Baseline comparison of three health utility measures and the feeling thermometer among participants in the action to control cardiovascular risk in diabetes trial

BACKGROUND Health utility (HU) measures are used as overall measures of quality of life and to determine quality adjusted life years (QALYs) in economic analyses. We compared baseline values of three HUs including Short Form 6 Dimensions (SF-6D), and Health Utilities Index, Mark II and Mark III (HUI2 and HUI3) and the feeling thermometer (FT) among type 2 diabetes participants in the Action to Control Cardiovascular Risk in Diabetes (ACCORD) trial. We assessed relationships between HU and FT values and patient demographics and clinical variables. METHODS ACCORD was a randomized clinical trial to test if intensive controls of glucose, blood pressure and lipids can reduce the risk of major cardiovascular disease (CVD) events in type 2 diabetes patients with high risk of CVD. The health-related quality of life (HRQOL) sub-study includes 2,053 randomly selected participants. Interclass correlations (ICCs) and agreement between measures by quartile were used to evaluate relationships between HU's and the FT. Multivariable regression models specified relationships between patient variables and each HU and the FT. RESULTS The ICCs were 0.245 for FT/SF-6D, 0.313 for HUI3/SF-6D, 0.437 for HUI2/SF-6D, 0.338 for FT/HUI2, 0.337 for FT/HUI3 and 0.751 for HUI2/HUI3 (P  <  0.001 for all). Common classification by quartile was found for the majority (62%) of values between HUI2 and HUI3, which was significantly (P  <  0.001) higher than between other HUs and the FT: SF-6D/HUI3  =  40.8%, SF-6D/HUI2  =  40.9%, FT/HUI3  =  35.0%, FT/HUI2  =  34.9%, and FT/SF-6D  =  31.9%. Common classification was higher between SF-6D/HUI2 and SF-6D/HUI3 (P  <  0.001) than between FT/SF-6D, FT/HUI2, and FT/HUI3. The mean difference in HU values per patient ranged from -0.024  ±  0.225 for SF-6D/ HUI3 to -0.124 ± 0.133 for SF-6D/HUI2. Regression models were significant; clinical and demographic variables explained 6.1% (SF-6D) to 7.7% (HUI3) of the variance in HUs. CONCLUSIONS The agreements between the different HUs were poor except for the two HUI measures; therefore HU values derived different measures may not be comparable. The FT had low agreement with HUs. The relationships between HUs and demographic and clinical measures demonstrate how severity of diabetes and other clinical and demographic factors are associated with HUs and FT measures. TRIAL REGISTRATION ClinicalTrials.gov Identifier: NCT00000620.