OBJECTIVE: To evaluate two different methods to obtain a dead (0)--full health (1) scale for EQ-5D-5L valuation studies when using discrete choice (DC) modeling. METHOD: The study was carried out among 400 respondents from Barcelona who were representative of the Spanish population in terms of age, sex, and level of education. The DC design included 50 pairs of health states in five blocks. Participants were forced to choose between two EQ-5D-5L states (A and B). Two extra questions concerned whether A and B were considered worse than dead. Each participant performed ten choice exercises. In addition, values were collected using lead-time trade-off (lead-time TTO), for which 100 states in ten blocks were selected. Each participant performed five lead-time TTO exercises. These consisted of DC models offering the health state 'dead' as one of the choices--for which all participants' responses were used (DCdead)--and a model that included only the responses of participants who chose at least one state as worse than dead (WTD) (DCWTD). The study also estimated DC models rescaled with lead-time TTO data and a lead-time TTO linear model. RESULTS: The DC(dead) and DCWTD models produced relatively similar results, although the coefficients in the DCdead model were slightly lower. The DC model rescaled with lead-time TTO data produced higher utility decrements. Lead-time TTO produced the highest utility decrements. CONCLUSIONS: The incorporation of the state 'dead' in the DC models produces results in concordance with DC models that do not include 'dead'.
OBJECTIVE: To evaluate two different methods to obtain a dead (0)--full health (1) scale for EQ-5D-5L valuation studies when using discrete choice (DC) modeling. METHOD: The study was carried out among 400 respondents from Barcelona who were representative of the Spanish population in terms of age, sex, and level of education. The DC design included 50 pairs of health states in five blocks. Participants were forced to choose between two EQ-5D-5L states (A and B). Two extra questions concerned whether A and B were considered worse than dead. Each participant performed ten choice exercises. In addition, values were collected using lead-time trade-off (lead-time TTO), for which 100 states in ten blocks were selected. Each participant performed five lead-time TTO exercises. These consisted of DC models offering the health state 'dead' as one of the choices--for which all participants' responses were used (DCdead)--and a model that included only the responses of participants who chose at least one state as worse than dead (WTD) (DCWTD). The study also estimated DC models rescaled with lead-time TTO data and a lead-time TTO linear model. RESULTS: The DC(dead) and DCWTD models produced relatively similar results, although the coefficients in the DCdead model were slightly lower. The DC model rescaled with lead-time TTO data produced higher utility decrements. Lead-time TTO produced the highest utility decrements. CONCLUSIONS: The incorporation of the state 'dead' in the DC models produces results in concordance with DC models that do not include 'dead'.
Authors: Christopher McCabe; John Brazier; Peter Gilks; Aki Tsuchiya; Jennifer Roberts; Anthony O'Hagan; Katherine Stevens Journal: J Health Econ Date: 2006-02-24 Impact factor: 3.883
Authors: Joanna Coast; Terry N Flynn; Lucy Natarajan; Kerry Sproston; Jane Lewis; Jordan J Louviere; Tim J Peters Journal: Soc Sci Med Date: 2008-06-21 Impact factor: 4.634
Authors: Federico Augustovski; Lucila Rey-Ares; Vilma Irazola; Osvaldo Ulises Garay; Oscar Gianneo; Graciela Fernández; Marcelo Morales; Luz Gibbons; Juan Manuel Ramos-Goñi Journal: Qual Life Res Date: 2015-08-05 Impact factor: 4.147
Authors: Fredrick Dermawan Purba; Joke A M Hunfeld; Aulia Iskandarsyah; Titi Sahidah Fitriana; Sawitri Supardi Sadarjoen; Juan Manuel Ramos-Goñi; Jan Passchier; Jan J V Busschbach Journal: Pharmacoeconomics Date: 2017-11 Impact factor: 4.981
Authors: Juan M Ramos-Goñi; Jose L Pinto-Prades; Mark Oppe; Juan M Cabasés; Pedro Serrano-Aguilar; Oliver Rivero-Arias Journal: Med Care Date: 2017-07 Impact factor: 2.983