INTRODUCTION

Loss of balance is one of the chief complaints after stroke.1,2 The weakness and numbness on one side of the body that result from stroke can cause difficulty in maintaining stability for sitting and standing. It has been revealed that 2 months after stroke onset, more than half of patients still have severe balance deficits and have not regained the ability to stand independently.3 Loss of balance is associated with increased fall risk, dependence in activities of daily living, and decreased walking ability.4,5 Thus, a primary goal of stroke rehabilitation is improvement in balance function.

Responsiveness can be defined as the ability of an outcome measure to detect change over time in the construct to be measured.6 It is a necessary characteristic of an outcome measure that is used for monitoring changes over time within individuals and for demonstrating the effects of particular treatments of patients' balance deficits.6 Because patients with severe balance deficits may have slower recovery,7-9 they need to be assessed with balance measures that have high responsiveness if subtle improvements in balance function are to be detected.

The Postural Assessment Scale for Stroke (PASS)10 and the Berg Balance Scale (BBS)11 are 2 commonly used and well-validated outcome measures of balance function in patients with stroke.12,13 The PASS was specifically designed for patients with stroke. It contains 12 items assessing the abilities to maintain or change a lying, sitting, or standing posture. The BBS was originally designed to assess balance function in community-dwelling older adults.11 According to a Canadian National Survey,14 the BBS has been recognized as one of the most commonly used measures across the continuum of stroke rehabilitation. One of the 14 items in the BBS assesses static sitting balance, and the rest assess sit-to-stand, postural transitions during standing, or stability of a standing posture. Both the PASS and the BBS have good reliability, validity, and responsiveness in patients with stroke, and both are promising outcome measures for all stroke populations in clinical and research settings.15-19

Previous studies showed that the PASS and the BBS both had good responsiveness in heterogeneous samples with a wide range of stroke severity levels.15,20,21 That is, the 2 measures indicated similarly large changes in the study samples. However, the similarity of responsiveness between the PASS and the BBS is doubtful. The PASS has more items assessing basic balance abilities, such as bed mobility and postural transition during lying and sitting, whereas the BBS has only one item assessing sitting balance, which is the minimum requirement of balance function in the BBS. The greater number of low-difficulty items in the PASS would be useful for demonstrating patients' changes in basic balance function, especially those of patients with severe balance deficits. For instance, if patients have progressed from "unable to roll on the bed" to "able to sit up from lying," the PASS may be able to detect the improvement in balance function, but the BBS may show no change in balance function. Additionally, in terms of the items assessing the same postural transition, the passing criteria of the score categories in the PASS are lower than those in the BBS. For example, when a patient improves from "unable to stand up" to "able to stand up with maximal assistance," the score will change from a 0 to a 1 on the PASS but remain a 0 on the BBS. Two studies also found a floor effect when the BBS was administered to patients at 14 days after stroke,15,22 implying that the BBS may have difficulty in distinguishing differences in balance function within individual patients whose pretest and posttest scores both cluster at the bottom of the scale.

To date, no studies have tested the differences in responsiveness between the PASS and the BBS exclusively in patients with stroke who have severe balance deficits, which complicates the selection of an optimal outcome measure for this population. Thus, the purpose of this study was to compare the responsiveness of the PASS with that of the BBS in patients with stroke having severe balance deficits.

METHODS

Participants

The data were retrieved from a previous longitudinal study,23 wherein a consecutive sample of 301 patients with stroke was recruited from the inpatient rehabilitation ward in a medical center. Among them, 212 completed the assessments of the PASS and the BBS at both 14 and 30 days after stroke. Patients were recruited for the previous study if they (1) had first onset of stroke within 14 days before hospitalization and (2) could follow instructions. Patients who had a recurrent stroke during the period 14 to 30 days after stroke or other major diseases (eg, arthritis and Parkinson's disease) which might have affected their balance function were excluded. The study was approved by the institutional review board of the medical center. Written informed consent was obtained from all participants.

In this study, the data of those patients who could not stand independently were selected from the dataset of 212 patients for analysis of the responsiveness of the PASS and the BBS in patients with severe balance deficits. This study only included patients who were scored 0 on both the item of "standing without support" in the PASS and that of "standing unsupported" in the BBS at 14 days after stroke.

Procedure

In the previous study,23 patients with stroke who were consecutively admitted or transferred to the inpatient rehabilitation ward were approached by a research assistant and evaluated for eligibility based on the selection criteria. Each eligible patient was assessed with both the PASS and the BBS by a trained occupational therapist, who was blind to the study purpose. The assessments were administered at both 14 and 30 days after stroke by the same therapist in the inpatient rehabilitation ward or the patient's home. All patients received regular inpatient rehabilitation programs, including balance training, during the study period.

Outcome Measures

PASS

The PASS has 12 items specially designed for evaluating balance function in patients with stroke.10 Each item is scored on a 4-point (0-3 points) scale. The total score ranges from 0 to 36, and higher scores indicate better balance function. The PASS has good internal consistency (Cronbach's [alpha] = 0.94-0.96), interrater reliability (interclass correlation coefficient = 0.97), concurrent validity (high correlation [r = 0.95] with the BBS), and both group-level (effect size = 0.87-0.89) and individual-level responsiveness (53% of the patients with clinically significant change) in patients with subacute stroke who have different severity levels of balance deficit.15-17

BBS

The BBS has 14 items designed to assess balance function in community-dwelling older adults.11 Each item is scored on a 5-point (0-4 points) scale. The total score ranges from 0 to 56, and higher scores indicate better balance function. The BBS has good internal consistency (Cronbach's [alpha] = 0.92-0.98), interrater reliability (interclass correlation coefficient = 0.97), concurrent validity (high correlations [r = 0.91-0.95] with the PASS and the Static Balance Test), and both group-level (effect size = 0.79-0.80) and individual-level responsiveness (47% of the patients with clinically significant change) in patients with subacute stroke who have different severity levels of balance deficits.15,18,19

Data Analysis

The responsiveness of the PASS was compared with that of the BBS at both the group and individual levels. At the group level, the responsiveness of both measures was examined with the standardized response mean (SRM). The SRM was used to quantify the size of the change scores between 14 and 30 days after stroke and was calculated by dividing the mean change score by the standard deviation of the change scores. SRM values 0.20 or more, 0.50 or more, and 0.80 or more indicated small, moderate, and large changes, respectively.24 Previous evidence showed that patients achieved moderate improvement in balance function within the first month after stroke.25 Thus, if the PASS or the BBS captured a moderate or large change, this result could indicate that the measure had good group-level responsiveness.

At the individual level, the percentages of patients shown by the PASS and the BBS to have clinically significant improvement were calculated. Clinically significant improvement within each individual patient was defined by showing improvements (change scores) beyond the minimal detectable change with 95% confidence (MDC₉₅) of the corresponding balance measure.16 That is, a patient's change score of more than 3.2 points on the PASS or 6.7 points on the BBS indicated clinically significant improvement.26 The MDC₉₅ is the 95% confidence interval (CI) of the distribution of change scores that can be expected if no actual change has occurred between 2 successive assessments.27 Namely, a difference between 2 successive assessments exceeding the MDC₉₅ indicates a clinically significant change in an individual patient.

The responsiveness of the PASS was further compared with that of the BBS using the bootstrap method.28 The pairwise difference of the SRMs and that of the percentages of patients with clinically significant improvement were calculated in each of the 10 000 bootstrap samples. The 95.0% bootstrap percentile CI of differences was computed to examine whether the difference of responsiveness between the 2 measures was significant. A 95% CI that did not contain 0 indicated a significant difference in the group-level or individual-level responsiveness between both measures. If both the lower and upper limits of a 95% CI were less than 0, this result indicated that the PASS was significantly less responsive than the BBS. In contrast, if both limits of a 95% CI were greater than 0, this result indicated that the PASS was significantly more responsive than the BBS.

RESULTS

A total of 49 patients' PASS and BBS scores at 14 and 30 days after stroke were retrieved in this study. The demographic and clinical characteristics of these patients are shown in Table 1. The mean age of the patients was 69.8 years, and the majority of the patients (85.7%) were 60 years or older. Most patients were severely dependent in activities of daily living, as shown by the scores of the Barthel Index. In addition, their motor functions (upper extremity and lower extremity), on average, were moderately to severely impaired.

Table 1. Demographic and Clinical Characteristics of the Participants (N = 49)

Table 2 shows the group-level and individual-level responsiveness of both measures. At 14 and 30 days after stroke, the mean scores of the PASS were 2.2 and 6.4, respectively, and those of the BBS were 0.3 and 2.4, respectively. At the group level, the PASS indicated moderate changes (SRM = 0.79), while the BBS indicated small changes (SRM = 0.39). The 95.0% CI of the difference (0.18-0.59) of the group-level responsiveness between the PASS and the BBS was greater than 0.

Table 2. Group- and Individual-Level Responsiveness Between 14 and 30 Days After Stroke of the PASS and the BBS in Patients With Stroke Having Severe Balance Deficits (N = 49)

At the individual level, the PASS and the BBS showed that 42.9% and 6.1% of the patients, respectively, achieved clinically significant improvements in balance function. The comparison of the individual-level responsiveness showed that the 95.0% CI of the percentage differences ranged from 24.5% to 51.0%, which was greater than 0.

DISCUSSION

Older adult patients with stroke, especially those with severe impairments,29 have higher risks of falling and fall-related injuries than older community-dwelling adults without stroke,30,31 and these risks can have significant adverse effects on rehabilitation. Thus, assessment of balance is critical in patients with severe balance deficits following stroke. In this study, the responsiveness of the PASS and that of the BBS were compared in patients with stroke having severe balance deficits. The results showed that at both the group and individual patient levels, the responsiveness of the PASS was significantly better than that of the BBS. At the group level, the PASS indicated moderate changes (SRM = 0.79), while the BBS indicated only small changes (SRM = 0.39). At the individual level, the PASS showed that 42.9% of the patients had clinically significant improvements in balance function, while the BBS showed that only 6.1% of the patients had clinically significant improvements. The findings imply that the PASS is able to detect improvements in balance in a group of patients or in individual patients with severe balance deficits better than the BBS.

Our finding showed a similar trend to 2 previous validation studies comparing the group-level responsiveness of the PASS and that of the BBS in different subgroups of stroke severity.15,21 In those 2 studies, the PASS indicated large group changes (effect sizes were 1.01-1.10), while the BBS indicated moderate changes (effect sizes were 0.64-0.75) in patients with severe stroke between 14 and 30 days after stroke.15,21 Additionally, when used to assess balance function in patients with mild or moderate stroke, both the PASS and the BBS could detect similar magnitudes of group changes in these populations. The effect sizes of the PASS and the BBS were 0.60 to 0.61 and 0.67 to 0.75 in mild stroke, and 1.00 to 1.14 and 1.21 to 1.24 in moderate stroke, respectively.15,21 Thus, as compared with the BBS, the PASS is in general a similarly responsive or more responsive outcome measure of balance in patients with stroke across different levels of severity.

The better responsiveness of the PASS than the BBS in patients with severe balance deficits can be explained by the spread of item difficulties. The PASS was designed to assess balance in patients with stroke, and it has more items with low balance demands. The BBS only has one item (unsupported sitting) that does not require assessing balance in a standing position. In addition, patients with initially severe stroke have shown slower and poorer recovery.7-9 Therefore, if those patients have not regained their ability to stand independently, their scores may be clustered at the bottom of the BBS scale at admission and discharge, and they may show no balance improvements during rehabilitation.

Although the BBS is one of the most widely used balance outcome measures in stroke rehabilitation,14 our findings imply that the BBS, as compared with the PASS, may underestimate the effect of a particular balance treatment in patients with severe balance deficits at both the group and individual patient levels. At the group level, the BBS would be less able to detect the average change in balance function of a study sample and thereby may mislead researchers to underestimate the magnitude of the treatment effect. At the individual level, the BBS would be less able to detect the clinically significant improvement of an individual patient and thereby may mislead clinicians to underestimate the number of patients that would benefit from a specific balance treatment. Consequently, we recommend the PASS, rather the BBS, as a responsive outcome measure of balance for individuals with stroke, especially those who have severe balance deficits.

Selecting a responsive outcome measure for a target population is particularly important for clinicians and researchers in long-term follow-up of the changes within individual patients. Secondary to personal or institutional preference, some clinicians may be accustomed to using a uniform outcome measure in the whole patient population regardless of their different severities of balance deficits. However, this may lead to underestimation of the treatment effects. Moreover, when assessing the change in balance within an individual patient, clinicians may be unlikely to employ different outcome measures that are respectively suited to the patient's levels of balance function at different time points secondary to the difficulty of score transformations between different rating scales. Thus, based on our findings, if the patients have severe balance deficit (such as the inability to stand independently) at baseline, clinicians and researchers should choose the PASS, rather than the BBS, as a pretreatment and posttreatment outcome measure in order to better capture their changes in balance.

This study examined the responsiveness at both the group and individual levels. Outcome measures with group-level responsiveness are able to detect the average treatment effects in a group of patients in research or in an entire unit. On the other hand, outcome measures with individual-level responsiveness are able to show treatment effects in individual patients in both clinical and research settings. Individual-level indices can provide unique information that researchers can use to translate research evidence into clinical practice. Particularly, group- and individual-level responsiveness indices may show different results for competing measures.16,18,32,33 Thus, comprehensive evidence of responsiveness at both levels is critical for users in choosing outcome measures.

One limitation of this study is noted. Although the sample size of 49 in this study seems acceptable for examining responsiveness,34 the participants were recruited from only one medical center. This limitation might have threatened the generalizability of our findings in other institutions. There is a need to further validate our findings with a larger and more representative sample.

CONCLUSIONS

This is the first study to compare the responsiveness of the PASS with that of the BBS exclusively in patients with severe balance deficits following stroke. Our results showed that the responsiveness of the PASS was significantly better than that of the BBS both in a group of patients and in individual patients with severe balance deficits. These findings imply that at the group level, the PASS is better able than the BBS to detect the magnitude of the treatment effect in a study sample, and at the individual level, the PASS is better able than the BBS to capture the number of patients that would benefit from a specific balance treatment. Therefore, the PASS is recommended as an outcome measure for detecting change in balance in patients with stroke having severe balance deficits.

REFERENCES