Measurement in Health and Disease exercise: Reliability of balance performance measures in chronic stroke patients

This website is for students following the M.Sc. in Evidence Based Practice at the University of York.

The following is the (slightly edited) abstract of a paper:

Purpose. To examine the relative reliability and absolute reliability of the Berg Balance Scale (BBS) and the Postural Assessment Scale for Stroke Patients (PASS) in chronic stroke patients.

Method. A total of 52 mild to moderate stroke patients, who had a stroke more than 6 months previously, participated in the study. Both balance measures were administered twice, seven days apart, to the patients. A relative reliability index (intra-class correlation coefficient, ICC) was used to examine the level of agreement between test and retest. Absolute reliability indices, including the the standard error of measurement (SEM), and the smallest real differences (SRD), were used to define the extent to which a balance score varies on test-retest measurements.

Results. Test-retest agreements were high (ICC: BBS 0.98; PASS = 0.97), indicating excellent agreement from a relative perspective. The SEM of the BBS and PASS were 2.4 and 1.1, respectively. The SRD of the BBS and PASS were 6.7 and 3.2, respectively, exhibiting the smallest change threshold that indicates a real improvement for a single individual.

Conclusions. The test-retest agreements of the BBS and PASS were high in mild to moderate chronic stroke patients. The thresholds of both measures to detect real change are acceptable in research and clinical settings.

(Source: Liaw LJ, Hsieh CL, Lo SK, Chen HM, Lee S, Lin JH. The relative and absolute reliability of two balance performance measures in chronic stroke patients. Disability and Rehabilitation 2008; 30: 656-661.)


This paper uses two terms which are different to those in the lecture:

The BBS and PASS measures

The Berg Balance Scale is a staff-completed assessment scale of the ability to maintain balance, either statically or while performing various functional movements, to help make decisions about the patient's mobility and level of care needed. It comprises 14 observable tasks measured on a scale from 0 to 4. The tasks are described as sitting unsupported, change of position from sitting to standing, change of position from standing to sitting, transfers, standing unsupported, standing with eyes closed, standing with feet together, tandem standing, standing on one leg, turning trunk with feet fixed, retrieving objects from the floor, turning 360 degrees, stool stepping, and reaching forward while standing. The total score is between 0 and 56.

The Postural Assessment Scale for Stroke Patients (PASS) contains 12 similar items scored 0 to 3 that grade the ability to maintain or change a given lying, sitting, or standing posture. The total score is between 0 and 36.

Questions about this abstract:

  1. Why do the authors refer to the intra-class correlation coefficient as a "relative" reliability index?
    Check suggested answer.

  2. What kind of reliability are the authors measuring here?
    Check suggested answer.

  3. What can we conclude from the ICC = 0.98 for BBS?
    Check suggested answer.

  4. For PASS the SEM was 1.1 and the SRD was 3.2. What do these numbers tell us?
    Check suggested answer.

  5. Why are the ICCs for BBS and PASS very similar, but the SEM and SRD are quite different?
    Check suggested answer.

  6. Standard deviations relate to populations, samples, and distributions, standard errors are attached to estimates Why is the within subject standard deviation also called the standard error of measurement? (This is a difficult one!)
    Check suggested answer.

Back to Measurement in Health and Disease index.

To Martin Bland's home page.

This page maintained by Martin Bland.
Last updated: 15 July, 2008.

Back to top.