Level of agreement between patient and proxy responses to the EQ-5D health questionnaire 12 months after injury.

BACKGROUND: Health-related quality of life represents a patient's experiences and expectations and should be collected from the patient. In trauma, collection of information from the patient can be challenging, particularly for subgroups where cognitive impairment is prevalent, increasing reliance on proxy reporting. This study assessed the agreement between patient and proxy reporting of health-related quality of life 12 months after injury.
METHODS: The Victorian State Trauma Registry and Victorian Orthopaedic Trauma Outcomes Registry collect EQ-5D data at 12 months after injury. Cases where data were collected from the patient and proxy were extracted. Agreement between patient and proxy responses was compared using kappa (K) coefficients for the individual EQ-5D items, and Bland-Altman plots and Wilcoxon signed-rank tests for the EQ-5D summary score and visual analog scale (VAS).
RESULTS: Agreement between patient and proxy respondents was substantial for the mobility (K = 0.61) and personal care items (K = 0.67) and moderate for the usual activities (K = 0.50), pain/discomfort (K = 0.42), and anxiety/depression items (K = 0.47). The mean difference between proxy and patient-reported scores for the VAS (0.74, 95% confidence interval: -2.73, 4.21) and the EQ-5D summary score (-0.02, 95% confidence interval: -0.07, 0.03) was small, but the limits of agreement were wide (-34.22 to 35.71 for VAS and -0.55 to 0.51 for summary score), suggesting no systematic bias.
CONCLUSIONS: Although proxy and patient responses for the EQ-5D VAS may differ, the differences show random variability rather than systematic bias. Group comparisons using proxy responses are unlikely to be biased, but proxy responses should be used with caution when assessing individual patient recovery.