OBJECTIVES

To compare assessments of health-related quality of life outcomes of care home residents reported by residents and care staff acting as proxies.

DESIGN

Linear regression and bivariate modelling of paired assessments from care home residents and care staff.

SETTING

78 care homes in 2 regions in England.

PARTICIPANTS

556 care home residents aged 65 years or older and care staff.

MAIN OUTCOME MEASURES

EQ-5D utility scores and responses to individual EQ-5D dimensions.

RESULTS

The depression status, cognitive function, physical function, activities of daily living, social engagement, pain and dementia diagnosis of care home residents all predicted discrepancies in EQ-5D reporting. For residents with no depressive symptoms, care staff underestimated residents' mean EQ-5D utility score by 0.134 (95% CI 0.097 to 0.171) and for those with severe depressive symptoms they overstated mean utility scores by 0.222 (95% CI 0.104 to 0.339). With increasing levels of pain in residents the care staff progressively estimated EQ-5D utilities above self-reported values; by 0.236 (95% CI 0.003 to 0.469) in those with the second highest pain scores. For those with no cognitive impairment, proxies overstated mean utility scores by 0.097 (95% CI 0.049 to 0.146), while for those with severe cognitive impairment they underestimated mean utility scores by 0.192 (95% CI 0.143 to 0.241).

CONCLUSIONS

Care home residents and staff appear to differ fundamentally in their assessment of the health-related quality of life, as measured by the EQ-5D, of residents with different levels of depression, pain and/or cognitive impairment. This could lead to interventions evaluated using proxy-based quality-adjusted life year estimates being wrongly rejected on cost-effectiveness grounds and may also make it difficult for carers to act as advocates with health and social care professionals for certain groups of residents. A more resident-focussed approach to assessment of health-related quality of life is needed.