Psychometric properties of 2 simplified 3-level balance scales used for patients with stroke.

The Berg Balance Scale (1) (BBS), a 5-level scale, is commonly used to evaluate balance performance in elderly people and in patients with stroke. In 1999, Benaim et al (2) developed a new 4-level balance scale especially for patients with stroke called the Postural Assessment Scale for Stroke Patients (PASS). In a recent study, Mao et al (3) compared the psychometric properties of these 2 scales in 112 patients with stroke and found that both had good interrater reliability (intraclass correlation coefficient [ICC] [greater than or equal to] .95), high concurrent validity (the BBS scores were highly correlated with the PASS scores, with Spearman [rho] [greater than or equal to] .92), and moderate to high responsiveness (effect size >.63). Although either the BBS or the PASS is suggested for measuring balance in patients with stroke, (3,4) the usefulness of the levels of scaling of each measure was rarely investigated. Whether the level of scaling of the balance measures would influence their psychometric properties remains unknown.

Some people believe that a scale with 4 or 5 levels of performance (eg, balance scale) has better psychometric properties than a 3-level measure, (5) although data for such a claim are lacking. Evidence, however, often shows that increasing the level of scaling may not improve the psychometric properties of a measure. (6,7) Thus, it might be reasonable to try to simplify a measure by reducing the levels without sacrificing its psychometric properties. We therefore proposed 2 simplified measures with reduced levels of scaling, a 3-level Berg Balance Scale (BBS-3P) rather than a 5-level scale and a 3-level Postural Assessment Scale for Stroke Patients (PASS-3P) rather than one that calls for 4 levels of measurement. The purpose of our study was to investigate whether these two 3-level measures have psychometric properties that are comparable to those of the measures on which they were based for patients with stroke.

Method

Our study was conducted in 2 parts. In the first part, we examined the distribution, concurrent validity, and convergent validity of data obtained with the simplified and original measures. The BBS and PASS-3P were administered by one physical therapist, the BBS-3P and PASS were administered by another therapist, and the Barthel Index (BI) was administered by a third therapist. To minimize the effects of any possible fatigue on the subjects, the 3 therapists administered the measures to each subject in a random order within 24 hours. The therapists were masked to each other's results during the study period.

In the second part, additional psychometric properties (including predictive validity and responsiveness) of these measures were evaluated in subjects with stroke who were followed from the acute stage up to 90 days after the onset of stroke. The original BBS and PASS were administered to the subjects at 14, 30, and 90 days after the onset of stroke. The BI, which assesses the performance of activities of daily living (ADL), was administered 90 days after the onset of stroke. The degrees of responsiveness of the balance measures were calculated based on the changes occurring between 14 and 30 days after the onset of stroke, between 30 and 90 days after the onset of stroke, and between 14 and 90 days after the onset of stroke. All of these measures were administered by an occupational therapist who was not informed of the purpose of this study. Both of the simplified 3-level measures in this part of this study were achieved via the recoding schemes (the BBS-3P [0-2-4] was recoded from the BBS [0-1-2-3-4]; the PASS-3P [0-1.5-3] was recoded from the PASS [0-1-2-3]), and the resulting scores were used in the data analysis.

Instruments

The BBS (1) is designed to evaluate a person's performance on 14 items (1 sitting item and 13 standing items) related to balance function tasks that are frequently encountered in everyday life. The scoring method is based on a 5-level scale of 0 to 4, with the total score ranging from 0 to 56. The BBS was originally developed for screening elderly people who are at risk for falling, (1) but the psychometric properties of the BBS have been well supported for use in patients with stroke. (3-5,8)

The PASS (2) was developed to be applicable to all patients with stroke, even for those with poor postural performance. The PASS contains twelve 4-level (0-3) items that are used to grade performance for situations of varying difficulty in maintaining or changing a given lying, sitting, or standing posture. The total score ranges from 0 to 36. The psychometric properties of the PASS have been reported to be satisfactory in patients with stroke. (2,3)

The BBS-3P was simplified from the original measure. In the first part of our study, the original scoring criteria for the BBS were used, except that the third and fourth levels of scoring on each BBS item were not used. That is, subjects who could fulfill the scoring criteria for the second easiest level but who could not pass the highest scoring criteria were awarded the middle level on each item of the measures. In the second part of our study, the BBS-3P scores were retrieved from the original BBS scores; the 3 levels in the center of the BBS were collapsed to a single level. To facilitate comparisons of versions, each item of the BBS-3P was recoded as 0-2-4, making the lowest and highest scores of the BBS-3P the same as those of the original BBS in both parts of the study.

The PASS-3P also was simplified from the original measure. In the first part of our study, the original scoring criteria for the PASS were used, except that the third level of scoring on each PASS item was not used. In the second part of our study, the PASS-3P scores were retrieved from the original PASS scores by collapsing the 2 levels in the center of the PASS to a single level. To facilitate comparisons of the versions, each item of the PASS-3P was recoded as 0-1.5-3, making the lowest and highest scores of the PASS-3P the same as those of the original PASS.

The BI is a measure of the severity of disability. (9) The BI is used to evaluate 10 basic ADL items, and the total score ranges from 0 to 100. It has been shown to yield good interrater reliability (ICC=.94), high concurrent validity (Spearman [rho] [greater than or equal to.] 92) measurements of ADL in patients with stroke. (10,11) The BI was used as an external criterion to examine convergent validity of data obtained with the balance measures in the first part of the study and to examine predictive validity for disability of data obtained with the measures in the second part of the study.

The Fugl-Meyer motor assessment (FM) (12) is used mainly to measure motor impairment after stroke. The FM consists of 50 items of upper- and lower-extremity motor function. Each item is graded on a 3-level scale. The total possible score ranges from 0 to 100 points. It has been shown to have good interrater reliability (ICC [greater than or equal to.] .92) and high concurrent validity (r [greater than or equal to.] .99) (13-15) We used the FM score as an index of neurological severity for our subjects.

Subjects for the First Part of the Study

To select patients with broad range of balance deficits, subjects were recruited from 5 rehabilitation units in Taichung, Taiwan, from July to August 2002. Patients were included in the study if they met the following criteria: (1) diagnosis (International Classification of Diseases, 9th Revision, Clinical Modification (16) [ICD-9] codes) of cerebral hemorrhage (ICD-9: 431) or cerebral infarction (ICD-9: 434) according to the medical records, (2) first onset of stroke without other major disease and the absence of a pre-existing disability, and (3) ability to follow instructions. Only patients who had a single stroke and were able to give informed consent personally or by proxy were included in this part of the study.

A total of 77 subjects with a wide spectrum of balance deficits, ranging from subjects without obvious neurological symptoms to subjects who were bedridden, participated in the concurrent validity and convergent validity parts of the study. Table 1 shows the characteristics of the subjects in the study.

Subjects for the Second Part of the Study

This part of the protocol was a study of predictive validity and responsiveness. Data were retrieved from the records of subjects participating in an ongoing study, the Quality of Life After Stroke Study in Taiwan, initiated on December 1, 1999. Individuals enrolled in the study were evaluated at 14 days after onset of stroke and were reassessed at other points after onset of stroke for up to 3 years to characterize their recovery of neurological impairments, balance ability (as measured by the BBS and PASS), functional abilities, and health-related quality of life. Further selection and exclusion criteria of the study sample are described elsewhere. (3)

A total of 226 subjects were included in this part of the study. A total of 685 patients were excluded because onset of their stroke was more than 14 days prior to admission, because the patient did not live in the greater Taipei area, because of recurrent stroke, or because of communication difficulties. The participants were 103 women and 123 men with a mean age of 69 years (SD=11.3, range=33-95). More than half (72%) of these subjects had cerebral infarction, and the others had cerebral hemorrhage. The FM scores indicated that the subjects had a wide range of motor impairments (median FM score=54.5, interquartile range= 16.8-91). Twenty-four subjects were lost to follow-up at 30 days after onset of stroke, and another 35 subjects were lost to follow-up at 90 days after onset of stroke. A total of 167 subjects completed follow-up at 90 days after onset of stroke.

Data Analysis

Concurrent validity. The Bland-Altman method (17) was used to examine agreement between data obtained with the original measures and data obtained with the simplified measures. In this method, the average scores of the original and simplified measures are plotted against the difference between scores. The extent of agreement was examined, like examining interrater agreement, using the ICC (1,1). (18) Wilcoxon matched-pairs signed-rank tests were performed to determine the statistical significance of the differences between scores. In addition, the Spearman [rho] was used to examine the interrelationships between data obtained with the simplified measures and data obtained with the original measures.

Convergent validity. Convergent validity was determined by examining the relationships between data obtained with the measures and data obtained with instruments measuring similar constructs. The relationships between the total scores of the BBS-3P and PASS-3P and the BI were examined using the Spearman [rho].

Predictive validity. The predictive validity, of data obtained with both simplified balance measures was assessed by comparing the results at 14 and 30 days after onset of stroke with the results obtained with the BI at 90 days after onset of stroke using the Spearman [rho] to determine whether the level of disability could be predicted by the data obtained from the balance measures.

Responsiveness. Because there is no consensus regarding how best to assess the responsiveness of measurement instruments, we used 2 different approaches. First, the standardized response mean (SRM), one type of effect size, was calculated by dividing the mean change in scores by the standard deviation of the change in scores in the same subjects. Wilcoxon matched-pairs signed-rank tests were performed to determine the statistical significance of the change in scores. Furthermore, to determine whether the responsiveness of the measures varied depending on the initial stroke-induced deficits, we separated subjects into one of the following 3 groups based on their FM scores: 0 to 35 (severe stroke), 36 to 79 (moderate stroke), and 80 or greater (mild stroke). (19)

Results

Tables 2 and 3 show the distributions of data obtained with both simplified and original balance measures. The tables show that, on average, 27.1% of the subjects were awarded the middle levels (11.5% and 15.6% on levels 1 and 2, respectively) of each item on the PASS (26.7% on the PASS-3P) and 27.9% were awarded the middle levels (5.2%, 6.6%, and 16.1% on levels 1, 2, and 3, respectively) of each item on the BBS (22.6% on the BBS-3P).

The extent of agreement between data obtained with the simplified 3-level measures and the original measures is shown in Figures 1 and 2. At most, 4 observations (5.2%) had a 2-standard deviation difference between the simplified measure scores and the original measure scores. The difference scores were not different from zero. The extent of agreement was excellent between the BBS and BBS-3P scores (ICC=.99, 95% confidence interval [CI]=.98-.99) and between the PASS and PASS-3P scores (ICC=.97, 95% CI=.96-.98). (20) Table 4 shows that the interrelationships of data obtained with the PASS, PASS-3P, BBS, and BBS-3P were [rho] [greater than or equal to.] .91 (P<.0001). These results indicated that both simplified measures demonstrated high concurrent validity. The correlations of the PASS-3P and BBS-3P scores with the BI scores were [rho]=.82 and .87, respectively. The convergent validity of data obtained with both simplified balance measures was as good the original measures.

[FIGURES 1-2 OMITTED]

Table 5 shows that the scores of the BBS-3P and PASS-3P on the 14th and 30th days after onset of stroke were correlated with the BI scores on the 90th day after onset of stroke ([rho] [greater than or equal to.] .75, P<.0001). The predictive validity of data obtained with the simplified 3-P measures was similar to that of data obtained with the original measures.

Table 6 shows that the simplified balance measures were moderately to highly responsive in detecting changes before 90 days after onset of stroke (14-30 days after onset of stroke: SRM [greater than or equal to.] 0.82; 30-90 days after onset of stroke: SRM [greater than or equal to.] 0.67). The changes of the BBS-3P and PASS-3P at each stage were all significant (P [less than or equal to.] .0001). As shown in Table 7, the SRM of the simplified measures at different stages for subjects with different levels of stroke severity were larger than 0.37. The responsiveness of the recoded BBS-3P and PASS-3P measures was very similar to that of the original measures (Tabs. 6 and 7).

Discussion

A psychometrically sound balance measure enables both clinicians and researchers to identify, monitor, and manage balance ability following stroke. In our study, 2 simplified balance measures (BBS-3P and PASS-3P) using only 3 levels of measurement were proposed, and their psychometric properties were compared with the original measures with more levels of measurement in patients with stroke. The psychometric properties of both 3-level balance measures were essentially identical to those of the original measures. The 3-level balance measures may meet the needs of both clinicians and researchers for measures of balance because of their simplicity of administration. Our findings may shed light on future measure development and revision.

There are other possible ways to simplify the level of scaling for both measures, such as changing 5- or 4-level scales to 2-level scales (eg, collapsing the lowest levels or highest levels). We believe this would make the results of the measures too prone to floor or ceiling effects and result in a loss of information. Furthermore, it could limit the discriminative power of the measures. Thus, we selected a 3-level scale.

There are several schemes to simplify a 4- or 5-level measure to a 3-level measure, such as collapsing the 3 center levels, or collapsing the 2 lowest levels and 2 highest levels, of a 0-1-2-3-4 scale. The central levels of the BBS and PASS can be viewed generally as "partially completes the task (individual item)." In contrast, the lowest level means "cannot perform the task," and the highest level means "fully completes the task." Furthermore, as collapsing the lowest or highest levels would increase the floor or ceiling effects of the measures, we collapsed the middle levels of the 4- or 5-level scale in our simplifications.

The concurrent validity and convergent validity of data obtained with the BBS-3P and PASS-3P were as high as for the original measures. Furthermore, the high validity scores of both simplified balance scales were generally in accordance with the findings of previous studies examining scores from the original scales. (2,3,5) The findings of our study demonstrate the validity of data obtained with the 2 simplified balance measures.

Responsiveness can be defined as the ability to detect minimal clinically important differences. (21,22) The results of the SRM indicated that the simplified measures had fair to good levels of responsiveness before 90 days after onset of stroke for all of the subjects studied. The responsiveness of both simplified measures at different stages, in our opinion, also was acceptable for subjects with different levels of stroke severity. There are only 29 possible points for the BBS-3P and only 25 possible points for the PASS-3P, as compared with the scoring possibilities for their original counterparts of 57 (0-56) and 37 (0-36), respectively. Our results indicate that the simplified scoring for the BBS-3P and PASS-3P resulted in the same level of responsiveness as for the original scales. These simplified 3-level balance measures appeared, in our opinion, not to lose any sensitivity to clinically important changes of in contrast to the original measures.

The range of scores for the simplified 3-level measures was the same as for the original measures (eg, 0-2-4 for the BBS-3P and 0-1-2-3-4 for the BBS) to facilitate comparisons of the versions. In clinical or research settings, the scoring scheme of each item could be changed to 0-1-2 for both the BBS-3P and the PASS-3P. Because such a change of scoring scheme is presumably linear, the psychometric properties of the measures should not be altered. In the future, analysis of these simplified measures using methods such as Item Response Theory (23,24) may provide insight into their clinical applicability.

Conclusion

Both simplified 3-level balance measures, in our opinion, demonstrated acceptable validity and responsiveness. More importantly, the psychometric properties of both simplified balance measures and the original measures were very, similar. One possible explanation for these findings is that both original measures have too many middle levels. We found that about a quarter of the subjects were awarded the middle levels of each item on the PASS and the BBS. About three quarters of the subjects in both groups, however, had the highest or lowest score on each item of the measures. This distribution of scores suggests that collapsing the middle levels of the items of the BBS and PASS will result, at worst, only in the loss of a limited amount of clinically important information. Furthermore, our results demonstrated that the simplified balance measures and the original measures were psychometrically similar, especially with respect to sensitivity to clinically important changes. Some recent studies (6,11,25) also have shown that increased grading levels do not improve the psychometric characteristics of the ADL measures. These findings suggest that 3-level scaling is may be adequate for the balance measures in people with stroke and that all measures based on scales with more than 3 levels should be re-evaluated in the interest of simplifying the measures.

In our study, the interrater and intrarater reliability of data obtained with the simplified measures was not examined. Agreement between the simplified and original balance measures, however, was excellent (ICC [greater than or equal to.] .97). Furthermore, in a previous study, (3) it was reported that the BBS and PASS had high interrater reliability (ICC [greater than or equal to.] .95). Therefore, we believe the interrater and intrarater reliability of data obtained with the simplified measures not to be a major concern. Because the data for the predictive validity and responsiveness parts of the study were recoded from the BBS and PASS according to the same set of data, the predictive validity and responsiveness of the simplified measures might have been overestimated. According to the results of the first part of the study, however, direct assessment of the simplified measures and the original measures showed excellent agreement (ICC [greater than or equal to.] .97). Recoding data from the original measures, therefore, might not be an issue. In addition, the raters in this study reported that the simplified measures were easier to use because of simplicity of scoring level than the original measures. According to Wade, (26) a simple-to-use measure will increase the patient and rater adherence to a simple scoring system. Future study is needed to investigate how much meaningful utility (eg, saving administration time and improving patient/rater adherence) can be gained with the simplified balance measures as compared with the original measures.

In summary, our results indicate that the level of scaling of the BBS and the PASS can be simplified without sacrificing their psychometric properties. A 3-level scale was found to be adequate for assessing balance in patients with stroke.

Table 1.
Characteristics of the Subjects With Stroke Who Participated in the
First Part (Examining Reliability and Concurrent and Convergent
Validity) of the Study (n=77) (a)

Characteristic

Sex (male/female)                 52/25
Age (y)
  [bar]X                          59.8
  SD                              11.9
  Range                           22-80

Diagnosis
  Cerebral hemorrhage             37
  Cerebral infarction             40

Side of hemiplegia (right/left)   35/42

PASS
  Median                          29
  Interquartile range             19-33

PASS-3P
  Median                          28.5
  Interquartile range             18.8-33

BBS
  Median                          46
  Interquartile range             14.5-53

BBS-3P
  Median                          46
  Interquartile range             13-52

BI
  Median                          85
  Interquartile range             60-95

(a) BBS=original 5-level (0-1-2-3-4) Berg Balance Scale,
BBS-3P=simplified 3-level (0-2-4) Berg Balance Scale,
PASS=original 4-level (0-1-2-3) Postural Assessment Scale
for Stroke Patients, PASS-3P=simplified 3-level (0-1.5-3)
Postural Assessment Scale for Stroke Patients, BI=Barthel
Index.

Table 2.
Distributions of the Original 5-Level (0-1-2-3-4) Berg Balance Scale
(BBS) and Simplified 3-Level (0-2-4) Berg Balance Scale (BBS-3P)
Scores for Subjects With Stroke Who Participated in the First Part
(Examining Reliability and Concurrent and Convergent Validity) of
the Study (n=77)

                                         BBS

                                         0      1      2      3
Item                                     (%)    (%)    (%)    (%)

Sitting to standing                      16.9    3.9    3.9   13.0
Standing unsupported                     22.1    1.3    3.9    7.8
Sitting unsupported                       5.2    3.9    0.0    7.8
Standing to sitting                      15.6    5.2    3.9   18.2
Transfers                                11.7   11.7    7.8   16.9
Standing with eyes closed                24.7    0.0    3.9   15.6
Standing with feet together              27.3    3.9    2.6   13.0
Reaching forward with outstretched arm   26.0    1.3    5.2   14.3
Retrieving object from floor             28.6    1.3    5.2   18.2
Turning to look behind                   29.9    1.3    7.8   10.4
Turning 360[degrees]                     31.2   10.4   10.4   22.1
Placing alternate foot on stool          36.4   11.7   10.4   18.2
Standing with one foot in front          31.2    5.2   13.0   33.8
Standing on one foot                     37.7   11.7   14.3   16.9
Mean percentage                          24.6    5.2    6.6   16.1

                                         BBS    BBS-3P

                                         4      0      2      4
Item                                     (%)    (%)    (%)    (%)

Sitting to standing                      62.3   15.6   23.4   61.0
Standing unsupported                     64.9   22.1   14.3   63.6
Sitting unsupported                      83.1    5.2    9.1   85.7
Standing to sitting                      57.1   18.2   20.8   61.0
Transfers                                51.9   24.7   10.4   64.9
Standing with eyes closed                55.8   24.7   10.4   64.9
Standing with feet together              53.2   27.3   13.0   59.7
Reaching forward with outstretched arm   53.2   26.0   19.5   54.5
Retrieving object from floor             46.8   32.5   11.7   55.8
Turning to look behind                   50.6   27.3   24.7   48.1
Turning 360 [degrees]                    26.0   31.2   44.2   24.7
Placing alternate foot on stool          23.4   35.1   39.0   26.0
Standing with one foot in front          16.9   31.2   42.9   26.0
Standing on one foot                     19.5   41.6   32.5   26.0
Mean percentage                          47.5   25.9   22.6   51.5

Table 3.
Distributions of the Original 4-Level (0-1-2-3) Postural Assessment
Scale for Stroke Patients (PASS) and Simplified 3-Level (0-1.5-3)
Postural Assessment Scale for Stroke Patients (PASS-3P) Scores for
Subjects With Stroke Who Participated in the First Part (Examining
Reliability and Concurrent and Convergent Validity) of the Study
(n=77)

                                         PASS

                                         0      1      2      3
Item                                     (%)    (%)    (%)    (%)

Sitting without support                  0.0     6.5    7.8   85.7
Standing with support                     6.5    6.5   23.4   63.6
Standing without support                 23.4    5.2   15.6   55.8
Standing on nonparetic leg               29.9   26.0   16.9   27.3
Standing on paretic leg                  70.1   23.4    3.9    2.6
Supine to affected side lateral           1.3    7.8   15.6   75.3
Supine to nonaffected side lateral        1.3    7.8   22.1   68.8
Supine to sitting up on the edge of       1.3   20.8   16.9   61.0
  the table
Sitting on the edge of the table to       1.3   10.4   15.6   72.7
  supine
Sitting to standing up                    9.1   11.7   14.3   64.9
Standing up to sitting down               9.1    7.8   22.1   61.0
Standing, picking up a pencil from the   31.2    3.9   13.0   51.9
  floor
Mean percentage                          15.4   11.5   15.6   57.5

                                         PASS-3P

                                         0      1.5    3
Item                                     (%)    (%)    (%)

Sitting without support                  0.0    16.9   83.1
Standing with support                     6.5   16.9   76.6
Standing without support                 15.6   26.0   58.4
Standing on nonparetic leg               31.2   42.9   26.0
Standing on paretic leg                  70.1   28.6    1.3
Supine to affected side lateral           1.3   24.7   74.0
Supine to nonaffected side lateral        1.3   22.1   76.6
Supine to sitting up on the edge of       2.6   33.8   63.6
  the table
Sitting on the edge of the table to       5.2   28.6   66.2
  supine
Sitting to standing up                   11.7   24.7   63.6
Standing up to sitting down              11.7   31.2   57.1
Standing, picking up a pencil from the   28.6   24.7   46.8
  floor
Mean percentage                          15.5   26.7   57.8

Table 4.
Concurrent and Convergent Validity of Scores for the Balance
Measures (n=77) (a)

                                           Convergent
          Concurrent Validity ([rho])      Validity ([rho])

          BBS   BBS-3P   PASS    PASS-3P   BI

BBS             .96 *    .94 *   .91 *     .85 *
BBS-3P                   .94 *   .92 *     .87 *
PASS                             .94 *     .84 *
PASS-3P                                    .82 *

(a) BBS=original 5-level (0-1-2-3-4) Berg Balance Scale,
BBS-3P=simplified 3-level (0-2-4) Berg Balance Scale,
PASS=original 4-level (0-1-2-3) Postural Assessment Scale
for Stroke Patients, PASS-3P=simplified 3-level (0-1.5-3)
Postural Assessment Scale for Stroke Patients, BI=Barthel
Index. Asterisk indicate P<.0001.te.

Table 5.
Predictive Validity for Disability of Scores for the Simplified
3-Level Balance Measures at 3 Time Points After Stroke (a)

                                Predictive Validity ([rho])

                                BBS-3P        PASS-3P
                                (BBS) vs BI   (PASS) vs BI

14 days after onset of stroke   .75 (.76)     .78 (.78)
  (n=226)
30 days after onset of stroke   .81 (.81)     .82 (.82)
  (n=202)

(a) BBS-3P=simplified 3-level (0-2-4) Berg Balance Scale,
PASS-3P=simplified 3-level (0-1.5-3) Postural Assessment
Scale for Stroke Patients, BI=Barthel Index. The BI was
administered at 90 days after onset of stroke. The values
shown in parentheses are the [rho] values for the original
BBS and PASS measures.

Table 6.
Responsiveness of the Simplified 3-Level Balance Measures at
Different Stages of Stroke Recovery (a)

                            Standardized Response Mean

                            BBS-3P (BBS)   PASS-3P (PASS)

14-30 days after onset of   0.82 (0.79)    0.86 (0.84)
  stroke (n=202)
30-90 days after onset of   0.70 (0.69)    0.67 (0.65)
  stroke (n=167)
14-90 days after onset of   1.11 (1.07)    1.04 (1.02)
  stroke (n=167)

                            Wilcoxon Z

                            BBS-3P (BBS)        PASS-3P (PASS)

14-30 days after onset of   10.16 * (10.37 *)   10.27 * (10.35 *)
  stroke (n=202)
30-90 days after onset of   8.47 * (8.60 *)     8.10 * (8.17 *)
  stroke (n=167)
14-90 days after onset of   10.35 * (10.37 *)   10.41 * (10.50 *)
  stroke (n=167)

(a) BBS-3P=simplified 3-level (0-2-4) Berg Balance Scale,
PASS-3P=simplified 3-level (0-1.5-3) Postural Assessment
Scale for Stroke Patients. Asterisk indicates P<.0001. The
values in parentheses are for the original BBS and PASS measures.

Table 7.
Responsiveness of the Simplified 3-Level Balance Measures at
Different Stages of Stroke Recovery for Subjects With Different
Levels of Severity (a)

                                   Mild Stroke (n=67)
                                   (b)

                                   BBS-3P        PASS-3P
                                   (BBS)         (PASS)

14-30 days after onset of stroke   0.67 (0.65)   0.61 (0.59)
  (n=202)
30-90 days after onset of stroke   0.37 (0.35)   0.46 (0.43)
  (n=167)
14-90 days after onset of stroke   0.82 (0.78)   0.78 (0.78)
  (n=167)

                                   Moderate Stroke (n=50)
                                   (b)

                                   BBS-3P        PASS-3P
                                   (BBS)         (PASS)

14-30 days after onset of stroke   1.21 (1.19)   1.00 (0.93)
  (n=202)
30-90 days after onset of stroke   0.68 (0.67)   0.56 (0.52)
  (n=167)
14-90 days after onset of stroke   1.35 (1.29)   1.19 (1.12)
  (n=167)

                                   Severe Stroke (n=85)
                                   (b)

                                   BBS-3P        PASS-3P
                                   (BBS)         (PASS)

14-30 days after onset of stroke   0.75 (0.73)   1.01 (1.02)
  (n=202)
30-90 days after onset of stroke   1.00 (0.99)   0.92 (0.92)
  (n=167)
14-90 days after onset of stroke   1.29 (l.26)   1.34 (1.35)
  (n=167)

(a) BBS-3P=simplified 3-level (0-2-4) Berg Balance Scale,
PASS-3P=simplified 3-level (0-1.5-3) Postural Assessment
Scale for Stroke Patients. The values shown in parentheses
are for the original BBS and PASS measures.

(b) Number of subjects tested at 14 to 30 days after onset
of stroke; number of subjects varied at each stage of stroke
recovery.

References

(1) Berg K, Wood-Dauphinee S, Williams JI, Gayton D. Measuring balance in the elderly: preliminary development of an instrument. Physiother Can. 1989;41:304-311.

(2) Benaim C, Perennou DA, Villy J, et al. Validation of a standardized assessment of postural control in stroke patients: the Postural Assessment Scale for Stroke Patients (PASS). Stroke. 1999;30:1862-1868.

(3) Mao HF, Hsueh IP, Tang PF, et al. Analysis and comparison of the psychometric properties of three balance measures for stroke patients. Stroke. 2002;33:1022-1027.

(4) Berg K, Wood-Dauphinee S, Williams JI. The Balance Scale: reliability assessment with elderly residents and patients with an acute stroke. Scand J Rehabil Med. 1995;27:27-36.

(5) Hsueh IP, Mao HF, Huang HL, Hsieh CL. Clinical applications of balance measures in stroke inpatients. Formosan Journal of Medicine. 2001;5:261-268.

(6) Hobart JC, Lamping DL, Freeman JA, et al. Evidence-based measurement: which disability scale lot neurologic rehabilitation? Neurology. 2001;57:639-644.

(7) Hocking C, Williams M, Broad J, Baskett J. Sensitivity of Shah, Vanclay and Cooper's modified Barthel Index. Clin Rehabil. 1999;13: 141-147.

(8) Wood-Dauphinee S, Berg K, Bravo G, Williams JI. The balance scale: responsiveness to clinically meaningful changes. Can J Rehabil. 1997; 10:35-50.

(9) Mahoney FI, Barthel DW. Functional evaluation: the Barthel Index. Md State Med J. 1965;14:61-65.

(10) Hsueh IP, Lee MM, Hsieh CL. Psychometric characteristics of the Barthel activities of daily living index in stroke patients. J Formos Med Assoc. 2001;100:526-532.

(11) Hsueh IP, Lin JH, Jeng JS, Hsieh CL. Comparison of the psychometric characteristics of the functional independence measure, 5-item Barthel index, and 10-item Barthel index in patients with stroke. J Neurol Neurosurg Psychiatry. 2002;73:188-190.

(12) Fugl-Meyer AR, Jaasko L, Leyman I, et al. The post-stroke hemiplegic patient, 1: a method for evaluation of physical performance. Scand J Rehabil Med. 1975;7:13-31.

(13) Duncan PW, Propst M, Nelson SG. Reliability of the Fugl-Meyer assessment of sensorimotor recovery following cerebrovascular accident. Phys Ther. 1983;63:1606-1610.

(14) Berglund K, Fugl-Meyer AR. Upper extremity function in hemiplegia: a cross-validation study of two assessment methods. Scand J Rehabil Med. 1986;18:155-157.

(15) Sanford J, Moreland J, Swanson LR, et al. Reliability of the Fugl-Meyer assessment for testing motor performance in patients following stroke. Phys Ther. 1993;73:447-454.

(16) Public Health Service and Health Care Financing Administration. International Classification of Diseases, 9th Revision, Clinical Modification. 4th ed. Washington, DC: Public Health Service; 1991.

(17) Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet. 1986;1:307-310.

(18) Shrout PE, Fleiss JL. Intraclass correlations: uses in assessing rater reliability. Psychol Bull. 1979;86:420-428.

(19) 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.

(20) Hobart JC, Thompson AJ. The five item Barthel index. J Neurol Neurosurg Psychiatry. 2001;71:225-230.

(21) Deyo RA, Centor RM. Assessing the responsiveness of functional scales to clinical change: an analogy, to diagnostic test performance. J Chronic Dis. 1986;39:897-906.

(22) Guyatt G, Walter S, Norman G. Measuring change over time: assessing the usefulness of evaluative instruments. J Chronic Dis. 1987; 40:171-178.

(23) Hays RD, Morales LS, Reise SP. Item response theory and health outcomes measurement in the 21st century. Med Care. 2000;38 (suppl9):II28-II42.

(24) Cella D, Chang CH. A discussion of item response theory and its applications in health status assessment. Med Care. 2000;38(suppl 9):II66-II72.

(25) Wallace D, Duncan PW, Lai SM. Comparison of the responsiveness of the Barthel Index and the motor component of the Functional Independence Measure in stroke: the impact of using different methods for measuring responsiveness. J Clin Epidemiol. 2002;55:922-928.

(26) Wade DT. Measurement in Neurological Rehabilitation. Oxford, United Kingdom: Oxford University Press; 1992.

CH Wang, PT, BS, is Associate Professor, School of Physical Therapy, College of Medical Technology, Chung Shan Medical University, and Department of Physical Therapy, Chung Shan Medical University Rehabilitation Hospital, Taichung, Taiwan, Republic of China.

IP Hsueh, OT, MA, is Lecturer, School of Occupational Therapy, College of Medicine, National Taiwan University, Taipei, Taiwan, Republic of China.

CF Sheu, PhD, is Associate Professor, Department of Psychology, DePaul University,, Chicago, Ill.

G Yao, PhD, is Associate Professor, Department of Psychology, College of Science, National Taiwan University.

CL Hsieh, OT, PhD, is Associate Professor and Chair, School of Occupational Therapy, College of Medicine, National Taiwan University, Chung-Shan S Rd, Taipei 100, Taiwan, Republic of China ([e-mail]). Address all correspondence to Dr Hsieh.

Mr Wang and Dr Hsieh provided concept/idea/research design, writing, and institutional liaisons. Ms Hsueh provided data collection, and Mr Wang and Dr Yao provided data analysis. Mr Wang provided subjects and clerical support. Ms Hsueh and Dr Hsieh provided project management. Mr Wang and Dr Hsieh provided fund procurement and institutional liaisons. Ms Hsueh, Dr Sheu, Dr Yao, and Dr Hsieh provided consultation (including review of manuscript before submission).

This study was approved by the institutional review board of Chung Shan Medical University Rehabilitation Hospital.

This study, was supported by research grants from the National Science Council (NSC-90-2314-B-002-325) and Chung Shan Medical University (CSMU 91-OM-B-018).

This article was received August 24, 2003, and was accepted November 20, 2003.