Elsevier

Journal of Health Economics

Volume 30, Issue 5, September 2011, Pages 1057-1063
Journal of Health Economics

Revisiting United States valuation of EQ-5D states

https://doi.org/10.1016/j.jhealeco.2011.07.009Get rights and content

Abstract

In the original US valuation study of EQ-5D states, all worse-than-dead time trade-off responses (26% of the sample) were divided by 39 to increase the QALY estimates. This transformation has no theoretical justification and motivates this re-examination. Using the publically available dataset, we compared three alternative random utility models: instant (IRUM), angular (ARUM), and episodic (ERUM) models. Each leads to a distinct econometric estimator: mean ratio, ratio of means, and coefficient, respectively. IRUM suggests that 203 of the 243 EQ-5D states are worse-than-dead, which has little face validity compared to ARUM and ERUM (42 and 3 WTD states). ARUM and ERUM estimates are proportionally related such that losses in QALYs are approximately 37% larger under ARUM than ERUM. Compared to ERUM, economic evaluations using ARUM estimates emphasize quality of life, and this difference may influence policy decisions. Either ERUM or ARUM values sets are recommended over the original, transformed set.

Section snippets

Methods

The original US valuation study applied a multi-stage probability sampling design with over-sampling of Hispanics and Black adults. A complete description of this nationally representative interview survey can be found elsewhere (Coons et al., 2005, Shaw et al., 2005b). Five respondents with no TTO responses were excluded from this analysis (N = 4043).

Results

According to the 2000 US Census, Hispanic and Black, non-Hispanic adults compose 10.7% and 10.9% of all adults in the US; however, these sub-populations compose 30% and 27.7% of the sample respectively (Table 1) (Coons et al., 2005). In addition to over-sampling these two sub-populations, the sample differs from the US population by age, sex, and rural residence, which may be attributed to the design of the household interview survey. Average age of a male respondent is 42.76 and average age of

Discussion

In this paper, we respond to repeated criticisms of the original US valuation study and its ad hoc adjustment of WTD TTO responses by re-estimating EQ-5D health state values under three alternative random utility models. The IRUM, ARUM, and ERUM provide theoretical support for three distinct econometric estimators (mean of ratios, ratio of means, and coefficient), none of which involve the arbitrary transformation of WTD responses or the bounding of health state values above −1. IRUM values

References (26)

  • B.M. Craig et al.

    From a different angle: a novel approach to health valuation

    Soc. Sci. Med.

    (2009)
  • P. Dolan

    Modeling valuations for EuroQol health states

    Med. Care

    (1997)
  • M.F Drummond

    Methods for the Economic Evaluation of Health Care Programmes

    (2005)
  • Cited by (12)

    • Quality-Adjusted Life-Years without Constant Proportionality

      2018, Value in Health
      Citation Excerpt :

      Governmental agencies have the responsibility to allocate public resources toward health interventions that maximize population health, namely, health-related quality of life (HRQOL) and life span of the general population [1,2].

    • The Value Adults Place on Child Health and Functional Status

      2015, Value in Health
      Citation Excerpt :

      Health problems for a given duration may be summarized by their equivalence in lost QALYs using various methods. Although standard gamble and time trade-off (TTO) responses were commonly used in the past [1], these methods have serious limitations, including cognitive difficulty, scaling biases, and practical considerations [2–5]. A recent alternative that addresses these limitations and facilitates the measurement across a much wider domain of problems is to use preference questions on a health valuation survey, such as paired comparisons.

    • US valuation of health outcomes measured using the PROMIS-29

      2014, Value in Health
      Citation Excerpt :

      Alternatively, some studies, particularly those based on time trade-off (TTO) tasks, have applied the instant random utility model, which first divides HRQOL by duration before including an additive error term, U(h/t) + ε, simplifying episodes to health states (i.e., an instantaneous experience, h/t). Instant random utility model describes the relationship between health states and choices and becomes unstable when the duration becomes small (e.g., t → 0) [12,14]. Although the concepts are sometimes used interchangeably [15], this article differentiates between value and utility.

    • Italian population-based values of EQ-5D health states

      2013, Value in Health
      Citation Excerpt :

      First, similarly to the approach used in the analogous past valuation studies [14,15,17,18,20–28], a monotonic transformation was applied to the values for states worse than dead. Recent methodological researches, however, propose new models to manage the limitation of a monotonic transformation [46–48]. Despite these recent suggestions, we preferred to adopt the approach that is still accepted as official from the EuroQoL group, before new findings from ongoing research conducted using the new five-level (5L) version of the EQ-5D questionnaire, and likely adopting some changes in the methods of preferences elicitation, will be made available [49].

    View all citing articles on Scopus
    1

    Tel.: +31 10 7043807/4306; fax: +31 10 7044695.

    View full text