AB022. Validity of the five-level new version of the EQ-5D in asthma patients

Background: A previous study evaluating the psychometric properties of the traditional EQ-5D in asthmatic patients showed a high ceiling effect (59% of patients with perfect health) questioning its usefulness in these patients. Therefore, the EuroQol Group developed a new EQ-5D version increasing the number of responses from 3 to 5 levels. This new version, the EQ-5D-5L, has never been tested in asthmatic patients. The aim of this study is to examine the distribution and construct validity of the new EQ-5D-5L in a European cohort of asthmatic adolescents and adults.

Methods: A subgroup of 316 patients between 12–40 years included in the ASTROLAB cohort who completed EQ-5D-5L in the online questionnaire were analysed. It is a brief, multi-attribute, generic, preference-based health status measure consisting of five dimensions of health-five response options. The index value ranges from 1 (best health possible) to −0.594 (negative values indicate health states worse than death), where 0 is the value assigned to death. Index values were calculated using the preference values from the Rasch Model developed from EQ-5D-3L French. To examine the distribution of the index measures of central tendency, dispersion, ceiling and floor effect, and observed range were calculated. Construct validity was examined by their ability to differentiate between known groups defined by the Asthma Control Questionnaire (ACQ-5) by ANOVA. ACQ-5 measures the presence of asthma symptoms during the previous week in 7 Likert scale response options, with a score ranging from 0 to 6 (lower score better asthma control). Three groups of asthma control were defined according to tertiles: good (ACQ-5 <0.4), intermediate (ACQ-5 0.4–1.2) and bad (ACQ-5 >1.2).

Results: Mean EQ-5D-5L index was 0.80 (SD=0.17). The observed range was from −0.03 to 1, floor and ceiling effects were 0% and 22.8%, respectively. Mean EQ-5D-5L index for patients with well-controlled asthma was 0.91 (95% CI, 0.89–0.93); 0.84 (95% CI, 0.81–0.87) for those with intermediate and 0.73 (95% CI, 0.69–0.78) for patients with poorly controlled asthma. The differences were statistically significant (P<0.001) and of great magnitude between the end groups. Most subjects reported ‘no problems’ in mobility (82%) and self-care (98.1%) dimensions, while around half of the subjects endorsed this category in pain/discomfort (46.5%) and anxiety/depression (50.9%) dimensions.

Conclusions: The distribution of the new EQ-5D-5L in patients with asthma is adequate, with an acceptable percentage of patients in perfect health. Relevant differences across known asthma control groups indicate good construct validity for assessing health related quality of life in these patients.

Keywords: Asthma; EQ-5D-5L; distribution; construct validity; Asthma Control Questionnaire (ACQ-5)