Expanding the evidence base in transplantation: more and better randomized trials, and extending the value of observational data.

Transplant registries, despite being complete and large, will always be observational data and subject to potential biases, such as selection bias. Consequently conclusions about the effects of interventions will almost always be uncertain, except when the magnitude is very large, and natural history of a condition homogeneous. Rather than re-running old debates of the outcomes-research movement, it is time for improvements in registries and randomized trials to occur so that effective interventions can be administered to individual patients in whom benefits will outweigh harms.