Presentation (CTI)

OBJECTIVES: There is disagreement about whether EQ-5D is appropriate for measuring caregiver’s health-related quality of life (HRQoL) in Alzheimer’s Disease, because studies report small changes in carers’ EQ-5D as patient severity increases. We aimed to understand if EQ-5D is appropriate in this context by systematically examining if it is statistically significantly associated with factors related to patient health and caregiving burden.
METHODS: We consolidated evidence from two National Institute for Health and Care Excellence appraisals and conducted a systematic literature review using 4 databases to identify studies that analysed the relationship between caregivers’ EQ-5D scores and disease-related or caregiving variables. We extracted data from studies that used multivariate regression or longitudinal analyses to control for potential confounding.
RESULTS: We included 46 studies: 5 were longitudinal, 15 reported multivariate analyses, 1 used both, and 26 were only univariate analyses. We grouped studies which analysed the effect of similar variables on carers’ HRQoL. 6 studies analysed the effect of caregiving burden (4 using Zarit Burden Interview (ZBI)): all found significant results. 3 studies analysed proxy-reported patient HRQoL: 2 found significant results. 5 studies analysed the effect of patient disease severity: 3 found significant results. 5 studies analysed the effect of patient symptoms/needs: 3 found significant results. Carers’ EQ-5D was not statistically significantly associated with care setting (5 studies), caring duration (2 studies), or patient-reported HRQoL (2 studies). A one unit increase on the ZBI 12-item 0-48 scale corresponded to an EQ-5D disutility of 0.002-0.006.
CONCLUSIONS: EQ-5D demonstrates validity in caregivers via its responsiveness to changes in caregiving burden. Predicting caregivers’ HRQoL from patient severity alone oversimplifies the caregiving experience and is inaccurate and insufficient for economic evaluation. Standard psychometric tests using univariate analysis are poorly suited to analysing caregivers’ HRQoL data, as they do not control for the multiple inter-related factors that affect caregivers.