Lowering the ‘Floor’ of the Sf‐6d Scoring Algorithm Using a Lottery Equivalent Method

Abstract

This paper presents a new scoring algorithm for the SF-6D, one of the most popular preference-based health status measures. Previous SF-6D value sets have a minimum (a floor), which is substantially higher than the lowest value generated by the EQ-5D model. Our algorithm expands the range of SF-6D utility scores in such a way that the floor is significantly lowered. We obtain the wider range because of the use of a lottery equivalent method through which preferences from a representative sample of Spanish general population are elicited.

Inverse probability weighted estimation of social tariffs: An illustration using the SF-6D value sets

“…If we look at the studies which have derived the EQ-5D tariffs for the United Kingdom, Spain, the Netherlands and Japan, we conclude that this overlap requirement in the levels of severity of the dimensions of health is met in six out of the 10 estimated coefficients in the case of the UK (Dolan, 1997) and Spain (Badia et al, 2001) estimations, whereas the condition is satisfied in just one out of 10 estimated coefficients in the Dutch (Lamers et al, 2006) and Japanese (Tsuchiya et al, 2002) tariffs. Consequently, the validity of the Dutch and Japanese estimates rests on whether the corresponding regression models were correctly specified or not.…”

“…Consequently, the validity of the Dutch and Japanese estimates rests on whether the corresponding regression models were correctly specified or not. That is, the Dutch and Japanese tariffs are less likely to be robust to the misspecification of the regression model than those in Dolan (1997) and Badia et al (2001). The expression of the large-sample variance of ˇk w in the case where and ı are known is A −1 33 B 33 (A −1 33 ) T .…”

“…Abellán-Perpiñán et al (2011), after discussing other possible reasons, concluded that the wider range of their SF-6D value set was mainly due to the usage of this (probability) lottery equivalent method. Both theoretical and empirical arguments supporting this finding are provided in their paper.…”

“…New SF-6D algorithms have been estimated by using the standard regression approach for other countries apart from the UK (Lam et al, 2008;Brazier et al, 2009;Ferrerira et al, 2010;Abellán-Perpiñán et al, 2011). The Spanish SF-6D value set derived in Abellán-Perpiñán et al (2011) uses the same database as in this paper.…”

“…The Spanish SF-6D value set derived in Abellán-Perpiñán et al (2011) uses the same database as in this paper. The novelty of that value set is that has a minimum (a floor) which is significantly lower than those estimated in previous studies, making the range of SF-6D values more similar to the EQ-5D scores range.…”