Centro de Estudos e Investigação em Saúde da Universidade de Coimbra

Funder

Organizational Unit

Publications

Valuing health states : is the MACBETH approach useful for valuing EQ-5D-3L health states?

Publication . Oliveira, Mónica Duarte; Agostinho, Andreia; Ferreira, Lara; Nicola, Paulo; Bana e Costa, Carlos

Background: Quality Adjusted Life Years (QALYs) are a key outcome measure widely used within health technology assessment and health service research studies. QALYs combine quantity and quality of life, with quality of life calculations relying on the value of distinct health states. Such health states’ values capture the preferences of a population and have been typically built through numerical elicitation methods. Evidence points to these value scores being influenced by methods in use and individuals reporting cognitive difficulties in eliciting their preferences. Evidence from other areas has further suggested that individuals may prefer using distinct elicitation techniques and that this preference can be influenced by their numeracy. In this study we explore the use of the MACBETH (Measuring Attractiveness by a Categorical Based Evaluation Technique) non-numerical preference elicitation approach for health states’ evaluation. Methods: A new protocol for preference elicitation based on MACBETH (only requiring qualitative judgments) was developed and tested within a web survey format. A sample of the Portuguese general population (n=243) valued 25 EQ-5D-3L health states with the MACBETH protocol and with a variant of the time trade-off (TTO) protocol, for comparison purposes and for understanding respondents’ preference for distinct protocols and differences in inconsistent evaluations. Respondents answered to a short numeracy test, and basic socio-economic information collected. Results: Results show that the mean values derived from MACBETH and the TTO variant are strongly correlated; however, there are substantial differences for several health states’ values. Large and similar numbers of logical inconsistencies were found in respondents’ answers with both methods. Participants with higher levels of numeracy according to the test preferred expressing value judgments with MACBETH, while participants with lower levels were mostly indifferent to both methods. Higher correlations between MACBETH and TTO variant evaluations were observed for individuals with higher numeracy. Conclusion: Results suggest that it is worth researching the use of non-numerical preference elicitation methods. Numeracy tests more appropriate for preference elicitation when no explicit considerations of uncertainty are made need to be explored and used. Further behavioural research is needed to fully understand the potential for using these methods in distinct settings (e.g. in different evaluation contexts and in face-to-face and non-face-to-face environments), as well as to explore the effect of literacy on assessments and on respondents’ preferences.

2018Journal article

Open access