Trying to estimate a monetary value for the QALY

https://doi.org/10.1016/j.jhealeco.2009.02.003Get rights and content

Abstract

In this paper we study the feasibility of estimating a monetary value for a QALY (MVQ). Using two different surveys of the Spanish population (total n = 892), we consider whether willingness to pay (WTP) is (approximately) proportional to the health gains measured in QALYs. We also explore whether subjects’ responses are prone to any significant biases. We find that the estimated MVQ varies inversely with the magnitude of health gain. We also find two other (ir)regularities: the existence of ordering effects; and insensitivity of WTP to the duration of the period of payment. Taken together, these effects result in large variations in estimates of the MVQ. If we are ever to obtain consistent and stable estimates, we should try to understand better the sources of variability found in the course of this study.

Introduction

Much cost-effectiveness and cost-utility analysis has been undertaken to guide health care resource allocation. However, if we wish to go further and conduct cost-benefit analysis – and thereby make health care resource allocation more directly comparable with decision making in other areas of public policy – we need to find some way of attaching monetary values to health benefits. Since much of the health benefit measurement to date has been conducted in terms of quality adjusted life years (QALYs), one solution – were it to be feasible – would be to estimate the monetary value of a QALY (henceforth, the MVQ).

In the UK, the National Institute for Health and Clinical Excellence (NICE) recognised the desirability of having such an estimate and recently commissioned a study to explore its feasibility. Others, too, have recognized the potential value of such a figure: for example, Johannesson and Meltzer (1998) considered that obtaining this information “should be a research priority” (p. 4). They identified two possible strategies for deriving the willingness to pay (WTP) per QALY gained. One would be based on direct elicitation of WTP for some marginal health change(s), while the other would derive this figure from estimates of the value of statistical life (VSL) in the literature. Some work of this latter kind has been attempted,1 but the present study focuses instead on the possibility of estimating the MVQ on the basis of eliciting people's WTP for a range of health benefits.

We acknowledge the possibility that estimating a unique MVQ may not be feasible. Several papers, including Bleichrodt and Quiggin (1999) and Dolan and Edlin (2002), have shown that the conditions for a unique MVQ to exist are quite restrictive and are unlikely to hold. However, our objective was to explore just how stable – or variable – such an estimate might be. Our strategy was to use changes that were small relative to those used in other studies in this area2 in the hope that budget constraints would not cause significant non-linearities.3

Abellan-Perpiñan et al. (2006) and Bleichrodt and Pinto (2005) have suggested that a non-linear QALY model may be better than the linear model. If non-linearities are important, this may result in estimates of the MVQ varying considerably according to the basis upon which they are derived. So we wished to check for the possible impact of various non-linearities with respect to severity, duration and the size of risk reduction.

Our empirical work involved two surveys. The first, and larger, of these was designed to investigate several issues fundamental to the robust estimation of a reasonably stable MVQ. The second survey was designed to follow-up and clarify certain issues raised by the results of the first.

Section snippets

Broad structure of the study

Before describing the particular features of each study, we set out the general framework within which we are working and according to which we shall derive the MVQ estimates.

The basic idea of a QALY is to provide a measure that facilitates comparisons across a broad spectrum of health benefits. Ideally, it allows health care decision makers to weigh the total benefit of an intervention that alleviates short-term conditions involving moderate adverse effects against other interventions

First survey

Respondents were 560 members of the general population. They were contacted by telephone through random dialing and those who agreed to be interviewed were visited by an interviewer. They were then assigned at random to one of the seven groups described below (n = 80 each group).

Second survey: design, implementation and results

Given the presence of order effects in our first survey, we tried to estimate MVQ for some of these health gains using a between-Group design. This gave rise to a second survey. The second study took the four most straightforward scenarios from the first survey (Types 1, 3, 5 and 7) and added two more (Types 8 and 9) in order to study the derivation of MVQ estimates from even smaller health gains. These two scenarios were always presented as the second question, so they were vulnerable to being

Conclusions

The main objective of this study was to test the robustness of various conditions that would need to hold if we are to obtain a reliable and reasonably all-purpose MVQ from questions involving different combinations of QoL, duration and risk.

Our results cast quite serious doubt on the possibility of obtaining such value. Besides finding insufficient sensitivity to the duration of health states and the size of QoL improvements, we also found clear and substantial violations of procedural

Acknowledgements

We gratefully acknowledge financial support from Ministerio de Educación (project SEJ2007-67734/ECON), Junta de Andalucía (project P07-SEJ-02936), Fundación BIOEF and Agència d’Avaluació de Tecnologia i Recerca Mèdiques (project 081/27/02).

References (35)

  • H. Bleichrodt et al.

    The validity of QALYs under nonexpected utility

    The Economic Journal

    (2005)
  • K. Blumenschein et al.

    Eliciting willingness to pay without bias: evidence from a field experiment

    The Economic Journal

    (2008)
  • A.H. Briggs et al.

    Pulling cost-effectiveness analysis up by its bootstraps: a non-parametric approach to confidence interval estimation

    Health Economics

    (1997)
  • M.M. Byrne et al.

    Willingness to pay per quality-adjusted life year in a study of knee osteoarthritis

    Medical Decision Making

    (2005)
  • Donaldson, C., Baker, R., Jones-Lee, M., Loomes, G., Mason, H., Robinson, A., et al, 2008. Weighting and valuing...
  • B. Effron et al.

    An Introduction to Bootstrap

    (1993)
  • B. Efron

    Better bootstrap confidence intervals

    Journal of the American Statistical Association

    (1987)
  • Cited by (64)

    • The demand for a COVID-19 vaccine

      2022, Economics and Human Biology
    • A Systematic Review of Demand-Side Methods of Estimating the Societal Monetary Value of Health Gain

      2021, Value in Health
      Citation Excerpt :

      Evidently, the monetary values for a QALY tended to be higher when the size of QALY gains was smaller. This finding is consistent with previous studies.23,63,65 We found that some studies had very high WTP per QALY estimates, which might be explained by the very few QALYs gained.24,34,44,51,55

    • The Quality Adjusted Life Year: A Total-Utility Perspective

      2018, Cambridge Quarterly of Healthcare Ethics
    View all citing articles on Scopus
    View full text