Drs. Asch and Hershey point out two errors, one arithmetic and one conceptual. The arithmetic error, transposing a decimal point (turning 0.25% into 2.5%), does not affect the model. Thus, we focus here on the conceptual error: We added lives saved (by not performing surgery) for patients who were not being considered for intervention in the first place. We concur with the calculations of Asch and Hershey but not concur that our miscalculations understate the persuasiveness of quantitative health policy. When the calculations of Asch and Hershey are used, cost-effectiveness still fails to act as a deterministic decision point. Network A provides the most cost-effective point for the 4-cm threshold, but it no longer does so for the 5-cm threshold. Asch and Hershey agree that few could support using the 4-cm threshold at network A after viewing the associated costs, even though doing so would improve societal outcome. This underscores our main point, which is that perverse decisions (such as choosing a network of lesser quality) can occur if one focuses on cost-effectiveness without considering other important factors. Rigorous cost-effectiveness analyses have a role in health care decisions, but we are not convinced that the needed rigor will always be present within a competitive marketplace where coverage decisions may be influenced by financial incentives.

Ms. Jung's comments bring out the varied uses of medical necessity. However, we do not share her view that medical necessity is a strong factor in clinical decision making. First, there is little to suggest that providers routinely incorporate the concept of medical necessity in their clinical decisions. Second, medical necessity decisions are primarily made by payers and are often predicated on contractual requirements rather than on clinical judgment. Third, medical necessity does not always correlate with good public policy; this is true, for example, for immunization to prevent re-emergence of an infectious disease, such as diphtheria. Finally, interpretations by courts and payers vary, and medical necessity is thus not currently a viable foundation on which to prioritize clinical actions. Whether it can be is controversial, and some have argued that cost-effectiveness should become the basis for coverage decisions [1].

We believe that cost-effectiveness will not readily solve the medical necessity conundrum. Nonetheless, developing better methods to target those who truly need a medical service and those who do not is paramount. Our primary concern was to demonstrate the intrinsic problems that arise when medical necessity and cost-effectiveness are used. We hope that our model was helpful in this regard.