BMC Medical Research Methodology (May 2020)

Similar responses to EQ-5D-3L by two elicitation methods: visual analogue scale and time trade-off

  • Xiuying Wang,
  • Lin Zhuo,
  • Yifei Ma,
  • Ting Cai,
  • Aviva Must,
  • Ling Xu,
  • Lang Zhuo

DOI
https://doi.org/10.1186/s12874-020-01008-9
Journal volume & issue
Vol. 20, no. 1
pp. 1 – 10

Abstract

Read online

Abstract Background Health-related quality of life (HRQoL) is often measured using EQ-5D-3L by the elicitation methods of visual analogue scale (VAS) and time trade-off (TTO). Although many countries have constructed both national VAS and TTO value sets, the fact that VAS and TTO value sets produces different values bewilders researchers and policymakers. The aim of this study is to explore certain conditions which could yield similar value sets using VAS and TTO. Methods A homogeneous sample of medical school students was selected to value 18 hypothetical health states using VAS and TTO methods. The 18 hypothetical health states were produced by orthogonal design (L18, 2*3^7). The range of rescaled values was transformed into − 1 ~ 0 ~ 1. The investigations via different methods were carried out by computer-assisted personal interviewing with a wash-time interval of 72 h. Value sets for VAS and TTO were constructed using general least square regression models. Independent variables were composed of 10 dummy variables from 5 dimensions and including or omitting both constant and N3 terms. Results Three hundred thirteen medical students participated. The mean age was 21.03 ± 0.44 years and 56.2% were female. The four regression models (for each method with and without constant and N3 terms) were all statistically significant (P 0.94 and MAE < 0.033). Differences between the coefficients of the 10 dummy variables corresponding to each model were all less than 0.059. Pearson correlation coefficients between observed means and predicted values exceeded 0.981. Fitted curves of VAS and TTO largely coincided. Conclusions VAS and TTO can generate similar responses under certain conditions, suggesting that the two valuation methods could be equivalent intrinsically. The VAS method appears a more valid approach for valuation in the general population due to its greater simplicity and feasibility.

Keywords