Surgical Therapy for Coronary Disease: Randomized and Observational Studies-Reply

In Reply.—  Dr Mark's letter recalled Gehan's¹ quotation of Dr Emil Frei: "People who insist that all studies must be randomized, on the one hand, and those who insist that all studies should be historical control on the other, share one thing in common—they are both equally wrong." No form of inductive reasoning can deliver a demonstratively certain conclusion. All inductive evidence has limitations, as well as subjectivity, and produces an inference supporting a conclusion at a given probability. Both randomized and observational studies are useful, or even complementary, and the key to either is understanding advantages and limitations for a given question, performing meticulous analysis, and properly interpreting results. To quote Gehan¹: "No clinical trial—whether randomized or not—is an entity unto itself. It should instead be considered as one stage in an evolving research programme."Randomized trials minimize selection bias and differences between treatment groups. They are