Uncertainty around the Incremental Cost-Utility Ratio Accounting for Mapping Interpolation

Presenter: Carole Siani, Université Claude Bernard Lyon I

Abstract

An important debate has recently occurred in health economics about how to measure the cost-effectiveness of a new medical treatment relative to a standard.
In cost-effectiveness analysis (CEA), comparing one or more medical treatment(s) with a standard treatment on the two-fold basis of cost and medical effectiveness,
the results are generally expressed in terms of an incremental cost-effectiveness ratio (ICER) by decision-makers, or more recently in terms of, an ICER adjusted by the quality of life (expressed as a cost per QALY gained).

De Peretti and Siani have shown that the truncated Filler's method is the most appropriate method for building a confidence region for the standard ICER (not adjusted by the quality of life) and usable for decision-making purpose, overcoming the mirror decision problem.
The truncated Filler's method is based on the Gaussian asymptotic approximation of the pair composed by the mean costs difference and the mean effects difference.
In this case, the effectiveness of a treatment is measured simply by the number of years of life gained.

However, in the case of QALY, the effectiveness is the sum of durations in a life state times the utility of this life state.
Consequently, it is more difficult to handle the uncertainty around the ICER.
In practice, the utility is assessed by the questionnaire EuroQol EQ-5D (EuroQol Group 1990; Kind, 1996; Rosser and Sintonen, 1993).
However, the EuroQol is rarely available for the entire clinical trial.
In practice, the EuroQol is interpolated from a purely functional status (disability) questionnaire: the Health Assessment Questionnaire (HAQ).
This is an additional source of uncertainty on the estimate of the ICER.
This additional uncertainty is not accounted for in the studies of the literature and in the cost effectiveness analysis made in the pharmaceutical industry, and decisions are made with respect to these results.

The purpose of this paper is to construct a confidence region around the ICER adjusted by the quality of life, accounting for the uncertainty coming for the EuroQol interpolation.
We wish to enlighten that the EuroQol interpolation increases dramatically the uncertainty around the ICER adjusted by the quality of life so that the conclusion are not reliable.

The first step of the paper is to construct a confidence region around the ICER adjusted by the quality of life (ICUR).

The second step of the paper, is to integrate EuroQol interpolation procedure in the procedures to construct the confidence region of the ICUR: in the theoretical procedure as well as in the bootstrap procedure.
This permits to recompute the confidence region of the ICUR and then to reassess the uncertainty.

Monte Carlo experiments are carried out to assess the performance of the methods.
Skewed and leptokurtic data are generated to check the robustness of the methods.