This is, to our knowledge, the first study of the association between gross motor function (GMFCS) and manual ability (MACS) in a total population of children with CP. We found a poor overall correlation between the two systems as evaluated by the kappa statistics. However, in the CP subtypes, different associations were found. In hemiplegic CP, manual ability was more limited than gross motor function. The opposite was found in children with diplegic CP where gross motor function was more limited. We found a closer association between levels in children with dyskinetic CP.
In a previous study by Beckung and Hagberg , a strong correlation was found between GMFCS level and Bimanual Fine Motor Function (BFMF) level in a population of 178 children with CP. The BFMF has not been tested for reliability and validity and has more focus on manipulation and grip than on manual ability in daily activities. It is therefore difficult to compare BFMF levels with MACS levels, as the two systems are almost certainly describing very different aspects of function, with BFMF looking more at "impairment" and MACS more at "activity".
In the present study, the proportion of children with minor functional limitations, GMFCS I and MACS I, was higher than in some earlier studies [11, 13, 22]. We believe that this is mainly due to our careful inventories in order to identify all children with CP living in our region, to offer them participation in the CPUP programme. The prevalence of CP in the study area is 2.4/1.000 .
Gross motor function and manual ability in cerebral palsy are not equivalent entities. Hand function is very closely dependent on cognitive ability and voluntary motor control, and there is often a significant difference between maximal capacity and spontaneous performance, i.e., what the child can do and what he or she really does. The MACS aims to evaluate the latter and give a picture of how well the child can manage in daily manual activities. Performance and capacity are often more closely related in gross motor function, i.e., if a child can walk or sit, he or she usually does .
Another difference between gross motor function and upper limb function is the consequences of a unilateral impairment. A person with unilateral CP will probably walk, or try to walk, on the plegic leg but will sometimes not use the affected arm at all, leading to a lack of bimanual function. To be classified as MACS level I, bimanual hand function is required, and since many children with spastic hemiplegia use alternative strategies to compensate for poor bimanual function they will be evaluated as MACS level II.
These differences may explain why many children with hemiplegia in the present study were evaluated as functionally more limited in manual ability than in gross motor function. In children with spastic hemiplegic CP 53 of 121 (44%) were more limited in manual ability than in gross motor function.
Among children with diplegia 39% were more limited in gross motor function (GMFCS) than manual ability (MACS). The variability was large in this group, with a distribution across all levels of both the MACS and the GMFCS. This variability indicates that the CP subtype spastic diplegia alone gives us insufficient information about the child's gross motor function and about manual ability . Almost all children with spastic tetraplegia were classified as GMFCS V and MACS V, and this subtype seems to be well defined in the Swedish classification. The term bilateral CP, suggested by the SCPE group, joins together spastic diplegia and tetraplegia, which further stresses the need for additional functional grading in order to correctly evaluate each child. A structured model for such a functional grading was also suggested by the SCPE group and may be very useful when describing populations of children with CP.
Children with dyskinetic CP in the present study (n = 51) had large functional limitations and 67% were at levels GMFCS and MACS IV and V. In this group of children, the association between GMFCS and MACS levels was strong. Children with ataxia constituted a small group in the present study (7%). Most had good motor function, 9 of 17 were GMFCS and MACS level I.
There were few outliers. Two children with MACS IV and GMFCS I had spastic hemiplegia, probably with decreased cognitive ability. Two children with MACS I and GMFCS IV had spastic diplegia, and one boy with MACS I and GMFCS V had dyskinetic CP.
The two different non-parametric tests yielded consistent results regarding the association between GMFCS and MACS in the CP subtypes. The Sign test is adequate for paired ordinal data, but generally has low statistical power. The Wilcoxon signed ranks test uses more information, but its applicability for ordinal data could be questioned.