Preference Based Measurement of Vision-Related Quality of Life: Are We Mixing Apples and Oranges?

Purpose: Frequently, a standard of cost-effectiveness of $50,000-100,000/QALY is used to evaluate the cost-effectiveness of the intervention. This is based upon utility measured from patients or the public using the standard gamble or time trade-off on a scale with death as the lower anchor and perfect health as the upper anchor (i.e., perfect health-current health-death: the "policy scale" or PS). In evaluation of vision-related interventions, the utility measures have traditionally been based upon a scale anchored by death and perfect vision (i.e, i.e., perfect vision-current vision-death; the "vision truncated scale" or VTS). This approach assumes the two scales are additive, with the vision truncated scale as a subset of the policy scale. However, this assumption has not been tested empirically. Methods: We interviewed participants with diabetic retinopathy (DR) =59, glaucoma=99, and macular degeneration (AMD)=44. Utilities were estimated using the standard gamble for two scenarios: 1) assuming current health status including current visual function, what risk of death would be accepted to gain perfect health (PH scenario); 2) assuming current health status and assuming perfect visual function what risk of death would be accepted to gain perfect health (CP scenario). We also administered the SF-36 and NEI-VFQ (an instrument to measure vision-related quality of life). We tested two hypothesis to determine if there was evidence of additivity of the two scales: 1) that the two trades (PH and CP) would differ by an amount that roughly approximated the vision-related utility related to the person's condition; and 2) that a conceptual model based on the International Classification of Functioning, Disability, and Health (ICF) will define health-related quality of life (HRQoL) by three constructs measured by the SF-36 and NEI-VFQ. Results: Overall, there was no difference seen between the PH and CP scenarios (see Table). Among specific eye conditions there were extremely modest differences between the scales for those with DR and RE. There was no difference for glaucoma and cataract and the difference seen in AMD may indicate cognitive difficulties in the trade. Similar relationships were seen when disease was stratified by severity (not shown) In comparing the SF-36 and VFQ (not shown) most correlations between domains were weak. In evaluating the measurement models, preliminary results suggest that compared with the conceptual model, the latent constructs are well explained by items from the SF-36, but not the NEI-VFQ. This indicates that the hypothesized latent constructs are only measured by non-visual functions; visual function measures a separate construct. Exploratory factor analysis shows that the measures of visual function may form two constructs, vision and vision-related pain. An additional measurement model assessing only pain items indicate that vision related pain is not correlated to bodily pain. Conclusions: This investigation provides no evidence that the VTS and PS have an additive relationship. Indeed, it appears that the two trades might be measuring different constructs. Thus where cost-utility analyses is used in evaluation of vision-related interventions, it may be inappropriate to use standards of cost-effectiveness developed on the policy scale. Condition N PH Scenario(SD) CP Scenario(SD) Diabetic Retinopathy 58 0.87 (0.20) 0.89 (0.18) Glaucoma 99 0.89 (0.23) 0.89 (0.24) Macular Degeneration 44 0.83 (0.27) 0.81 (0.30)