Quality of life before intensive care using EQ-5D: patient versus proxy responses

Abstract

Objective: To compare patients' retrospectively reported baseline quality of life before intensive care hospitalization with population norms and proxy reports.

Design: Prospective cohort study.

Setting: Thirteen ICUs at four teaching hospitals in Baltimore, MD.

Patients: One hundred forty acute lung injury survivors and their designated proxies.

Interventions: Around the time of hospital discharge, both patients and proxies were asked to retrospectively estimate patients' baseline quality of life before hospital admission using the EQ-5D quality-of-life instrument.

Measurements and main results: Mean patient-rated EQ-5D visual analog scale scores and utility scores were significantly lower than population norms but were significantly higher than proxy ratings. However, the magnitude of difference in average utility scores between patients and either population norms or proxies was not clinically important. For the five individual EQ-5D domains, κ statistics revealed slight to fair agreement between patients and proxies. Bland-Altman plots demonstrated that for both the visual analog scale and utility scores, proxies underestimated scores when patients reported high ratings and overestimated scores for low patient ratings.

Conclusions: Patients retrospectively reported worse baseline health status before acute lung injury than population norms and better status than proxy reports; however, the magnitude of these differences in health status may not be clinically important. Proxies had only slight to fair agreement with patients in all five EQ-5D domains, attenuating patients' more extreme ratings toward moderate scores. Caution is required when interpreting proxy retrospective reports of baseline health status for survivors of acute lung injury.

Figure 1. Bland-Altman Plots for EQ-5D Utility and VAS scores

1a: EQ-5D utility score 1b: EQ-5D visual analogue scale score These figures display the relationship between patient-rated EQ-5D score and the difference between patient- and proxy-rated scores. The dashed line is the fitted linear regression of the relationship, while the upper and lower lines represent the 95% confidence interval.

Figure 1. Bland-Altman Plots for EQ-5D Utility and VAS scores

1a: EQ-5D utility score 1b: EQ-5D visual analogue scale score These figures display the relationship between patient-rated EQ-5D score and the difference between patient- and proxy-rated scores. The dashed line is the fitted linear regression of the relationship, while the upper and lower lines represent the 95% confidence interval.