Bad medicine: using surrogate markersBMJ 2014; 348 doi: http://dx.doi.org/10.1136/bmj.g2012 (Published 10 March 2014) Cite this as: BMJ 2014;348:g2012
Spence reports that nalmefene has been approved by the Scottish Medicines Consortium (and is being considered by the National Institute for Health and Care Excellence) for alcohol misuse on the basis of criteria such as a reduction in “high drinking levels” and increases in “non-drinking days,” with minimal improvements in liver function tests.1 Although we agree that improvements in liver function tests are debateable, surely the use of such surrogate markers in harm reduction is the bigger problem.
When using an intervention to reduce the harm posed by a risk, hard endpoint outcomes are crucial in deciding how and when to use that intervention. If, for example, we wanted to reduce the harm from high cholesterol, the intervention ought to have a significant absolute (with less weight to relative) risk reduction on hard outcomes such as death and cardiovascular events. The mechanism of action (say through HMG-CoA enzyme reductase inhibition) might be useful to the intervention’s developer, and the biological surrogate markers (reduction in low density lipoprotein cholesterol) might be of interest to clinicians and patients.
Explaining the rationale for preventive treatment to patients can be challenging at the best of times and becomes more difficult when a treatment’s benefit is measured only in the abstract concepts of surrogate markers. We can’t always have hard endpoint data, but we should not mistake easily measured surrogate endpoints as being equivalent to clinical relevance. Approving an alcohol misuse drug on the basis of an increase in non-drinking days is far from evidence based medicine. It is similar to approving a weight reduction intervention on the basis of reducing the number of cakes consumed daily rather than the weight loss achieved (outcome), blood pressure and cholesterol improvements (surrogate markers), or ideally obesity related complications or obesity related death (hard endpoints).
Cite this as: BMJ 2014;348:g2012
Competing interests: None declared.