Neurol Ther. 2026 Feb 18. doi: 10.1007/s40120-026-00891-6. Online ahead of print.
ABSTRACT
INTRODUCTION: The spinocerebellar ataxia composite score (SCACOMS) comprises items from the functional Scale for the Assessment and Rating of Ataxia (f-SARA) and the Clinician Global Impression of Change (CGI-C). In the derivation of SCACOMS, weights reflecting 1-year responsiveness were assigned to each item using partial least squares (PLS) regression modeling. The current objective was to incorporate patient-feedback into the SCACOMS item weights, examine corresponding responsiveness of the composite scale, and discuss potential implications for future use.
METHODS: Item weights derived by PLS regression were compared to each item’s relative importance as assigned by 16 patients with SCA during semi-structured interviews. SCACOMS item weights were adjusted using the following combinations: (1) 50/50 weighted combination of PLS and patient weights and (2) reducing the weight of CGI-C to 20% and averaging individual item weights obtained from each perspective. The 1-year mean to standard deviation ratios (MSDRs) for the resulting reweighted scales were compared, with larger MSDRs indicating greatest sensitivity to disease progression.
RESULTS: The PLS-derived SCACOMS had the highest MSDR (0.99). When item weights were averaged across the two sources, the resulting MSDR was 0.91. When the weight of CGI-C was set to 20%, reflecting patient preferences for higher weights on the discrete symptoms, the MSDR was 0.79.
CONCLUSIONS: This study took a novel approach to enhance the face validity of SCACOMS by incorporating patient feedback into the statistically optimized item weights. The result is the merging of objectively derived item weightings (reflecting optimal scale responsiveness) with patient-assigned relevance. While this update may increase the patient centricity of a composite measure, this comes at the expense of reduced sensitivity. This potential trade-off in sensitivity to detect change should be evaluated in the context of the composite measure’s intended use.
PMID:41706362 | DOI:10.1007/s40120-026-00891-6
