Determining correspondence between scores on the EQ-5D "thermometer" and a5-point categorical rating scale

BACKGROUND. "Feeling thermometers" and category rating scales (CRS) can be used to measure self-rated health, though the interpretation of scores on t he feeling thermometer may be problematic. OBJECTIVES. To analyze correspondence between scores of self-rated health o n the EuroQol-5D "feeling thermometer" and a five-Point CRS. To determine t he influence of sociodemographic variables on correspondence. RESEARCH DESIGN. Correspondence between EQ-5D "thermometer" scores and CRS categories was determined in a within-subjects design using data from a lar ge-scale health survey. SUBJECTS. Twelve thousand, two hundred and forty-fi ve members of the Spanish general population interviewed in the Catalan Hea lth Interview Survey. MEASURES. Instruments used were the EQ-5D "feeling thermometer" and a five- Point CRS with categories from "excellent" to "poor." RESULTS. Median scores on the thermometer corresponding to CRS categories w ere as follows: poor = 40; fair = 53; good = 76; and very good = 80; excell ent = 90; differences were statistically significant (P < 0.05). Ranges in thermometer scores covered by CRS categories varied from 23.5 points for th e category "poor" to 13 points for the category "very good" (range, 25-75 p ercentiles), with considerable overlap between categories at the upper end of the scale. Median thermometer scores corresponding to the categories "go od," "fair," and "poor" were lower in older respondents and in those with a lower educational level. CONCLUSIONS. Determining correspondence between CRS and thermometer scores is useful in interpreting and categorizing thermometer scores. The age and level of education in particular affect the interpretation of CRS categorie s, and should be taken into account when analyzing results obtained with su ch instruments.