Abstract
INTRODUCTION: The aim of this study was to analyze discrepancies between self- and proxy-rated health-related quality of life (HRQoL), measured with the EuroQol 5 Dimension 5 Level survey (EQ-5D-5L), in people living with dementia (PlwD) and their caregivers on an individual response level. METHODS: EQ-5D-5L, sociodemographic and clinical data were obtained from baseline data of n = 174 dyads of a cluster-randomized, controlled intervention trial. Self- and proxy-rated EQ-5D-5L health profiles were evaluated in terms of response distribution and agreement (weighted Kappa), and discrepancies in individual dimension level were analyzed using the Paretian Classification of Health Change (PCHC) as well as the presence and degree of inconsistencies between ratings. RESULTS: PlwD had a mean age of 80.1, nearly the half were female and 82.3% were mildly to moderately cognitively impaired. PlwD reported a higher utility index than caregiver proxies (mean 0.75 vs. 0.68, 83% of PlwD > 0.5). According to the PCHC and inconsistency approach, 95% of PlwD rated their health differently compared to proxies; 66% with divergent responses in at least three EQ-5D-5L dimensions. Nine dyads (5%) showed identical ratings. Discrepancies of one higher or lower EQ-5D-5L response represented the most frequent discrepancy (35.4%). Caregivers were two times more likely to report "moderate problems," representing the middle of the 5-point Likert scale. Usual activities had the lowest agreement between ratings (weighted kappa = 0.23). In PlwD reporting no or some problems in EQ-5D-5L-dimensions, proxies were more likely to report more problems and vice versa, especially in the more observable dimension usual activities and less likely in the less observable domains pain/discomfort and anxiety/depression. DISCUSSION: The central tendency bias observed in proxy-ratings could be associated with assessment uncertainties, resulting in an underestimation (overestimation) in PlwD reporting better (worse) health. This diverging trend extends the knowledge from previous studies and underlines the need for more methodological research in this area. HIGHLIGHTS: People living with dementia (PlwD) rate their health differently than proxies.Proxy-ratings over- or underestimate PlwD health when self-ratings are low or high.Proxies indicate a possible central tendency bias.Further research is needed to understand influencing factors.