Evaluate the psychometric and clinimetric properties of the modified-functional Scale for the Assessment and Rating of Ataxia (f-SARA).

International experts developed the SARA to provide semi-quantitative scoring of disorders of motor control in cerebellar ataxia, notably spinocerebellar ataxia (SCA).  Accounting for feedback from FDA, a subset of items were selected (gait, stance, sitting, speech) and response options collapsed to a uniform 5-point scale to create f-SARA, a putatively clinically meaningful measure of disease progression.

Two data sources were examined:  SCA subjects (ranging in disease severity) enrolled in the PROM-Ataxia validation study (MGH-cohort, cross-sectional) and SCA subjects enrolled in BHV4157-206, 48-week study (NCT03701399, longitudinal).  Psychometric properties evaluated were data acceptability (ceiling/floor effects), internal consistency (using Cronbach’s alpha) and test-retest reliability, convergent/divergent validity, and responsiveness.  Intra-individual meaningful changes were examined using distribution-based (0.5-SD and SEM) and anchor-based methods (anchor: CGI-I).

In subjects enrolled in the MGH-cohort (N=33), ceiling effects were absent while floor effects were observed.  Excellent internal consistency was demonstrated (α_total=0.90; α_{items-removed}= 0.86-0.90) and item-to-total correlations were strong (r=0.82-0.91, per item). Convergent and divergent validity were supported with stronger correlations observed between the f-SARA and scales of similar construct (p<0.001) (e.g., FARS-FUNC (r=0.92), BARS (r=0.88), PROM-ADL (r=0.83), and FARS-ADL (r=0.69)), while weaker correlations were observed amongst measures of differing constructs.  Mean scores (item and total scores) increased with disease severity (p<0.001).  Leveraging the BHV4157-206 cohort, high test-retest reliability was demonstrated with intra-class correlation coefficients (ICCs) of 0.91 (total) and 0.73-0.92 (items).  A 1-point threshold for meaningful change is supported with 0.5-SD=0.89, SEM=1.19, and mean changes from baseline for patients classified as “improved”, “no change” or “deteriorated” on CGI-I were -0.68, 0.02 and 0.58, respectively.

We show that f-SARA has psychometric validity and can detect meaningful change over a 1-year time-period, with a 1-point intra-individual threshold supported.