Reliability and discriminant validity of ataxia rating scales in early onset ataxia

Objective: To determine observer-agreement and discriminantvalidity of ataxia rating scales.Background: In children and young adults, Early Onset Ataxia(EOA) is frequently concurrent with other Movement Disorders,resulting in moderate inter-observer agreement among MovementDisorder professionals. To investigate whether subjective phenotypicassessment is replaceable by quantitative measures, we aimed todetermine inter-observer agreement and discriminant validity ofataxia rating scales.Methods: In 40 EOA patients (15 (5-34) years; mean (range)),three independent pediatric neurologists assessed quantitative ataxiarating scales (ICARS, SARA and BARS), and phenotyped the pri-mary Movement Disorder characteristic (i.e. ataxic, dystonic, myo-clonic, choreatic, tics) in each patient. We determined inter- andintra-observer agreement and specified outcomes for “primary” (i.e.unanimous ataxia identification as the primary feature by all asses-sors and/or genetic ataxic diagnosis (n526)) and “secondary” (i.e.incomplete identification of ataxia as the primary Movement Disor-der (n512)) subgroups.Results: Inter- and intra-observer agreement of ataxia ratingscales revealed high intra-class correlation coefficients (ICC: .92 -.99; for ICARS, SARA and BARS), with no significant differencesbetween “primary” and “secondary” subgroups. Total ataxia ratingscale scores revealed higher outcomes in the “primary” than the“secondary” subgroup (p<.001). A multivariable regression analysisrevealed that the severity of the primary Movement Disorder charac-teristic (b .57- 1.0; p .011 - <.001) predicted these higher quantita-tive outcomes. The primary movement characteristic itself did notpredict higher outcomes.Conclusions: In EOA, quantitative ataxia rating scales revealhigh inter- and intra-observer reliability, reflecting reliable applicabil-ity. However, multivariable regression analysis revealed low discrim-inant validity between ataxia and other Movement Disordercharacteristics. Despite high reliability of quantitative ataxia scores,these data implicate that preceding phenotypic characterizationremains irreplaceable.