Improved Precision of a Lower Extremity Osteoarthritis Functional Outcome Instrument

Purpose: A limitation of traditional functional outcome instruments, like the WOMAC, is reduced precision at different points along the functional continuum. An advantage of outcome instruments built with contemporary Item Response Theory (IRT) models is their ability to achieve optimal precision across the entire content range. Administering IRT instruments using computer adaptive test methodology (CATs), which uses computer algorithms to administer a subset of items from an underlying bank of items based on a patient’s functional level, enables the user to more precisely estimate a person’s location along a functional continuum without sacrificing feasibility of doing the assessment.
Methods: We constructed a functional item bank by asking 309 patients with confirmed knee or hip OA to rate the difficulty and pain they experienced in performing 126 functional tasks & activities. The item banks serve as a foundation for an OA-CAT functional outcome instrument. IRT methods were used to calibrate items into two distinct scales: 1. Functional Difficulty and 2. Functional Pain. Precision, defined in IRT analysis as how well scale scores reflect a patient’s true score at different points along the functional scale, was estimated as a standard error of measurement (SE) around each score for the OA-CAT item bank scales and compared with WOMAC difficulty and pain scale scores estimated from the same sample. Rasch analysis was used to calculate SEs. In contrast to Classical Test Theory, which estimates the same SE associated with the entire scale, IRT methods such as Rasch analysis provide a unique SE at each score level.
Results: Precision of the OA-CAT item bank scales was superior to the WOMAC at 5 pre-determined locations along each functional scale (logit -4,-2, 0, 2, 4). At each location in the Functional Difficulty scale, OA-CAT had SEs that were all < 0.36 while the corresponding WOMAC SEs ranged from 0.31 - 1.19. At each point along Functional Pain scale, SEs for the OA-CAT were under 0.35 in contrast to the corresponding WOMAC SEs, which ranged from 0.57 - 1.83.
Conclusions: Findings from this analysis demonstrate that a person’s pain and difficulty in performing a functional task can be measured precisely across the entire functional continuum using IRT-derived functional item banks in contrast to the WOMAC. Improved measurement precision is likely to lead to improved sensitivity to change.

 A.M. Jette, None; S.M. Haley, None; P. Ni, None; S. Olarsch, None; D.J. Hunter, None; D.T. Felson, None.