Assessing the agreement between self- and proxy-reported responses for measuring health-related quality of life in people with dementia using the Alzheimer’s Disease Five Dimensions instrument

BACKGROUND: The level of agreement between utility values derived from self- and proxy-reports, assessed using the preference-based health-related quality of life (HRQoL) instrument, the Alzheimer's Disease Five Dimensions (AD-5D), remains unclear. We aimed to investigate the agreement between self- and proxy-reported HRQoL, measured using the AD-5D, a dementia-specific, preference-based HRQoL instrument. METHODS: The data comprise 77 Australian dyads of people with dementia and their caregivers. The agreement between AD-5D utility values derived from self- and proxy-reports was analysed using a Bland-Altman plot. The ordinary least squares regression technique was employed to identify factors associated with the AD-5D utility value sets and to assess agreement between the resulting AD-5D utility values derived from self- and proxy-reports. RESULTS: The mean AD-5D utility value derived from self-reports (0.667) was higher than the value derived from proxy-reports (0.523). The Bland-Altman plot shows that 7.79% of the differences in AD-5D utility values fell outside the limits of agreement. The regression results indicated that the AD-5D utility value derived from self-reports for people with dementia aged 80 years or older was, on average, 0.20 points lower (β = 0.20, SE = 0.10) than that for people with dementia aged less than 70 years. CONCLUSION: Utility values derived from self-reports were higher than those derived from caregiver proxy-reports. While proxy reporting is a necessary alternative when self-reporting is not feasible, these perspectives are not interchangeable. Future economic evaluations should incorporate sensitivity analyses to account for this systematic 'proxy-gap'.