Abstract
Birch and Gafni (1992) claim in a recent article that the decision rules of cost-effectiveness/utility analysis (CEA) fail to achieve their stated objectives, namely the maximization of health gains for a given amount of resources. This critique includes the following objections to CEA: First, they argue that CEA does not guarantee improvements in social welfare in situations where multiple health objectives exist (e.g., survival and functional status). Second, they argue that CEA does not consider the health gains forgone by reallocating resources from existing programs to fund new programs. Third, they argue that incremental CEA can lead to inefficient resource allocation when there are alternative levels of programs which compete for budgetary resources. Finally, they argue that the decision rules of CEA are incorrect in the presence of program indivisibilities, and that integer programming techniques are needed. These arguments are illustrated by using hypothetical examples.
The analysis by Birch and Gafni is critically examined in this paper. First, we review the optimal decision rules in cost-effectiveness analysis. Second, we show that most of the objections to CEA raised by Birch and Gafni in their examples have no basis if CEA is used in an appropriate way. In fact, they are led to misleading conclusions because they fail to interpret the incremental cost-effectiveness ratios in their examples properly. The inconsistent results in their analysis arise due to (1) their failure to recognize the basis of QALYs as a measure of social welfare; (2) their failure to distinguish between optimal decision rules for independent programs versus mutually exclusive programs; and (3) their failure to exclude dominated alternatives from consideration in analysis of competing programs. Third, we address the valid but well-known point about program indivisibilities. We end with some concluding remarks.