Reporting of results among drug tests registered in ClinicalTrials.gov
Annals of Internal Medicine
Florence T. Bourgeois, Srinivas Murthy, Kenneth D. Mandl
2010
An analysis published in 2010 on Annals of Internal Medicine, carried out by three researchers from the Children's Hospital of Boston and publicly funded, describes the characteristics of the trials registered on ClicalTrials.gov regarding the efficacy and safety of 5 classes of drugs and verifies whether the source of the funding was associated with the publication of favorable results to the drug. Among the 78.276 trials registered by 31.08.2009, those concerning 5 classes of drugs were selected: anticolesterolemics, antidepressants, antipsychotics, proton pump inhibitors, vasodilators. Of the 4.825 trials thus highlighted, 546 were included in the study, those concerning efficacy and safety started and concluded between 01.01.2000 and 31.12.2006. For each selected trial, study design, funding source, registration and publication date, and final outcome were described. The main source of funding was classified as industry, government agencies, non-profit or non-federal bodies (universities, hospitals, foundations). Of the 546 trials, 346 (63%) were funded by industry, 74 (14%) by government agencies, and 126 (23%) by non-profit or non-federal organizations. Industry-funded trials are more likely to be phase 3 or 4 (88.7%; P <0.001), to use an active comparator (36.8%; P = 0.010), to be multicentric (89.0%; P <0.001), to enroll a greater number of participants (median sample size, 306 participants versus 78 and 50 of the trials funded by other sources; P <0.001), to enroll at least 75% of the expected participants (84,9%; P <0.001). In contrast, only the trials funded by government agencies included a significant number of children (37,8%; P <0.001). Of all trials, only 362 (66,3%) published results: 230 (66,5%) of those funded by industry, 41 (55,4%) by government agencies, and 91 (72,2%) by non-profit or non-federal entities. Industry-funded trials reported positive results in 85.4% of publications, compared to 50.0% of government trials and 71.9% of those funded by non-profit or non-federal organizations (P <0.001). Among the latter, those in which there was a contribution from the industry reported positive results more frequently than those without contributions (85.0% vs. 61.2%; P = 0.013). The trial publication rate within 24 months of the end of the study ranges from 32.4% for those funded by industry to 56.2% for those funded by non-profit or non-federal organizations with no industry contributions (P = 0.005). The authors point out that it was not always possible to confirm whether a study had been published and this could have led to some classification errors. Furthermore, further information, which is not yet available, on the protocol and on the results of the trials in general could shed light on the factors underlying the link between the source of funding and the results. The authors close their article by stating that industry-funded trials are less likely to be published within 2 years of their completion and more likely to report positive results.