Interrater reliability of the Physical Ability Scale for patients after stroke: a pilot study

This study aimed to evaluate the interrater reliability of the Physical Ability Scale (PAS) and secondly to test the statistical methods to analyse the scoring. Williamson ( 1993a ) stated that the PAS measures the degree of compatibility of the trunk with gravity and the supporting surface through investigation of limb, head, and posture in five different positions, each with six levels of ability. Four experienced physiotherapists observed a video of 10 inpatients from a general hospital stroke unit being assessed with the PAS up to 5 months after a stroke. The intraclass coefficient (ICC) for the total scores was 0.7 (95% CI 0.35-0.91) with the highest individual position scores in standing 0.86 (95% CI 0.68-0.96) and long sitting 0.74 (95% CI 0.46-0.92). The ICC for the levels of ability indicated good agreement in standing 0.79 (95% CI 0.55-0.93) and long sitting 0.77 (95% CI 0.52-0.93). Kappa values for each body part yielded the highest values in long sitting (0.62) and standing (0.56). Interrater reliability is variable, with the greatest level of agreement in long sitting and standing but with poor to fair agreement in sitting, prone, and supine. The study has also shown that the scoring can be analysed to yield meaningful repeatable results. Further investigation is needed to examine validity and reliability with a larger sample of diverse ability, therapists with varying clinical experience, and with direct observation of the assessment.