European coronary surgery study.

To the Editor: Serious flaws in the design of the European Coronary Surgery Study' raise questions as to the validity of its conclusions.Patients were retained in their original randomly allocated treatment groups, creating a strong bias against surgery: Once randomized to surgery, patients who did not undergo operation were analyzed as though they had.Likewise, patients randomized to medical treatment were analyzed as "medical" even if they underwent surgery (22.3% of the medical group).The unreasonable nature of such a study becomes clear if one applies this design to the study of a hypothetical drug:Once a decision is made to administer drug A to a patient, any untoward outcome is called a toxic effect of the drug even if the patient never received it.Furthermore, although 22% of patients in the control group are being given drug A, they are still counted as controls.Such a design violates the basic principle of the controlled clinical trial in which "the two series, under simultaneous investigation, are as alike as possible in every respect except that in one series, the patients receive the new drug or procedure, while in the other, they receive the control drug or procedure."2 It is not sufficient for the authors to state that the "policy of surgery" is being compared with the "policy of medical treatment," as this does not address the study question: whether coronary bypass surgery prolongs life in patients with stable angina that in past practice has not been considered severe enough to warrant surgery.'A poorly designed study is worse than no study at all, as it gives unsupported bias the authority of scientific fact.If a study design is inadequate to answer the question at hand, no conclusions at all may be drawn.