Psychometric properties of the caregiver assessment of movement participation scale for children with developmental coordination disorder

Empirical Research Background: Developmental coordination disorder (DCD) in children is defined as marked impairment in the development of motor coordination (American Psychiatric Association, 1994, 2000). Four diagnostic criteria are used: (i) the child‘s motor performance is substantially below expected; (ii) the disturbance significantly interferes with academic achievement or activities of daily living; (iii) the disturbance is not due to other medical conditions; (iv) it excludes mental retardation with an intellectual quotient (IQ) score less than 70 on a standardized IQ test. As most children with DCD can walk, jump like their peers and attain developmental milestones, they often fell out of the loop at routine health surveillance checks and would therefore not be referred to a more detailed evaluation using criterion-referenced assessment. The second criterion, therefore, becomes essential for early screening of children who are at risk of having deficient movement participation significantly enough to cause atypical interference on activities of daily living. The Caregiver Assessment of Movement Participation (CAMP) was therefore developed as an objective measure of deficient movement participation at home to justify the need to initiate further professional assessment. Empirical Research Aims: The Caregiver Assessment of Movement Participation (CAMP) was designed as a caregiver-proxy questionnaire to measure problems of movement participation in children with DCD aged between 5 and 10 years in daily living contexts. The aim of this research was to investigate the psychometric properties of CAMP, with the following specific focus: (1) Is the 1-5 rating scale appropriate for screening purposes? (2) Do the items constitute a unidimensional model? (3) Are the items too easy or too difficult for children having DCD? Empirical Research Sample: Participants were recruited between September 2006 and January 2007 from mainstream schools. It consisted of a convenience sample of 465 children in two primary schools and two kindergartens (192 females and 273 males) aged 5-10 (94 children aged 5-6, 241 aged 7-8, 130 aged 9-10; and K.3 = 30, P.1-2 = 236, P.3-4 = 199). The inclusion criteria for the school samples were: (1) no known sensory, motor, neurological or intellectual impairment, as in compliance with the diagnostic criterion adopted by the American Psychiatric Association and World Health Organization (American Psychiatric Association, 1994; World Health Organization, 1992); (2) aged between 5 and 10; and (3) studying currently in mainstream primary schools from grade one to grade four, or, in the last year of kindergarten. Children who were found to have any past medical history of neurologically related diagnosis such as Attention Deficits Disorder, Autism, Cerebral Palsy or Mental Deficiency, as reported by their caregivers in the CAMP questionnaire, were excluded. Empirical Research Method: The ethics committee of the University, as well as the Institutional Review Board of the participating kindergartens and schools, approved the study protocols. All of the participants' parents/caregivers signed informed consent forms prior to data collection. Every caregiver was asked to fill out the CAMP. In order to determine the presence or absence of DCD, we used the Movement Assessment Battery for Children (M-ABC) Test (Henderson & Sugden, 1992) as the gold standard. Those who scored at or below the 5th percentile on the M-ABC Test were categorized into the DCD group while those with scores above the 5th percentile were included in the non-DCD group. Based on these criteria, there were 33 children with DCD and 432 non-DCD children in the sample. Empirical Research RASCH: Partial credit model Rasch analysis was performed using WINSTEPS version 3.66.0 for data analyses of the following psychometric properties of the CAMP. In the first part of the study, the Rasch model was employed to analyze the appropriateness of the 1-5 ratings. The second part of the study examined the extent to which the CAMP items with revised rating categories contributed to a unidimensional construct. The third part of the study examined the model fit and item fit including point-measure correlation coefficient, the person/item reliability and the analysis of item map. Empirical Research Results: As a result of collapsing the 5th category downward into the 4th category, no more misfit categories were found as the MnSq of all rating categories were less than +2.0. Not only was there a higher percentage of category frequency observed for the rating categories (6%), but also the distribution of category frequency appeared more uniform with total observed count increased from 662 and 298 to 960 for the highest categories. This implies that the revision of the CAMP rating scale into 1-4 reduced the chances of scoring inconsistency and unreliability. Collapsing the 4th and 5th categories together improved the representation and interpretation of the CAMP rating scale. When the 31 children with poor misfit statistics were excluded from the item fit and dimensionality analysis, misfit items were deducted stepwise from the item pool when their fit statistics exceeded our predefined infit/outfit MnSq and/or ZSTD values. As a result, after removing five misfit items in a stepwise manner, the final item statistics showed that all remaining 30 items in the CAMP had their MnSq<1.4 fitting the Rasch model. The dimensionality of the 30-item CAMP was explored using principal contrast analysis (PCA). Measured in eigenvalue units, among the 60.4(100%) total raw variance in observations, 30.4 raw variance (50.4%) of which can be explained by the CAMP items; 21.7 raw variance (36.0%) of which can be explained by participants’ responses. 30.0 (49.6%) raw variance was unexplained variance which was further scattered out into many small clusters. Since no significant clusters was found, this helped to confirm that the CAMP is uni-dimensional and its items have no meaningful other dimensions. None of the items in the 30-item CAMP had negative point-measure correlation. All of the CAMP items had a point measure correlation higher than 0.5 ranging from 0.5 to 0.8 when item scores were correlated with the overall scores. The Rasch model standardized residual variance statistics therefore confirmed the unidimensionality of the construct measured by the CAMP. That is, the 30 items form a single construct and the total or summary score can be used as an indicator of the degree of deficient movement participation in home contexts for children aged between 5 and 10. The item constructs as reviewed on the item map were analyzed according to the three aspects of task contexts: (1) task relevancy of context – task relevant or irrelevant; (2) stability of the environment – stable or unstable environment; and (3) variability in time – variable or invariable time. Looking at the person measure on the item map, the children who perform between average mean and one standard deviation below, that is, children who had least movement problems scored mostly 1 or 2. The easiest items that were scored 1 are mostly task relevant, stable environment and time invariable including items like “cutting, folding and pasting”, “buttoning and zipping” and “assembling construction”. The task difficulty increases as one or more task characteristics change to the other side of the continuum. As for instances, in “fastening”, the parameter of force adjustment changes as the laces are pulled tighter and tighter making the environment unstable but the time remains invariant for the task. It can be seen that most of these items were scored 2 for an average performing child. For children who perform from average to one standard deviation above the mean, their scores for most of those items mentioned earlier were from 2 to 3; as the items included “organizing schoolbag”, “carrying multiple items”, “competitive games”, “skip steps in multiple action task”, “grasping pencil”, “handwriting”. Items with higher logits ranging from 1.08 to 0.10 become more significant to measure performance for this level of person performance and were scored 1 or 2. Items include: “hopping and jumping”, “putting on long pants without leaning”, “using keys and sharpeners”, “on tip toes or on one foot”, “bump into furniture”, “clapping to rhythm”, “break fall when tripped” and “new action games”, etc. Empirical Research Conclusions: Our findings support deficient movement participation as an accurate quantifier of DCD in caregiver assessment. The total CAMP score can be served as an objective and valid evidence for measuring the impact of movement participation problems at home due to the impact of DCD. Future research should could determine a cut-point for classification purposes after a more thorough item map analysis to screen children having DCD.