1 Modelling valuations for the EQ-5D health states: an alternative model using differences in valuations Jennifer Roberts and Paul Dolan Sheffield Health Economics Group
2 The EQ-5D Mobility Self Care Usual Activities Pain/Discomfort Anxiety/Depression – 3 levels defining problems on each dimension (none, some, a lot).
3 Motivation the original EQ-5D study preference based single index – 42 states (from 243) valued by TTO – i = 2997, j = 12, n = 35,964 – ‘tariff’ produced (Dolan 1997) – predicts mean health state values quite well despite – enormous variation across individual valuations of states
4 Descriptive Statistics for TTO Values StateMeanS.D
5 Differences in Health States State AState BNDIF(A-B)SD% Neg99%CI
6 Agreement with pair wise ranking of states
7 The data for the new tariff model all respondents valued plus 11 others – 2 very mild (n 1200) – 3 each from mild, moderate and severe (n ~ 750) consider each respondents value of each state as the difference from their value for – i = 2997, j = 11, n = 32,967 – random effects model used
8 Tariff Results C DIFMOB10.320DIFPAIN DIFMOB20.391DIFPAIN DIFSC10.179DIFMOOD DIFSC20.280DIFMOOD DIFUA10.084ANY DIFUA R 2 within0.554No > |0.05)8 MAE0.03No > |0.10|1
9 Comparing the old and new tariffs
10 Conclusion Lack of consensus on ordering of states has implications for policy decisions … Two models are similar in structure and findings New model has better predictive ability Some important differences arise – difference in health change given by the two tariffs could be as much as Which (if any) is the right model?