OBJECTIVES: This study aims to develop bolt-on value sets using the parameters obtained in the US EQ-5D-5L valuation study and anchor the bolt-on estimates elicited through the application of the SPP and DCE methods. The study uses a vision bolt-on and develops independent value sets, derived from the data obtained from the application of the SPP and DCE methods, and compares them in terms of precision and accuracy.

METHODS: Data was collected through self-completed online surveys(1008 respondents). Respondents completed blocks of 7 DCE and 7 SPP tasks, with the order of presentation of the choice tasks (i.e., SPP first or DCE first) randomly assigned on a 1:1 basis. Bolt-on modeling followed two distinct approaches: an experimental modeling design was used involving masking or fixation of known coefficients and a conventional approach, in which all parameters were empirically estimated using observed data modeling.

RESULTS: An initial examination of SPP response patterns suggested the presence of ordering effects, which led to the development of new approaches that parsed the SPP data in ways that allowed the estimation of models disregarding (SPPNoPA) and accounting for (SPPOPA) improvement order, in order to account for these effects. When ordering effects are taken into consideration, SPP models yield accurate and precise estimates for the vision bolt-on item and all EQ-5D core dimensions.

CONCLUSIONS: This study presents the first large-scale application of the SPP method to value bolt-on dimensions and demonstrates a set of methods well suited to future low-cost bolt-on valuation. According to our results, values for bolt-on health states can be accurately estimated based on standard EQ-5D value sets using data obtained through the application of ordinal preference elicitation methods like the DCE and the SPP. Furthermore, while the findings of this study indicate that further improvements can be made to the SPP method, SPPNoPA models yielded promising results.