Purchase this article with an account.
David A Leske, Sarah R Hatt, Laura Liebermann, Jonathan M Holmes; Comparison of Rasch Methods for Responsiveness Analysis of the Adult Strabismus 20 Questionnaire. Invest. Ophthalmol. Vis. Sci. 2014;55(13):4088.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
The Adult Strabismus 20 (AS-20) Questionnaire is a validated strabismus-specific health-related quality of life (HRQOL) instrument. Rasch analysis provides a linear measure of HRQOL, and for ease of use in the ophthalmic community, we have previously provided Rasch lookup tables online. We compared responsiveness of the AS-20 to surgical intervention when raw scores are converted to a Rasch measure using the lookup table vs a more rigorous method of performing a de novo stacked Rasch analysis.
147 adult strabismus patients completed the AS-20 before surgery and again at their 6-week postoperative visit (median 7 weeks, range 4-21 weeks). AS-20 scores were analyzed using 2 methods 1) using Rasch lookup tables 2) using a de novo stacked Rasch analysis. Pre- and postoperative scores were compared for each of the 4 AS-20 domains (Self-Perception, Interactions, Reading Function, and General Function) by comparing distributions (signed-rank), effect sizes, and the proportion of subjects exceeding 95% limits of agreement (LOA) derived from a previous test-retest study (McNemar’s).
Significant improvement was observed for all 4 AS-20 domains using lookup methods and stacked analysis (P<0.0001 for all comparisons). Effect sizes were comparable whether obtained using lookup tables or stacked analysis for all 4 AS-20 domains (0.60 vs 0.60 respectively for Self-Perception, 0.46 vs 0.51 for Interactions, 0.66 vs 0.72 for Reading Function, and 0.93 vs 0.88 for General Function). A higher proportion of subjects exceeded 95% LOAs using the lookup table vs stacked analysis for Interaction (30% vs 22%, P=0.02), Reading Function (39% vs 25%, P<0.0001), and General Function domains (39% vs 33%, P=0.04).
Similar effect sizes with each method supports using lookup tables to represent patient domain scores. An advantage of using lookup tables is the ability to quantify true change of a single patient using LOAs, which we found problematic when using a stacked analysis due to the open-ended nature of the logit scale. Using the AS-20 Rasch lookup table is a reasonable approach to analyzing response of the AS-20 to an intervention for adult strabismus and may avoid the need to perform a more rigorous de novo stacked Rasch analysis for every study, under the assumption that the population doesn’t markedly differ from that used to create the lookup table.
This PDF is available to Subscribers Only