Mihir Gandhi1, Kim Rand2, Nan Luo3. 1. Department of Biostatistics, Singapore Clinical Research Institute, Singapore; Centre for Quantitative Medicine, Duke-NUS Medical School, Singapore; Tampere Center for Child Health Research, University of Tampere and Tampere University Hospital, Tampere, Finland. Electronic address: mihir.gandhi@scri.edu.sg. 2. Health Services Research Centre, Akershus University Hospital, Lørenskog, Norway; Department of Health Management and Health Economics, University of Oslo, Oslo, Norway. 3. Saw Swee Hock School of Public Health, National University of Singapore, Singapore.
Abstract
OBJECTIVES: To evaluate the discriminative ability of negative values measured in 5-level EuroQol 5-dimensional questionnaire (EQ-5D-5L) value set studies. METHODS: This is a secondary analysis of EQ-5D-5L value set studies from Singapore, the Netherlands, China, Thailand, and Canada in which health state values were elicited from a general population sample using a composite time trade-off (TTO) method. Mean values were calculated for health states with same severity. The association between the mean values and severity was evaluated using Pearson correlation (r). A linear mixed model using severity as the fixed effect was fitted for values. The analyses were performed separately for positive values (from a conventional TTO for health states considered "better than death") and negative values (from a lead time TTO for health states considered "worse than death"). RESULTS: In Singapore (N = 1000; negative values 32.6%), the mean decreased with severity from 0.89 to 0.21 for positive values and increased with severity from -0.98 to -0.89 for negative values. The correlation between values and severity was much lower for negative values (r = -0.016) than for positive values (r = -0.614). The coefficient of severity in the linear mixed model for negative values was much smaller (coefficient = -0.009; pseudo-R2 < 0.001) compared with the model for positive values (coefficient = -0.041; pseudo-R2 = 0.337). Results using data sets from the other countries were similar. CONCLUSIONS: Negative values are not associated with severity of health states in EQ-5D-5L valuation studies, suggesting poor discriminative ability of the lead time TTO method in valuing health states considered worse than death.
OBJECTIVES: To evaluate the discriminative ability of negative values measured in 5-level EuroQol 5-dimensional questionnaire (EQ-5D-5L) value set studies. METHODS: This is a secondary analysis of EQ-5D-5L value set studies from Singapore, the Netherlands, China, Thailand, and Canada in which health state values were elicited from a general population sample using a composite time trade-off (TTO) method. Mean values were calculated for health states with same severity. The association between the mean values and severity was evaluated using Pearson correlation (r). A linear mixed model using severity as the fixed effect was fitted for values. The analyses were performed separately for positive values (from a conventional TTO for health states considered "better than death") and negative values (from a lead time TTO for health states considered "worse than death"). RESULTS: In Singapore (N = 1000; negative values 32.6%), the mean decreased with severity from 0.89 to 0.21 for positive values and increased with severity from -0.98 to -0.89 for negative values. The correlation between values and severity was much lower for negative values (r = -0.016) than for positive values (r = -0.614). The coefficient of severity in the linear mixed model for negative values was much smaller (coefficient = -0.009; pseudo-R2 < 0.001) compared with the model for positive values (coefficient = -0.041; pseudo-R2 = 0.337). Results using data sets from the other countries were similar. CONCLUSIONS: Negative values are not associated with severity of health states in EQ-5D-5L valuation studies, suggesting poor discriminative ability of the lead time TTO method in valuing health states considered worse than death.