Aversion to health inequalities and priority setting in health care

Abstract

Traditionally aversion to health inequality is modelled through a concave utility function over health outcomes. Bleichrodt et al. [Bleichrodt, H., Diecidue E., Quiggin J., 2004. Equity weights in the allocation of health care: the rank-dependent QALY model. Journal of Health Economics 23, 157–171] have suggested a “dual” approach based on the introduction of explicit equity weights. The purpose of this paper is to analyze how priorities in health care are determined in the framework of these two models. It turns out that policy implications are highly sensitive to the choice of the model that will represent aversion to health inequality.

Introduction

The most common approach to aggregate health benefits in economic evaluations of health care is by unweighted summation, also referred to as quality-adjusted life-year (QALY)-utilitarianism. In this approach each individual in society gets the same weight and the aim is to maximize the benefits accruing from health care. Many authors have criticized QALY-utilitarianism for being blind to equity considerations (Nord, 1995, Williams, 1997, Williams and Cookson, 2000) and, indeed, empirical studies have typically observed that people care about equity in the distribution of health (for an overview see Dolan and Tsuchiya, 2006).

Two approaches have been put forward to incorporate equity considerations into economic evaluations of health care. The first approach defines a social utility function over health and uses health utilities rather than health itself as input in the medical decision process. The parameters in the social utility function then reflect the weight that different individuals should receive. The utility approach to equity was pioneered by Wagstaff (1991), who argued in favour of the iso-elastic utility function suggested by Atkinson (1970), and was also advocated by Dolan (1998), who suggested a Cobb–Douglas utility function, which is a special case of the iso-elastic utility function.

Bleichrodt et al. (2004) proposed a different approach to introduce equity weighting into economic evaluation. Instead of using a utility function over health they applied equity weights in the decision process. An advantage of modelling equity through explicit equity weights is that attitudes towards outcomes are separated from attitudes towards inequality. In the utility model attitudes towards outcomes and attitudes towards inequality are inseparable. At the level of fundamental principles, they reflect different notions, however. The former reflects an attitude towards health (extra health is more desirable when in poor health than when in close to perfect health), whereas the latter expresses an attitude towards inequality (inequality is undesirable). Because they reflect different concepts, it is desirable to separate attitudes towards outcomes from attitudes towards inequality. The model of Bleichrodt et al. (2004) allows this separation.

Bleichrodt et al. (2004) gave an axiomatic foundation for their model and argued that the conditions on which the model depends are plausible. An important advantage of their model is that the empirical estimation of the equity weights is very simple, which fosters the practical applicability of the model. Obviously, if we want to apply equity weights in practice, we need a way to measure these. The approach of Bleichrodt et al. (2004) can be considered dual to the utility-based approach of Wagstaff (1991) and Dolan (1998) in the following sense. In the utility-based approach, the health outcomes are transformed but the proportion of people in each category of health outcome is not transformed, while in the approach of Bleichrodt et al. (2004) the proportion of people in each category of health outcome is transformed but the health outcomes themselves are left untransformed.

It may appear at first sight that the two approaches, the utility model and the dual model, are essentially equivalent ways of incorporating equity considerations into economic evaluation. This paper will argue that this is not true. As we will show, the two models have rather different implications for priority setting in health and the choice between the two models is not trivial. Hence, there is really something to choose between the two approaches and our analysis will clarify the differences between the two approaches.

Alternative models are often compared in terms of the appeal of their underlying axioms. However, they may also be evaluated in terms of their implications for choices and decisions. The purpose of this paper is to compare the utility approach to model health inequality with the dual approach in terms of their implications for priority setting in health. Intuitively, the following requirements make sense for an inequality-averse social planner. First, it seems plausible that the more inequality averse the social planner is, the more priority she will give to the worse-off. This requirement catches the basic idea behind inequality aversion and is a basic requirement for any model that seeks to capture inequality aversion. Second, the priority given to the worse-off should increase with the degree of health inequality. The more unequal the distribution of health, the more the inequality hurts and the more attention should be paid to the worse-off by the inequality-averse social planner. This idea already appears in Nord (1993), where he argues that the priority given to the worse-off should increase with the severity of their illness. The lower the initial health of the worse-off, the more unequal the distribution of health and the more priority they should get. The notion that the degree of health inequality is important also implies that priority setting should depend on the prevalence of serious illnesses, i.e. the proportion of patients in the worst condition. This proportion plays a central role in the all measures of health inequality, e.g. the Gini index, the concentration index, the squared coefficient of variation, and the mean relative dispersion. Given the widespread use of these measures, in particular the Gini index and the concentration index in empirical health inequality measurement, it is important to examine to what extent the utility model and the dual model are sensitive to changes in prevalence.

We will evaluate the utility model and the dual model in terms of the above requirements. It turns out that both approaches satisfy the first requirement, that more inequality aversion entails more priority for the worse-off. When the degree of health inequality varies the models have different implications. Nord’s (1993) requirement, that more priority should be given to the worse-off when their situation deteriorates, is satisfied by the utility model, but not by the dual model. This weakness of the dual model is partially offset by the fact that the dual model can account for changes in the prevalence of the worst disease, but the utility model not. An implication of this last finding is that it is impossible to reconcile the utility model with the Gini index or the concentration index: under the utility model it is “irrational” to evaluate health distributions by the Gini index or the concentration index. On the other hand, the Gini index and concentration index equality orderings are special cases of the dual model (Bleichrodt and van Doorslaer, 2006) and, hence, they can be rationalized under the dual model.

In what follows, Section 2 describes the decision problem considered in this paper, which is close to the one adopted by Hoel (2003) and subsequently by Bui et al. (2005) to explore the impact of treatment risks on health care priorities. Section 3 analyzes the benchmark case where the social planner is inequality neutral. Sections 4 The utility model, 5 The dual model analyze the utility and the dual model of inequality measurement, respectively. In the conclusion, we contrast the results obtained in each model and some potential extensions are suggested. All formal derivations of results presented throughout the paper are in the appendices.