CLINICAL APPLICABILITY AND TEST-RETEST RELIABILITY OF AN EXTERNAL PERTURBATION TEST OF BALANCE IN STROKE SUBJECTS

We address the test-retest reliability and clinical applicability of a n adapted external perturbation balance assessment, ie, the Postural S tress Test (PST). Repeated-measures were designed to assess the clinic al features of a component of balance disorder in stroke. Twenty ambul atory stroke patients and 20 age-, gender-, height-, and weight-matche d healthy control subjects participated in this study. Stroke patients were tested (using the adapted PST) on 4 separate days; matched contr ol subjects were tested on one occasion. With the subject standing, ba ckward perturbation forces were applied at the level of the center of gravity. Postural reactions to the test were scored in real-time and f rom videotape, from two different viewing angles, ie, 45 degrees and 9 0 degrees to the saggital plane. Scores (out of a maximal of 81) were ascertained using a 10-point subjective-observational scale. None of t he control subjects fell during testing; four of the hemiplegic subjec ts fell. Subjects were protected from potential injury by a custom-des igned safety harness system. For the hemiplegic subjects, intraclass c orrelation coefficients (ICCs), calculated as the reliability of any o ne occasion, ranged from 0.71 to 0.77, whereas those calculated as the reliability of the mean of the first two occasions ranged from 0.83 t o 0.93. Although scores on the fourth occasion were significantly grea ter than those on the third occasion, both being significantly greater than those on the first and second test occasions (p < .05), differen ces were less than 5 points on the 81-point scale. Results suggested a learning effect over time, beginning on the third occasion, and indic ated that data acquired over the first two occasions could provide a s uitable baseline. Whether the 5-point difference might be clinically m eaningful, is currently unclear. Data averaged over the four occasions for the stroke subjects were used to compare hemiplegic and control s ubjects. The two angles of viewing for the videotaped assessment produ ced similar scores for the stroke (t = 1.38; p > .05) and the healthy (t = 0.65; p > .05) subjects. Similarly, real-time and videotaped scor es (at the 90 degrees observation angle) were similar for the stroke ( t = 0.56; p > .05) and control subjects (t = 0.13; p >.05). However, v ideotaped (p < .01) and real-time scares (p < .01) (both at the 90 deg rees observation angle) were significantly lower for the stroke in com parison with the control subjects. Left (n = 10) and right (n = 10) he miplegic subjects did not exhibit a difference in adapted-PST scores ( at 90 degrees observation angle and using videotaped data; p > .05). T he adapted-PST was reliable when data were averaged for the stroke pat ients over at least two test occasions, It differentiated between a hi gh-functioning stroke group and a healthy elderly group. Both angles o f viewing produced similar results indicating that clinicians may choo se their preferred patient observation angle. The assessment can be sc ored in real-time, eliminating the need for expensive videotaping equi pment for assessment. (C) 1995 by the American Congress of Rehabilitat ion Medicine and the American Academy of Physical Medicine and Rehabil itation