The resources available for the public provision of health care are not unlimited. Cost-effectiveness evidence on new healthcare interventions can help us prioritise in order to use scarce resources wisely, but to interpret cost-effectiveness evidence, it may appear as if we must make trade-offs between life and money. This is not so. If we are able to quantify the health improvements that resources would or could have generated in alternative use, a decision about providing or denying treatment can instead be framed as a trade-off between health gained and health forgone. In this thesis, I seek to provide a more robust basis for this way of reporting and interpreting cost-effectiveness evidence.

In Chapter II, I discuss the definition of opportunity cost in economic evaluation. The opportunity cost of providing an intervention is what we must give up to provide it. More precisely, it is typically defined as the value of the best alternative forgone. In economic evaluation of health care, opportunity cost has been understood in terms of the least cost-effective, currently funded intervention, which should be displaced when funding new interventions subject to a fixed budget. I show that alternative uses forgone may be neither currently funded nor well-defined, which implies that we should not look to cost-effectiveness evidence on specific interventions for information on opportunity cost. Further, identifying a best alternative use assumes that priority setting is based on objectives that can be summarised into a single measure of value. If economic evaluation is used to inform trade-offs between one measure of value (e.g., quality-adjusted life years, QALYs) and other, unquantified objectives, I suggest that it would be more appropriate to define opportunity cost as value in expected alternative use.

To quantify opportunity cost as health forgone, we need evidence on the health that resources would or could have generated in alternative use. In Chapter III, I use panel data on health spending and life expectancy in Swedish regions to estimate the marginal cost of producing a QALY. My findings imply that Swedish health care can produce health at a marginal cost of SEK 180,000 per QALY, which could be used as an expectation on how productive health spending would be in alternative use. I discuss methodological issues with this approach and identify some credibility problems with selection-on-observables strategies plaguing this and similar research to date. I address (some of) these problems by assessing coefficient stability and the causal mechanisms between healthcare resource use and health outcomes, using a second panel on hospital bed capacity and mortality. This analysis finds that health could be gained at a cost of SEK 420,000 per QALY by providing more hospital beds.

To illustrate the role of this evidence in healthcare priority setting, Chapter IV considers how it could have been used to inform decision making in a case of pharmaceutical reimbursement. I propose that economic evaluation report cost-effectiveness evidence as QALYs forgone per QALY gained. This frames a decision about providing or denying treatment as a judgement on the relative priority of QALYs gained and QALYs forgone, which is more transparent about a trade-off between equity and efficiency than deciding whether the monetary cost per QALY is too high. Framing decisions as health gained versus health forgone could also lead to better decision making by making opportunity costs more salient to decision makers and the reason for sometimes denying costly treatments easier to communicate.

In summary, cost-effectiveness evidence can be used to achieve the theoretical objective of health maximisation, but economic evaluations rarely report opportunity costs explicitly as health forgone. This thesis provides the practical means to be explicit and implications for the definition of opportunity cost and the interpretation of cost-effectiveness evidence when health maximisation is not the sole objective of healthcare priority setting.