Multifactorial predictors and outcome variables of vision restoration training in patients with post-geniculate visual field loss

Abstract

Purpose: Systematic stimulation of the visual field border in patients with visual field loss after cerebral lesions improves visual function even years after the onset of partial blindness. However, computer-based training programs like Vision Restoration Training (VRT) are not equally effective in all patients. We therefore tested which factors determine training outcome and which visual and cognitive functions are changed by VRT.

Methods: Multiple outcome measures were predicted using a multifactorial regression approach. Nineteen patients with post-geniculate visual system lesions performed six months of VRT and underwent extensive testing before and after treatment, including visual field measurements, attention functions, and subjective parameters.

Results: Visual field size increased significantly during training, but a number of cognitive, especially attentional, variables also improved, as did subjective visual function. The size of areas of residual vision was the strongest predictor variable for visual field increase. Demographic and lesion-related variables had little influence on training success.

Conclusions: With multivariate regression models, training outcome on different variables can be accurately predicted. Moreover, visual field increase is sufficiently predictable based on a set of variables readily available to the clinician: age of the patient, time since lesion, number of absolute perimetric defects, eccentricity of the visual field border, size of areas of residual vision, and average reaction time to perimetric stimuli.