Recent efforts by medical journal staffs to improve the quality of research papers have had mixed results. Examples are given to show that randomized, placebocontrolled trials are not free from bias and that the failure to include all-cause death rates can be extremely misleading, as can the use of relative risks in the absence of absolute risks. Other examples show how the conclusions in an abstract may not agree with the data in the body of the paper, or do not tell the whole truth. Still others use false surrogate endpoints or faulty trial protocols to favor a desired outcome. The whole picture may be seen as a breakdown of the peer-review system.  – Joel M. Kauffman, Ph.D.

Bias in Recent Papers on Diets and Drugs in Peer-Reviewed Medical Journals