summarize strength may be useful. For the purposes of research, multiple measures render statistical analyses highly complex because of nonindependence of these measurements. Studies that have assessed the strength of the lower extremity, among persons with stroke, have presented results for 1 to 7 different muscles in 1 to 3 different positions, along with 1 to 3 trials.⁵, ⁶, ⁷, ⁸, ¹⁰, ¹¹, ¹², ¹³, ¹⁴, ¹⁵, ¹⁶, ¹⁷, ¹⁸, ¹⁹, ²⁰, ²¹, ²², ²³, ²⁴, ²⁵, ²⁶, ²⁷

In studies of stroke recovery, strength can be either the outcome (dependent variable)⁶, ¹⁴, ¹⁷, ²¹, ²⁵ or the exposure (independent variable).¹⁰, ¹¹, ¹², ¹³, ¹⁶, ¹⁸, ¹⁹, ²⁰, ²², ²⁶ In both of these cases, the probability of finding an association increases with the number of associations tested; statistical methods available to deal with this problem, such as a Bonferroni adjustment, are very conservative and result in hypothesis testing at an overly small P value.²⁸ This type of correction for multiple comparisons increases the risk of a type II error and the study is likely to fail to reject the null hypothesis.

When strength is the outcome, the issue of multiple outcomes needs to be addressed statistically. There are numerous statistical methods for use with multiple outcomes; factor weights, z scores, average ranks, multivariate analysis of variance and generalized estimating equations are available techniques.²⁹ In the rehabilitation literature, 2 methods predominate for creating global indices: factor weights and z-score transformation. Factor weights arising from a factor analysis or principal component analysis provide a weighting for each element that reflects the importance of that element to the global construct that is theoretically measured by the collection of elements. By using the factor weight, a linear combination of the weighted elements can be used as a method of deriving a global index of the construct. The z-score approach provides an estimate of how many standard deviations an individual’s value on an element differs from the group mean for that element. This transformation expresses each element in the same units so that a global index can be created by simple addition. The z-score method is sample dependent and is useful for purposes of research. The factor weighting approach is more generalizable because the factor weightings from 1 sample could be applied to another sample provided the 2 samples are similar on important characteristics.²⁹ For example, factor weightings derived from a large population based sample of the U.S. general population are used to create 2 summary scales from 8 subscales of the Medical Outcomes Study 36-Item Short-Form Health Survey (SF-36). These factor weightings are applied when calculating individual scores for any patient using the SF-36.³⁰

Other than the SF-36, few rehabilitation studies deal statistically with the issue of multiple outcomes. When assessing muscle strength, researchers have either selected only a few muscles to test or have used all of the muscle strength variables without considering their nonindependence. When numerous trials have been carried out for 1 muscle, strategies have been to take the best trial or an average. There have been studies that attempted to summarize muscle strength data,⁸, ¹⁴, ²⁴ but neither of the 2 above mentioned statistical methods was used. In the study by Chu et al,¹⁴ the measures of strength for the muscles on 1 side of the body were summed. A study by Kim et al²⁴ performed a similar composite score in muscle strength, but they summed the change in muscle strength before and after an intervention for each side. Teixeira-Salmela et al⁸ summed the muscle strength of both sides of the body together. This assumes that each muscle contributes equally to the global strength index, which may not be the case.

This issue of multiple rating of strength for a number of muscles in different positions is a feature common to all muscle testing methods; hence, a statistical way of summarizing this complex data would be helpful for research and clinically for summarizing strength in order to monitor change over time. Therefore the objective of this study was to develop a position-specific index of muscle strength to be used for assessing persons with stroke.

Back to Article Outline

Methods 

Participants 

A total of 63 persons with chronic stroke participated in this cross-sectional study. We recruited the subjects from among a pool of persons participating in 2 ongoing research studies; subjects were at least 3 months and no more than 1 year poststroke. The studies from which subjects were drawn excluded persons with cognitive deficits and receptive aphasia. For this substudy, subjects were eligible if they could walk independently for 10m with or without an aid, and were discharged from active rehabilitation. Subjects were excluded if they had any other neuromuscular conditions, failed a cardiology stress test, or, at the time of assessment, had uncontrolled hypertension (currently diagnosed as hypertensive and not on medications, or on medications but still hypertensive).

Main Outcome Measure 

We assessed muscle strength using hand-held dynamometry with the Nicholas manual muscle tester.^a The muscle testing was performed by the same evaluator (CM) for all subjects. Six muscle groups of the lower extremity were tested bilaterally using 3 trials of the hip flexors, hip extensors, knee flexors, knee extensors, ankle plantarflexors (gastrocnemius, soleus), and dorsiflexors. Muscle strength was assessed in a combination of gravity related positions, following a standardized protocol; each muscle group was tested against gravity, alternate against gravity, and with gravity eliminated positions. The intrarater reliability was high, ranging from .87 to .99. Table 1 provides details regarding subject positions. The break test method was used. The evaluator must push against the subject’s limb until the maximal muscular effort is overcome and the joint gives way.³¹ A total of 96 data points (16 different positions × 3 trials × 2 sides) were obtained for each person. The best score of the 3 trials was used for the analysis.

Table 1. Description of Muscle Testing Position

Data Analysis 

We used basic descriptive statistics to describe the characteristics of the sample. Missing data arose because some persons were not comfortable in some of the testing positions. This occurred in less than 10% of the measures. In these circumstances values for the missing data elements were imputed using a linear regression model. For example, a regression equation was determined for the hip extensors in the against gravity, alternate against gravity, and gravity eliminated positions. If the subject’s against gravity position was missing, it was imputed based on the linear relationship to the other 2 positions. This was performed for each group of muscles. For each muscle group, paired t tests were performed to assess if there were differences in strength between the different positions.

We used principal component analysis (PCA), with varimax rotation, to create a global index of strength. The goal of PCA is to summarize the patterns of correlations of the elements, to reduce the number of elements to a minimal number of factors.³² It can be further defined as a linear combination of optimally weighted elements that explains a maximal amount of variance in the data set. There is the same number of principal components as elements, but ideally, the first principal component should explain at least 68% of the variability in the observed data.³²

PCA yields eigenvalues, which reflect the amount of variance that is being accounted for by each component. Every variable contributes 1 unit of variance to the total variance of the data set. A component with an eigenvalue of greater than 1 means that the component is explaining a greater amount of variability than had been contributed by 1 other element. Therefore, components with an eigenvalue of 1 and greater are retained. Each element in the component receives a loading. The high loading elements, in a given component, will tend to have something in common. These loadings are used to determine the weight that each element contributes for that component.³²

Back to Article Outline

Results 

A total of 76 persons were approached; 8 did not meet the eligibility criteria and 5 refused to participate. The average age of the sample was 67 years and 68% were men. Forty percent of subjects had right hemiparesis, 51% of subjects had left hemiparesis, and 9% had no hemiparesis. The average onset since stroke was 4 months. The average gait speed was .89m/s and distance walked was 277m on the 6-minute walk test (6MWT). The average Berg Balance Scale was 48 out of a score of 56.

The mean muscle strength for the 3 trials for each muscle group is presented in table 2, according to affected and unaffected side and position. The t tests comparing against gravity tests and alternate against gravity tests were not significant, except for the hip flexors on both sides. The t tests comparing against gravity tests to gravity eliminated tests were significant except for hip flexors on both sides and affected side ankle dorsiflexors.

Table 2. Average Muscle Strength of 63 Subjects According to Side and Position of Testing

NOTE. Values are mean kilograms ± standard deviation (SD) and range.

Abbreviations: DF, dorsiflexor; Gastroc, gastrocnemius; PF, plantarflexor.

For the tests in gravity related positions, a single factor for each side emerged with factor loadings ranging from .70 to .92. For the tests in gravity eliminated positions, 1 significant factor emerged for the affected side with factor loadings all around .80 except for dorsiflexion, which loaded more weakly on this factor (.59). For the unaffected side in gravity eliminated positions, 2 factors emerged, 1 for hip and knee, and 1 for the muscles of the ankle.

Based on these results, it is statistically justified to create 5 indices. Position-specific indices of muscle strength were created by weighting each muscle by the factor loading and adding them. Index 1 (affected gravity related) and index 3 (unaffected gravity related) include the muscle strength for the hip flexors and extensors in both the against gravity and alternate against gravity positions. Because it would not make sense to allow these muscle groups to contribute twice to the index and in a clinical setting, it is likely that only 1 of these positions would be chosen, index 1 was further divided into index 1a, which includes the muscle strength for the affected against gravity positions and index 1b, which includes the affected alternate against gravity positions of the hip flexors and extensors. The same is done for index 3, therefore resulting in a total of 7 different position-specific indices depending on which positions are used for muscle testing. Table 3 provides a summary of the indices that were developed. The average score, intercorrelations and factor loadings of the individual and the position-specific indices are provided in table 4. Essentially, for the affected side, the weights for the hip flexors and hip extensors in the against gravity and alternate against gravity positions are identical indicating that either of these positions could be used to assess a person with stroke. Therefore, to create index 1, the formula would be:

Furthermore, the indices were found to correlate moderately to highly with each other (r² range, .51–.99), allowing for the 7 indices to be further analyzed using the PCA. The PCA resulted in 1 common factor, indicating that the 7 indices can be combined into 1 global index.

Table 3. Equations for Creating the Position-Specific Indices of Muscle Strength of the Lower Extremity

Abbreviation: NA, not applicable.

Legend: Index 1a, affected gravity related against; index 1b, affected gravity related alternate; index 2, affected gravity eliminated; index 3a, unaffected gravity related against; index 3b, unaffected gravity related alternate; index 4, unaffected gravity eliminated hip and knee; index 5, unaffected gravity eliminated ankle.

Table 4. Mean Score, Correlation, and Factor Loadings of Indices

NOTE. Values are mean ± SD or as otherwise noted.

⁎Each index is created as outlined in table 3.

†P<.001.

The gravity related muscles of the affected side, were, on average, 85% (index 1 / index 3) of the muscle strength of the unaffected side with a range from 37% to 157%, depending on position. In other words, the affected gravity related muscles are, on average, 15% weaker than the unaffected independent of testing position. The affected gravity-eliminated muscle strength was, on average, 91% (index 2 / index 4 + index 5) of the strength of the unaffected side (range, 53%–121%). The affected muscles perform slightly better in gravity eliminated positions because they are only 9% weaker on the affected side in comparison to the gravity related muscles that are 15% weaker.

Back to Article Outline

Discussion 

This study addressed the methodologic issue of how to summarize multiple data points that relate to 1 construct, namely, strength of different muscle groups assessed in several positions. The statistical approach was a PCA and what emerged was that the 96 elements (16 different positions × 3 trials × 2 sides) could be summarized by 5 indices, 2 for the affected limb, and 3 for the unaffected limb. It can be further divided into 7 indices by separating the against and alternate against gravity muscle strength. For the affected side, index 1 summed the strength with the gravity related and either gravity position could be used interchangeably. Index 2 summed the gravity eliminated. Index 3 is a summary of the unaffected gravity related muscles. The gravity eliminated muscle strength of the unaffected side is separated into hip and knee (index 4) and ankle (index 5).

A second finding is that these indices could themselves be summarized into 1 global measure, a finding that is probably more relevant for research than for clinical practice. There are advantages of developing the indices. Therapists can decide which muscles to assess depending on the functional level of the persons assessed. For example, for persons with stroke who are not functional in the upright, against gravity, position, that is, in standing or walking, therapists could assess strength in the gravity-eliminated position and have a summary index of overall strength. For the unaffected side, the strength of the hip and knee can be considered together but strength of the ankle should be considered separately.

Another finding from this study is that the muscles tested contribute almost equally to the global index of muscle strength. It would follow that the assessment of any one muscle provides almost as much information as all of the muscles. Clinically, an evaluator could choose to assess, for example, only the strength of the quadriceps. However, it is known that different muscles are more or less implicated in different functional tasks,³¹ so that the choice would be task-specific. Research identifying those muscles that are most implicated in specific tasks is needed. The global index approach provides a form of “screening” for the relationship between strength and functional task. If the global index is significant, identifying the specific muscles implicated in the task could be the subject of research investigation. This study provides a global index which could be used by other investigators interested in the link between specific muscles and functional tasks.

Two issues affect strength in different positions. In general, more force is generated in positions where gravity is eliminated because all of the motor units are used to generate force against the testing device, whereas with gravity some of the motor units are used to maintain the upright position. However, gravity eliminated positions are also less stable and so the tester must be vigilant in ensuring stability. Our data demonstrate the phenomenon of muscle force being lower when tested in the gravity related positions rather than the gravity eliminated, with the exception of the knee extensors. This is likely due to the difficulty of stabilizing the knee extensor in the gravity eliminated position.

The sample of this study is somewhat typical of patients with stroke at this stage, because the distance walked during the 6MWT is comparable to subjects participating in other studies.³³, ³⁴, ³⁵ The average gait speed of this sample was .89m/s and the average distance walked during the 6MWT was 277m. The average degree of weakness for the affected gravity related muscle strength was 15% and 9% for the gravity eliminated muscle strength in comparison to the unaffected side. The results and conclusions from this study can be generalized to the stroke population who have the capacity to walk up to a distance of 623m in 6 minutes.

Back to Article Outline

Conclusions 

Creating summary indices for muscle strength may be a statistically efficient way of dealing with multiple data points. Clinically, this method would also provide a way of providing a measure of strength of the entire lower limb at 1 or more points in time. This was a small study, so the factor loadings may not be stable, and therefore it would be important to repeat this method in other samples and eventually accumulate enough data to provide population muscle position-specific weights.

Supplier

Back to Article Outline

References 

1. Shamay S, Shepherd R. Weakness in patients with stroke: implications for strength training in neurorehabilitation. Phys Ther Rev. 2000;5:227–238
   • View In Article
2. Bourbonnais D, Vanden Noven S. Weakness in patients with hemiparesis. Am J Occup Ther. 1989;43:313–319
   • View In Article
   • MEDLINE
3. Mayo NE. Epidemiology and recovery. Phys Med Rehabil State Art Rev. 1993;7:1–25
   • View In Article
4. Patten C, Lexell J, Brown HE. Weakness and strength training in persons with poststroke hemiplegia: rationale, method, and efficacy. J Rehabil Res Dev. 2004;41:293–312
   • View In Article
   • MEDLINE
   • CrossRef
5. Adams RW, Gandevia SC, Skuse NF. The distribution of muscle weakness in upper motoneuron lesions affecting the lower limb. Brain. 1990;113(Pt 5):1459–1476
   • View In Article
6. 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
7. Nadeau S, Arsenault AB, Gravel D, Bourbonnais D. Analysis of the clinical factors determining natural and maximal speeds in adults with stroke. Am J Phys Med Rehabil. 1999;78:123–130
   • View In Article
   • MEDLINE
   • CrossRef
8. Teixeira-Salmela LF, Olney SJ, Nadeau S, Brouwer B. Muscle strengthening and physical conditioning to reduce impairment and disability in chronic stroke survivors. Arch Phys Med Rehabil. 1999;80:1211–1218
   • View In Article
   • Abstract
   • Full-Text PDF (1203 KB)
   • CrossRef
9. Bohannon RW. Reference values for extremity muscle strength obtained by hand-held dynamometry from adults aged 20 to 79 years. Arch Phys Med Rehabil. 1997;78:26–32
   • View In Article
   • Abstract
   • Full-Text PDF (787 KB)
   • CrossRef
10. Hsu AL, Tang PF, Jan MH. Test-retest reliability of isokinetic muscle strength of the lower extremities in patients with stroke. Arch Phys Med Rehabil. 2002;83:1130–1137
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (179 KB)
    • CrossRef
11. Eng JJ, Kim CM, Macintyre DL. Reliability of lower extremity strength measures in persons with chronic stroke. Arch Phys Med Rehabil. 2002;83:322–328
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (98 KB)
    • CrossRef
12. Sakai T, Tanaka K, Holland GJ. Functional and locomotive characteristics of stroke survivors in Japanese community-based rehabilitation. Am J Phys Med Rehabil. 2002;81:675–683
    • View In Article
    • MEDLINE
    • CrossRef
13. Bohannon RW, Andrews AW. Interrater reliability of hand-held dynamometry. Phys Ther. 1987;67:931–933
    • View In Article
    • MEDLINE
14. Chu KS, Eng JJ, Dawson AS, Harris JE, Ozkaplan A, Gylfadottir S. Water-based exercise for cardiovascular fitness in people with chronic stroke: a randomized controlled trial. Arch Phys Med Rehabil. 2004;85:870–874
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (100 KB)
    • CrossRef
15. Pohl PS, Duncan PW, Perera S, et al. Influence of stroke-related impairments on performance in 6-minute walk test. J Rehabil Res Dev. 2002;39:439–444
    • View In Article
16. Pohl PS, Perera S, Duncan PW, Maletsky R, Whitman R, Studenski S. Gains in distance walking in a 3-month follow-up poststroke: what changes?. Neurorehabil Neural Repair. 2004;18:30–36
    • View In Article
    • MEDLINE
17. Andrews AW, Bohannon RW. Distribution of muscle strength impairments following stroke. Clin Rehabil. 2000;14:79–87
    • View In Article
    • MEDLINE
    • CrossRef
18. Hsu AL, Tang PF, Jan MH. Analysis of impairments influencing gait velocity and asymmetry of hemiplegic patients after mild to moderate stroke. Arch Phys Med Rehabil. 2003;84:1185–1193
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (92 KB)
    • CrossRef
19. Newham DJ, Hsiao SF. Knee muscle isometric strength, voluntary activation and antagonist co-contraction in the first six months after stroke. Disabil Rehabil. 2001;23:379–386
    • View In Article
    • MEDLINE
    • CrossRef
20. Suzuki K, Nakamura R, Yamada Y, Handa T. Determinants of maximum walking speed in hemiparetic stroke patients. Tohoku J Exp Med. 1990;162:337–344
    • View In Article
    • MEDLINE
    • CrossRef
21. Tripp EJ, Harris SR. Test-retest reliability of isokinetic knee extension and flexion torque measurements in persons with spastic hemiparesis. Phys Ther. 1991;71:390–396
    • View In Article
    • MEDLINE
22. Bohannon RW. Gait performance of hemiparetic stroke patients: selected variables. Arch Phys Med Rehabil. 1987;68:777–781
    • View In Article
    • MEDLINE
23. Nadeau S, Gravel D, Arsenault AB, Bourbonnais D. Plantarflexor weakness as a limiting factor of gait speed in stroke subjects and the compensating role of hip flexors. Clin Biomech. 1999;14:125–135
    • View In Article
24. Kim CM, Eng JJ, Macintyre DL, Dawson AS. Effects of isokinetic strength training on walking in persons with stroke: a double-blind controlled pilot study. J Stroke Cerebrovasc Dis. 2001;10:265–273
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (105 KB)
    • CrossRef
25. Pang MY, Eng JJ, Dawson AS. Relationship between ambulatory capacity and cardiorespiratory fitness in chronic stroke: influence of stroke-specific impairments. Chest. 2005;127:495–501
    • View In Article
    • MEDLINE
    • CrossRef
26. Canning CG, Ada L, Adams R, O’Dwyer NJ. Loss of strength contributes more to physical disability after stroke than loss of dexterity. Clin Rehabil. 2004;18:300–308
    • View In Article
    • MEDLINE
    • CrossRef
27. Bohannon RW. Strength of lower limb related to gait velocity and cadence in stroke patients. Physiother Can. 1986;38:204–206
    • View In Article
    • CrossRef
28. Moore D, McCabe G. Introduction to the practice of statistics. 4th ed. New York: W.H. Freeman; 2003;
    • View In Article
29. Fairclough DL. Summary measures and statistics for comparison of quality of life in a clinical trial of cancer therapy. Stat Med. 1997;16:1197–1209
    • View In Article
    • MEDLINE
30. Ware JE, Sherbourne CD. The MOS 36-item Short-Form Health Survey (SF-36): I. Conceptual framework and item selection. Med Care. 1992;30:473–481
    • View In Article
    • MEDLINE
    • CrossRef
31. Bohannon RW. Make tests and break tests of elbow flexor muscle strength. Phys Ther. 1988;68:193–194
    • View In Article
    • MEDLINE
32. Tabachnick BG, Fidell LS. Using multivariate statistics. 4th ed. Boston: Allyn & Bacon; 2001;
    • View In Article
33. 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
34. Duncan P, Studenski S, Richards L, et al. Randomized clinical trial of therapeutic exercise in subacute stroke. Stroke. 2003;34:2173–2180
    • View In Article
    • CrossRef
35. Salbach NM, Mayo NE, Wood-Dauphinee S, et al. A task-oriented intervention enhances walking distance and speed in the first year post stroke: a randomized controlled trial. Clin Rehabil. 2004;18:509–519
    • View In Article
    • MEDLINE
    • CrossRef

• a Model 01160; Lafayette Instrument, PO Box 5729, 3700 Sagamore Pkwy N, Lafayette, IN 47903.