Eight physiotherapists and four physiotherapy assistants participated in the study. The physiotherapists ranged in experience from one

to 14 years post-graduation and the physiotherapy assistants had between two and 10 years of experience. Physiotherapists were managing caseloads of a mean of 8 patients (SD 2). The participants had a mean (SD) age of 68 (13) years, 9 (64%) were male, 7 (50%) had a right-sided stroke lesion, 6 (43%) had a left-sided lesion and 1 (7%) had a bilateral stroke. The average duration of physiotherapy sessions was 55.6 (23.4) minutes (range 19 to 90) (Table see more 1). There was strong agreement between therapist-estimated and video-recorded total therapy times (ICC = 0.90, see Table 1), however there was a systematic overestimation of total

therapy time by the therapists, mean difference 7.7 (SD 10.5) minutes (95% CI Ibrutinib chemical structure 4.6 to 10.8). The Bland-Altman plot (Figure 1) for total therapy time presents this systematic overestimation. Similarly, there was strong agreement between therapistestimated and video-recorded time for total active time in therapy sessions (ICC = 0.83, see Table 1) with a systematic overestimation of total active time by the therapists, mean difference 14.1 (SD 10.3) minutes, 95% CI 11.1 to 17.1 ( Figure 2). However, there was less agreement between therapist-estimated and video-recorded inactive time (ICC = 0.62, see Table 1), and therapists systematically underestimated the amount of time patients were inactive during therapy sessions, mean difference –6.9 Oxalosuccinic acid (SD 9.5) minutes, 95% CI –9.7 to –4.1 ( Figure 3). Comparing the influence of session type (individual versus group) using percentage mean difference,

there was no difference in the accuracy of estimations of total active time between individual (28%) and circuit class therapy (28%) sessions, but therapists tended to underestimate inactive time in circuit class therapy sessions (37%) to a greater extent than in individual therapy sessions (29%) (Table 2). In terms of the various subcategories of activity, ICC scores ranged from 0.73 to 0.99 for all of the categories except for ‘transfers and sit-to-stand practice’, which had a low ICC score of 0.37, indicating only a fair agreement between therapists’ estimations and video recordings (Table 3). As with the total active time, therapists tended to overestimate the time patients spent engaged in the various physical activity categories. The magnitude of this overestimation varied, but in some cases was as high as 63%. This is the largest study to date to investigate the accuracy of therapists in recording therapy time, and the only such study to involve multiple data collection centres and to include group therapy as well as individual therapy sessions.