DEVELOPMENT AND VALIDATION OF AN APPROACH TO INTERPRET COMPUTERIZED ADAPTIVE TEST SCORES FOR ASSESSING PATIENT OUTCOMES
Author(s)
Sinha R1, Slavin MD1, Ni P1, Houlihan B1, Skeels SE2, Zazula J1, Jette A1
1Boston University School of Public Health, Boston, MA, USA, 2Tufts University, Medford, MA, USA
OBJECTIVES: There is great interest in using computerized adaptive tests (CAT) to assess health outcomes. CATs are based on Item Response Theory (IRT) and yield a numeric estimate, but it is difficult to interpret the health status and clinical meaning from these scores. Item maps generated from IRT analyses and bookmarking procedures facilitate development of levels, and provide a meaningful context for CAT scores. We aim to develop and validate the levels for Spinal Cord Injury Functional Index (SCI-FI) used as a functional outcome measure for persons with SCI. The specific objectives are:(i) Develop levels for five SCI-FI domains: basic mobility, self-care, fine motor, wheelchair and ambulation; (ii) Determine if outcome scores for the levels are significantly different; (iii) Ascertain whether levels distinguish people of varying clinical profile (injury level and completeness). METHODS: A three-phase analytic approach involving quantitative and qualitative methodology was deployed: (a) SCI-FI items were ordered along a continuum of difficulty using IRT, and empirical item maps were generated from calibration study data (N=855); (b) Delphi approach was employed with expert panel (n=6) reviewing item maps and arriving at consensus for cut-off scores for level development; and (c) Developed levels were described and refined for meaningful outcome interpretation. One-way ANOVAs were performed (levels as factor, SCI-FI CAT scores as dependent variable). Chi square analyses were performed to compare actual to expected number of persons at each level for the varied clinical profiles, paraplegia-complete, paraplegia-incomplete, tetraplegia-complete and tetraplegia-incomplete. RESULTS: Five levels representing varying range of functional outcomes were identified for all the SCI-FI domains except one having four levels. ANOVA (pair-wise comparisons) results revealed significant score differences across levels. Chi-square tests were significant in the hypothesized direction. CONCLUSIONS: Developed levels define patient functional outcomes that provide meaningful interpretation of CAT scores for use in research and patient monitoring.
Conference/Value in Health Info
2014-05, ISPOR 2014, Palais des Congres de Montreal
Value in Health, Vol. 17, No. 3 (May 2014)
Code
UT4
Topic
Patient-Centered Research
Topic Subcategory
Health State Utilities, Patient-reported Outcomes & Quality of Life Outcomes
Disease
Musculoskeletal Disorders