Registering Clinical TrialsRegistering Clinical Trials

To the Editor: Drs Dickersin and Rennie¹ stated that a comprehensive register of clinical trials would address the problem of publication bias. However, they omitted an important aspect of this problem. Every researcher has had the experience of a manuscript being rejected. This sometimes occurs with well-designed and well-executed studies with good data simply because the results were not considered to be newsworthy by journal editors. Although this is a hazard of academic medicine, it nevertheless should be noted that journals as well as researchers are responsible for publication bias. Therefore, if a future researcher or patient looking in a registry finds a clinical trial that otherwise seemed to have disappeared, the results may have been very important and close to the researcher's heart, but not to an editorial board of a medical journal.

Unless the outcomes of such studies were made part of the registry, it would be impossible to determine what happened to studies that "disappear"—the "why" may be as important as the result. Well-designed studies may become outdated because of other information that becomes available before their completion. Similarly, their results may have been duplicative of others or they simply may not have been interesting enough for the readership of journals to which they were submitted. To serve all the purposes stated by Dickersin and Rennie, a registry should also contain study results and the reasons for the outcome. However, the rigorous review required to support and justify study researchers' conclusions may make an accurate registry that fulfills all these purposes beyond reach at the current time.