Generating population weights for the Well-being of older people measure (WOOP)

Background & Objective

There is a growing need for accurately measuring well-being of older people to inform public policy. To account for the large heterogeneity in quality of life, multi-attribute instruments are required. To be suitable for economic evaluations, the instrument's levels have to be weighted. These weights are commonly based on population preferences. This study aims to assess the suitability of two alternative approaches in estimating such population weights for a newly developed nine-dimensional well-being measure, the WOOP (well-being of older people) in an elderly population. Furthermore, we will present first results for the population weights.

Methods

In order to assess the cognitive burden of best-worst scaling (BWS) and discrete choice experiment (DCE) for obtaining weights for the WOOP, we created an online survey, which was tested in six think-aloud interviews and administered to a sample of 469 individuals aged 65 and above. Respondents were randomised to one of three randomised study arms with 13 choice tasks each. One arm included a DCE, where levels were colour-coded and overlapped. The second arm consisted of a case 2 BWS exercise. To test, whether colour coding would also be useful for BWS, we included a third arm, which presented respondents with a colour-coded BWS exercise. Cognitive burden and quality of the different tasks are assessed in terms of stability, monotonicity and continuity, as well as self-reported cognitive burden and answering time. The preferred method will then be used to create population weights using a sample of 1,000 Dutch citizens aged 65 and above.

Results

Think-aloud interviews and survey responses indicate that both BWS and DCE for valuing a nine-dimensional instrument are feasible for most respondents. Drop-out rate and completion time were lower in the DCE, while there was mixed evidence on the self-reported cognitive burden of the tasks itself. However, a larger share of respondents expressed that they could have answered more tasks in the DCE arm. In terms of stability, monotonicity and continuity, the answers to DCE tasks outperform the BWS answers. Preliminary results on the importance of the dimensions indicate that besides dimensions like mental and physical health, being able to cope with the current situation and feeling useful is also relevant to respondents.

Discussion

Results of the conducted analysis hint towards DCE being the more suitable methodological choice for estimating the population weights of a nine-dimensional quality of life instrument in the elderly. This not only informs the decision on how to proceed with the valuation of the WOOP but the more general discussion on the appropriateness of BWS and DCE exercises in health preference research.