Response to the Letter to the Editor

Abstract

I greatly appreciate the comments offered by Drs Krug and Crott in ‘The “How” of Cost-Effectiveness Analysis is not so straightforward’. While I agree with several points made in their response to my original article, there are others I disagree with strongly. There are several potential sources of bias in any economic analysis, including cost-effectiveness analyses (CEA) of a care pathway. However, comparing implementation of a care pathway to an existing situation where a care pathway is not always used, and therefore current practice is ‘inappropriate’, is not one of them. Clearly, care pathways should be evidence based and it should be widely accepted that they lead to the best patient results given current knowledge. What I proposed in my original article was analysis of the incremental costs and effects of increasing the use of a care pathway from its current level in a given setting, which could be as low as no use, to as high a level of use as possible with the implementation being tested. I cannot see where the issue of bias comes into play with this basic comparison. However, I agree that generalizability of the results is often an issue and that researchers should bear this in mind when designing studies and communicating their results. Krug and Crott make a valid point on the importance of considering perspectives. Researchers should be completely explicit when stating the perspective used in the analysis, explain the reasoning behind their choices, and what would happen to the results if other costs and consequences considering different perspectives were included. I also agree with the authors on the sensitivity of results to the time horizon used for the analysis. What bears closer examination is what resources are needed to keep compliance with care pathways at a maximum level for the length of time chosen for the time horizon or, alternatively, what will happen to compliance with the care pathway beyond initial implementation if no resources are dedicated to maintaining it. Quite often these data are not available. Non-medical costs are clearly difficult to measure in almost all circumstances. Whether or not the researcher includes these costs is entirely dependent on the specific research question, which in turn depends of the needs of those commissioning the study. Given the choice, and assuming research resources are available, the societal perspective including patient and family costs should be included. The issue of what to use for an effectiveness measure in many cases involves a trade-off as presented in the original paper. To label my original discussion of this a ‘flaw’ is unwarranted. Using more accurate intermediate or process measures in the absence of good epidemiological data linking that outcome to something more tangible, such as life-years saved, means that the researcher is trading off the ability to compare results with future studies of effectiveness for a defensibly accurate result. However, if future research does accurately link process measures to tangible outcomes, the original cost-effectiveness study could be re-analyzed to yield results with a common denominator using the newly available data. Beginning with a ‘thorough and systematic’ review of relevant literature for these or any studies is de riguer and should require no reiteration. The purpose of my original paper was, as clearly stated, to provide elementary information for CEAs for care pathways in the hope that it would lead to greater appreciation and support for this kind of research. It was not written to delve deeply into the details of performing such a study. There are several reference books that serve this purpose admirably in the expansive ways necessary for such detail. However misguided several of the criticisms given by Krug and Crott were, I appreciate the attention it brings to this very important topic and hope that it contributes to making economic analyses of care pathway implementation more numerous, more rigorous and more widely used.