| To Register SMDM Homepage | |
|
|
|
||||
Purpose: Clinical journals increasingly present uncertainty about the cost and effect of health care interventions using cost-effectiveness acceptability curves (CEACs). CEACs present the probability that each competing alternative is optimal for each value of the cost-effectiveness threshold. Our objective is to evaluate the limitations of CEACs for presenting uncertainty in cost-effectiveness analyses.
Methods: We explored many hypothetical parametric distributions of incremental cost and effect, allowing asymmetrical distributions as well as correlations between cost and effect. For each joint distribution we compared different presentations of uncertainty, such as: CEACs, credible intervals on the incremental net benefit, and value of information.
Results: (1)Maximizing the probability of cost-effectiveness may result in a different ranking of interventions than maximizing expected benefits when distributions of net benefits are asymmetric. A risk-neutral decision maker is interested in the latter ranking. (2)A risk-averse decision maker cares about both the probability and the consequences of making the wrong decision. Therefore, he may prefer an intervention with a low probability of cost-effectiveness to avoid a small probability of a catastrophe. CEACs do not inform about the consequences of making the wrong decision. (3)CEACs mix the magnitude and the precision of the mean incremental net benefit. Consequently, the medical importance suggested by, for example, a 90% probability of cost-effectiveness is ambiguous: it can reflect a huge, though imprecise difference in net benefit (e.g., a promising new cancer treatment) or a small but precisely defined difference hardly justifying the cost of implementation. (4)Because magnitude and precision are mixed, evidence presented as CEACs is difficult to synthesize with other qualitative, quantitative or subjective evidence and risk-attitude. Credible intervals on the incremental net benefit present magnitude and precision separately. (5)CEACs are typically equivocal about the value of information: the exact same CEAC can represent a decision with a high or a low value of information.
Conclusion: Both for guiding immediate decisions and for prioritizing information collection, these considerable drawbacks of CEACs should make us rethink their use in communicating uncertainty. A more informative presentation of uncertainty would be the credible/confidence intervals of incremental cost, effect and net benefit, together with the total expected value of perfect information.
See more of Oral Concurrent Session A - Cost Effective Analysis: Methods
See more of The 26th Annual Meeting of the Society for Medical Decision Making (October 17-20, 2004)