Postural Assessment Scale for Stroke Patients (PASS)

Overview

A literature search was conducted to identify all relevant publications on the psychometric properties of the Postural Assessment Scale for Stroke Patients (PASS). Eleven papers were reviewed.

Floor/Ceiling Effects

Benaim et al. (1999) examined the frequency distribution and density trace of PASS scores in 58 patients at 30 and 90 days post-stroke. While a uniform distribution was noted at 30 days post-stroke, there was a pronounced peak around the highest values at 90 days as 38% of patients had achieved the maximum score. Testing on 30 age-matched healthy subjects revealed that 90% of participants achieved the maximum score.

Mao et al. (2002) examined the floor and ceiling effect of the PASS on a sample of patients with stroke at 14, 30, 90 and 180 days post-stroke. Floor and ceiling effects were adequate at all time points (floor effect range 2.2-3.8%; ceiling effect range 3.3-17.5%).

Chien et al. (2007b) examined the floor and ceiling effects of the PASS among 287 patients at 14 days post-stroke, and a second cohort of 197 patients. The PASS demonstrated no significant floor effects (6.3%, 6.1% respectively) or ceiling effects (2.8%, 1.7% respectively) in either cohort.

Chien et al. (2007b) also examined the floor and ceiling effects of the 5-item SFPASS among 287 patients at 14 days post-stroke, and a second cohort of 197 patients. The SFPASS demonstrated no significant ceiling effect in either cohort (7.0%, 8.4%). A poor floor effect (20.2%) was seen in the first cohort, but was not evident in the second cohort (16.2%).

Yu et al. (2012) reported no significant floor or ceiling effect of the PASS among 85 patients with stroke (most with acute stroke and severe disability) on admission to (<15%) or discharge from (<10%) a rehabilitation ward.

Reliability

Internal constancy

Benaim et al. (1999) reported excellent internal consistency of the PASS (?=0.95) when examined on a sample of 58 patients with stroke using Cronbach ?-coefficient. The authors concluded that the PASS is homogenous and is likely to produce consistent responses. Further, there was a strong correlation between the sums of maintaining-position and changing-position items (r=0.86, p<0.001).

Mao et al. (2002) examined the internal consistency of the PASS on a sample of 112 patients with stroke at 14, 30, 90 and 180 days post-stroke, using Cronbach’s ? coefficient. Excellent internal consistency was reported at all time points ( range = 0.94-0.96).

Hsieh et al. (2002) examined the internal consistency of the trunk control items of the PASS (PASS-TC: items 1, 6, 7, 8, 9) in a sample of 182 patients with acute stroke. Results from two raters indicated excellent internal consistency of the PASS-TC (?=0.93-0.94), measured using Cronbach’s ?.

Chien et al. (2007b) examined the internal consistency of the PASS on a sample of 287 patients at 14 days post-stroke, and a second cohort of 197 patients. Excellent internal consistency (?=0.96) was reported in both cohorts, measured using Cronbach’s ?.

Chien et al. (2007b) also examined the internal consistency of the SFPASS on a sample of 287 patients at 14 days post-stroke, and a second cohort of 197 patients. Internal consistency was adequate in the first cohort (=0.66) and excellent in the second cohort (=0.93), as measured using Cronbach’s ?.

Chien et al. (2007b) noted that the high internal consistency of the PASS may indicate redundancy among items.

Inter-rater

Benaim et al. (1999) measured inter-rater reliability of the PASS using ?-coefficient for individual item reliability and Pearson product moment correlation for total score reliability. Two clinicians assessed patients with stroke on the same day, with a total sample of 12 patients. The authors reported adequate to excellent inter-rater reliability for individual items (average ?=0.88, range 0.64-1) and excellent inter-rater reliability for the total score (r=0.99, p<0.001). Further, a Bland and Altman plot for inter-rater reliability showed that differences between scorings were weak (0.5) and homegenous (differences were within or very near the 95% confidence limits of the mean).

Mao et al. (2002) examined the inter-rater reliability of the PASS using ?-coefficient for individual item reliability and Pearson product moment correlation for total score reliability. Two clinicians assessed patients at 14 days post-stroke on the same day, with a total sample of 112 patients. Inter-rater reliability for individual items was adequate to excellent (median ?=0.88, range 0.61-0.96) and inter-rater reliability for the total score was excellent (ICC=0.97, 95% CI 0.95-0.98).

Hsieh et al. (2002) examined the inter-rater reliability of the trunk control items of the PASS (PASS-TC: items 1, 6, 7, 8, 9) in a sample of 182 patients with acute stroke. Results indicated excellent inter-rater reliability of the PASS-TC (ICC=0.97), measured using intraclass correlation coefficient.

Persson et al. (2011) examined the same-day inter-rater reliability of the SwePASS in a sample of 114 patients with acute stroke, using Spearman’s rank correlation. The inter-rater reliability of items was excellent (r=0.77-0.99).

Intra-rater

Persson et al. (2011) examined the same-day intra-rater reliability of the SwePASS in a sample of 114 patients with acute stroke. The inter-rater reliability of items was excellent (r=0.88-0.98) when measured using Spearman’s rank correlation, and adequate to excellent (?=0.70-0.93) when measured using Kappa coefficient.

Test-retest

Benaim et al. (1999) measured test-retest reliability of the PASS on 12 patients with stroke, using ?-coefficient for individual items and Pearson product moment correlation for the total score. The authors reported adequate to excellent reliability for individual items (average ?=0.72, range 0.45 – 1) and excellent reliability for the total score (r=0.98, p<0.001). Further, a Bland-Altman plot showed that differences between scorings were weak (0.5) and homegenous (differences were within or very near the 95% confidence limits of the mean).

Chien et al. (2007a) examined the 2-week test-retest reliability of the PASS among 20 patients with chronic stroke, and reported excellent test-retest reliability (ICC=0.84), measured using a 1-way random effects model intraclass correlation coefficient (ICC).

Liaw et al. (2008) examined 7-day test-retest reliability of the PASS in a sample of 52 patients with chronic stroke, using the intraclass coefficient for relative reliability (i.e. the degree to which individuals maintain their position in a sample with repeated measures), and Bland-Altman plots and standard error of measurement (SEM) for absolute reliability (i.e. the degree to which repeated measurements vary for individuals). Relative test-retest reliability was excellent (ICC=0.97). Bland-Altman plots revealed small limits of agreement (-2.72 to 3.52), indicating high stability with low natural variation. The SEM was small (1.14%), indicating that the PASS is useful to identify real change.

Liaw et al. (2012) examined the 7-day test-retest reliability of the SFPASS among a sample of 52 patients with chronic stroke, using the weighted ? statistic for individual items and intraclass correlation coefficient (ICC) for the total score. The authors reported adequate to excellent test-retest reliability for individual items (mean ?=0.78, range 0.66 – 0.84) and excellent reliability for total scores (ICC=0.93, 95% CI 0.88-0.96). Bland-Altman plots of the differences between measurements from the two test sessions against the mean of the two test sessions for each patient revealed small limits of agreement (1.99 to -2.33), indicating high stability with low natural variation. Standard error of measurement (SEM) was 5.2%, representing a small and acceptable level of measurement error.

Note: When performing a Bland and Altman analysis, a mean difference close to zero indicates higher agreement between measurements.

Validity

Content validity

No studies have reported on the content validity of the PASS.

Criterion

Predictive

Benaim et al. (1999) examined the predictive validity of the PASS by comparing PASS scores at 30 days post-stroke with FIM scores at 90 days post-stroke on a sample of 58 patients. Correlations between the PASS and FIM total score (r=0.75, p<0.001), transfer items (r=0.74, p<0.006) and locomotion items (r=0.71, p<0.001) indicate that it is possible to predict functional recovery from PASS scores at 30 days post-stroke.

Mao et al. (2002) examined the predictive validity of the PASS, Berg Balance Scale and the Fugl-Meyer Assessment modified balance scale at 14, 30 and 90 days post-stroke by comparison with the Motor Assessment Scale walking subscale score at 180 days post-stroke, in a sample of 123 patients. The PASS demonstrated excellent predictive validity at all time points (?=0.86-0.90), as measured using Spearman’s p correlation coefficient.

Hsieh et al. (2002) examined the predictive validity of the trunk control items of the PASS (PASS-TC: items 1, 6, 7, 8, 9) in a sample of 169 patients by comparing PASS-TC scores taken at 14 days post-stroke with Barthel Index (BI) and Frenchay Activities Index (FAI) scores taken at 6 months post-stroke. The PASS-TC demonstrated excellent predictive validity (r=0.68, p<0.001), as measured using Pearson correlation coefficient. The PASS-TC was found to be a stronger predictor of comprehensive ADL function than the Barthel Index or the Fugl-Meyer motor test.

Wang et al. (2004) examined the predictive validity of the PASS and a modified version of the PASS that used a 3-level scale (12-item PASS-3P) in patients with stroke by comparing scores taken at 14 (n=226) and 30 days (n=202) post-stroke with Barthel Index (BI) scores taken at 90 days post-stroke, using Spearman’s ? correlation coefficient. Both versions of the PASS demonstrated excellent predictive validity at 14 days (p=0.78) and 30 days (p=0.82) post-stroke.

Chien et al. (2007a) examined the predictive validity of the PASS in a sample of 32 patients with stroke by comparing PASS score taken within 3 months of the stroke with Barthel Index and Frenchay Activities Index scores taken approximately 1 year later. Results indicated poor predictive validity (r2=0.24), as measured using Pearson correlation coefficient.

Chien et al. (2007b) examined the predictive validity of the PASS and SFPASS in a sample of 218 patients with stroke by comparing scores on each test at 14 days post-stroke with Barthel Index (BI) scores at 90 days post-stroke. Results indicated adequate predictive validity of the PASS (r=0.49) and the SFPASS (r=0.48). The authors replicated the process on a second cohort of 179 patients by comparing PASS and SFPASS scores on admission to rehabilitation with BI scores on discharge from hospital. Results revealed excellent predictive validity of the PASS (r=0.83) and the SFPASS (r=0.82), as measured using product-moment correlations.

Di Monaco et al. (2010) examined the predictive validity of the PASS and the Trunk Impairment Scale (TIS) by comparing scores on admission to inpatient rehabilitation with FIM discharge scores, in a sample of 60 patients with stroke. Results indicated excellent predictive validity of the PASS (?=0.687, p<0.001), as measured using Spearman rank correlation. PASS admission scores were also significantly associated with FIM change scores (P<0.001), FIM effectiveness (P=0.017) and destination at discharge (P=0.032).

Yu et al. (2012) examined the predictive validity of the PASS and the Balance Computerized Adaptive Test (Balance CAT) by comparing scores on admission to a rehabilitation ward with Barthel Index (BI) and Stroke Rehabilitation Assessment of Movement mobility subscale (MO-STREAM) discharge scores in a sample of 85 patients with stroke. Correlations between PASS and BI scores (=0.62, r2=0.39, p<0.001) and between PASS and MO-STREAM scores (=0.80, r2=0.63, p<0.001) indicated sufficient predictive validity of the PASS, as measured using  and r2 from a simple linear regression analysis. This indicates that PASS scores at admission can predict discharge function and mobility.

Concurrent

Mao et al. (2002) examined the concurrent validity of the PASS, Berg Balance Scale (BBS) and the Fugl-Meyer Assessment modified balance scale (FMA-B), using Spearman’s ? correlation coefficient. A sample of 123 patients with stroke was followed at 14, 30 (n=110), 90 (n=93), and 180 (n=80) days after stroke onset. There was excellent concurrent validity between the PASS and the FMA-B (?=0.95-0.97) and between the PASS and the BBS (?= 0.92-0.95) at all time points.

Wang et al. (2004) examined the concurrent validity of the PASS, Berg Balance Scale (BBS) and modified versions of both assessments that used 3-level scales (12-item PASS-3P, 14-item BBS-3P) in a sample of 77 patients with stroke, using Spearman’s ? correlation coefficient and Intraclass Correlation Coefficient. There was excellent concurrent validity between all measures (rho ? 0.91, P<0.0001); of note, agreement between the PASS and the BBS (?=0.94, P<0.0001), and between the PASS and the PASS-3P (?=0.94, P<0.0001; ICC=0.97, 95% CI 0.96-0.98.) was excellent.

Chien et al. (2007b) examined the concurrent validity of the PASS and the 5-item SFPASS in a sample of 287 patients at 14 days post-stroke, using a random effects model intraclass correlation coefficient (ICC). There was excellent concurrent validity between the PASS and the 5-item SFPASS (ICC= 0.98; 96% variance). This result was repeated in a subsequent cohort of 179 patients (ICC=0.98). Further, Bland-Altman plots revealed no systematic trend between the difference and mean score of the PASS and the 5-item SFPASS (mean difference 1.6, limits of agreement range from -3.7 to 6.8). This suggests that the PASS and the SFPASS can be used interchangeably.

Di Monaco et al. (2010) reported excellent concurrent validity between the PASS and the Trunk Impairment Scale (TIS) (?=0.849, P<0.001), measured in a sample of 60 patients on admission to inpatient rehabilitation.

Construct

Known Group Validity

No studies have reported on the known-groups validity of the PASS.

Convergent/Discriminant Validity

Benaim et al. (1999) examined correlations between PASS performance and clinical scales of functional status, motricity, spasticity, spatial inattention and somatosensory threshold among 58 patients at 30 days post-stroke, using Pearson correlation coefficients. Excellent correlations were found with FIM total score (r=0.73), transfer tasks (r=0.82) and locomotor tasks (r=0.73); and motricity scores of the lower limb (r=0.78) and upper limb (r=0.63). Adequate negative correlations were found with the star cancellation test of spatial inattention (r=-0.53) and pressure sensitivity of the lower limb (r=-0.45) and upper limb (r=-0.42). There was no significant correlation with spasticity, measured using the Ashworth Scale. The authors also examined the correlation between PASS performance and equilibrium, measured using a rocking platform, among a smaller sample of 31 patients at 90 days post-stroke, and reported adequate negative correlations with measurement of postural stabilization (r=-0.48) and postural orientation with respect to gravity (r=0.36).

Mao et al. (2002) examined the convergent validity of the PASS, Berg Balance Scale and the Fugl-Meyer Assessment modified balance scale by comparison with the Barthel Index (BI), using Spearman’s p correlation coefficient. A sample of 123 patients with stroke was followed at 14, 30 (n=110), 90 (n=93), and 180 (n=80) days after stroke onset. There was excellent convergent validity between the PASS and the BI (?=0.88-0.92) at all time points.

Hsieh et al. (2002) examined the convergent validity of the trunk control items of the PASS (PASS-TC: items 1, 6, 7, 8, 9) by comparison with the Barthel Index (BI) and Fugl-Meyer balance test (FM-B), in a sample of 182 patients at 14 days post-stroke. The PASS-TC demonstrated excellent convergent validity with the BI (r=0.89) and the FM-B (r=0.73), using Pearson correlation coefficient.

Wang et al. (2004) examined the convergent validity of the PASS and the PASS-3P by comparison with the Barthel Index (BI), in a sample of 77 patients with stroke. The PASS and the PASS-3P both demonstrated excellent convergent validity with the BI (?=0.84, (?=0.82 respectively), measured using Spearman p correlation coefficient.

Chien et al. (2007b) examined the convergent validity of the PASS and the 5-item SFPASS by comparison with the Barthel Index (BI) and FIM in a sample of 287 patients at 14 days post-stroke. The PASS and the SFPASS both demonstrated excellent correlations with the BI (PASS r=0.87; SFPASS r=0.86) and the FIM (PASS r=0.75; SFPASS r=0.75).

Responsiveness

Mao et al. (2002) examined the responsiveness of the PASS in a sample of 123 patients with stroke by comparing scores taken at 14, 30 (n=110), 90 (n=93) and 180 (n=80) days post-stroke. There was a significant change in PASS scores at all stages (14-30 days, 30-90 days, 90-180 days, 14-90 days and 14-180 days post-stroke), measured using Wilcoxon matched-pairs signed-rank tests. Effect size was large at the interval between 14-30 days post-stroke (ES=0.89), became moderate in the interval between 30-90 days (ES=0.64) and low in the interval 90-180 days (ES=0.31). The overall effect size (14-180 days) was large (ES=1.12). These results indicate that the PASS demonstrates good responsiveness before 90 days post-stroke but low responsiveness at later stages of recovery. The authors also examined responsiveness according to stroke severity and found that the PASS is more responsive to detecting change in moderate to severe stroke than mild stroke across most time intervals. The overall effect size (14-180 days) was largest among patients with severe stroke (ES=1.54).

Wang et al. (2004) examined the responsiveness of the PASS and the PASS-3P in patients with stroke by comparing scores taken at 14 days (n=202), 30 days (n=167) and 90 days (n=167) post-stroke. There was a significant change in PASS and PASS-3P scores at all stages (14-30 days, 30-90 days and 14-90 days post-stroke), measured using Wilcoxon matched-pairs signed-rank tests. Both measures demonstrated a large effect size in the interval 14-30 days post-stroke (SRM=0.84 and 0.86 respectively) and 14-90 days post-stroke (SRM=1.02 and 1.04 respectively), but only a moderate effect size in the interval 30-90 days post-stroke (SRM=0.65 and 0.67 respectively), measured using standardized response mean (SRM). The authors examined responsiveness of the PASS and the PASS-3P according to severity of stroke – mild (Fugl-Meyer Assessment score ? 80), moderate (FMA score 36-79) and severe stroke (FMA score 0-35). Both measures showed a moderate effect size among patients with mild stroke (PASS SRM range 0.43-0.78; PASS-3P SRM range 0.46-0.78), moderate to large effect size among patients with moderate stroke (PASS SRM range 0.52-1.12; PASS-3P SRM range 0.56-1.19), and a large effect size among patients with severe stroke (PASS SRM range 0.92-1.35; PASS-3P SRM range 0.92-1.34). The effect size of both measures was consistently larger in the intervals 14-30 days post-stroke and 14-90 days post-stroke, than 30-90 days post-stroke.

Chien et al. (2007a) examined the responsiveness of the PASS in a sample of 40 patients with subacute stroke, measured using Cohen’s effect size. The PASS was administered twice over a 2-week interval, during which time patients received an intensive rehabilitation program that comprised postural training and weight shift exercises for more than 2 hours per day, 5 days a week. The effect size after 2 weeks was small (d=0.41). The minimal detectable change (MDC) (i.e. the threshold value that determines whether score changes are beyond chance) was 2.22 (95% CI) at an individual score level, and 0.50 (95% CI) at a group score level.

Chien et al. (2007b) examined the responsiveness of the PASS in a sample of 262 patients with stroke. The change score from 14 days post-stroke to 30 days post-stroke was significant (4.9, p<0.01) and the effect size was small (ES=0.42). A small effect size (ES=0.43) was also seen in a subsequent cohort of 179 patients who were assessed at admission to rehabilitation and again on discharge from hospital.

Chien et al. (2007b) examined the responsiveness of the 5-item SFPASS in a sample of 262 patients with stroke. The change score from 14 to 30 days post-stroke was significant (5.4, p<0.01) and the effect size was small (ES=0.44). A small effect size (ES=0.42) was also seen in a subsequent cohort of 179 patients who were assessed at admission to rehabilitation and again on discharge from hospital.

Chien et al. (2007b) reported on the Standard Error of Measurement (i.e. an estimate of the dispersion of scores that would be obtained if the measure was administered to a patient multiple times) of the PASS and the 5-item SFPASS in a cohort of 287 patients at 14 days post-stroke. The SEM of the PASS was 2.4 (4.7, 95% CI). The SEM of the 5-item SFPASS in the same cohort was 3.4 (6.7, 95% CI). This score is lower than 10% of the highest possible score of 36, which indicates that the measurement error does not exceed clinical importance.

Liaw et al. (2008) examined the smallest real difference (SRD – the smallest change threshold that indicates a real improvement for a single individual) of the PASS in a sample of 52 patients with chronic stroke. Participants were assessed by the same clinician on 2 occasions, 7-days apart. The SRD was 3.2, indicating that a change of more than 4 points in the total score for the PASS in chronic stroke patients is not likely to be attributable to chance variation or measurement error.

Liaw et al. (2012) examined the minimal detectable change (MDC) of the Short Form PASS (SFPASS) in a sample of 52 patients with chronic stroke. Participants were assessed by the same clinician on two occasions, 7 days apart. Results indicate that a change in an individual’s SFPASS scores greater than 2.16 points can be interpreted as true change (95% CI).

Yu et al. (2012) examined the internal and external responsiveness of the PASS in a sample of 85 patients with stroke. There was a significant change in PASS scores from admission to discharge (Wilcoxon Z=7.7, p<0.001), and the effect size was large (d=0.87), indicating adequate internal responsiveness. External responsiveness was calculated by comparing PASS change scores (admission to discharge) with change scores from the Barthel Index (BI) and the mobility subscale of the Stroke Rehabilitation Assessment of Movement (MO-STREAM), using  and r2 from a simple linear regression analysis. Results revealed a fair association between PASS and BI changes scores (=0.44, r2=0.20, p<0.001) and a moderate association between PASS and MO-STREAM change scores ((=0.77, r2=0.59, p<0.001) indicating sufficient external responsiveness of the PASS to changes in function and mobility following stroke.

Sensitivity

No studies have reported on the sensitivity of the PASS.

References
  • Barros de Oliveira, C., Torres de Medeiros, I.R., Ferreira Frota, N.A., Greters, M.E., & Conforto, A.B. (2008). Balance control in hemiparetic stroke patients: main tools for evaluation. Journal of Rehabilitation Research and Development, 45(8), 1215-26.
  • Benaim, C., Perennou, D.A., Villy, J., Rousseaux, M., & Pelissier, J.Y. (1999). Validation of a standardized assessment of postural control in stroke patients: The Postural Assessment Scale for Stroke Patients (PASS). Stroke, 30, 1862-8.
  • Blum, L. & Korner-Bitensky, N. (2008). Usefulness of the Berg Balance Scale in stroke rehabilitation: a systematic review. Physical Therapy, 88, 559-66.
  • Chien, C-W., Hu, M-H., Tan, P-F., Sheu, C-F., & Hsieh, C-L. (2007a). A comparison of psychometric properties of the Smart Balance Master system and the Postural Assessment Scale for Stroke in people who have had mild stroke. Archives of Physical Medicine and Rehabilitation, 88, 374-80.
  • Chien, C-W., Lin, J-H., Wan, C-H., Hsueh, I-P., Sheu, C-F., & Hsieh, C-L. (2007b). Developing a short form of the Postural Assessment Scale for People with Stroke. Nuerorehabilitation and Neural Repair, 21, 81-90.
  • Di Monaco, M., Trucco, M., Di Monaco, R., Tappero, R., & Cavanna, A. (2010). The relationship between initial trunk control or postural balance and inpatient rehabilitation outcome after stroke: a prospective comparative study. Clinical Rehabiltiation, 24, 543-54.
  • Hseih, C-L., Sheu, C-F., Hsueh, I-P., & Wang, C-H. (2002). Trunk control as an early predictor of comprehensive activities of daily living function in stroke patients. Stroke, 33, 2626-30.
  • Liaw, L-J., Hsieh, C-L., Hsu, M-J., Chen, H-M., Lin, J-H., & Lo, S-K. (2012). Test-retest reproducibility of two short-form balance measures used in individuals with stroke. International Journal of Rehabilitation Research, in press. Epub ahead of print retrieved from http://journals.lww.com/intjrehabilres/Abstract/publishahead/Test_retest_reproducibility_of_two_short_form.99818.aspx
  • Liaw, L-J., Hsieh, C-L., Lo, S-K., Chen, H-M., Lee, S., & Lin, J-H. (2008). The relative and absolute reliability of two balance performance measures in chronic stroke patients. Disability and Rehabilitation, 30(9), 656-61.
  • Mao, H-F., Hsueh, I-P., Tang, P-F., Sheu, C-F., & Hsieh, C-L. (2002). Analysis and comparison of the psychometric properties of three balance measures for stroke patients. Stroke, 33, 1022-7.
  • Persson, C.U., Hansson, P-O., Danielsson, A., & Sunnerhagen, K.S. (2011). A validation study using a modified version of Postural Assessment Scale for Stroke Patients: Postural stroke study in Gothenburg (POSTGOT). Journal of NeuroEngineering and Rehabilitation, 8, 57-64.
  • Wang, C.H., Hsueh, I.P., Sheu, C.F., Yao, G., & Hsieh, C.L. (2004). Psychometric properties of 2 simplified 3-level balance scales used for patients with stroke. Physical Therapy, 84(5), 430-8.
  • Yu, W-H., Hsueh, I-P., Hou, W-H., Wang, Y-H., & Hsieh, C-L. (2012). A comparison of responsiveness and predictive validity of two balance measures in patients with stroke. Journal of Rehabilitation Medicine, 44, 176-80.