‘The relationship of cerebral palsy subtype and functional motor impairment: a population‐based study’

SIR–We write to address several concerns after reading the recent paper by Shevell et al. We have questions about the authors’ interpretation of their data to say that neurological subtype is a powerful predictor of functional status related to ambulation. We are most concerned if, as is suggested, type of cerebral palsy (CP) and limb distribution are going to be used by professionals as a basis for counselling individual families on the functional prognosis of their child with CP. It is surprising that the findings were not contrasted with existing knowledge on the use of classification systems and functional prognosis in children with CP. In 2004, Gorter et al. published a study of 657 children and young people with CP that examined the relationships among topographical distribution, type of motor impairment, and functional status as described by the Gross Motor Function Classification System (GMFCS). The estimated rate and limit of gross motor function as measured by the Gross Motor Function Measure (GMFM) were noted to vary substantially within subgroups of limb distribution (i.e. hemiplegia, diplegia, and quadriplegia). Considerable evidence was provided to support the conclusion that ‘Although classification of CP by impairment level is useful for clinical and epidemiological purposes, the value of these subgroups as an indicator of mobility is limited in comparison with the classification of severity with the GMFCS.’ It was reported elsewhere that ‘The GMFCS seems to provide meaningful distinctions in gross motor development between five functional subgroups. In contrast, grouping by limb distribution or type of motor impairment does not provide the clinician additional prognostic information in terms of gross motor abilities.’ This latter finding was recently confirmed in a study by Wichers et al. who concluded that, notwithstanding the value of limb distribution, ‘... activity limitations are determined only partly by the mere presence of motor impairments.’ Further, we are unaware of clear and meaningful descriptions of the distinctions between, for example, ‘severe’ diplegia and quadriplegia, or between asymmetrical hemisyndromes (with very few signs on the contralateral side) and bilateral CP. Differentiation in limb distribution beyond unilateral and bilateral syndromes is still controversial for its reliability. This might be an explanation for the fact that Shevell et al. found that the subtype assignment in young children with CP – based on the child’s impairment – changed over time in 28% of the cases. Thus, if physicians use these data to counsel parents at the time of disclosure of diagnosis of CP, they are likely to be wrong for a significant group of children, especially for those with a non-spastic subtype, if they say that these children will not walk. The findings reported by Shevell et al. are hardly surprising when the five-level GMFCS is dichotomized to two levels – walking and not walking. There is an important statistical issue of loss of power in analyses where a five level ordinal grouping is dichotomized. Wood and Rosenbaum observed that in offering a prognosis for ‘walking’ it was important to define what one meant by the term. Depending on whether one grouped children in levels I to III versus IV and V, or included children in level III with those in levels IV and V, the predictive validity changed considerably when those predictions were made with young children. Shevell et al. also combined children with dyskinetic and spastic quadriplegic CP and argued that children with dyskinetic CP were significantly less likely to walk. In a recent paper from the Surveillance of Cerebral Palsy in Europe (SCPE) group more than 40% of the 578 children with dyskinetic CP were recorded as walking with or without aids. The SCPE paper describes the profile of associated impairments and walking ability as being very different from bilateral spastic CP, making it clear that it is not appropriate to combine these subgroups in analyses as Shevell et al. have done. Shevell et al. studied associations between limb distribution and GMFCS in a cohort of 243 children in Canada. Their report is described as a population-based study, but it is relevant to note that only one child out of 52 (2%) with spastic diplegia was classified in the combined GMFCS level IV and V group, as compared to 8.3% of 657 children with spastic diplegia in another community sample in Canada. Although both studies found that the overall association between classification on the impairment level (limb distribution and type of motor impairment) and classification according to function (by GMFCS) was statistically significant, Gorter et al. showed that the explanatory power of these associations was low. The same would almost certainly apply if the Shevell et al. data from Table III were analysed in the same way.