Dear Editor


Angelis and Kanavos’s comment [1] on our Editorial “Does MCDA trump CEA?” [2] fails to address the three main points of our paper.

They fail completely to address the inadequate treatment of opportunity cost in MCDA and, in particular, its failure to treat cost (whether accounting cost or opportunity cost) in the multidimensional way that is the key characteristic of MCDA. This means that MCDA not only does not attempt to measure the best alternative use of resources but also fails to be consistent with its own principles in its lack of multidimensionality in cost.

They misunderstand our reference to double counting, particularly the MCDA treatment of cost as a criterion additional to cost-effectiveness. Given the size of the cost element in most empirical analyses, this is a tremendous case of double counting. Distinguishing between types of cost and even different concepts of cost is of great importance in cot-effectiveness analysis (CEA) [3].

Similarly, they sidestep our criticism of ‘independence of criteria’ by introducing an irrelevant confusion with ‘statistical independence’ and ‘preference-independence’ to which we did not refer. Our objection was to the blurred definition of many of the criteria and variables, which greatly magnifies the risk of overlap and overcounting. There is a risk of bias in all methods, but a method that is as imprecise and jumbled as MCDA introduces biases that are virtually impossible to allow for, even when well-designed deliberative decision-making procedures are employed.

Finally, as we noted in our Editorial, how holistic a CEA ought to be is always contextual [4,5,6].