Cost-Minimization Analysis

Cost-minimization analysis is a special form of cost-effectiveness analysis where the health outcomes can be considered to be equivalent between two treatment alternatives and therefore the interest is only on which of the two strategies has the lower cost. Cost-minimization analysis appears to have much to commend it: in particular, it embodies an apparently simplified approach to decision making by looking at only the cost side of the equation. However, there are a number of potential pitfalls that exist in terms of the practical use of cost-minimization analysis.

The first of these represents a problem of definition. Many apparent examples of cost-minimization studies fail to present any justification of the [Page 227]equivalence of health outcomes between two treatments and are therefore more accurately described as cost analysis. A simple cost analysis should not be considered a true cost-minimization study without some form of evidence for the equivalence of health outcomes being presented. Note that these cost analyses are also often incorrectly described as “cost-benefit analyses” due to the net-benefit approach to decision making, particularly in the early health economic evaluation literature.

More recently, as economic evaluation alongside clinical trials has become more common, the problem has become one of interpretation. It is all too common to see “cost-minimization analyses” presented that turn out to be based on the interpretation of lack of significance of an effect measure in a clinical trial as evidence of equivalence. In the clinical trial field, there is a well-known adage that “absence of evidence is not evidence of absence.” To interpret the lack of a significance as evidence of no effect is to place the importance of the Type I error (concluding a difference exists when the null hypothesis of no difference is true) above that of the Type II error (concluding that no difference exists when in fact the alternative hypothesis of a difference is true). To properly show that two treatments are no different (within a small margin of error) requires an appropriately designed equivalence study that typically requires a greater sample size to reliably demonstrate equivalence than is recruited to many superiority (difference) trials.

Furthermore, clinical trials typically are powered to detect differences in only a single effect measure (primary trial endpoint). In contrast, health economic analyses are multidimensional, often trading off different effects (risks and benefits) to obtain a composite measure of outcome. It would be very rare indeed for two treatments to be truly equivalent on all measures of outcome and rarer for a clinical trial to be adequately powered to demonstrate such a multidimensional equivalence.

As a consequence of these difficulties, examples of true cost-minimization studies are rare. One of the most popularly cited (though rather old) examples relates to a cost-minimization study of alternative oxygen delivery methods, with the underlying assumption that the treatment (oxygen) is truly equivalent between alternative delivery systems. It is worthy of note that the original analysis (in common with the healthcare perspective of many economic studies) did not include any convenience to the patient in the analysis.

Although conceptually appealing, due to the simplified approach to decision making, the practical problems associated with cost-minimization analysis have led some commentators to argue the “(near) death of cost-minimization analysis.” The appropriate framework for analysis of most studies will be the estimation of cost-effectiveness. It is clear that the use of separate and sequential tests of hypothesis of cost and effect based on superior study designs does not constitute appropriate grounds for using cost-minimization as a decision-making tool.

...