Two Balance Measures as Poststroke Predictors of Ambulation Status at Discharge From Inpatient Rehabilitation : Journal of Acute Care Physical Therapy

Secondary Logo

Journal Logo


Two Balance Measures as Poststroke Predictors of Ambulation Status at Discharge From Inpatient Rehabilitation

Berry, Olivia; Voigtmann, Christina; Curran, Christopher; Dawson, Nicole; Dominguez, Jose; Beato, Morris

Author Information
Journal of Acute Care Physical Therapy: July 2022 - Volume 13 - Issue 3 - p 126-134
doi: 10.1097/JAT.0000000000000186
  • Free


On average, every 40 seconds, someone in the United States has a stroke.1 Impaired mobility is characterized as the most severe consequence after a stroke, with less than half of stroke survivors returning to independent community ambulation.2,3 Only a portion of stroke survivors achieve the ability to walk outdoors and within their communities safely. Early assessment of walking speed is a strong indicator of future walking ability after stroke, wherein decreased velocity indicates greater mobility impairment.3 Previous work by Perry et al3 determined that the ability to walk household or community distances could be predicted by gait velocity. Schmid et al4 found that a change to a higher class of ambulation, such as household ambulators to limited community ambulators as measured by gait speed, results in better function and quality of life in subacute stroke survivors. Unassisted walking at 6 months can be predicted within days after a stroke using balance and strength measures5–8; however, our ability to predict the degree to which patients' walking abilities reach speeds appropriate for household or community mobility requires further investigation.

Most stroke survivors require specialized care at an inpatient rehabilitation facility (IRF).1 Inability to walk without physical assistance is the primary reason for referral to inpatient rehabilitation.9 The purposes of neurological rehabilitation in the acute stages after a stroke are to increase independence with activities of daily living, maximize functional recovery in the period of most rapid neurological recovery, and reintegrate patients back into their environment. Clinicians use standardized outcome measures, such as the Functional Independence Measure (FIM), to assess functional mobility, design intervention paradigms, and evaluate progress. In stroke rehabilitation, frequently used clinical outcome measures include those recommended by the Academy of Neurologic Physical Therapy (ANPT) Stroke Evidence Database to Guide Effectiveness (StrokEDGE) and Core Set of Outcome Measures for Adults with Neurologic Condition Clinician Practice Guideline. The Berg Balance Scale (BBS) and the Postural Assessment Scale for Stroke (PASS) are used to assess balance, and the 10-m walk test (10MWT) to assess walking ability.10 Measures such as these are used to anticipate medical and equipment needs early in a patient's inpatient rehabilitation program, allowing for a smoother transition of care upon discharge. In the United States, the average length of stay for patients poststroke in a comprehensive IRF is decreasing within the Medicare model. Currently, patients remain in an inpatient setting for about 12 to 13 days, with an average annual change of −0.4 days.11,12 Considering the shorter lengths of stay, maximizing the efficiency of services is essential as is providing selective skilled therapy interventions with the patient's functional prognosis in mind. Using assessment results at admission to predict walking ability at discharge can be useful to clinicians to begin discharge preparations early on in a patient's stay and to better structure a targeted treatment plan.13

The primary aim of this study was to compare the admission scores of 2 commonly used balance outcome measures in poststroke rehabilitation, the BBS and the PASS, regarding their ability to predict ambulatory capacity and classification of walking performance upon discharge from the IRF. Furthermore, this study aimed to determine potential scores on each of these outcome measures that could assist clinicians in differentiating between patients who will be nonambulatory versus those who will be able to walk at a speed suitable for the community at the time of discharge from the IRF.


Participants and Setting

Participants in this study were adults with an admitting diagnosis of stroke admitted to a comprehensive IRF from February 2019 to November 2019. The rehabilitation unit includes a stroke rehabilitative program certified by the Commission on Accreditation of Rehabilitation Facilities. Patients admitted to the unit receive 3 hours of therapy (any combination of physical, occupational, and speech therapy) for a minimum of 5 days per week. Therapy evaluations, including outcome measures, are performed within 24 hours of admission. Interventions were determined by the treating physical therapists and included but were not limited to conventional therapy interventions (balance training, transfer training, muscle strengthening, etc), various gait training techniques (overground gait training, bodyweight-supported treadmill training, and robotic-assisted gait training), functional electrical stimulation, and use of ankle-foot orthoses.

Inclusion criteria were as follows: 18 years and older with an admitting diagnosis of cerebral vascular accident confirmed by imaging. Exclusion criteria included unexpected discharge from the inpatient rehabilitation unit to a higher level of care due to medical instability and individuals with a length of stay beyond 30 days because of social or environmental barriers preventing discharge when functionally appropriate. This retrospective study was deemed exempt under Category 4, secondary research for which consent is not required by the Orlando Health and University of Central Florida Institutional Review Board. Further, the Orlando Health Institutional Review Board approved the HIPAA (Health Insurance Portability and Accountability Act) waiver of authorization for this project.


All subjects were assessed with the BBS, 10MWT, and PASS within 2 days of admission to the inpatient rehabilitation unit and reassessed within 2 days of discharge from the facility. Admission and discharge assessments were completed by physical therapists or physical therapist assistants who were formally trained in completing the standardized assessments according to published protocols.

Gait Speed

Gait speed was assessed using the 10MWT. This test is a responsive assessment tool for evaluating walking speed in individuals with hemiparesis in the acute stages of stroke.14 In this testing procedure, the total time taken to ambulate 6 m is recorded with an additional 2 m on each end to account for acceleration and deceleration.15 Two trials are administered at an individual's comfortable, or self-selected, walking speed and 2 trials at an individual's fast walking speed. The 2 trials, for each speed, are averaged and reported in meters per second.15 Patients may walk with an assistive device, bracing, and no more than minimal assistance from the therapist. Gait speed was reported as “0 m/s” if the patient could not ambulate or the amount of assistance required affected the speed of forward propulsion.10

Following data collection, patients were further classified into ambulation classification using the calculated gait speeds based on Perry et al.3 Gait speed was determined to be efficient in predicting ambulation classification. Patients were stratified using the following classification: nonambulators (0 m/s); household ambulators (<0.4 m/s); limited community ambulators (0.4-0.8 m/s); and community ambulators (>0.8 m/s).

Balance Measures

The ANPT, a section of the American Physical Therapy Association, published stroke-specific outcome measure recommendations for the inpatient and outpatient settings via the StrokEDGE II task force in 2018. Within this recommendation were 2 of the balance measures used in this study, the PASS and the BBS. Both measures are commonly used in individuals poststroke to assess postural control with static and dynamic tasks. Both measures are highly correlated with the FIM, a functional assessment widely used in IRFs. The FIM has excellent internal consistency and validity in the poststroke population.11 It measures 18 items, including 13 motor tasks and 5 cognitive tasks that are deemed essential to activities of daily living. It is generally assessed at admission and discharge, with higher scores indicating greater independence in function.11

The BBS is a clinician-rated scale consisting of 14 balance items designed to assess the ability to maintain posture in static positions and adapt posture during functional movements. Sixteen items are scored on a 5-point ordinal scale ranging from 0 to 4 for a maximum score of 56, with higher scores indicating better ability to complete the task. Scoring is performed relative to time and level of assistance required. This test requires less than 20 minutes to administer and requires a stopwatch, standard height chair with and without armrests, stool, ruler, and shoe. The BBS has been shown to have excellent psychometric properties in the poststroke population.16–19

The PASS is a disease-specific outcome measure developed to assess postural control after a stroke. It is a highly recommended measure by the ANPT StrokEDGE II for use in the inpatient rehabilitation setting, with the greatest sensitivity in the first 3 months.20 Both the interrater and intrarater reliabilities of the PASS are high in patients with acute stroke.20,21 The PASS consists of 12 items assessed in sitting, standing, and supine positions; and is further divided into 2 subsections: maintaining a posture and changing a posture. Scoring is performed on a 4-point scale from 0 to 3 for a total score of 36, with higher scores indicating better performance with criteria. This measure takes approximately 10 minutes to administer and requires an examination table, stopwatch, and pen. The PASS was designed to be applicable to all poststroke patients regardless of the level of functional mobility. The test assesses postural control both statically and during positional changes with increasing levels of difficulty throughout the test.18 Differences between these 2 measures lie in the range of difficulty of items leading to ceiling effects on the PASS in individuals 90 days poststroke,20 and floor effects on the BBS in individuals 14 days poststroke.18,19 The BBS cannot adequately assess balance in positions other than standing, while the PASS assesses balance tasks in both sitting and supine positions.

Statistical Analysis

Data analysis was performed using SPSS Statistical Software (Version 22.0, IBM Statistics). Only patients who completed the full battery of outcome measures were included. After the initial data review, simple descriptive statistics (standard deviation, percentage, mean, median, and mode) were calculated for patient demographics, admission gait speed, and admission balance measures. To allow for sample representation, frequencies were calculated for gender, stroke hemisphere, discharge location, and admission gait classification.

Paired t tests were used to identify differences between admission and discharge balance and gait speed measures and the FIM admission and discharge mobility change scores. The correlation was assessed using the Spearman rank correlation coefficient for (1) admission BBS and admission PASS, (2) admission BBS and admission self-selected gait speed (SSGS) and fast gait speed (FGS), and (3) admission PASS and admission SSGS and FGS.

Multinomial logistic regression analysis was then used to identify predictors of functional gait classification at discharge. The analysis model is presented with significance (P < .05) and odds ratio with 95% confidence interval (CI). The Spearman rank correlation coefficient (ρ) was used to control multicollinearity between independent variables. Variables with a strong correlation (ρ≥ 0.7) were not used in the same logistic regression analysis.

Finally, receiver operating characteristic (ROC) curves were calculated to identify balance measure scores that would predict community ambulators and nonambulators at discharge from the facility. These 2 gait classifications were chosen over household and limited community ambulation taking into consideration the significance of the functional implications and effect on patient quality of life.22 In an ROC curve, the true-positive rate (sensitivity) is plotted against the false-positive rate (1-specificity). The area under the curve (AUC) was calculated with a 95% CI and has shown to be an effective measure of sensitivity and specificity that can describe the validity of tests.23 The AUC was used to assess the prediction model of the admission balance scores for determining the classification of gait upon discharge from inpatient rehabilitation. An AUC of 0.5 indicates no discrimination beyond chance, and a maximum AUC of 1 indicates perfect discrimination between the test results of both groups. Goodness-of-fit is presented with Kolmogorov-Smirnov statistical values. To further calculate an optimal cut-off value on each balance measure in predicting gait classification, the Youden Index was used by selecting the coordinates on the ROC curve that maximized the sensitivity and specificity for distinguishing those who were nonambulatory (0 m/s) from those who were ambulatory (>0 m/s).24 This was repeated on another ROC curve to distinguish those who ambulated at a speed suitable for community ambulation (>0.8 m/s) from those who did not at discharge (<0.8 m/s).


Sample Characteristics

The sample consisted of 180 individuals (79 female; ages 64.6 ± 14.1 (SD) years; time since onset (days) 11.1 ± 12.7). Baseline characteristics, admission, and discharge outcome measure data are shown in Table 1. The average admission BBS score was 18.7 ± 15.4 points, and the average admission PASS score was 22.5 ± 8.4 points. Ten individuals were missing gait speed measurements and were therefore excluded from the analysis.

TABLE 1. - Baseline Characteristics, Admission and Discharge Scores of All Participants (N = 180)
Variable Value
Age, mean (SD), y 64.6 (14.1)
CVA onset duration, mean (SD), d 11.1 (12.7)
Gender, n (%)

101 (56.1)
79 (43.9)
Length of stay, mean (SD), d 15.3 (8.7)
Lesion side, n (%)
No paresis

89 (49.4)
80 (44.4)
9 (5.0)
2 (1.1)
Discharge location, n (%)
Home + home health

120 (66.7)
41 (22.8)
1 (0.6)
18 (10)
FIM, mean (SD)a

62.1 (19.3)
90.3 (19.3)
PASS, mean (SD)a

22.5 (8.4)
28.3 (6.1)
BBS, mean (SD)a

18.7 (15.4)
32.8 (16.9)
Self-selected gait speed, mean (SD), m/s

0.27 (0.29)
0.46 (0.35)
Fast gait speed, mean (SD), m/s

0.38 (0.41)
0.63 (0.48)
Admission gait category, n (%)b
Limited community

63 (35.0)
50 (27.8)
46 (25.6)
11 (6.1)
Discharge gait category, n (%)b
Limited community

28 (15.6)
40 (22.2)
58 (32.2)
28 (15.6)
BBS, Berg Balance Scale; CVA, cerebral vascular accident; FIM, Functional Independence Measure; PASS, Postural Assessment Scale for Stroke; SD, standard deviation; SNF, skilled nursing facility.
aPaired t test for change from admission to discharge, P < .001.
bAs measured by self-selected gait speed.

TABLE 2. - Multinomial Logistic Regression Analysis for Admission PASS Scores and Other Predictors of Gait Classification at Discharge From Inpatient Rehabilitationa
Variables OR (95% CI) P Value
Household ambulator
Stroke onset, d
Lesion location
ADM self-selected gait speed
Admission PASS

1.0 (0.95-1.05)
1.49 (0.41-5.34)
0.99 (0.95-1.03)

1.35 (0.22-8.47)
1.25 (1.10-1.41)

Limited community ambulator
Stroke onset, d
Lesion location
ADM self-selected gait speed
Admission PASS

1.00 (0.95-1.06)
0.91 (0.21-3.95)
0.99 (0.94-1.05)
7.17 (1.06-48.6)
1.30 (1.12-1.50)

Community ambulator
Stroke onset, d
Lesion location
ADM self-selected gait speed
Admission PASS

0.95 (0.88-1.02)
2.14 (0.32-14.47)
0.97 (0.85-1.12)

25.9 (2.89-232.6)
1.65 (1.30-2.10)

ADM, admission; BBS, Berg Balance Scale; CI, confidence interval; OR, odds ratio; PASS, Postural Assessment Scale for Stroke.
aReference category: Did not ambulate.
bP < .01.
cP < .10.
dP < .05.

A statistically significant improvement in BBS, PASS, SSGS, FGS, and FIM mobility change scores was found between admission and discharge from inpatient rehabilitation (P < .0001). Based on the SSGS as measured by the 10MWT at admission, 63 individuals were in the “nonambulators” category (35%), 50 were “household ambulators” (27.8%), 46 were limited community ambulators (25.6%), and 11 individuals were in the “community ambulators” category (6.1%).

Stroke incidence is 33% higher among men than women,25 and incidence is 60% lower for women than men ages 55 to 64 years.26 The distribution of gender and stroke incidence in this sample population aligns with the above statistics. There were a greater proportion of men in the sample of patients with stroke (56.1%), with an average age of 64.6 years.

Admission PASS Versus Admission BBS

The admission BBS and the admission PASS were highly correlated (r = 0.8, P < .0001). Therefore, 2 parallel multinomial analyses were performed. Interestingly, we found that both the admission PASS and the admission BBS demonstrated a correlation with the length of stay. The PASS and length of stay relationship was stronger (r = 0.8, P < .0001) than the admission BBS and length of stay (r = 0.7, P < .0001). Therefore, these variables were excluded during further logistic regression analysis to control for multicollinearity.

Predictors of Gait Classification

Patients were stratified into the previously mentioned gait categories based on the discharge SSGS and FGS. First, categorization and analysis using the SSGS will be discussed. At discharge, as measured by the SSGS, 28 individuals did not ambulate (15.6%), 40 were “household ambulators” (22.2%), 58 were limited community ambulators (32.2%), and 28 individuals were in the “community ambulators” category (15.6%). As depicted in Table 2, the admission PASS score was significant in predicting those who ambulated compared with nonambulators (P < .01), as well as differentiating between each gait classification (P < .01). The admission BBS score was also significant in predicting those who ambulated at discharge (P < .01) and differentiation between gait classifications, as seen in Table 3. For every point scored higher on each assessment, patients were 1.7 times more likely to discharge from inpatient rehabilitation as a “community ambulator.” Admission SSGS was found to be moderately predictive of community ambulation alone (P < .05).

TABLE 3. - Multinomial Logistic Regression Analysis for Admission BBS Scores and Other Predictors of Gait Classification at Discharge From Inpatient Rehabilitationa
Variables OR (95% CI) P Value
Household ambulator
Stroke onset, d
Lesion location
ADM self-selected gait speed
Admission BBS

1.01 (0.96-1.06)
1.30 (0.36-4.69)
0.99 (0.95-1.03)
1.24 (0.20-7.72)
1.44 (1.10-1.87)

Limited community ambulator
Stroke onset, d
Lesion location
ADM self-selected gait speed
Admission BBS

1.02 (0.97-1.08)
0.61 (0.13-2.77)
0.99 (0.93-1.05)
7.99 (1.23-51.96)
1.47 (1.12-1.92)

Community ambulator
Stroke onset, d
Lesion location
ADM self-selected gait speed
Admission BBS

0.98 (0.90-1.06)
1.03 (0.14-7.85)
0.87 (0.73-1.04)

39.78 (3.31-478.10)
1.66 (1.25-2.19)

ADM, admission; BBS, Berg Balance Scale; CI, confidence interval; OR, odds ratio.
aReference category: Did not ambulate.
bP < .01.
cP < .05.

TABLE 4. - Receiver Operation Characteristic Curve Analysis for Predicting Ambulation Category of Patients With Stroke at Discharge
Variable Scores AUC SE P Value 95% CI K-S Sensitivity Specificity
Predicting nonambulation








Predicting community ambulation








AUC, area under the curve; BBS, Berg Balance Scale; CI, confidence interval; K-S, Kolmogorov-Smirnov statistic; PASS, Postural Assessment Scale for Stroke; SE, standard error.

Next, categorization and analysis using the discharge FGS will be discussed, as the authors feel it is important to highlight similarities of the results using both measures of the SSGS and the FGS. At discharge from the inpatient rehabilitation unit, 28 individuals did not ambulate (15.6%), 29 were “household ambulators” (16.1%), 35 were limited community ambulators (19.4%), and 62 individuals fell in the “community ambulators” category (34.4%) based on categorization using the FGS data. Using FGS data, the results were similar to those found utilizing SSGS data. The admission PASS and BBS were effective in differentiating between nonambulators and those who did ambulate, and between each gait category (P < .01). The initial FGS and stroke hemisphere appeared to be moderately predictive of community ambulation when analyzed using the discharge FGS (P < .05). Age, gender, and days since onset of stroke were not predictive in differentiating between nonambulators and ambulators at discharge from inpatient rehabilitation using FGS data.

Cut-off Scores

The ROC curve and the Youden index were used to determine optimal cut-off scores for the admission BBS and the PASS for predicting functional classification of gait at discharge. Clinically, the authors felt that predicting whether a patient will be nonambulatory or walking at a speed suitable for community mobility upon discharge would be most important. When predicting those who did not ambulate at discharge based on SSGS, the optimal ROC curves were obtained with a cut-off score of less than or equal to 6 out of 56 on the BBS (sensitivity 96%; specificity 83%) and less than or equal to 17 out of 36 on the PASS (sensitivity 92%; specificity 90%). Figure 1 illustrates the ROC curves for the BBS (AUC = 0.939) and the PASS (AUC = 0.933). When predicting those who were community ambulators at discharge based on SSGS, the optimal ROC curves were obtained with a cut-off score of greater than or equal to 29 out of 56 on the BBS (sensitivity 92%; specificity 86%) and greater than or equal to 30 out of 36 on the PASS (sensitivity 80%; specificity 87%). Figure 2 illustrates the ROC curves for the BBS (AUC = 0.931) and the PASS (AUC = 0.918). Table 4 depicts the statistical analysis of both ROC curves.

Receiver Operating Characteristic (ROC) Curve Analysis of Nonambulatory Patients With Stroke (0 m/s) at Discharge With Admission Postural Assessment Scale for Stroke (PASS) and Berg Balance Scale (BBS) Scores.
Receiver Operating Characteristic (ROC) Curve Analysis of Patients With Stroke With Full Community Ambulation Ability (>0.8 m/s) at Discharge With Admission Postural Assessment Scale for Stroke (PASS) and Berg Balance Scale (BBS) Scores.


Results from the logistic regression analysis indicate that admission scores on 2 different types of balance measures, the PASS and the BBS, may similarly predict the ambulation status poststroke at discharge from a comprehensive IRF. The PASS and the BBS are alike in that they both objectively measure postural control in various functional positions. However, they differ in the degree of difficulty and task demands. Walking velocity as measured by SSGS on admission was identified as a potential predictor of community ambulation at discharge. The result of the study supports the findings from previous studies indicating that balance and walking velocity outcome measures on admission may predict the walking capacity of patients with stroke at discharge.6,8,27–31 The results of this study support findings from others that also analyzed the use of the BBS in predicting household and community ambulators at discharge.8,13 It also supports an additional study investigating the PASS as a predictor of independent walking at discharge.8 This investigation of the PASS and the BBS further defines the predictive gait qualities of these measures by using 4 different categories of ambulation: (1) nonambulators; (2) household ambulators; (3) limited community ambulators; and (4) community ambulators.

Walking speed is a quick, simple, and reliable measure that classifies patients' walking ability.32 Improvements in gait speed that result in a transition to a higher classification of ambulation category (ie, limited community ambulators with gait speed of 0.4-0.8 m/s improving to community ambulators walking >0.8 m/s) are associated with improved quality of life in patients poststroke.4 This is especially true for individuals who fall in the lower gait classifications.4 As measured by SSGS, the number of individuals in this study considered “household ambulators” decreased by 28% from admission to discharge. The number of “limited community ambulators” increased 15%, with nearly double the number of individuals considered full “community ambulators” from admission to discharge. Community ambulation is an important outcome after a stroke, with 75% of patients rating the ability to go out to places of interest in their communities as “essential.”22 Poststroke survivors who report a greater level of community ambulation demonstrate faster walking speeds.22

Outcome measure cut-off scores to predict independent ambulation have great utility in clinical practice. Clinicians can use these values at admission to a comprehensive IRF to guide their choice of intervention. Therapists may also use these scores at admission to discuss discharge recommendations with the health care team, patients, and their families. Gait training is an essential part of every patient's plan of care after a stroke. However, emphasizing basic functional mobility or family training may be more important for those scoring less than or equal to 6 on the BBS or 17 on the PASS, as the likelihood of functional ambulation at discharge is low with conventional therapy paradigms. Therapists may also predict that these patients will require durable medical equipment such as wheelchairs and lifts. In contrast, therapists may choose to emphasize gait training earlier in the rehabilitation process with those patients scoring at least a 29 on the BBS or a 30 on the PASS, as those patients are likely to achieve full community ambulation.

Previous studies have established cut-off scores for both the PASS and the BBS. A study by Huang et al8 identified that a total score of 12.5 on the PASS predicts independent ambulation at discharge from inpatient rehabilitation. To the authors' knowledge, the study mentioned above is the only published article to investigate and establish a cut-off score for the PASS. Our finding of 30 points for community ambulation is much higher than previous reports. However, Huang et al used a dependent variable of independent ambulation with or without a device over 10 m and did not account for the walking speed. Our finding of a score less than or equal to 17 on the PASS predicting nonambulators at discharge is closer to the value established by Huang et al. It should also be noted that in their study, 2 samples of patients had an average length of time since stroke onset of 26 and 31 days compared with 11 days in this study. Time after stroke is known to reflect spontaneous recovery of motor function in the first 10 weeks.33 Therefore, a higher early PASS assessment score is possibly more significant in predicting gait capacity in the future. Several studies have established cut-off scores for the BBS in predicting community ambulation.13,34,35 Our finding of a cut-off score of at least 29 on the BBS for a community ambulator is identical to that of Louie and Eng.34

A previous study by O'Dell et al27 published in 2013 compared the BBS and the PASS in their ability to predict outcomes at discharge, including gait classification. In contrast to the results of this investigation, O'Dell et al found that neither the admission BBS nor the PASS was significantly effective in differentiating between limited and full community ambulation. The higher power of the sample population and difference in statistical analysis likely led to the greater value of both the BBS and the PASS in predicting higher-level gait categories in this study. Consistent with the results of both studies, the 2 selected balance measures were effective in predicting the “household ambulation” (<0.4 m/s) category. Of note, O'Dell et al chose to classify those individuals who could not perform the 10MWT as “household ambulators” (<0.4 m/s). We determined separating those who cannot functionally ambulate (0 m/s) due to limitations such as requiring greater than minimal assistance for ambulation, unable to ambulate the full 10 m, or use of ambulation devices specifically for therapeutic purposes (ie, Hi-Lo Walker) from those who are considered ambulatory within their homes (>0 m/s) was important. Furthermore, this study extends beyond the results of O'Dell et al to determine clinical prediction cut-off scores to further assist clinicians in determining discharge disposition based on admission PASS and BBS scores.

Projecting a patient's discharge functional abilities early in a comprehensive inpatient rehabilitation program can help clinicians, patients, families, and case managers better prepare for the patient's care needs after leaving the facility (ie, planning discharge support such as home health or outpatient therapy, estimating the level of assistance the patient will require from caregivers, to order equipment such as a wheelchair or walker, or to estimate the financial burden after discharge).36 Prediction guidelines using clinical outcomes at admission to inpatient rehabilitation can improve rehabilitation efficiency by reducing the length of stay, increasing therapists' confidence in interventions, and identifying a likely discharge location early on in the patient's stay.36–38 Both PASS and BBS scores were valuable predictors of ambulation classification. Scores established in this study can assist clinicians in determining walking outcomes and organizing discharge needs early in a patient's inpatient rehabilitation stay. The retrospective nature of this study increases the ability to generalize these results to many inpatient rehabilitation stroke populations, as few inclusion/exclusion criteria were used. The demographic characteristics are representative of a national sample with a similar length of stay and initial FIM mobility scores (see Table 1).39 The assessments were performed by physical therapists and physical therapist assistants on patients receiving evidence-based physical therapy interventions.

This study has several limitations. The retrospective design of this study creates an inherent selection bias. Our sample was limited to a single institution with certain therapeutic equipment within 1-time window and may not apply to other institutions or settings. Other commonly used assessments of gait performance, such as the 6-minute walk test,40 may have been better to assess ambulation abilities. Lastly, comorbidities were not recorded for all patients, and many factors such as chronic disease could impact walking ability. We were able to control for multiple variables within our logistic regression analysis; however, they may have been overlooked since these other factors were not recorded. This study investigated each measure in isolation and compared their individual predictive abilities. Future studies should investigate these commonly used balance measures as a predictive cluster.


In summary, this study further supports the evidence that early balance assessments predict ambulation ability in patients with stroke admitted to inpatient rehabilitation. The admission PASS and the BBS were equally effective in predicting gait classification upon discharge. Patients who scored less than or equal to 6 on the BBS or less than or equal to a 17 on the PASS were likely to be nonambulatory at discharge. Patients who scored greater than or equal to 29 on the BSS or 30 on the PASS were likely to be community ambulators. Both measures can serve as useful tools for decision-making in the inpatient rehabilitation setting.


We are grateful to all the physical therapy staff at the Orlando Health Orlando Regional Medical Center Institute for Advanced Rehabilitation, who assisted in performing the assessments.


1. Virani SS, Alonso A, Benjamin EJ, et al. American Heart Association Council on Epidemiology and Prevention Statistics Committee and Stroke Statistics Subcommittee. Heart Disease and Stroke Statistics-2020 Update: a report from the American Heart Association. Circulation. 2020;141(9):e139–e596.
2. Pound P, Gompertz P, Ebrahim S. A patient-centred study of the consequences of stroke. Clin Rehabil. 1998;12(4):338–347.
3. Perry J, Garrett M, Gronley JK, Mulroy SJ. Classification of walking handicap in the stroke population. Stroke. 1995;26(6):982–989.
4. Schmid A, Duncan P, Studenski S, et al. Improvements in speed-based gait classifications are meaningful. Stroke. 2007;38(7):2096–2100.
5. Feigin L, Sharon B, Czaczkes B, Rosin A. Sitting Equilibrium 2 weeks after a stroke can predict the walking ability after 6 months. Gerontology. 1996;42(6):348–353.
6. Veerbeek JM, Van Wegen EE, Harmeling-Van der Wel BC, Kwakkel G; EPOS Investigators. Is accurate prediction of gait in non-ambulatory stroke patients possible within 72 hours poststroke? The EPOS study. Neurorehabil Neural Repair. 2011;25(3):268–274.
7. Kollen B, van de Port I, Lindeman E, Twisk J, Kwakkel G. Predicting improvement in gait after stroke: a longitudinal prospective study. Stroke. 2005;36(12):2676–2680.
8. Huang Y, Wang W, Liou T, Liao C, Lin L, Huang S. Postural Assessment Scale for Stroke Patients Scores as a predictor of stroke patient ambulation at discharge from the rehabilitation ward. J Rehabil Med. 2016;48(3):259–264.
9. Dobkin BH. Clinical practice. Rehabilitation after stroke. N Engl J Med. 2005;352(16):1677–1684.
10. Moore JL, Potter K, Blankshain K, Kaplan SL, O'Dwyer LC, Sullivan JE. A core set of outcome measures for adults with neurologic conditions undergoing rehabilitation: a clinical practice guideline. J Neurol Phys Ther. 2018;42(3):174–220.
11. DeJong G, Horn SD, Conroy B, Nichols D, Healton EB. Opening the black box of post-stroke rehabilitation: stroke rehabilitation patients, processes, and outcomes. Arch Phys Med Rehabil. 2005;86(12, suppl 2):S1–S7.
12. Medicare Payment Advisory Commission. Inpatient rehabilitation facility services: assessing payment adequacy and updating. In: Report to the Congress: Medicare Payment Policy. Washington, DC; MedPAC. 2020:273–294.
13. Bland MD, Sturmoski A, Whitson M, et al. Prediction of discharge walking ability from initial assessment in a stroke inpatient rehabilitation facility population. Arch Phys Med Rehabil. 2012;93(8):1441–1447.
14. Vos-Vromans D, de Bie R, Erdmann P, van Meeteren N. The responsiveness of the ten-meter walking test and other measures in patients with hemiparesis in the acute phase. Phyiother Theory Pract. 2005;21(3):173–180.
15. Steffen T, Seney M. Test-retest reliability and minimal detectable change on Balance and Ambulation Tests, the 36-Item Short-Form Health Survey, and the Unified Parkinson Disease Rating Scale in people with Parkinsonism. Phys Ther. 2008;88(6):733–746.
16. Hiengkaew V, Jitaree K, Chaiyawat P. Minimal detectable changes of the Berg Balance Scale Fugl-Meyer Assessment Scale, Timed “Up & Go” Test, Gait Speeds, and 2-Minute Walk Test in individuals with chronic stroke with different degrees of ankle plantarflexor tone. Arch Phys Med Rehabil. 2012;93(7):1201–1208.
17. Blum L, Korner-Bitensky N. Usefulness of the Berg Balance Scale in stroke rehabilitation: a systematic review. Phys Ther. 2008;88:599–666.
18. Mao HF, Hsueh IP, Tang PF, Sheu CF, Hsieh CL. Analysis and comparison of the psychometric properties of three balance measures for stroke patients. Stroke. 2002;33(4):1022–1027.
19. Chou CY, Chien CW, Hsueh IP, Sheu CF, Wang CH, Hsieh CL. Developing a short form of the Berg Balance Scale for people with stroke. Phys Ther. 2006;86(2):195–204.
20. Benaim C, Pérennou D, Villy J, Rousseaux M, Pelissier J. Validation of a standardized assessment of postural control in stroke patients. Stroke. 1999;30(9):1862–1868.
21. Hsieh CL, Sheu CF, Hsueh IP, Wang CH. Trunk control as an early predictor of comprehensive activities of daily living function in stroke patients. Stroke. 2002;33(11):2626–2630.
22. Lord S, McPherson K, McNaughton H, Rochester L, Weatherall M. Community ambulation after stroke: how important and obtainable is it and what measures appear predictive? Arch Phys Med Rehabil. 2004;85(2):234–239.
23. Hajian-Tilaki K. Receiver operating characteristic (ROC) curve analysis for medical diagnostic test evaluation. Caspian J Intern Med. 2013;4(2):627–635.
24. Fluss R, Faraggi D, Reiser B. Estimation of the Youden Index and its associated cut-off point. Biom J. 2005;47(4):458–472.
25. Appelros P, Stegmayr B, Terént A. Sex differences in stroke epidemiology. Stroke. 2009;40(4):1082–1090.
26. Lofmark U, Hammarstrom A. Evidence for age-dependent education-related differences in men and women with first-ever stroke. Neuroepidemiology. 2007;28(3):135–141.
27. O'Dell M, Au J, Schwabe E, Batistick H, Christos P. A comparison of two balance measures to predict discharge performance from inpatient stroke rehabilitation. PM R. 2013;5(5):392–399.
28. Loewen SC, Anderson BA. Predictors of stroke outcome using objective measurement scales. Stroke. 1990;21(1):78–81.
29. Kwakkel G, Wagenaar RC, Kollen BJ, Lankhorst GJ. Predicting disability in stroke—a critical review of the literature. Age Ageing. 1996;25(6):479–489.
30. Paolucci S, Antonucci G, Grasso MG, et al. Functional outcome of ischemic and hemorrhagic stroke patients after inpatient rehabilitation: a matched comparison. Stroke. 2003;34(12):2861–2865.
31. Meijer R, Ihnenfeldt DS, de Groot IJ, van Limbeek J, Vermeulen M, de Haan RJ. Prognostic factors for ambulation and activities of daily living in the subacute phase after stroke. A systematic review of the literature. Clin Rehabil. 2003;17(2):119–129.
32. Bowden MG, Balasubramanian CK, Behrman AL, Kautz SA. Validation of a speed-based classification system using quantitative measures of walking performance poststroke. Neurorehabil Neural Repair. 2008;22(6):672–675.
33. Kwakkel G, Kollen B, Twisk J. Impact of time on improvement of outcome after stroke. Stroke. 2006;37(9):2348–2353.
34. Louie DR, Eng JJ. Berg Balance Scale score at admission can predict walking suitable for community ambulation at discharge from inpatient stroke rehabilitation. J Rehabil Med. 2018;50(1):37–44.
35. Lee G, An S, Lee Y, Park DS. Clinical measures as valid predictors and discriminators of the level of community ambulation of hemiparetic stroke survivors. J Phys Ther Sci. 2016;28(8):2184–2189.
36. Harari Y, O'Brien MK, Lieber RL, Jayaraman A. Inpatient stroke rehabilitation: prediction of clinical outcomes using a machine-learning approach. J Neuroeng Rehabil. 2020;17(1):71.
37. Stinear CM, Byblow WD, Ackerley SJ, Barber PA, Smith MC. Predicting recovery potential for individual stroke patients increases rehabilitation efficiency. Stroke. 2017;48(4):1011–1019.
38. Brauer SG, Bew PG, Kuys SS, Lynch MR, Morrison G. Prediction of discharge destination after stroke using the motor assessment scale on admission: a prospective, multisite study. Arch Phys Med Rehabil. 2008;89(6):1061–1065.
39. Granger CV, Markello SJ, Graham JE, Deutsch A, Ottenbacher KJ. The uniform data system for medical rehabilitation: report of patients with stroke discharged from comprehensive medical programs in 2000-2007. Am J Phys Med Rehabil. 2009;88(12):961–972.
40. Pohl P, Duncan P, Perera S, et al. Influence of stroke-related impairments in performance in 6-minute walk test. J Rehabil Res Dev. 2002;39(4):439–444.
© 2022 Academy of Acute Care Physical Therapy, APTA