If the purpose of systematic reviews is to encourage physicians to readily use new treatments in their own patients, I submit that the number needed to treat (NNT) [1] is not the correct method for reporting efficacy of treatments. From a clinical standpoint, NNT has the major shortcoming of representing efficacy only in the patients in whom the treatment is studied, because it depends on baseline risk. Therefore, it is not likely to represent efficacy in a physician's own patients, who will have different baseline risks. As a result, I do not think that physicians would readily use the treatment in their patients.

A relative risk reduction has the important advantage of being independent of baseline risk and thus has the same value in all patients for a given treatment [1]. Therefore, when efficacy derived from a clinical trial is reported as relative risk reduction (as is customary), it is directly applicable to other patients with different baseline risks and may make physicians more likely to use the same treatment in their own patients. An example of this phenomenon is the rapid worldwide adoption of thrombolytic therapy in patients with acute myocardial infarction; this has occurred even though the studies were performed primarily in Europe and United States.

I suggest that systematic reviews should report efficacy of treatments primarily as relative risk reductions. If NNT is needed for some reason, it can be easily calculated from the reported relative risk reduction and estimated baseline risk.