Sonja Mensch

147 They offered suggestions for improvement of the wording of items, for the score system and for additions to the training. Content validity, as judged according to COSMIN criteria, was adequate. Chapter 4. Reliability of Movakic After improving the instrument according to outcomes of the pilot, test-retest reliability and inter-rater reliability were assessed. Thirty-seven trained physiotherapists participated, all working in specialized children’s day-care centres andwith over 10 years of experience with the target population. Each of them selected one or two children of their own clients, resulting in a total of 60 participating children. Test-retest reliability could be assessed for 50 children: intraclass correlations for the total score (ICC=.95, 95%-confidence interval (CI) .92-.97) and for scores of the four groups of motor abilities (range .72-.98) were all excellent or good. In 19 children for whom a second therapist was available who was familiar with the child too, inter-rater reliability could be established. Intraclass correlations for the total score (ICC=.94, 95%-CI .85-.98) and scores for the four groups of motor abilities (range ICC=.85-.97) were also excellent or good. Therefore, we conclude that Movakic is a reliable instrument for children with severe multiple disabilities. Chapter 5. Construct validity and responsiveness of Movakic After establishing reliability, in a prospective cohort design, Movakic was completed six times during 18 months by the same 37 physiotherapists for the same 60 children. For each child, one situation had to be selected that was relevant to the child and was expected to remain stable during follow-up. In order to measure construct validity, scores were compared with the expert judgement of the physiotherapists bymeans of Pearson correlations. Prior to completion of Movakic, experts had to report their scores on a visual-analogue scale; these scores could not be seen or changed afterwards. Construct validity was good ( r =.50-.71). In order to measure responsiveness to change, score-changes in 3-month- periods, in which no relevant influences had been reported for the child, were compared with score-changes in periods during which events had occurred that might have influenced the scores positively or negatively, using intraclass correlations. To this end, after each 3-month-period, the physiotherapists reported the relevant events. The following events were used for analyses; surgery, increase of contractures and pain. Summary