gms | German Medical Science

Purpose: To supplement self report instruments for individuals with profound low vision with an assessment tool consisting of activities requiring minimal, yet some, vision, and to calibrate these activities.

Methods: 17 activities were modeled after self-report items in the 150-item version of our ULV questionnaire, with the following criteria: 1) feasible for indoor administration under controlled lighting; 2) requiring a limited set of props and instruments, 3) amenable to duplication across centers. Activities included object localization, movement discrimination, or the presence or absence of lights, in a variety of settings. All activities were 2-, 3-, or 4-alternative forced-choice, and each was presented at 3 levels of difficulty. Each condition was performed twice, in counterbalanced order. Sixteen participants with visual acuity worse than 20/500 in the better eye were enrolled; Rasch analysis was carried out to examine psychometric properties of the sample.

Results: Only correct/incorrect scores data are presented here; timing data were collected for later analysis. Percent correct data were scaled for equal chance and 100% levels, rank-ordered into 9 performance levels, and analyzed using an Andrich model, with CUTLO set at -2 to reduce the effect of excessive guessing. Person and item reliability were .76; and both person ability and task difficulty were evenly distributed and spanned a range of 1.2 logits, i.e. a factor 3.3, with the exception of a single outlying person measure at +3.5 logits, well outside the ULV range.

Conclusion: Our analysis demonstrates that construction of a calibrated activity set in the ULV range is possible. Such a set has the potential to add objective performance assessments to clinical trials involving minimally sighted participants.