The underlying assumption behind the recommendations in this chapter is that it is unethical to conduct poorly designed clinical trials. The basis for this assertion is that patients are placed at risk in a trial where potentially convincing data cannot be obtained. This premise is widely believed and has been stated many times by leading ethicists (e.g., Levine 1988), leading methodologists in the United States (e.g., T. Chalmers, personal communication, 1993) and Europe (e.g., Turner 1989), and leaders of Ethics Committees (ECs) (W. Rosinga, European Ethical Review Committee, personal communication, 1986).

If one accepts the premise that poorly designed clinical trials are unethical to conduct, then one of the first questions to consider is whether or not any, some, many, or most clinical trials conducted today are poorly designed. Although it is difficult to generalize about the quality of unpublished clinical trials, it is clear that many published trials are poorly designed. It is the author’s view that it is the responsibility of Institutional Review Boards (IRBs)/ECs to improve the standards of clinical trials by refusing to condone and approve clinical trial protocols that are not well designed. Many such trials are currently approved by IRBs/ECs on the basis that the trial does not ethically compromise the care and treatment of patients (i.e., the patients are not directly harmed). This attitude is similar to that of the IRBs/ECs only reviewing protocols to see whether there are any sins of commission and ignoring the fact that there are important sins of omission in the protocol. Patients who are enrolled in a poorly designed clinical trial are being exposed to risks unnecessarily and unethically because the data obtained will not convince most readers of the results obtained or may even mislead the readers by presenting false-positive or false-negative results. Most IRBs/ ECs would need to have protocols reviewed by statisticians and/or a clinical trial methodologist if they wished to adhere to the proposal of reviewing the trials to ensure they are adequately powered and are well designed. (Statistical reviews are only
sometimes conducted by an IRB/EC.) In addition, the IRB/EC would know that it was complete in its assessment of a protocol if it used a checklist to confirm that it has addressed each of the essential issues (e.g., Are there any therapies, diagnostics, or disease-prevention products in the protocol that are not approved for general use in the doses to be used? Are any advertisements to be used, and if so, are they included in the submission?) rather than relying on the reading of the protocol by two “designated reviewers.” A simple checklist has been published (Spilker 2002), and the author is aware that many IRBs/ECs have adopted the practice of using checklists.

In addition to the IRB/EC responsibility, the journals that publish clinical trial results also have an important responsibility that, until now, has never been adequately addressed. This responsibility involves publishing more information on clinical trials so that poorly designed trials cannot be dressed up to hide some of their serious methodological flaws (e.g., not listing tests where the data was negative, only presenting some of the data obtained, not presenting complete information in the Methods section of the publication). Although many flaws are identified by reviewers for the journal during the peer review process, there are others that are able to be easily hidden because very few details of a trial’s methodology (and conduct) are ever published.