All evaluation exercises involve ethical values, as they require some conception of the “good life.” Evaluation of health technologies is no exception. Because there is no consensus about what is a good life, we have to devise decision-making procedures in which citizens with different opinions are heard and treated fairly (1). The purpose of health technology assessment (HTA) is to offer useful input into this process so as to increase the quality of the deliberations and of the resulting decisions. How to bring ethical values into this process?
This question is most often inspired by some uneasiness about the dominance of economic evaluation. This has led to an unfortunate opposition of “economic” and “ethical values.” Everything that is not included in standard economic evaluation is then called “ethical”—ranging from distributional issues to bioethical concerns about the acceptability of specific technologies. One ends with a long list of ethical questions, with the added proviso that even this long list is not complete and with the explicit acceptance that different observers will answer them differently (2).
This leads to an unequal playing field. At one side, one has what looks like a “hard science” approach, with a well-defined methodology, involving sophisticated technical questions, giving unambiguous answers—and apparently supported by a large majority of analysts. At the other side, a bunch of soft arguments about disparate questions, without clear answers, and put forward by analysts who disagree about how to tackle the issues. No wonder that the first side dominates the scene. This is not to say that decision makers always follow the economic evaluation results. Yet, even if they have good reasons to diverge, this often gives an impression of adhockery.
If we want to change that situation, the first thing to do is to communicate that the “hard” approach is not as hard as it looks. Rather than opposing “economic” and “ethical” values, one should emphasize that economic evaluation implements a narrow set of ethical values. In fact, many agree that “maximizing the unweighted sum of QALYs” is at best an imperfect approximation of what would be an ethically desirable objective function for society. Sociologists of science can investigate the paradox that a consensus has grown to implement an objective function about which there is near consensus that it is not a good one. My hunch is that what makes the framework attractive are its strong analytical foundations.
Yet, one can keep these analytical foundations and still introduce broader ethical considerations into the analysis. One just has to interpret “outcomes” and “costs” in a broader way. Why limit the analysis to QALYs? It is perfectly possible to introduce other measures of health, to make the definition of health multidimensional, or even to go beyond health and consider the effects of the health technology on other dimensions of the good life (such as autonomy, the quality of social relations, and income). And why take an unweighted sum? Introducing refined distributional weighting is technically easy. Including equity and a broader measure of individual outcomes into the analysis would already take up a large part of what now are seen as “ethical” concerns.
If it is technically feasible, and if there is consensus that the unweighted sum of QALYs is too restrictive, why is this broader framework not used? First, data are missing on the effects of specific technologies on important dimensions of well-being, differentiated for different groups in society. Yet, there is no reason why such information could not be collected in a scientifically valid way. Putting richer information on the table would immediately change the format of the debate. The first requirement to improve the ethical content of the debate is to invest in richer data collection.
Second, if we broaden the scope of the analysis, unambiguous conclusions will no longer be possible. Opinions differ about what is equitable and about what dimensions of life to include. One then necessarily has to resort to a sensitivity analysis, in which one produces a series of results, each corresponding to a specific ethical position. The analyst can only show the range of possible choices and explain how they are driven by the choice of value judgments. Yet, this is exactly how it should be. If there is no consensus about crucial normative choices, HTA should not create the misleading impression that there is one (and certainly not create a feeling of safety by using an objective function that almost nobody likes).
All I said until now rested on the conviction that it is crucial not to narrow down “consequentialism” to “welfarism” (or, worse, utilitarianism), and that most ethical values can be integrated in an extended consequentialist framework. Indeed, extended consequentialism can go much further than what I described above. Even rights may be treated as goals and consequential analysis based on comprehensive outcomes can take note of processes and freedoms exercised (3). Of course some deontological arguments cannot be handled within extended consequentialism. Bioethicists may object to the use of certain technologies, whatever their positive side-effects. These absolute principles act as strict constraints on the choices and they can be analyzed as such.
I conclude. The best strategy to introduce ethical values is not to formulate a long list of questions to “complement” the economic evaluation. Nor is it helpful to question the idea of formal evaluation. Broader ethical values will only be taken seriously if they are integrated in the formal evaluation exercise itself. This is possible in an “extended consequentialist” perspective. Many reasons have been formulated to not follow that route. Some take a broad critical stance on the societal role of health technology and of HTA, some are afraid to reduce qualitative considerations into quantitative measures (this fear is based on a misunderstanding: ranking without measuring is often sufficient), some believe that too much technical analysis makes the debate nontransparent. All these reasons are understandable, but those who use them should then not be disappointed that their influence on the decisions remains minimal.
Two final pragmatic remarks. First, what I have sketched is a long-run perspective. In the short run, a kind of Socratic approach is probably the best we can do. Yet, the ambition to extend and improve the present techniques of economic evaluation and to collect richer data should become an essential part of the exercise. Second, I am well aware that convincing practitioners of economic evaluation to take up a broader perspective will require an uphill battle. Ethicists that talk the language of economic evaluation are badly needed.
