PATHOLOGY

strokearticle1.pdf

Home >Applied Sciences homework help >PATHOLOGY

The Stroke Rehabilitation Assessment of Movement (STREAM): A Comparison With Other Measures Used to Evaluate Effects of Stroke and Rehabilitation

Background and Purpose. The Stroke Rehabilitation Assessment of Movement (STREAM) is a relatively new measure of voluntary move- ment and basic mobility. The main objectives of this study were: (1) to examine the relationship of the STREAM to other measures of impairment and disability and (2) to compare its usefulness for evaluating effects of stroke and rehabilitation and for assessing change over time with that of other measures of impairment and disability. Subjects and Methods. The performance of 63 patients with acute stroke on the STREAM and other measures of impairment and disability was evaluated during the first week after stroke and 4 weeks and 3 months later. Results. Scores on the STREAM were associated with scores on the Box and Block test, Balance Scale, Barthel Index, gait speed, and the Timed “Up & Go” Test (with Pearson correlation coefficients ranging from .57 to .80) and were associated with catego- ries of the Barthel Index and Balance Scale. The STREAM’s ability to predict discharge destination from the acute care hospital, as well as to predict gait speed and Barthel Index scores at 3 months poststroke, was comparable to that of other commonly used measures. Standardized response mean estimates provided supporting evidence for the ability of the STREAM to reflect change over time. Discussion and Conclu- sion. The results obtained with the STREAM, as compared with other measures of impairment and disability in people with stroke, suggest that it may be useful in clinical practice and research. [Ahmed S, Mayo NE, Higgins J, et al. The Stroke Rehabilitation Assessment of Move- ment (STREAM): a comparison with other measures used to evaluate effects of stroke and rehabilitation. Phys Ther. 2003;83:617– 630.]

Key Words: Measurement, Motor recovery, Outcome measure, Psychometrics, Stroke.

Sara Ahmed, Nancy E Mayo, Johanne Higgins, Nancy M Salbach, Lois Finch, Sharon L Wood-Dauphinée

Physical Therapy . Volume 83 . Number 7 . July 2003 617

Re se

ar ch

Re po

rt �

��

D ow

nloaded from https://academ

ic.oup.com /ptj/article/83/7/617/2805284 by guest on 10 M

arch 2021

T he evaluation of motor recovery is a cornerstone of the assessment of people with stroke. Most measurement instruments have emerged from theoretical frameworks developed to fit patterns

of motor recovery observed in selected samples of peo- ple recovering from strokes. Some measures were based on theories that have been questioned, such as the assumption that recovery occurs in a predictable stereo- typed pattern performed within flexor and extensor synergies.1– 4 The theoretical basis of existing measures may explain why, according to a 1992 Canadian survey,5,6 published instruments for motor evaluation following stroke were used routinely in less than 5% of physical therapy departments. Other reasons for nonuse, we believe, may be related to practicality, including the time to administer the test, the dependence on equipment, and the complexity of the scoring scheme.

Given what we viewed as the limitations of many mea- sures of motor recovery, researchers and clinicians devel- oped what they believe is a more “user-friendly” instru- ment called the Stroke Rehabilitation Assessment of Movement (STREAM).7 The content and format of the instrument were created in an effort to minimize barri-

ers to routine clinical use. The STREAM was developed as an outcome measure that could be used to monitor the re-emergence of voluntary movement and basic mobility. Items in the original STREAM were based on clinical experience of physical therapists working in stroke rehabilitation and existing instruments. Further content validation was carried out by 2 consensus panels that made recommendations based on their collective clinical experience. The first and second panels con- sisted of 11 and 9 physical therapists, respectively, rep- resenting all phases of stroke rehabilitation, including acute care, inpatient and outpatient rehabilitation, and long-term care. All therapists had more than 1 year of clinical experience. The panels produced an intermedi- ate test version of the STREAM made up of 43 items. This intermediate version then underwent preliminary reliability and internal consistency testing, and item reduction was then carried out.7

The current version of the STREAM7,8 contains 30 items divided among 3 subscales: 10 items for voluntary motor ability of the upper extremity (UE), 10 items for volun- tary motor ability of the lower extremity (LE), and 10 items for basic mobility. Examples of items on the

S Ahmed, MSc, BSc (PT), is a doctoral student in the Department of Epidemiology and Biostatistics, McGill University, Montreal, Quebec, Canada.

NE Mayo, PhD, BSc (PT), is Quebec Senior Research Scientist, Royal Victoria Hospital, Montreal, Quebec, Canada, and Associate Professor, Faculty of Medicine, School of Physical and Occupational Therapy, McGill University. Address correspondence to Dr Mayo at Royal Victoria Hospital, Division of Clinical Epidemiology, 687 Pine Ave W, Ross 4.29, Montreal, Quebec, Canada H3A 1A1 ([email protected]).

J Higgins, MSc, BSc (OT), is a doctoral student in the School of Physical and Occupational Therapy, Faculty of Medicine, McGill University.

NM Salbach, MSc, BSc (PT), is a doctoral student in the Department of Epidemiology and Biostatistics, McGill University.

L Finch, MSc, BSc (PT), is a doctoral student in the School of Physical and Occupational Therapy, Faculty of Medicine, McGill University. She was Physical Therapist, Royal Victoria Hospital, at the time of this study.

SL Wood-Dauphinée, PhD, BSc (PT), is Professor, School of Physical and Occupational Therapy, and Professor, Faculty of Medicine, Department of Epidemiology and Biostatistics, McGill University.

Ms Ahmed, Ms Salbach, and Ms Higgins were graduate students in the School of Physical and Occupational Therapy, Faculty of Medicine, McGill University, when this study was completed.

Ms Ahmed, Dr Mayo, and Ms Finch provided concept/research design. Ms Ahmed and Dr Mayo provided writing and data analysis. Ms Ahmed, Ms Higgins, and Ms Salbach provided data collection. Dr Mayo provided fund procurement, subjects, facilities/equipment, and institutional liaisons. Dr Mayo, Ms Higgins, Ms Salbach, Ms Finch, and Dr Wood-Dauphinée provided consultation (including review of manuscript before submission). The authors thank the research nurses Susan Anderson, Angela Andrianakis, Rosemary Hudson, and Lisa Wadup for patient recruitment and Claudette Corrigan for assistance in running the study. They also acknowledge the assistance of Adrian Levy, Susan Scott, and Judy Soicher in the analysis of data.

Ethical approval for this study was obtained from the ethics committees of each hospital involved and from the Institutional Review Board of McGill University.

This research was presented, in part, in poster format at the Réseau Provincial de Recherche en Adaptaton–Réadaptation Conference; June 4 – 6, 1998; Quebec City, Quebec, Canada.

In support of this research, Ms Ahmed received a fellowship in gerontology from the Physiotherapy Foundation of Canada.

This article was submitted March 1, 2002, and was accepted March 24, 2003.

618 . Ahmed et al Physical Therapy . Volume 83 . Number 7 . July 2003

D ow

nloaded from https://academ

ic.oup.com /ptj/article/83/7/617/2805284 by guest on 10 M

arch 2021

STREAM include protraction of the scapula in a supine position, flexion of the hip and the knee in a supine position, and rolling onto one side from a supine position. A 3-point ordinal scale is used for scoring voluntary movement of the limbs, and a 4-point ordinal scale is used for basic mobility. The extra category for basic mobility was added to allow for one of the score choices to be independence in the activity without the help of an aid (eg, walking aid, splints). The quality of movement for the UE and LE is also scored on a 3-point scale, but it is not reflected in the final score. A total score for each subscale is calculated, out of 20 points for the UE and LE subscales and 30 points for basic mobility. To allow for the possibility that occasionally an item cannot be scored, the subscales are converted to a percentage score out of 100 even though the scores are not interval based, and the total score is calculated as an average of scores obtained for the 3 subscales. The STREAM requires approximately 15 minutes to admin- ister.

Internal consistency of the STREAM was assessed on 26 individuals with stroke (20 subjects from a rehabilitation center and 6 additional subjects with a low level of voluntary movement as indicated by a score of less than 30 on the STREAM). Cronbach alphas, which reflect how well the items of the scale relate to each other, were .98 for the subscales and the complete STREAM. Twenty experienced physical therapists (mean years of experi- ence�5, SD�2.1, range�2–9) who had worked with individuals with stroke participated in the interrater and intrarater reliability testing of the STREAM through videotaped assessments. For this reliability study, the therapists were trained in a 2-hour training session (led by K Daley, one of the original developers of the new version of the STREAM) where the STREAM test manual was discussed, a videotaped STREAM assessment was scored, and the scores were discussed. Generalizability coefficients, which indicate the extent to which a person can generalize results to any rater, subject, or occasion, for intrarater reliability ranged from .96 to .999 for the subscale and total scores, and generalizability coeffi- cients for interrater reliability ranged from .98 to .995.8

The STREAM was developed to fill what was perceived by its developers as a need in the early 1990s, and we believe it had good measurement properties. When used in a clinical trial that assessed the impact of body weight support on functional outcomes poststroke, it was able to reflect change mediated through treadmill training with body weight support.9 The potential of the STREAM as a clinical evaluation tool warranted further examination of how it relates to other stroke outcome measures commonly used in clinical practice.

The aim of our study was to assess how the STREAM compared with other measures of impairment and dis- ability in people following a stroke. The objectives were:

1. To determine the degree of association between the STREAM and other measures of impairment and disability during the first 3 months poststroke.

2. To determine if the STREAM could be used to differentiate different levels of performance on mea- sures of balance and independence of activities of daily living immediately following a stroke and at 5 weeks and 3 months poststroke.

3. To assess whether the STREAM scores obtained dur- ing the first week poststroke could be used to predict discharge destination and to compare this ability with that of the Barthel Index.

4. To assess whether the STREAM and other standard measures used to evaluate the effects of stroke and rehabilitation during the first week poststroke could be used to predict independence in activities of daily living and gait speed scores 3 months poststroke.

5. To examine the extent to which scores on the STREAM reflect change over time as compared with other measures used to evaluate the effects of stroke and rehabilitation.

Method

Study Design This investigation was part of a longitudinal cohort study designed to examine the recovery of UE and LE function following a stroke. The methods, as well as the profile of recovery poststroke, have been reported by Salbach et al10 for the LEs and by Higgins ( J Higgins, unpublished research) for the UEs. Recovery from stroke is most rapid in the first few weeks following stroke, but contin- ues up to 3 months following stroke.11,12 Because most people are expected to show improvements during the acute period after a stroke, we believe this is an impor- tant time period during which to assess the usefulness of the STREAM. A cohort of patients with residual physical deficits following an acute stroke was followed over a 3-month period. Patients were evaluated during the first week poststroke and 4 weeks and 3 months later. During each evaluation, patients were assessed with measures of impairment and disability.

Subjects Consecutive patients with a first-time stroke according to clinical and radiological criteria who had been admitted to 1 of 5 large urban acute care university teaching centers in Montreal, Canada, were identified. Patients

Physical Therapy . Volume 83 . Number 7 . July 2003 Ahmed et al . 619

��

D ow

nloaded from https://academ

ic.oup.com /ptj/article/83/7/617/2805284 by guest on 10 M

arch 2021

were considered eligible for participation in the study if they had no apparent cognitive impairment and if they demonstrated stroke-related physical deficits of the UEs and LEs. Patients were excluded if they had completely recovered from the stroke, if they had severe language deficits, or if they had comorbid conditions such as disabling arthritis, Parkinson disease, amputation, or severe cardiovascular disease. Each subject was required to sign a consent form before being enrolled in the study.

In total, 357 consecutive patients were identified for the study. Of these, 189 patients met the inclusion criteria and were considered eligible for participation in the study, 78 patients were approached, and 67 patients consented to participate. The remaining patients could not be recruited because they were already enrolled in other research projects or were unavailable at the time of the first evaluation. Of the 67 consenting patients, sufficient data were obtained for 63 subjects. Table 1 summarizes the clinical and demographic characteristics of the final participants and nonparticipants for this study.

Measurement Once consent was obtained, information related to the occurrence of the stroke, the patient’s medical history,

and sociodemographics was recorded directly from the medical records. Subjects were classified according to stroke severity using the Canadian Neurological Scale (CNS).13 Three of the investigators (SA, JH, and NMS) served as evaluators throughout the study. The evalua- tors participated in a training session, and any difficulties with scoring items were discussed to obtain a consensus. In addition to the STREAM, the following instruments were administered at each evaluation.

The CNS13 measures neurological status in patients with stroke and is divided into 2 sections: mentation, and motor function. Concurrent validity was evaluated by comparing the CNS with a standard neurologic evalua- tion, resulting in Spearman rank correlation coefficients ranging from .574 to .775 (P�.001). The CNS has been used to classify patients as having a mild (CNS�11), moderate (9�CNS�11), or severe (CNS�9) stroke.13 A Cronbach alpha of .79 was reported for the internal consistency of the CNS among a sample of 155 individ- uals with stroke.13 Interrater reliability testing resulted in a Spearman correlation coefficient of .924 for the entire measure, and kappa statistics ranged from .535 to 1.00 for the interrater reliability of individual items.13

The Barthel Index14 is a self-proxy questionnaire that is designed to measure 3 categories of function: self-care, continence of bowel and bladder, and mobility. It is composed of 10 items and has a maximum score of 100.15 There is support for the reliability14,16 –18 and validity15,19 –21 of Barthel Index scores. Wolfe et al22 reported a test-retest kappa value of .89 among 50 patients with stroke, and Roy et al18 reported a Pearson product moment correlation coefficient of .99 for inter- rater reliability for a sample of 25 inpatients in a neurorehabilitation unit. Granger and colleagues15 found that patients who scored between 5 and 40 were less likely to return home than those who scored in the range of 41 to 60. Individuals who scored between 60 and 100 had a shorter hospital length of stay. Cutoff scores of 60 (dependence) and 85 (independence) also have been reported.15

The Balance Scale, developed by Berg and col- leagues,23,24 is a measure that consists of 14 task-oriented items, each scored on a scale from 0 to 4. An intraclass correlation coefficient (ICC) of .97 for intrarater reli- ability was reported for a group of 18 elderly nursing home residents, and an ICC of .98 was reported for interrater reliability for the total score among a group of 35 individuals with stroke.23 There is supporting evi- dence for validity of Balance Scale scores in subjects with stroke.25,26 Pearson product moment correlation coeffi- cients between the Balance Scale and the Barthel Index, the Fugl-Meyer Stroke Assessment Scale, the Timed “Up & Go” Test (TUG), and the Tinetti Balance Scale ranged

Table 1. Demographic and Clinical Characteristics of Study Participants and Eligible Nonparticipantsa

Characteristic Participants (n�63)

Nonparticipants (n�122) P

Age (y) X 67 70 .561 SD 14 13 Range 25–95 34–100

Sex, n (%) Male 39 (62) 67 (55) .363 Female 24 (38) 55 (45)

Side of lesion, n (%)b

Right 31 (49) 46 (38) .623 Left 30 (48) 38 (31) Bilateral 2 (3) 1 (0) Missing 0 37 (30)

Type of stroke, n (%) Ischemic 59 (94) 66 (54) .046 Hemorrhagic 4 (6) 14 (11) Missing 0 42 (34)

Stroke severity, n (%) Mild 12 (19) 44 (36) .007 Moderate 33 (52) 38 (31) Severe 18 (29) 18 (15) Missing 0 22 (18)

a Student t test used for comparison of age; chi-square test used for all other comparisons. b Percentages may not add up to 100 because of rounding.

620 . Ahmed et al Physical Therapy . Volume 83 . Number 7 . July 2003

D ow

nloaded from https://academ

ic.oup.com /ptj/article/83/7/617/2805284 by guest on 10 M

arch 2021

from .67 to .94.26 Balance Scale scores have been divided into 3 groups that roughly correspond to ambulatory status: poor�0 –20, fair�21– 40, and good�41–56.27

Gait speed was timed over a 5-m distance. The starting mark was placed 2 m before the test section, and the stopping mark was placed 2 m after the test section to allow for acceleration and deceleration. Several researchers have estimated the reliability of gait speed measurements obtained in a clinical setting with a stopwatch over distances of 5 m,10 6 m,28 8 m,29 and 10 m.30 Intraclass correlation coefficients for test-retest and interrater reliability have ranged from .89 to 1.00.29 –34 A study conducted by Salbach and col- leagues,10 which compared comfortable and maximum gait speeds for 5- and 10-m distances, indicated that a 5-m comfortable walking speed was the most sensitive to change. Therefore, the 5-m walking distance at a com- fortable pace was chosen for our study.

The TUG35 is considered to be a test of functional mobility. The patient is seated in a chair with armrests, and the time taken to stand up, walk forward 3 m, and return to the seated position is measured. Test-retest reliability and interrater reliability (ICCs of �.98) were demonstrated in 10 elderly institutionalized people rated by 2 therapists on 3 occasions.36 Among 12 people with neurological disorders, coefficients for interrater and intrarater reliability also were high (ICC�.93–.99), and TUG measurements correlated with measurements of walking speed (r �.71–.96).37 Although the correla- tions were high, the numbers of subjects in these studies were very small. Values for elderly individuals ranged from 7 to 10 seconds.35

The Box and Block Test38 is used to measure unilateral gross manual dexterity. This test involves the patient moving as many blocks as possible, one by one, from one compartment of a box to another compartment of equal size within 60 seconds. Desrosiers and colleagues38 examined the test-retest reliability and construct validity of measurements taken with this instrument in a group of elderly people with UE impairment. The ICCs ranged from .89 to .97, and correlations were demonstrated among the Box and Block Test and an upper-limb performance measure (ICC�.80 –.82), and a measure of functional independence (ICC�.42–.54).38

Data Analysis To determine the degree of association between the STREAM and other measures of impairment and disabil- ity, Pearson correlation coefficients were used in our study. Scores of the subscales and the total STREAM were correlated with scores from the Box and Block Test,38 the Balance Scale,23 gait speed,10 the TUG,35 and the Barthel Index.14 Correlations between 0 and .25 were

considered low, those between .25 and .5 were consid- ered fair, those between .5 and .75 were considered moderate, and those greater than .75 were considered strong.39 We expected a moderate correlation between the STREAM and other measures of impairment and disability. All correlations were examined cross- sectionally on data obtained at entry to the study, 4 weeks later, and 3 months poststroke.

We also wanted to assess the ability of the STREAM to be used to differentiate among groups of individuals with stroke on the basis of performance on measures of balance and independence in activities of daily living immediately after stroke and 5 weeks and 3 months after stroke. For this analysis, subjects were grouped accord- ing to scores on the Barthel Index and the Balance Scale. As described in the “Measurement” section, 3 classifica- tion groups were formed (good, fair, and poor) for each measure based on cutoff points from the literature.27 As these groups are based on clinical criteria, the ability of mean scores on the STREAM to differentiate among these groups may reflect the clinical usefulness of the STREAM. An analysis of variance was used to test whether the mean STREAM scores differed across the 3 groups.

To assess the ability of the STREAM to be used to predict discharge destination from the acute care hospital, the probability of being discharged home was examined. Some authors40,41 have identified functional ability, as measured with the Barthel Index, as an important pre- dictor of discharge destination, and we therefore com- pared the predictive ability of the Barthel Index with that of the STREAM. For these analyses, each possible value of the initial STREAM and Barthel Index scores, from 0 to 100, was used as a cutoff point, and the probability of discharge home was calculated for individ- uals with values at or below the cutoff point. The probability of discharge home was then plotted against consecutive cutoff points on the STREAM and the Barthel Index.

To test the ability of initial poststroke STREAM scores to predict Barthel Index scores and gait speed 3 months poststroke, multiple linear regression was used. The predictive ability of the STREAM was compared with that of all other measures of impairment and disability in this study. We used standardized betas, which are interpreted in terms of standard deviation units; for every 1–stan- dard deviation change in the independent variable (initial scores on the STREAM or any of the other measures of impairment and disability), there is a mean increase (of beta) in the dependent variable (Barthel Index or gait speed scores at 3 months poststroke).42 Because a different scale is used for each measurement, the standardized regression coefficient provides an esti-

Physical Therapy . Volume 83 . Number 7 . July 2003 Ahmed et al . 621

��

D ow

nloaded from https://academ

ic.oup.com /ptj/article/83/7/617/2805284 by guest on 10 M

arch 2021

mate that is comparable between the measures. In addition, we examined the R2 , which is the variance in the dependent variable (Barthel Index or gait speed scores at 3 months poststroke) explained by the initial STREAM score (or by the initial score on one of the other measures of impairment and disability: the Box and Block Test, the Balance Scale, gait speed, the TUG, and the Barthel Index) In all models, potential con- founding variables such as age, sex, type and side of lesion, level of cognition, and perceptual neglect were included. Because of the large number of potential confounding variables, this group of variables was first modeled alone, and only the significant confounding variables (P�.05) were retained in the final regression model. Separate regression models were used for each scale (STREAM, Box and Block Test, Balance Scale, gait speed, TUG, and Barthel Index). Residual and diagnos- tic plots were examined to assess for problems with outliers, major deviations from normality, linearity, and multicollinearity for the multiple linear regression and other analyses.

To compare whether the STREAM could be used to reflect change compared with other outcome measures in people with strokes, we used the standardized response mean (SRM) and the 95% confidence interval, calculated using Liang and colleagues’ method.43 The SRM is calculated as the mean change found in a particular measure divided by the standard deviation of the change score for the measure. We made compari- sons with the entire sample as well as within each subgroup (ie, mild, moderate, and severe) classified by scores on the CNS. Comparisons among all of the measures were made for the 3 time intervals: (1) from the initial evaluation to the 5-week evaluation, (2) from the 5-week evaluation to the 3-month evaluation, and (3) from the initial evaluation to the 3-month evaluation. A measure that is useful for assessing change over time should have a small ceiling effect. The ceiling effect of each measure used in this study was estimated by calcu- lating the percentage of individuals who had attained the maximum score for each evaluation.

Thirteen participants were not able to perform any of the items on the Balance Scale during the first scheduled visit. For these subjects, there was no information on the Balance Scale at the subsequent visits as well. Thus, the values for the initial evaluation were set at 0, because this score correctly reflects the subjects’ inability to perform any of the items. For these 13 subjects, values were created for the second and third evaluations by giving them the mean Balance Scale values of the subgroup of subjects who had complete data on the Balance Scale with a similar range of gait speed scores (because the Balance Scale scores were highly correlated with mea- surements of gait speed).

At the initial, 5-week, and 3-month evaluations, there were 17, 7, and 2 individuals, respectively, who were unable to perform the TUG because it was difficult for them to stand from a seated position. Therefore, for most analyses, we considered only whether or not they could perform the test by using a variable called “TUG ability,” which had 2 values (1 and 0) depending on whether the person was able or not able to perform the test. To assess the ability of the TUG to be used to assess change, values for those participants who were unable to perform this test were replaced with twice the maximum score of the entire study sample at each evaluation, because a high score on the TUG reflects worse func- tional mobility.

To assess the impact of the created values, all analyses were performed with and without the created scores. The Statistical Analysis System (Windows version 6.12)* was used.

Results The first evaluation was performed an average of 8 days poststroke (SD�3, range�3–14). There was a mean of 29 days (SD�5, range�19 –50) between the first and second evaluations and a mean of 85 days (SD�17, range�37–124) between the second and last evalua- tions. The mean and median scores for all measures are presented in Table 2. This table shows that mean scores improved over time.

The correlations between the total score of the STREAM and the scores for other measures of impairment and disability ranged from r �.36 to r �.80 for the 3 evalua- tions (Tab. 3). The total and subscale STREAM scores for the 3 evaluations also were correlated with severity of the stroke as measured by the CNS, with correlations ranging from r �.66 to r �.77.

At the time of the initial evaluation, we could use the STREAM to differentiate different levels of performance on the Balance Scale (Tab. 4). The mean scores of the STREAM for the 3 classification groups (good�41–56, fair�21– 40, and poor�0 –20 on the g Balance Scale) were different from each other (P�.05), with a mean difference in the STREAM score of 13.4 between good and fair, and 33.6 between fair and poor (Tab. 4). At 5 weeks poststroke, all means were different (good-fair mean difference�22, P�.05) except between the partic- ipants in the fair and poor groups (fair-poor mean difference�7, P�.05). The sample sizes for these 2 groups at 5 weeks were much smaller due to improve- ments in subjects’ scores, and therefore the power to detect a difference also was reduced.39 After 3 months, there were no patients classified as poor, and a differ-

* SAS Institute Inc, PO Box 8000, Cary, NC 27511.

622 . Ahmed et al Physical Therapy . Volume 83 . Number 7 . July 2003

D ow

nloaded from https://academ

ic.oup.com /ptj/article/83/7/617/2805284 by guest on 10 M

arch 2021

ence of 23.7 was found between those classified as fair and good (P�.0001). For the Barthel Index classifica- tion, the mean difference in the STREAM score between the fair and poor groups was 28.6 and 28.9 at the initial and 5-week evaluations, respectively (P�.05). The mean difference in the STREAM score between the fair and poor groups was 19.1 (P�.05) at the initial evaluation

and 4.1 (P�.05) at the 5-week evaluation. As the sample sizes were much smaller for the fair and poor groups at the 3-month evaluation, none of the mean differences were different from each other.

We examined our ability to use the STREAM to predict discharge home from the acute care hospital (Fig. 1).

Table 2. Performance of Study Subjects (n�63) on Measures of Impairment and Disability at 3 Points in Time Poststrokea

Measure

Initial Evaluation 5-Week Evaluation 3-Month Evaluation

X SD Median Range X SD Median Range X SD Median Range

STREAM Total 75 26.7 86 7–100 86 19.1 94 18–100 89 18.0 97 21–100 UE subscale 73 33.3 90 0–100 85 26.2 100 0–100 88 24.0 100 0–100 LE subscale 75 28.9 85 0–100 86 22.3 95 0–100 90 19.0 100 15–100 Mobility subscale 74 25.9 83 10–100 88 16.4 97 39–100 91 15.0 97 33–100

Box and Block Test Affected UE 25 21.0 27 0–77 36 22.8 43 0–80 41 22.2 46 0–80 Unaffected UE 49 13.8 47 21–81 56 11.8 56 27–84 56 12.7 58 24–83

Barthel Index 72 27.9 85 5–100 86 20.4 100 30–100 92 14.0 100 40–100

Balance Scale Missing values

imputed 34 21.2 39 0–56 44 15.9 52 0–56 48 10.0 52 22–56 Missing values

removed (n�50) 37 18.1 41 0–56 47 11.5 52 9–56 49 8.3 52 22–56

TUG(s) Missing values

imputed 73 87.1 21 7–106 34 49.2 13 7–83 21 25.9 11 7–72 Missing values

removed (n�46) 21 16.5 22 8–106 12 5.3 13 7–83 12 4.0 11 7–72

Gait speed (m/s) 0.55 0.38 0.58 0–1.33 0.82 0.43 0.90 0–1.60 0.85 0.36 0.93 0.13–1.45

a STREAM�Stroke Rehabilitation Assessment of Movement, UE�upper extremity, LE�lower extremity, TUG�Timed “Up & Go” Test.

Table 3. Pearson Correlations for the Stroke Rehabilitation Assessment of Movement (STREAM) Total and Subscale Scores With Other Measures of Impairment and Disability at 3 Points in Time Poststroke (n�63)a

STREAM Evaluation

Box and Block Test (Affected UE)

Box and Block Test (Unaffected UE)

Barthel Index

Balance Scale

TUG Ability

Gait Speed

Total Initial .73 .36 .78 .75 .80 .74 5 weeks .77 .37 .71 .68 .64 .62 3 months .78 .44 .75 .65 .57 .73

UE Initial .78 .31 .67 .57 .69 .56 5 weeks .79 .36 .66 .61 .49 .53 3 months .76 .31 .67 .53 .60 .64

LE Initial .53 .40 .71 .73 .75 .74 5 weeks .64 .29 .59 .55 .59 .55 3 months .70 .30 .63 .55 .51 .65

Mobility Initial .66 .55 .84 .88 .85 .83 5 weeks .69 .40 .75 .71 .57 .65 3 months .66 .40 .82 .78 .62 .76

a UE�upper extremity, LE�lower extremity, TUG�Timed “Up & GO” Test. All correlations significant at the P�.0001 level except for the unaffected UE during the Box and Block Test at all 3 evaluations (P �.025).

Physical Therapy . Volume 83 . Number 7 . July 2003 Ahmed et al . 623

��

D ow

nloaded from https://academ

ic.oup.com /ptj/article/83/7/617/2805284 by guest on 10 M

arch 2021

For this analysis, the probability of being discharged home versus being discharged to a rehabilitation center was plotted against cutoff values for the STREAM. Below a score of 63, the probability of being discharged home was zero. As the STREAM score increased beyond 63, the probability of discharge home increased in almost a linear fashion. For example, 20% of the participants who had a score of 80 or less on the STREAM at the time of the initial evaluation were discharged home after the acute care hospital. As shown in Figure 1, the ability of the STREAM to predict discharge home was similar to that for the Barthel Index.

The ability of the STREAM and all other measures during the first week poststroke to predict gait speed and Barthel Index scores after 3 months was assessed (Tab. 5). The only confounding variables were age and cognition (P�.05), and these variables were included in each of the regression models. The parameter estimates were significant (P�.002–.0001), and the STREAM, dur- ing the first week poststroke, was able to be used to predict gait speed and Barthel Index scores after 3 months. The standardized parameter estimates indi- cated that a 1–standard deviation change on the STREAM resulted in an 8-point increase on the Barthel Index and a 0.22-m/s increase in gait speed at 3 months. The Balance Scale was the strongest predictor of gait speed at 3 months poststroke, followed by initial gait speed measurements, the STREAM, and the Barthel Index. Initial Barthel Index performance was the stron- gest predictor of Barthel Index scores at 3 months

poststroke, followed by the TUG, the STREAM, and the Balance Scale. At the time of the initial evaluation, the STREAM could be used to explain a large proportion of the variability in gait speed (R2�61%), and the Barthel Index (R2�44%) at 3 months, second only to the Balance Scale and the TUG, respectively.

All these analyses also were performed without imputing missing values for the Balance Scale. When the scores of individuals with missing data were removed from the analysis, the mean of the Balance Scale scores increased by 2 to 4 points and the standard deviation decreased by 2 to 4 points for the 3 evaluation periods. When imputed values were left out of the analysis, the pattern of the correlations between the 3 subscales of the STREAM and the Balance Scale at the initial evaluation changed slightly. In addition, our ability to use the Balance Scale to predict gait speed and Barthel Index scores decreased tremendously, with the standardized beta coefficients decreasing to 4 and 0.18, respectively.

The SRMs and their confidence intervals for all mea- sures are shown in Table 6. Over the entire 3 months of the study, the 3 measures most able to reflect change were gait speed (1.15), the total score on the STREAM (0.96), and the Balance Scale (0.94). The measure least able to reflect change was the Box and Block Test for the affected UE (0.24). Over the first 5 weeks, the measures that reflected the largest amount of change were the Box and Block Test for the affected UE (1.3), followed by gait speed (1.05) and the Balance Scale (1.0), and then the

Table 4. Relationship Between Mean Stroke Rehabilitation Assessment of Movement (STREAM) Scores and Balance Scale and Barthel Index Score Classifications as Good, Fair, and Poor

Classification of Balance Scalea and Barthel Indexb

Scores

Initial Evaluation 5-Week Evaluation 3-Month Evaluation

Balance Scale

Barthel Index

Balance Scale

Barthel Index

Balance Scale

Barthel Index

Good X 91.3 88.14 93.5 91.7 94.7 91.6 95% CIc 88–95 84–92 90–97 88–95 92–97 88–95 n 30 42 45 52 49 59

Fair X 77.9 59.5 71.5 62.8 70.1 60.8 95% CI 68–87 41–77 55–88 35–90 56–86 �35–156 n 15 8 10 7 14 3

Poor X 44.3 40.4 64.5 58.7 47.8 95% CI 31–58 24–57 38–92 11–107 n 18 13 8 4 0 1

a Good�41–56, fair�21– 40, poor�0 –20. Significance test for the 3 Balance Scale classifications performed using analysis of variance. P �.05 except between the fair and poor classifications at 5 weeks (P�.05). Significance was calculated using the F statistic with 2 degrees of freedom. These results were the same when the Bonferroni correction (type I error/number of tests [0.05/3]) was used to adjust for multiple tests of comparison. b Good�61–100, fair�41– 60, poor�0 – 40. Significance test for the 3 Barthel Index classifications performed using analysis of variance. P �.05 except for between the fair and poor classifications at 5 weeks and all mean comparisons at 3 months (P�.05). Significance was calculated using the F statistic with 2 degrees of freedom. These results were the same when the Bonferroni correction (type I error/number of tests [0.05/3]) was used to adjust for multiple tests of comparison. c CI�confidence interval.

624 . Ahmed et al Physical Therapy . Volume 83 . Number 7 . July 2003

D ow

nloaded from https://academ

ic.oup.com /ptj/article/83/7/617/2805284 by guest on 10 M

arch 2021

Barthel Index (0.97) and the total score on the STREAM (0.94). The measure that reflected the least amount of change was the LE subscale (0.62) of the STREAM. The SRMs were much lower between 5 weeks and 3 months for all measures, ranging from �0.007 to 0.47. The STREAM had a lower ceiling effect as compared with the Barthel Index and the TUG (Tab. 7). Table 6 also shows the effect of the substitution strategy on the estimates of change for participants who were unable to perform the TUG or the Balance Scale.

Figure 2 presents the SRMs of all measures of outcome between the initial and 5-week evaluations of the study in subjects who sustained a mild, moderate, or severe stroke. In all severity groups, the STREAM was 1 of the 3 measures best able to be used to reflect change. The STREAM was able to reflect the most change throughout the 3 months of the study for individuals who were classified as having severe stroke, compared with those who were classified as having mild or moderate strokes. In the subjects with mild or moderate strokes, gait speed

and Box and Block Test scores, which are measured on true continuous scales, showed the greatest amount of change compared with the other measures.

Discussion and Conclusions Patients with motor and functional def- icits following an acute stroke are expected to improve11,44 and thus, we believe, provide an ideal population in which to assess the performance of the STREAM relative to other measures used to evaluate effects of stroke and rehabilitation. The STREAM showed a moderate to high correlation with the other measures used in this study. This finding is expected because the ability to perform functional activities is dependent on a person’s motor abili- ty.45,46 The correlations between data obtained for the STREAM and its sub- scales and data obtained for the Box and Block Test, the Balance Scale, gait speed, the TUG, and the Barthel Index were always less than .9, indicating that the STREAM may be related to these scales, but is reflecting a different com- ponent of recovery. In addition, mean STREAM scores could be used to dis- tinguish between different levels of per- formance on the Barthel Index and the Balance Scale. The cutoff criteria for the Barthel Index were based on what we considered clinically relevant vari- ables, including the probability of going home and hospital length of stay,

and the cutoff criteria for the Balance Scale taken from the literature15,17 were based on ambulatory status. The ability of mean STREAM scores to be used to distinguish among the group classifications based on Balance Scale and Barthel Index scores reflects its potential relation- ship to these clinical variables.

The ability to predict discharge destination and func- tional ability in individuals with stroke admitted to an acute care hospital allows prompt discharge planning, which may minimize hospital length of stay. The STREAM showed a usefulness comparable to that of the Barthel Index for predicting discharge destination from an acute care hospital. Compared with the Box and Block Test, the STREAM during the initial evaluation was better able to predict gait speed and functional ability 3 months poststroke, but its prognostic ability for these outcomes was similar to that of the Balance Scale, gait speed, the TUG, and the Barthel Index.

Figure 1. The probability of being discharged home after the acute care hospital given an initial score on the Stroke Rehabilitation Assessment of Movement (STREAM) or the Barthel Index � cutoff value (n�63).

Physical Therapy . Volume 83 . Number 7 . July 2003 Ahmed et al . 625

��

D ow

nloaded from https://academ

ic.oup.com /ptj/article/83/7/617/2805284 by guest on 10 M

arch 2021

The STREAM was one of the 3 measures with scores that changed the most over the entire 3 months of the study (Tab. 6). Interestingly, the STREAM, the Balance Scale, and the Barthel Index, all scored on an ordinal scale, were able to reflect the most change throughout the first 5 weeks for individuals we classified as having severe problems as compared with those who we classified as having mild or moderate problems (Fig. 2). In the subjects we classified as having mild or moderate prob- lems, gait speed, which is measured on a continuous

scale, was found to show the greatest amount of change compared with the other measures. It may be that in individuals who have had a severe stroke, changes need to occur in the components necessary for walking, which can be assessed with the STREAM and the Balance Scale, before recovery of gait speed is seen. These results agree with those of Richards et al,47 who found that, in a group of patients who were low-level performers and walked at very slow speeds, the Balance Scale, and to a lesser extent the Barthel Index ambulation score and the Fugl-Meyer

Table 5. Relationship of Initial Scores on Measures of Impairment and Disability to Gait Speed and Barthel Index Scores at 3 Months Poststroke (n�63)

Measure

Gait Speeda Barthel Index

Standardized Parameter Estimateb (P Value) R2c

Standardized Parameter Estimateb (P Value) R2

STREAMd 0.22 (.0001) 61 8 (.0001) 44

Balance Scale 0.24 (.0001) 62 8 (.0001) 39

Box and Block Test (affected extremity) 0.13 (.0001) 44 5 (.002) 26

TUGe ability 0.21 (.0001) 60 9 (.0001) 49

Gait speed 0.23 (.0001) 60 7 (.002) 31

Barthel Index 0.22 (.0001) 56 10 (.0001) 50

a Measured in meters per second. b Standardized regression coefficient��standard deviation. Adjusted for age and cognition. c The percentage of variability in gait speed or the Barthel Index accounted for by the corresponding measures in the first column, adjusted for age and cognition. The R2 for age and cognition was 30% for gait speed and 13% for the Barthel Index. d STREAM�Stroke Rehabilitation Assessment of Movement. e TUG�Timed “Up & Go” Test.

Table 6. Standardized Response Means (95% Confidence Interval) for All Measures (n�63)a

Measure

Initial Evaluation to 5-Week Evaluation

5-Week Evaluation to 3-Month Evaluation

Initial Evaluation to 3-Month Evaluation

STREAM Total score 0.94 (0.72 to 1.11) 0.37 (0.10 to 0.52) 0.96 (0.74 to 1.13) UE score 0.72 (0.54 to 0.87) 0.22 (�0.05 to 0.48) 0.72 (0.55 to 0.87) LE score 0.62 (0.36 to 0.80) 0.32 (0.08 to 0.55) 0.83 (0.57 to 1.02) Mobility score 0.81 (0.59 to 0.97) �0.22 (�0.02 to 0.46) 0.85 (0.66 to 1.00)

Box and Block Test Affected UE 1.30 (0.86 to 1.33) �0.20 (�1.93 to 1.15) 0.24 (�1.93 to 1.15) Unaffected UE 0.89 (0.54 to 1.04) �0.007 (�0.27 to 0.24) 0.82 (0.48 to 0.98)

Barthel Index 0.97 (0.76 to 1.14) �0.42 (0.07 to 0.62) 0.91 (0.60 to 1.10)

Balance Scale Missing values imputed 0.95 (0.73 to 1.14) �0.47 (0.23 to 0.69) 0.94 (0.71 to 1.14) Missing values removed (n�50) 1.04 (0.64 to 1.05) �0.31 (0.03 to 0.51) 0.89 (0.54 to 0.90)

TUG (s)b

Missing values imputed �0.63 (�0.50 to �0.75) �0.37 (�0.24 to �0.50) �0.69 (�0.53 to �0.84) Missing values removedc �0.68 (�0.20 to �0.95) �0.16 (�0.21 to 0.49) �0.61 (�0.11 to �0.89)

Gait speed (m/s) 1.05 (0.79 to 1.24) �0.17 (�0.13 to 0.43) 1.15 (0.80 to 1.43)

a STREAM�Stroke Rehabilitation Assessment of Movement, UE�upper extremity, LE�lower extremity, TUG�Timed “Up & Go” Test. b Values are negative because a lower number reflects a better score. c Initial evaluation to 5-week evaluation, n�46; 5-week evaluation to 3-month evaluation, n�56; initial evaluation to 3-month evaluation, n�46.

626 . Ahmed et al Physical Therapy . Volume 83 . Number 7 . July 2003

D ow

nloaded from https://academ

ic.oup.com /ptj/article/83/7/617/2805284 by guest on 10 M

arch 2021

leg subscore, were more discriminative than gait speed for the amount of physical assistance needed to ambu- late. In comparison, for subjects achieving 50% of nor- mal gait speed values, these scales became less discrimi- native and plateaued, whereas gait speed continued to improve 6 and 12 weeks following stroke.47 In our study, after gait speed and the Box and Block Test, the

STREAM was able to reflect the most change in the subjects we classified as having mild or moderate problems during the 5 weeks poststroke.

To our knowledge, only 2 other mea- sures, the Fugl-Meyer Stroke Assess- ment Scale46,48 and the disability inventory of the Chedoke-McMaster Stroke Assessment,49,50 have been studied for their ability to reflect change. In one study,46 the Fugl-Meyer scale scores were found to change much less than Barthel Index and Balance Scale scores. The Barthel Index demonstrated a greater ability to reflect a treatment effect than other measures of neurological status, stroke severity, and motor recovery in patients with acute stroke.46 The Bar- thel Index is considered by some authors48 to be a “gold standard” against which new instruments may be evaluated. In our study, the correla- tions between the STREAM scores and the Barthel Index scores for each eval- uation ranged from .75 to .78. In addi-

tion, the SRM of the STREAM was slightly higher than that of the Barthel Index over the first 3 months poststroke.

A concern with any outcome measure is that a great proportion of individuals would be at the high end of

Figure 2. Standardized response means for stroke outcomes among individuals with mild, moderate, and severe strokes (initial evaluation to 5-week evaluation). STREAM�Stroke Rehabilitation Assess- ment of Movement, TUG�Timed “Up & Go” Test.

Table 7. Number (%) of Subjects Reaching Maximum Scores on Stroke Measuresa

Measure Maximum Score

Evaluation 1 Evaluation 2 Evaluation 3

No. (%) With Maximum Score

STREAM Total score 100 5 (8) 15 (24) 21 (33) UE score 100 19 (30) 32 (51) 36 (57) LE score 100 17 (27) 31 (49) 37 (58) Mobility score 100 12 (19) 20 (32) 28 (44)

Box and Block Test (affected UE) . . .b 4 (7) 4 (7) 4 (7)

Barthel Index 100 17 (27) 32 (51) 38 (60)

Balance Scale Missing values removed 56 4 (8) 13 (26) 15 (30) Missing values imputed 56 18 (29) 16 (25) 18 (29)

TUG Missing values imputed 7–10 s 8 (13) 18 (29) 21 (38) Missing values removed 7–10 s 9 (20) 20 (44) 21 (53)

Gait speed . . .b 2 (3) 13 (21) 13 (21)

a STREAM�Stroke Rehabilitation Assessment of Movement, UE�upper extremity, LE�lower extremity, TUG�Timed “Up & Go” Test. b Mean sex- and age-specific normal values were used as maximum scores.

Physical Therapy . Volume 83 . Number 7 . July 2003 Ahmed et al . 627

��

D ow

nloaded from https://academ

ic.oup.com /ptj/article/83/7/617/2805284 by guest on 10 M

arch 2021

the scale and further improvements would be difficult to assess. The STREAM, however, was among the 3 mea- sures with the smallest ceiling effect for the first 2 evaluations but not the third evaluation (Tab. 7). During the acute period after a stroke, when the impact of the disability may not yet be manifested, fewer individuals would be expected to reach the maximum score on the STREAM. However, by 6 weeks after a stroke, 80% of patients would have reached their highest level of motor recovery.51 In our study, after 3 months, less than 40% of individuals had reached the maximum score on the total STREAM, and less than 60% had reached the maximum score on the UE and LE subscales. The Barthel Index had the greatest ceiling effect at 3 months poststroke, with 60% of the subjects reaching the maximum score. By 3 months poststroke, the subjects may have used all compensatory techniques for accomplishing activities of daily living, and better function can only be achieved with further motor recovery. The ceiling effect of the Barthel Index has been previously documented as one of the limitations of this measure.21,52

The results from our study have provided some indica- tions as to when the STREAM may be preferred over other measures to monitor recovery from a stroke. In our study, 26% of the subjects were not able to perform the TUG, and 21% were not able to perform the Balance Scale at the time of the initial evaluation. When people are unable to complete measures that require high levels of functioning during this acute period, a measure of voluntary movement such as the STREAM could play an important role in monitoring recovery. In addition, for more severe strokes, variables measured by outcomes such as the Box and Block Test, the Balance Scale, gait speed, the TUG, and the Barthel Index may not repre- sent the focus of immediate therapy and, therefore, may not be appropriate for monitoring changes in recovery during this acute period.53 During the time immediately after a stroke, the focus may be on restoring voluntary movement and basic mobility. These are assessed by the STREAM and we believe are necessary for further func- tional recovery. Moreover, for individuals who are unable to perform high-level functional tests, the STREAM can be used within the first few days poststroke to predict the probability of discharge home from an acute care hospital and functional potential 3 months poststroke. This is important as the length of stay in an acute care hospital becomes shorter.

Another important aspect of selecting an outcome mea- sure is its clinical utility. The STREAM is relatively simple to score as compared with other instruments of motor recovery, it requires little equipment, and it only takes 15 minutes to administer. The ease of use is comparable to that for the other measures examined in this study. However, because the calculation of the final score on

the STREAM requires several steps, it may be difficult to arrive at the total score in the presence of the patient.

Our study provides information about the relationships of the STREAM to other commonly used measures of stroke impairment and disability, as well as the measure- ment properties of the STREAM. Information on the construct validity54 of the STREAM, we believe, was demonstrated by its correlations with the other measures of impairment and disability12,25,45– 47,51,55– 62 and its abil- ity to differentiate levels of performance on measures of balance and independence in activities of daily living. The ability to use STREAM scores to predict important outcomes of stroke, including discharge home, func- tional independence, and gait speed, has resulted in information about its predictive validity.63 The longitu- dinal validity,6 which is the ability of an instrument to reflect change, of the STREAM was demonstrated by its capacity to monitor changes in recovery of voluntary movement and basic mobility during the first 3 months poststroke.

The question could be raised as to why there is a need for yet another measure of motor recovery. When the STREAM was developed, it filled a gap. According to a Canada-wide survey conducted in 1992,6,64 the complex- ity of existing measures of motor recovery was a barrier to their use. The underlying factor that appears to be driving use of these measures is the type of therapy being used and the need to evaluate patients’ progress along these therapeutic lines. The STREAM is not strongly linked to any one theoretical framework of how recovery occurs but rather provides a sampling of items that its developers believe reflect the re-emergence of move- ment and basic mobility. The STREAM, because of its independence from a treatment philosophy, its demon- strated measurement properties, and its ease of use, provides therapists with an option that may emerge as a measure of choice for the evaluation of different treat- ment approaches.

Limitations of the Study There were differences between the study sample and the nonparticipants in terms of the type and severity of stroke. The results of this study cannot be generalized to patients who are not similar to the study sample. This includes patients with severe cognitive impairment and substantial comorbidities.

When testing the ability of the STREAM to predict discharge destination, the outcome was dichotomized as “home” versus “not home” because most patients either went home or to rehabilitation. Testing the STREAM in a sample of patients with stroke where discharge desti- nation is more variable would provide more information

628 . Ahmed et al Physical Therapy . Volume 83 . Number 7 . July 2003

D ow

nloaded from https://academ

ic.oup.com /ptj/article/83/7/617/2805284 by guest on 10 M

arch 2021

with regard to its ability to discriminate between patients on this important outcome.

Another limitation of our study was that the first evalu- ation was done an average of 8 days poststroke. Motor recovery can take place during the first 10 days post- stroke; therefore, some patients may have experienced recovery before the first evaluation. Not having evalu- ated this early recovery may have reduced the variability in scores for all measures used in this study and may have underestimated their ability to predict the level of inde- pendence in functional activities of daily living and gait speed. In addition, this may have resulted in lower estimates of SRMs.

The methods we used to handle the missing data were only estimates of the true level of recovery. Some of the results may have been overestimated or underestimated. During the data analysis, however, care was taken to ensure that the imputed values did not cause large influences on the distribution of scores, and the results were compared with and without these values. Missing data on the TUG limited our ability to examine the relationship of this measure to the STREAM.

References 1 Carr JH, Shepherd RB, Nordholm L, Lynne D. Investigation of a new motor assessment scale for stroke patients. Phys Ther. 1985;65:175–180.

2 Brosseau L, Potvin L. Inter-rater reliability and construct validity of the LaVigne Motor Recovery Assessment Scale following a cerebrovas- cular accident. Physiotherapy Theory and Practice. 1993;9:91–102.

3 Carey JR, Burghardt TP. Movement dysfunction following central nervous system lesions: a problem of neurologic or muscular impair- ment? Phys Ther. 1993;73:538 –547.

4 Corcos DM. Strategies underlying the control of disordered move- ment. Phys Ther. 1991;71:25–38.

5 Herman JM, Culpepper L, Franks P. Patterns of utilization, disposi- tion, and length of stay among stroke patients in a community hospital setting. J Am Geriatr Soc. 1984;32:421– 426.

6 Cole B, Finch E, Gowland C, Mayo N. Physical Rehabilitation Outcome Measures. Toronto, Ontario, Canada: Canadian Physiotherapy Association–Publishing Supply and Services Canada; 1994.

7 Daley K, Mayo NE, Wood-Dauphinée SL, et al. Verification of the Stroke Rehabilitation Assessment of Movement (STREAM). Physiother- apy Canada. 1997;49:269 –278.

8 Daley K, Mayo NE, Wood-Dauphinée SL. Reliability of scores on the Stroke Rehabilitation Assessment of Movement (STREAM) measure. Phys Ther. 1999;79:8 –19.

9 Visintin M, Barbeau H, Korner-Bitensky N, Mayo NE. A new approach to retrain gait in stroke patients through body weight support and treadmill stimulation. Stroke. 1998;29:1122–1128.

10 Salbach NM, Mayo NE, Richards C, et al. Responsiveness and predictability of gait speed and other disability measures in acute stroke. Arch Phys Med Rehabil. 2001;82:1204 –1212.

11 Mayo NE. Epidemiology and recovery. Physical Medicine and Rehabil- itation: State-of-the-Art Reviews. 1993;7:1–25.

12 Wade DT, Wood VA, Hewer RL. Recovery after stroke: the first 3 months. J Neurol Neurosurg Psychiatry. 1985;48:7–13.

13 Cote R, Battista RN, Wolfson C, et al. The Canadian Neurological Scale: validation and reliability assessment. Neurology. 1989;39: 638 – 643.

14 Mahoney FI, Barthel DW. Functional evaluation: the Barthel lndex. Maryland State Medical Journal. 1965;14:61– 65.

15 Granger CV, Dewis LS, Peters NC, et al. Stroke rehabilitation: analysis of repeated Barthel Index measures. Arch Phys Med Rehabil. 1979;60:14 –17.

16 Loewen SC, Anderson BA. Reliability of the Modified Motor Assess- ment Scale and the Barthel Index. Phys Ther. 1988;68:1077–1081.

17 Shinar D, Gross CR, Mohr JP, et al. Interobserver variability in the assessment of neurologic history and examination in the Stroke Data Bank. Arch Neurol. 1985;42:557–565.

18 Roy CW, Togneri J, Hay E, Pentland B. An inter-rater reliability study of the Barthel Index. Int J Rehabil Res. 1988;11:67–70.

19 Gresham GE, Phillips TF, Labi ML. ADL status in stroke: relative merits of three standard indexes. Arch Phys Med Rehabil. 1980;61: 355–358.

20 Wade DT, Hewer RL. Functional abilities after stroke: measure- ment, natural history and prognosis. J Neurol Neurosurg Psychiatry. 1987;50:177–182.

21 Wellwood I, Dennis MS, Warlow CP. A comparison of the Barthel Index and the OPCS disability instrument used to measure outcome after acute stroke. Age Ageing. 1995;24:54 –57.

22 Wolfe CD, Taub NA, Woodrow EJ, Burney KD. Assessment of scales of disability and handicap for stroke patients. Stroke. 1991;22: 1242–1244.

23 Berg KO, Wood-Dauphinée SL, Williams JI. The Balance Scale: reliability assessment with elderly residents and patients with an acute stroke. Scand J Rehabil Med. 1995;27:27–36.

24 Berg KO. Balance and its measure in the elderly: a review. Physio- therapy Canada. 1989;41:240 –246.

25 Berg KO, Maki BE, Williams JI, et al. Clinical and laboratory measures of postural balance in an elderly population. Arch Phys Med Rehabil. 1992;73:1073–1080.

26 Berg KO, Wood-Dauphinée SL, Williams JI, Maki B. Measuring balance in the elderly: validation of an instrument. Can J Public Health. 1992;83(suppl 2):S7–S11.

27 Berg KO, Wood-Dauphinée SL, William JI, Gayton D. Measuring balance in the elderly: preliminary development of an instrument. Physiotherapy Canada. 1989;41:304 –311.

28 Evans MD, Goldie PA, Hill KD. Systematic and random error in repeated measurements of temporal and distance parameters of gait after stroke. Arch Phys Med Rehabil. 1997;78:725–729.

29 Bohannon RW, Andrews AW. Correlation of knee extensor muscle torque and spasticity with gait speed in patients with stroke [published erratum appears in Arch Phys Med Rehabil. 1990;71:464]. Arch Phys Med Rehabil. 1990;71:330 –333.

30 Wade DT, Wood VA, Heller A, et al. Walking after stroke: measure- ment and recovery over the first 3 months. Scand J Rehabil Med. 1987;19:25–30.

31 Holden MK, Gill KM, Magliozzi MR, et al. Clinical gait assessment in the neurologically impaired: reliability and meaningfulness. Phys Ther. 1984;64:35– 40.

32 Collen FM, Wade DT, Bradshaw CM. Mobility after stroke: reliability of measures of impairment and disability. Int Disabil Stud. 1990;12:6 –9.

Physical Therapy . Volume 83 . Number 7 . July 2003 Ahmed et al . 629

��

D ow

nloaded from https://academ

ic.oup.com /ptj/article/83/7/617/2805284 by guest on 10 M

arch 2021

33 Day RA, Moore KN. Evaluation of a self-medication program: pilot project. Can J Rehab. 1992;5:205–216.

34 Nelson AJ. Functional ambulation profile. Phys Ther. 1974;54: 1059 –1065.

35 Podsiadlo D, Richardson S. The timed “Up & Go”: a test of basic functional mobility for frail elderly persons. J Am Geriatr Soc. 1991;39: 142–148.

36 Newnham J, St Pierre D, Rousseau M, Sampalis J. Reliability of performance-oriented outcome measures in an elderly institutional- ized population [abstract]. Physiotherapy Canada. 1994;46:101.

37 Venturini A, Wilford C, Finch L, et al. The Timed Up and Go: reliability and validity in an acute neurological population [abstract]. In: Proceedings of the 12th International Congress of the World Confederation for Physical Therapy; June 25–30, 1995; Washington, DC. Alexandria, Va: American Physical Therapy Association; 1995:201.

38 Desrosiers J, Bravo G, Hebert R, et al. Validation of the Box and Block Test as a measure of dexterity of elderly people: reliability, validity, and norms studies. Arch Phys Med Rehabil. 1994;75:751–755.

39 Colton T. Statistics in Medicine. Boston, Mass: Little, Brown and Co; 1974:210 –211.

40 Granger CV, Hamilton BB, Gresham GE, Kramer AA. The Stroke Rehabilitation Outcome Study, part II: relative merits of the total Barthel Index score and a four-item subscore in predicting patient outcomes. Arch Phys Med Rehabil. 1989;70:100 –103.

41 Heinemann AW, Linacre JM, Wright BD, et al. Prediction of rehabilitation outcomes with disability measures. Arch Phys Med Rehabil. 1994;75:133–143.

42 Kleinbaum DG, Kupper LL, Muller KE. Applied Regression Analysis and Other Multivariable Methods. Boston, Mass: PWS-Kent Publishing Co; 1988.

43 Liang MH, Larson MG, Cullen KE, Schwartz JA. Comparative measurement efficiency and sensitivity of five health status instruments for arthritis research. Arthritis Rheum. 1985;28:542–547.

44 Bonita R, Beaglehole R. Recovery of motor function after stroke. Stroke. 1988;19:1497–1500.

45 Duncan PW, Goldstein LB, Matchar D, et al. Measurement of motor recovery after stroke: outcome assessment and sample size require- ments. Stroke. 1992;23:1084 –1089.

46 Wood-Dauphinée SL, Williams JI, Shapiro SH. Examining outcome measures in a clinical study of stroke. Stroke. 1990;21:731–739.

47 Richards CL, Malouin F, Dumas F, Tardif D. Gait velocity as an outcome measure of locomotor recovery after stroke. In: Craik RL, Oatis CA. eds. Gait Analysis: Theory and Application. St Louis, Mo: Mosby; 1995:355–364.

48 Wood-Dauphinée SL, Berg KO, Bravo G, Williams JI. The Balance Scale: responsiveness to clinically meaningful changes. Can J Rehab. 1997;10:35–50.

49 Gowland C, Stratford PW, Ward M, et al. Measuring physical impairment and disability with the Chedoke-McMaster Stroke Assess- ment. Stroke. 1993;24:58 – 63.

50 Huijbregts MPJ, Gowland C, Gruber RA. Measuring clinically- important change with the Activity Inventory of the Chedoke- McMaster Stroke Assessment. Physiotherapy Canada. 2000;52:295–304.

51 Jorgensen HS, Nakayama H, Raaschou HO, et al. Outcome and time course of recovery in stroke, part II: time course of recovery—the Copenhagen Stroke Study. Arch Phys Med Rehabil. 1995;76:406 – 412.

52 Task Force on Standards for Measurement in Physical Therapy. Standards for tests and measurements in physical therapy practice. Phys Ther. 1991;71:589 – 622.

53 Chae J, Johnston M, Kim H, Zorowitz R. Admission motor impair- ment as a predictor of physical disability after stroke rehabilitation. Am J Phys Med Rehabil. 1995;74:218 –223.

54 Streiner DL, Norman GR. Health Measurement Scales: A Practical Guide to Their Development and Use. 2nd ed. New York, NY: Oxford University Press; 1995:150 –157.

55 Brandstater ME, de Bruin H, Gowland C, Clark BM. Hemiplegic gait: analysis of temporal variables. Arch Phys Med Rehabil. 1983;64: 583–587.

56 Duncan PW, Goldstein LB, Horner RD, et al. Similar motor recov- ery of upper and lower extremities after stroke. Stroke. 1994;25: 1181–1188.

57 Dettman N, Linder MT, Sepic SB. Relationships among walking performance, postural stability, and functional assessments of the hemiplegic patient. Am J Phys Med. 1992;66:77–90.

58 Chyou PH, White LR, Yano K, et al. Pulmonary function measures as predictors and correlates of cognitive functioning in later life. Am J Epidemiol. 1996;143:750 –756.

59 Badke MB, Duncan PW. Patterns of rapid motor responses during postural adjustments when standing in healthy subjects and hemiplegic patients. Phys Ther. 1983;63:13–20.

60 Witte US, Carlsson JY. Self-selected walking speed in patients with hemiparesis after stroke. Scand J Rehabil Med. 1997;29:161–165.

61 Keenan MA, Perry J, Jordan C. Factors affecting balance and ambulation following stroke. Clin Orthop. January-February 1984: 165–171.

62 Stevenson TJ, Garland J. Standing balance during internally pro- duced perturbations in subjects with hemiplegia: validation of the balance scale. Arch Phys Med Rehabil. 1996;77:656 – 662.

63 Payton OD. Research: The Validation of Clinical Practice. Philadelphia, Pa: FA Davis Co; 1994:75–76.

64 Mayo NE, Cole B, Dowler J, et al. Use of outcome measurement in physiotherapy: survey of current practice. Can J Rehab. 1993;7:81– 82.

630 . Ahmed et al Physical Therapy . Volume 83 . Number 7 . July 2003

D ow

nloaded from https://academ

ic.oup.com /ptj/article/83/7/617/2805284 by guest on 10 M

arch 2021