Comparing the Postural Assessment Scale for Stroke and Berg Balance Scale for predicting community walking ability at discharge in subacute stroke: a prospective cohort study

Introduction

Stroke is a leading cause of long-term disability worldwide, often resulting in significant mobility impairments and dependence on caregivers for daily activities (Katan & Luft, 2018; Lee et al., 2024). These challenges are primarily due to a combination of motor weakness, sensory deficits, motor incoordination, spasticity, and balance issues (Lee, 2019). Previous studies have shown that balance and trunk control are key predictors of post-stroke walking ability, making the selection of appropriate balance assessment tools crucial for optimizing rehabilitation (Preston et al., 2021).

Three widely used tools for balance assessment in stroke rehabilitation are the Berg Balance Scale (BBS), the Postural Assessment Scale for Stroke Patients (PASS), and the Mini-Balance Evaluation Systems Test (Mini-BESTest). The BBS is commonly used for balance assessment, given its capacity to evaluate fall risk and functional stability through a series of tasks (Blum & Korner-Bitensky, 2008; Inoue et al., 2023). It measures both static and dynamic balance, providing comprehensive insights into a patient’s functional stability, and typically takes about 15–20 min to administer (Dos Santos et al., 2023; Inoue et al., 2023). However, its utility is limited by floor and ceiling effects, which reduce its sensitivity for patients with severe impairments or near-complete recovery (Blum & Korner-Bitensky, 2008; Louie & Eng, 2018). In contrast, the PASS is specifically designed for patients with stroke and emphasizes postural control during static and dynamic tasks relevant to daily activities (Benaim et al., 1999; Chien et al., 2007; Yoshimoto et al., 2016). It is quicker to administer (10–15 min) and particularly suited for individuals with significant functional limitations, demonstrating predictive validity with scores above 12.5 correlating with independent ambulation (Huang et al., 2016). The Mini-BESTest is another commonly used balance assessment tool that focuses on dynamic balance components, including anticipatory postural adjustments, reactive postural control, sensory orientation, and dynamic gait (Tsang et al., 2013). It has demonstrated excellent reliability and predictive validity for fall risk across various neurological conditions, including stroke (Tsang et al., 2013). However, Mini-BESTest is more comprehensive and time-intensive, making it less practical for routine assessments in rehabilitation settings. Additionally, its primary focus on dynamic balance may not fully capture the postural control deficits that are critical for early-stage stroke rehabilitation. Given these considerations, this study compares the predictive accuracy of PASS and BBS for community walking ability at discharge.

While both tools offer valuable insights into balance and postural control, accurately predicting community ambulation after stroke rehabilitation remains challenging. Walking speed alone is insufficient to reliably predict community walking capability (Alvarenga et al., 2023; Fulk et al., 2017). Combining the 6-Minute Walking Distance (6MWD) and Functional Ambulation Categories (FAC) provides a more comprehensive evaluation. The 6MWD, a simple and safe measure of walking endurance, distinguishes between house-bound and community ambulators with a threshold of 205 meters (71% specificity, 79% sensitivity, 74% overall accuracy) (Fulk & He, 2018; Fulk et al., 2017; Kubo et al., 2018). The FAC, which categorizes ambulation across six levels based on assistance required, complements this by assessing functional independence in walking (Mehrholz et al., 2007). However, neither tool fully accounts for balance deficits or other factors influencing community walking, highlighting the need for a multifaceted approach to prediction.

The subacute phase of stroke recovery, spanning from 7 days to 6 months post-onset, is a critical period for intervention, where timely and accurate assessments can guide targeted rehabilitation strategies to maximize functional outcomes (Bernhardt et al., 2017). While the BBS and PASS are widely used, there is limited evidence directly comparing their utility during this phase to predict walking ability at discharge. The PASS was specifically developed to assess both static and dynamic balance in patients with stroke. It is particularly effective in detecting balance improvements in individuals with severe balance deficits, an area where the BBS and Mini-BESTest may have limitations (Huang et al., 2020). The Mini-BESTest, though highly reliable and useful in assessing dynamic balance and fall risk, is more commonly used across various neurological conditions rather than specifically for post-stroke rehabilitation. Furthermore, PASS has demonstrated strong predictive validity for ambulation and functional independence, making it an appropriate choice for evaluating community walking ability post-stroke (Huang et al., 2016). Given these considerations, this study compares PASS and BBS to determine their predictive accuracy for community walking ability at discharge. The BBS’s susceptibility to floor and ceiling effects contrasts with the PASS’s strength in assessing patients with greater impairments, suggesting that their predictive accuracy may differ depending on the patient’s functional level and stage of recovery.

Given the strengths and limitations of both tools, a comparative study is warranted to determine which tool more accurately predicts community walking ability in patients with subacute stroke at discharge from rehabilitation. By integrating balance assessments with established measures such as 6MWD and FAC, this study aims to provide evidence-based insights to enhance clinical decision-making, optimize tool selection, and ultimately improve rehabilitation outcomes for patients with stroke.

Materials and Methods

Results

Of the 47 participants enrolled, 46 completed both admission and discharge assessments. One participant dropped out due to a COVID-19 infection at the time of discharge, and no outcome was recorded. This dropout was unrelated to the study or rehabilitation program and is unlikely to have introduced bias into the results. Thus, 46 patients completed the study. Of these, 18 were classified as community ambulators and 28 as home ambulators. Baseline demographic and clinical characteristics of participants are presented in Table 1. The characteristics include variables such as age, gender, time since stroke onset, type of stroke, affected side, number of rehabilitation sessions, and Modified Rankin Scale (mRS) scores to reflect stroke severity. The majority of participants had moderate to severe disability at baseline, with 47.8% at mRS level 4 and 15.2% at mRS level 5, indicating significant functional limitations.

Mean scores for the PASS and BBS assessments showed significant differences between community and home ambulators, with community ambulators scoring higher on both scales. Details of the mean scores, standard deviations, and statistical comparisons are provided in Table 2.

Both the PASS and BBS demonstrated high accuracy in predicting community walking abilities, with AUC values of 0.955 (95% CI [0.850–0.994]) and 0.991 (95% CI [0.906–1.000]), respectively, indicating excellent discriminative power (Fig. 1 and Table 3). The optimal cut-off scores were ≥28 for PASS and ≥46 for BBS, both showing high sensitivity (94.44%, 95% CI [72.7–99.9]) and specificity values of 85.71% (95% CI [67.3–96.0]) for PASS and 96.43% (95% CI [81.7–99.9]) for BBS. The BBS had a stronger specificity, suggesting it may be slightly more reliable for identifying community ambulators. The positive likelihood ratios (LR+) of 6.61 (95% CI [2.65–16.49]) for PASS and 26.44 (95% CI [3.85–181.81]) for BBS indicate that BBS has a greater utility in confirming community ambulation. Both scales had negative likelihood ratios (LR-) below 0.1, emphasizing their effectiveness in ruling out non-community ambulators, thus supporting their strong clinical decision-making utility. Further details on these metrics can be found in Table 3.

Analysis of floor and ceiling effects showed that neither the PASS nor the BBS had significant floor effects, with both scoring 0%. This suggests that both scales are effective in evaluating lower levels of postural control and balance ability. The PASS had a slight ceiling effect of 4.35%, well below the 20% threshold that would indicate a significant issue. The BBS had no ceiling effect (0%), suggesting it may be more effective in distinguishing higher levels of balance performance.

A pairwise comparison of ROC curves showed that the AUCs of PASS and BBS differed by 0.04, with a standard error of 0.02. The 95% confidence interval for this difference ranged from −0.01 to 0.08, yielding a z statistic of 1.68, which was not statistically significant (p = 0.093). This result indicates that both instruments perform similarly in predicting community ambulation among patients with subacute stroke after rehabilitation. High AUC values and favorable likelihood ratios further support the utility of PASS and BBS for predicting walking ability at discharge.

Discussion

This study demonstrated that both the BBS and PASS are effective tools for predicting walking ability at discharge in patients with subacute stroke undergoing rehabilitation, consistent with prior research on their role in balance assessment and functional outcome prediction. While both tools demonstrated high overall accuracy, the BBS showed slightly higher specificity. This specificity can be attributed to its inclusion of advanced balance tasks that assess critical balance and coordination abilities necessary for community ambulation.

The BBS, with its high specificity (93.33%) and a robust LR+ of 26.44, was particularly effective at identifying community ambulators. Patients scoring above the cut-off of 46 were 26 times more likely to achieve community ambulation, highlighting the BBS as a valuable clinical tool. This specificity can be attributed to the advanced balance tasks included in the BBS, which assess critical balance and coordination abilities necessary for community ambulation (Kim & Oh, 2023). For instance, tandem stance challenges the patient’s ability to maintain balance with a narrow base of support, mimicking scenarios such as walking in crowded spaces. Similarly, single-leg stance requires significant lower limb strength and trunk control, which are crucial for tasks like stair climbing or recovering from balance disturbances (Delfa-de-la-Morena et al., 2024; Labanca et al., 2021). These tasks collectively provide a comprehensive evaluation of both static and dynamic balance abilities, making the BBS highly effective for identifying community ambulators. However, these advanced tasks may also lead to floor effects, limiting the BBS’s applicability for patients with severe impairments.

In contrast, the PASS focuses on fundamental postural control and includes tasks such as sitting without support, rolling, and transitioning from sitting to standing (Estrada-Barranco et al., 2021). These foundational tasks are particularly effective for evaluating patients with severe impairments, as they target core stability and postural control essential for recovery, including those in the early subacute phase (1 week to 3 months post-stroke) when neuroplasticity is heightened. For example, the “rolling” task assesses the ability to initiate movement and maintain control while changing positions, which is critical in the early recovery stages. “Sitting without support” measures core stability, enabling patients to safely perform seated tasks and progress to more advanced activities. Additionally, “transitioning from sitting to standing” evaluates lower limb strength and balance, a key milestone in regaining mobility. The inclusion of such basic tasks makes PASS sensitive to capturing small but clinically meaningful improvements in balance. Moreover, tasks like “standing without support” and “standing on one leg” allow the PASS to assess more advanced postural skills, bridging early and moderate recovery stages.

The findings from this study are consistent with previous research, including Louie & Eng (2018), who also demonstrated the predictive value of the BBS for walking improvement in patients with stroke. However, our study offers several novel contributions compared to their work. Employing a prospective design, our research enabled real-time data collection and monitoring of patients with stroke during their rehabilitation journey, reducing biases inherent in retrospective analyses and strengthening predictive validity. Unlike Louie and Eng’s study, which evaluated only the BBS, we directly compared the BBS and PASS, providing a head-to-head assessment of two commonly used tools to predict walking ability at discharge. This dual assessment offers clinicians a more comprehensive understanding of the relative effectiveness of each tool and their applicability for patients with varying functional levels. Furthermore, we used a combined outcome measure of the 6MWD and FAC at discharge to classify community walkers, integrating both endurance and functional mobility to provide a more nuanced understanding of walking readiness—factors not addressed in Louie and Eng’s study.

Compared to Wang, Chen & Wang (2022), our study provides a more immediate evaluation of walking ability. Wang, Chen & Wang (2022) assessed walking status 3 months post-stroke using telephone interviews, a method that, while valuable, is limited by its reliance on subjective reporting and delayed outcome collection. By contrast, our study assessed walking ability at discharge using objective measures, including the 6MWD and FAC. This approach offers a real-time assessment of walking capability, which is critical for tailoring post-discharge care plans and predicting readiness for community ambulation. Furthermore, Wang, Chen & Wang (2022) emphasized the correlation between PASS scores at admission and walking status at 3 months post-stroke, underscoring the importance of early balance assessments. Our findings expand on this by directly comparing PASS with BBS and highlighting their complementary roles across different recovery stages.

We also addressed concerns regarding ceiling and floor effects for these scales. While earlier research reported such effects for both PASS and BBS (Benaim et al., 1999; Blum & Korner-Bitensky, 2008; Mao et al., 2002), we found no significant floor effects for either tool (0%). The PASS had a slight ceiling effect (4.35%), well below the 20% threshold of concern, whereas the BBS had no ceiling effect. These findings suggest that both scales effectively measure a wide range of abilities, with the BBS offering a slight advantage in distinguishing higher levels of balance performance. The absence of significant floor effects highlights the suitability of these tools for severely impaired patients, particularly the PASS, while the minimal ceiling effects ensure their utility in capturing progress across recovery stages.

In clinical practice, the choice between PASS and BBS should be guided by the patient’s recovery stage and functional ability. PASS’s simplicity and minimal equipment requirements make it ideal for resource-limited settings or for evaluating patients with severe impairments and early subacute phase of stroke survivors. The trunk control items of the PASS (Item 1: sitting without support, Item 6: supine to affected side lateral, Item 7: supine to non-affected side lateral, Item 8: supine to sitting up on the edge of the table, and Item 9: sitting on the edge of the table to supine) have been shown to predict activities of daily living (ADL) function 1 year after stroke (Wang et al., 2005). However, the PASS-Trunk Control (PASS-TC) exhibited a notable ceiling effect at 30% of the sample and had limited responsiveness beyond the first 30 days post-stroke, suggesting that it is most effective for early recovery assessment rather than long-term evaluation. In contrast, the BBS provides a more detailed assessment of balance abilities required for community ambulation and may be better suited for discharge planning or outpatient rehabilitation. Together, these tools offer a continuum of assessment: PASS is effective for capturing early recovery progress, while BBS is more appropriate for evaluating readiness for community ambulation. The assessment of trunk balance using the PASS can detect small but clinically meaningful changes in postural control, even in patients with severe impairments or highly deteriorated postural control. Their sequential use allows clinicians to tailor rehabilitation strategies to individual patient needs, optimizing outcomes.

Our study closed a gap in previous research by directly comparing PASS and BBS in the subacute phase of stroke recovery upon discharge from rehabilitation. This head-to-head comparison highlights the practical implications of our findings, emphasizing both tools’ effectiveness. Although the BBS showed slightly better performance with a higher LR+ (26.44) than PASS (6.61), PASS’s ease of use and shorter administration time (10–15 min) make it a practical choice in clinical settings where time efficiency is crucial. In contrast, the BBS requires 20–30 min to administer, which may be less convenient in busy environments.

Conclusions

The high predictive accuracy of the PASS and BBS suggests that clinicians can effectively use these tools to evaluate and predict community walking ability after rehabilitation. Cut-off scores of >28 for PASS and >46 for BBS distinguish between those likely to achieve community ambulation and those suited for home ambulation. The PASS’s ease of use makes it ideal for quick assessments, while the BBS’s higher LR+ and accuracy support its use for more detailed evaluations. Clinicians should consider patient needs and clinical context when selecting the most appropriate tool to optimize rehabilitation outcomes.

Supplemental Information