Objective: To determine the effects of using questions with and without health attribution on scores derived from a self-report disability instrument.
Methods: We administered the disability component of the Late Life Function and Disability Instrument to 75 community-dwelling older adults. Then, we administered the same 16 questions with attribution to specific health conditions. We used a series of analytic methods including weighted Kappa coefficient, Bowker's Test of Symmetry and Rasch analysis to assess the effects of attribution formats.
Results: A higher prevalence of disability was reported in the non-health attributed compared with the health attributed questions (t = 5.76; p < 0.001, 95% CI 3.8-7.8). Item analyses indicated that participants were significantly more likely to report disability on the non-health attributed version on 4 of the 16 questions.
Conclusion: For community-dwelling older adults, the use of a non-health attribution format may be preferable in instruments designed to assess prevalence of disability from contributing factors other than just health.