Toward a Smaller Design for EQ-5D-5L Valuation Studies.

Area of Science:

• Health Economics
• Psychometrics
• Decision Science

Background:

• The EuroQol Group uses the EuroQol Valuation Technology (EQ-VT) protocol with composite time trade-off (cTTO) to value 86 health states for EQ-5D-5L.
• This standard approach models observed values to generate utility values for all 3125 possible health states.

Purpose of the Study:

• To evaluate if a smaller, 25-state orthogonal design could accurately predict health state utilities using time trade-off (TTO) data.
• To compare the predictive accuracy of TTO data using standard EQ-VT, orthogonal, and D-efficient designs.

Main Methods:

• Collected TTO values from 525 university students using EQ-PVT software for EQ-VT, orthogonal, and D-efficient designs.
• Each design involved valuing a random block of health states, with 100 observations per state.
• Modeled TTO data by design and compared root mean square error (RMSE) between observed and predicted values.

Main Results:

• The standard EQ-VT design yielded the lowest RMSE (0.052).
• Orthogonal and D-efficient designs showed slightly higher RMSEs (0.066 and 0.063, respectively).
• Prediction errors varied for more severe health states, and some model coefficients differed across designs.

Conclusions:

• Smaller valuation designs, like the 25-state orthogonal or D-efficient designs, do not significantly increase prediction errors for TTO data.
• These smaller designs present a viable, resource-efficient option for health state valuation studies, particularly in resource-constrained settings.
• Smaller designs can be integrated with other preference data, such as discrete choice experiments, for joint modeling.