Incorporating Patient Perspectives into a Composite Score for Measuring Disease Progression in Spinocerebellar Ataxia (SCA)
Michele Potashman1, Basia Rogula2, Lauren Powell2, Ellen Korol2, Samuel Dickson3, Suzanne Hendrix3, Maggie Heinrich4, Katja Rudell4, Linda Abetz Webb5, Melissa Beiner1, Vlad Coric1, Liana Rosenthal6, Susan Perlman7, Jeremy Schmahmann8, Gilbert L'Italien1
1Biohaven Pharmaceuticals, 2Broadstreet HEOR, 3Pentara Corporation, 4Parexel International, 5Patient-Centered Outcomes Assessments, 6Department of Neurology, Johns Hopkins School of Medicine, 7Department of Neurology, UCLA School of Medicine, 8Department of Neurology, Massachusettes General Hospital
Objective:
To incorporate patients’ perspectives into the statistically-derived spinocerebellar ataxia composite score (SCACOMS).
Background:
SCACOMS was derived using partial least squares (PLS) regression to objectively determine item weights for the candidate 5 items (4 items from the f-SARA and clinicians global impression of change [CGI-C]). This scale was optimized to detect disease progression in patients with SCA, with more progressive items given higher weights.
Design/Methods:
Item weights derived statistically were compared to relative importance assigned to the items by N=24 patients with SCA who participated in semi-structured interviews. SCACOMS item weights were adjusted by 1) 50/50 combination of statistically-derived and patient-stated weights and 2) reducing maximum weight of CGI-C to 20% before averaging the remaining item weights. The 1-year mean to standard deviation ratios (MSDRs) were compared for these approaches; larger MSDRs indicated greater sensitivity to measure progression.
Results:
SCACOMS items’ PLS regression weights and patient assigned median weights, respectively, are: gait (12% vs 30%), stance (17% vs 17%), sitting (8% vs 10%), speech (10% vs 25%), and CGI-C (53% vs 13%). The MSDR for the PLS-derived SCACOMS is 0.99. The MSDR obtained incorporating patient weights is 0.70. When the weights are averaged across the 2 sources, the resulting MSDR is 0.91. When CGI-C is set to a maximum of 20% of the total composite and the remaining f-SARA item weights averaged, the MSDR is 0.79, indicating that the CGI-C item is a driver of scale responsiveness.
Conclusions:
This study took a novel approach to enhance the validity of a composite measure SCACOMS. The resulting scale, with an updated set of SCACOMS item weights, effectively balances scale responsiveness with patient-relevance. Incorporating results of qualitative patient validation weighting with the PLS-derived weighting may increase acceptance and adoption of composite measures, which are more likely to accurately track disease progression and identify effective treatments.
Disclaimer: Abstracts were not reviewed by Neurology® and do not reflect the views of Neurology® editors or staff.