Abstract
Purpose: :
To evaluate the responsiveness of the EuroQoL (EQ-5D) to vision-related functional ability loss in patients seen in low vision clinics.
Methods: :
Health state questionnaires including the EQ-5D, GDS, TICS, SF-36 and Activity Inventory (AI) were administered by telephone to 113 low vision patients across 15 clinical centers within the US. Patients ranged in age from 18-93 years with an average age of 74. Separate Rasch analyses were performed on patient responses to the GDS, SF-36 physical (SF-PHY), SF-36 psychological (SF¬-PSYCH), and AI. EQ-5D responses were transformed to full health utilities to generate the EQ-5D US index. Raw scores were used for the TICS and the SF-36 general health (SF-GH). The correlation matrix was analyzed with exploratory and then confirmatory factor analysis using a varimax rotation.
Results: :
From confirmatory factor analysis, four independent factors were necessary and sufficient to explain 65% of variance. We provisionally identified these four factors as psychological(F1), physical(F2), cognitive(F3), and vision(F4). F1 explained 74% of the variance in the GDS measure, 62% in the SF-PSYCH, 16% in the SF-GH, and less than 2% in the TICS and AI. 12% of the variance in the EQ-5D is explained by F1. F2 explains 92% of the variance in the SF-PHY, 28% in the SF-GH, 7% in the SF-PSYCH, and 6% in the GDS. 2% or less of the variance in the TICS and AI load onto F2. 33% of the variance in the EQ-5D is explained by F2. F3 explains 34% of the variance in the TICS score, 8% in the GDS and 5% in the SF-PHY. Only 1% of the variance in the SF-GH and AI load onto F3. None of the variance in the EQ-5D or SF¬-PSYCH is explained by F3. F4 explains 97% of the variance in the AI. Only 2% of the variance in the EQ-5D and GDS could be explained by F4. Almost no variance in the SF¬-PSYCH, TICS and SF-GH loaded on to F4. For all factors, a similar pattern of loading is seen with the EQ-5D and SF-GH. Overall only 47% of the variance in the EQ-5D can be explained by the four factors.
Conclusions: :
The variance in the EQ-5D does not load heavily on any one of the four factors identified for low vision patients, while most of the variance in the EQ-5D can be explained by the physical and psychological domains. With less than 2% of the variance explained by the vision factor, the EQ-5D is not an adequate measure of the functional consequences of visual impairment. TICS appears to be a well-aimed measure of the cognition factor but much of its variance remains unexplained. Of the four examined, only the vision factor contributes to AI measures.
Keywords: quality of life • low vision • clinical (human) or epidemiologic studies: outcomes/complications