Arising from: Pulst SM (2007) Ataxia rating scales in the balance. Nat Clin Pract Neurol 3: 119 doi:10.1038/ncpneuro0446

Subramony SH (2007) SARA—a new clinical scale for the assessment and rating of ataxia. Nat Clin Pract Neurol 3: 136–137 doi:10.1038/ncpneuro0426

Since Pierre Marie suggested a clinical differentiation between Friedreich ataxia and other forms of hereditary ataxias in the 19th century, it has been clear that these entities affect multiple neurological subsystems.1 The development of molecular testing for inherited ataxias has had a considerable impact on the nosological taxonomy of these overlapping presentations. Nevertheless, their clinical heterogeneity was confirmed by demonstrating the presence and prognostic value of nonataxia features, such as pyramidal2 and extrapyramidal3 signs.

As pointed out in the editorial by Pulst and in the Practice Point commentary by Subramony in the March issue of Nature Clinical Practice Neurology, scales recently developed for the assessment of spinocerebellar ataxias (SCAs) do not include items on extracerebellar deficits. Evaluation of cerebellar symptoms without accounting for extracerebellar manifestations—which are partially responsible for the impairment and disability seen in patients with SCA—oversimplifies complex neurological presentations into single-system disorders. By focusing on the severity and distribution of ataxia signs, instruments provide a highly objective and sensitive measure of the hallmark aspect of SCAs, but ignore other elements relevant to the understanding of their natural history and to the development of therapeutic strategies.

An important methodological issue arises from this fact: the homogeneity of instruments is frequently evaluated by means of inter-item correlation coefficients such as Cronbach's α, but SCAs have a considerable degree of heterogeneity. Thus, the rationale of 'the higher the coefficient, the better' might be inadequate when one intends to appraise a complex phenomenon. Ideally, there should be a moderate correlation among items to ensure that a scale is not too narrow in its scope and that each item brings additional information. In order to avoid redundancy and thus loss of content validity, an α value between 0.70 and 0.90 would be desirable4 when measuring different aspects of the same attribute. The recently developed Scale for the Assessment and Rating of Ataxia (SARA) has an α value of 0.94.

Even with the improvements in patient evaluation achieved by ataxia scales, many issues remain unresolved. Challenges include the quantification of disease severity, accounting for the variable combination of qualitatively different signs and symptoms and for disease progression over time. We recently developed a scale based on the standardized neurological examination (including items on cerebellar and noncerebellar deficits), and this instrument proved to be reliable (α value 0.77) and valid in a group of 99 patients with SCA3 (Kieling C et al., unpublished data).

As suggested by Pulst, it seems logical that a future 'unified' ataxia rating scale could be built that combines features from different impairment and disability rating scales and that possibly also includes measures of quality of life. Statistical limitations of such a procedure include the difficulty of combining instruments with distinct metric profiles and the fact that internal consistency, as measured by Cronbach's α, has also the drawback of being easily inflated by the number of items in a scale. To encompass a polymorphic clinical picture such as that of the SCAs and to yield a comprehensive description of their natural histories, an instrument should be feasible to apply and multidimensional, covering a broad variety of features of ataxia as a disease and not only as a neurological sign.