MEASURING THE IMPACT OF SECONDARY PROGRESSIVE MULTIPLE SCLEROSIS (SPMS) IN THE ASCEND TRIAL- EQUATING THE MSIS-29, MSWS-12, ABILHAND-56 AND SF-36
Author(s)
Cano S1, Cleanthous S1, Marquis P2, Hobart J3, Naoshy S4, Mikol D4, Petrillo J4, Steiner D4, Watson C4
1Modus Outcomes, Stotfold, UK, 2Modus Outcomes, Newton, MA, USA, 3Plymouth University, Plymouth, Devon, UK, 4Biogen, Cambridge, MA, USA
OBJECTIVES: ASCEND is a phase 3, randomized, double-blind, placebo-controlled trial assessing whether natalizumab slows disability progression in SPMS. The objective of this study was to carry out an empirical post-hoc analysis of ASCEND blinded baseline data to equate items from the Multiple Sclerosis Walking Scale (MSWS-12), Multiple Sclerosis Impact Scale (MSIS-29), ABILHAND-56, and Short-Form 36 (SF-36) in order to provide a clearer understanding of the relationship between their respective items on “Physical” and “Psychosocial” impact measurement continua. METHODS: MSWS-12, MSIS-29, ABILHAND-56, and SF-36 data from 889 patients were combined based on a predefined conceptual framework (contrasting broad domains of physical and psychosocial impact). Rasch Measurement Theory (RMT) analyses were performed on the equating item banks using RUMM2030 software to examine: scale-to-sample targeting, item fit, local dependency, and reliability. RESULTS: The ‘Physical Impact’ item bank showed adequate scale performance except for some reversed thresholds (5 of 92) and misfit (5 of 92 fit residuals; 20 of 92 Chi-square). Relative item locations implied that, in general, lower limb related items represent the less impacted end of the continuum followed by upper limb related items; general limitations items tended to sit in the middle of the measurement continuum. The ‘Psychosocial Impact’ item bank performed psychometrically well, except for some misfit (7 of 15 fit residuals; 2 of 15 Chi-square). Relative item locations implied that ‘worries’ and ‘anxieties’ are at the less impacted end of the continuum followed by ‘mood’ and ‘depression’ items. These findings are consistent with clinical expectation. CONCLUSIONS: Psychometrically, the ‘Physical Impact’ and ‘Psychosocial Impact’ equating item banks proposed have strong measurement properties. However, issues related to response options and item fit still require consideration (in both banks), and coverage could still be improved in the ‘Psychosocial Impact’ equating bank due to the relatively small number of items.
Conference/Value in Health Info
2015-11, ISPOR Europe 2015, Milan, Italy
Value in Health, Vol. 18, No. 7 (November 2015)
Code
PRM174
Topic
Methodological & Statistical Research
Topic Subcategory
PRO & Related Methods
Disease
Neurological Disorders