Purchase this article with an account.
P. Demers-Turco, J. Stelmack, J.P. Szlyk, R.W. Massof, T.R. Stelmack, T. Williams; Measuring Visual Ability with the VA LV VFQ-48 . Invest. Ophthalmol. Vis. Sci. 2003;44(13):1283.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Purpose: The VA LV VFQ is a questionnaire designed to measure the difficulty visually impaired persons have performing daily activities. Rasch Analysis is used with this instrument to estimate the ability of persons to perform activities (person measure) and the visual ability required to perform each activity (item measure). These statistical analyses are important in outcomes research/instrument development, but burdensome for clinical practice. This report describes the development of a simple scoring system for the VA LV VFQ-48 that can be used by clinicians and researchers to estimate the person measure that would be calculated with Rasch analysis. Methods: Theoretically, functional reserve is the difference between the person measure and the item measure. Algebraically then, the person's ability is equal to functional reserve plus the item measure. Functional reserve for each item was estimated from the rating category as the average difference between the person and item measure for the rating used. We then calculated average functional reserve across items from the respondent's difficulty rating for each item. To correct for the rating scale ceiling and floor, we used the equation Ln([FR-2.34]/[-2.22-FR]), where FR is the average estimated functional reserve across items. We applied this approximation to the VA LV VFQ rating scale data obtained from 367 low vision patients. Results: The results indicate that the approximation of the person measure from a logarithmic transformation of average functional reserve is linear (slope =0.9, intercept =0.05) with the estimate of the person measure from Rasch analysis over the entire 4.5 logit range of data (r2=0.98). The approximation fails when average functional reserve falls outside the acceptable range (2.34 to –2.22). This situation can occur when there are missing data and the patient responds ‘Not Difficult' or responds ‘Impossible' to all of the remaining items, or if patients have a large number of ‘Impossible' responses to easy items and ‘Not Difficult' responses to the hard items. Conclusions: This study validates a simple method of approximating person measure estimations for the VA LV VFQ. This scoring approximation will facilitate using the VA LV VFQ-48 in daily practice to evaluate the visual ability of patients before rehabilitation and to assess the outcome of service delivery either for the patient or a low vision program.
This PDF is available to Subscribers Only