Walking speed on parquetry and carpet after stroke: effect of surface and retest reliability

Objective: At the transition stage from rehabilitation to home this study a imed to (1) investigate the effect of floor surface (carpet and parquetry) on walking speed; (2) investigate whether there was a difference between th ese surfaces as stroke patients voluntarily increased from comfortable To f ast pace; (3) investigate whether walking speed on parquetry was a predicto r of walking speed on carpet at the two paces; (4) investigate whether walk ing speed at a comfortable pace was a predictor of walking speed at a fast pace on the two surfaces; and (5) quantify systematic and random error in r epeated measurements for fast-paced walking trials. Design: Subjects walked 10 metres at comfortable and fast paces on carpet a nd parquetry on two consecutive days. Setting: Inpatient rehabilitation centre. Subjects: Twenty-four stroke patients. Main outcome measure: Walking speed. Results: Two-way analysis of variance confirmed that patients walked more s lowly on carpel than parquetry (F(1,23) = 5.3, p <0.05) at both paces; the interaction effect was not significant (p >0.05). Walking speed on parquetr y was a strong predictor of walking speed on carpet at a comfortable (r= 0. 92), and fast pace (r= 0.97). Walking speed at comfortable pace was a moder ately strong predictor of walking speed at fast pace on parquetry (r= 0.84) ,and on carpet (r= 0.88). Random error in repeated measurements was higher when walking fast on carpet (7.21 m/min) and parquetry (8.32 m/min) than wh en walking at a comfortable pace on carpet (4.63 m/min) and parquetry (3.48 m/min). Systematic error was negligible (p <0.05). Conclusion: Carpet surface was more challenging than parquetry surface, as evidenced by the systematic decrease in walking speed. This may have been d ue to lack of familiarity. Relative to the wide range of scores in the grou p, stroke patients showed consistency of walking speed across both surfaces . Likewise, stroke patients retained their relative position in the group a s they changed from a comfortable to a fast walking pace. The difference in random error between comfortable and fast-paced trials highlights the need to quantify error in the repeated measurement situation according to speci fic test conditions.