Letters to the Editor ‘The GMFCS does not produce a score’ SIR–The efforts of Nemer McCoy et al. (2006)1 to validate a family-assessed measurement scale for children with spasticity are laudable. There is certainly a need to measure outcomes of interventions which are important to children and their families, and to obtain parents’ evaluations of the changes that are associated with those interventions. Changes in health status or quality of life relevant to children in Gross Motor Function Classification System (GMFCS) Levels IV and V do not appear to be detected using extant instruments such as the Gross Motor Function Measure (GMFM) or the Child Health Questionnaire. However, our attention was drawn to the article because the authors report mean and standard deviation values for the GMFCS and also refer to GMFCS ‘scores’. The GMFCS is an ordinal system (one that ‘orders’ the levels of function from more to less able, without assuming that the intervals between levels are the same throughout the system). The GMFCS is widely used to classify children’s movement ability into one of five levels but, unlike the GMFM, the GMFCS does not produce a score. The concept of a mean or standard deviation for an ordinal variable is erroneous, as it has no meaning. Instead, the number of children in each GMFCS level should be used to describe study participants; this descriptive information is usually best displayed in a table. To avoid such errors in the future, we would like to propose that Mac Keith Press adopt an editorial policy of only using roman numerals to describe levels, as intended by the developers and already used by most authors. DOI: 10.1017/S0012162206211496
Chris Morris MSc DPhila* Peter Rosenbaum MDb aMRC Training Fellow in Health Services Research, University of Oxford, UK. bProfessor of Paediatrics, McMaster University, Canada.
*Correspondence to:
[email protected] Reference 1. Nemer McCoy R, Blasco PA, Russman BS, O’Malley JP. (2006) Validation of a Care and Comfort Hypertonicity Questionnaire Dev Med Child Neurol 48: 181–187. Mac Keith Press have now adopted this as editorial policy.
‘Nemer McCoy et al. reply’ SIR–The points made by Morris and Rosenbaum are well taken. The suggestion that roman numerals be adopted as a convention for ordinal data is interesting, and it would probably be helpful for the field in general. The nonparametric statistics we used (Kruskal-Wallis H test) were appropriate for ordinal data, so none of our conclusions are affected. As stated in the statistical analysis, the GMFCS is not measured on a continuous scale.
702
Developmental Medicine & Child Neurology 2006, 48: 702–704
Regarding Tables 1 and 2, we wanted to include in the table a concise indication of the degree of motor involvement of our patient population. We did not intend to be misleading in presenting the means and standard deviations for GMFCS. It is difficult to come up with a concise summary for ordinal data, so the ‘mean’ (which we probably should have written as ‘average’) was an attempt to summarize; however, standard deviations should not have been included. Perhaps a more appropriate way to have presented the results may have been as follows: in part 2 of the study, none of the 10 participants treated with baclofen were below GMFCS Level III compared with nine of the 25 participants treated with botulinum toxin (36%; p=0.049, Fisher’s exact test utilizing all GMFCS levels). However, it would be difficult to fit all this in a table, so whether it is really better is debatable. DOI: 10.1017/S0012162206221492
Robin Nemer McCoy MDa Peter Blasco MDa* Barry Russman MDb Jean O’Malley MPHc aDepartment of Pediatrics; bDepartments of Pediatrics and Neurology; cNeuroscience Institute, Oregon Health and Science University,
Portland, Oregon, USA. *Correspondence to:
[email protected]
‘Thinking over the consensus view on the Panayiotopoulos syndrome’ SIR–In order to reach a consensus view on Panayiotopoulos Syndrome (PS), Ferrie et al. recently published an exhaustive review in DMCN.1 In this review, the authors concluded that PS should be classified as autonomic epilepsy, rather than occipital epilepsy. Indeed, in PS, the autonomic symptoms are prominent and often represent the main early ictal manifestation, as we recently stressed in our article on ictal PS.2 Since PS is but one of a number of clinical pictures whose features do, to some extent, overlap,3 we suggested2 that the so-called benign childhood seizure susceptibility syndrome may include all types of benign idiopathic focal childhood epilepsies. The child described in our article was well until the age of 3 years, when he started having frequent (1–2/mo, exclusively nocturnal) attacks consisting of pallor, tachycardia, unilateral eye deviation, and vomiting, always followed by mildly impaired consciousness and severe postictal migraine. Moreover, at the age of 7 years, as previously reported in some children with PS by Lada et al.,4 he also started experiencing occasional visual seizures (4–6/y) of multicoloured spheric hallucinations, each lasting 20 to 40 seconds; this last unusual finding is known to be frequent in Gastaut-type childhood occipital epilepsy, which suggests that the clinical features of these two forms overlap. In addition, in our report we stressed that all benign focal childhood epilepsies may be due to age-related migrant cortical hyperexcitability which may explain the age-related shift
