Multiplicative modelling of EQ-5D-3L TTO and VAS values

Area of Science:

• Health Economics
• Psychometrics
• Biostatistics

Background:

• The EQ-5D-3L is a widely used instrument for measuring health status.
• Time-trade-off (TTO) and visual analogue scale (VAS) are common methods for eliciting health state values.
• Evaluating different modeling approaches is crucial for accurate health economic analyses.

Purpose of the Study:

• To compare the predictive accuracy of multiplicative and additive models for EQ-5D-3L TTO and VAS values.
• To assess the impact of multiplicative modeling on the distribution and importance of health state values.

Main Methods:

• A multi-stage sampling design was used in Northern Malaysia (2010-2012).
• Face-to-face interviews collected data on 13 TTO and 15 VAS valuation tasks.
• Cross-validation was employed to evaluate out-of-sample predictive accuracy for additive and multiplicative models.

Main Results:

• Multiplicative models showed slight superiority over additive models for TTO values.
• Additive and multiplicative models had similar predictive accuracy for VAS values.
• The best performing TTO model was a non-linear multiplicative model (MULT7_TTO); the best VAS model was a linear additive model (ADD11_VAS).

Conclusions:

• Multiplicative modeling enhances the predictive accuracy of EQ-5D-3L TTO values.
• Multiplicative modeling did not improve the predictive accuracy of EQ-5D-3L VAS values.
• The choice of modeling approach impacts TTO value prediction but not the underlying health state value structure.