Advertisement
(LOGO: http://www.newscom.com/cgi-bin/prnh/20080627/CLF051LOGO )
Drs. Scott Ramsey and John Scoggins of University of Washington and FredHutchinson Cancer Research Center, Seattle, searched the National Institutesof Health's ClinicalTrials.gov registry to identify 2,028 registered researchstudies of cancer treatment. Major medical journals now require that allstudies considered for publication be registered in a publicly accessibledatabase like ClinicalTrials.gov .
Advertisement
A subsequent search of the National Library of Medicine's PubMed databaseshowed that just 17.6 percent of the trials were eventually published inpeer-reviewed medical journals. "We find that less than 1 in 5 completedtrials registered since the inception of ClinicalTrials.gov are listed in theregistry or PubMed as published manuscripts," Drs. Ramsey and Scoggins write.
The publication rate was particularly low for "industry-sponsored"studies, such as those funded by pharmaceutical companies -- just 5.9 percent,compared to 59 percent for studies sponsored by collaborative researchnetworks.
Of published studies, nearly two-thirds had "positive" results-the studytreatment worked as hoped. The remaining one-third had "negative" results --the outcome was disappointing or did not merit further consideration of thetested treatment.
This finding raises concern about "publication bias" in cancer treatmenttrials. The researchers suspect that the rate of negative results is muchhigher in the studies that have gone unpublished. Drs. Ramsey and Scogginswrite, "It is likely that many unpublished studies contain importantinformation that could influence future research and present practice policy."
There are several reasons why registered trials may not be published.Some trials may fail to meet recruitment or follow-up goals. If the trial iscompleted and the results are negative, researchers may feel that negativestudies do little to advance scientific understanding, or to enhance theirprofessional reputation. Sponsors may not encourage researchers to publishnegative results. For their part, medical journal editors may be lessenthusiastic about negative studies.
However, publication of negative studies is important for several reasons-- not only to avoid repeating negative trials, but also for what can belearned from the lack of response. "Unpublished trials may have specialimportance in oncology, due to the toxicity and/or expense of many therapies,"the researchers add. The fact that so much cancer research goes unpublishedraises concerns about the completeness of available information on present andfuture cancer treatments.
A pair of accompanying editorials comment on the underpublication ofcancer treatment studies, and suggest some approaches to addressing theproblem. Dr. James H. Doroshow of the National Cancer Institute (NCI) reportson efforts to develop a database of administrative and outcomes data for allstudies performed at NCI-supported institutions.
Drs. Gregory A. Curt and Bruce A. Chabner, Senior Editor andEditor-in-Chief of "The Oncologist," echo the call for increased NCIinvolvement in ensuring publication of funded studies. Meanwhile, the editorsof "The Oncologist" are considering publication of a new, peer-reviewed, fullysearchable venue for cancer treatment studies that would otherwise gounpublished. "There is a need for a new venue for publishing allwell-executed trials that fail to meet positive endpoints: 'negative' in asense, but valuable nonetheless," Dr. Curt comments.
The article entitled "Practicing on the Tip of an Information Iceberg?Evidence of Underpublication of Registered C
