Open access to clinical trials data.

摘要: Well-conducted randomizedclinical trials (RCTs) are thegold standard for evaluating the safety and efficacy of medical therapeutics. Yet most often, a single group individuals who conducted trial only ones have access to raw data, conduct analysis, publish study results. This limited does not typically allow others replicate findings. Given time expense required an RCT, it is often unlikely that will independently repeat similar experiment. Thus, scientific community public accept results produced published by original research team without opportunity reanalysis. Increasingly, however, opinions empirical data challenging assumption analysis clinical straightforward any other would obtain same results.1-3 In this issue JAMA, Ebrahim et al4 report their findings based on rigorous search previously reanalyses RCTs. Their first surprising discomforting finding was just how infrequently reanalysis has occurred in research. Searching literature from 1966 present, authors found 37 reports met criteria as RCT Of these few performed, majority (84%) had overlapping report. rare, but were reported fully independent group. Despite overlap, al about half differed statistical or analytic approaches, third definitions measurements outcomes, important, led interpretations conclusions different than those article. While definition what constituted analyses, end points, findings, subjective, authors’ general consistent with emerging indicates can yield originally published. Even when investigators presenting evidence venues always consistent. For example, there presented US Food Drug Administration (FDA)may differ important ways at sessions journals. Rising al5 assessed information provided FDA 9% discordance between Not unexpectedly, all direction favoring drug. Another example ClinicalTrials.gov journal articles.Hartungetal2 showedthat inarandomsampleofphase 3 4 trials, 15% primary point themain article trialists ClinicalTrials.gov. Moreover, 22% primaryoutcomevalue inconsistently,withsomeevenhavingdifferences number deaths. Other studies rates discordance.6 When groups somewhat reach alternative conclusions, cause necessarily bias. Independent individual patient-level because every involves discretionary decisions. situation exemplified evaluation Medtronic’s bone morphogenetic protein 2 (BMP-2) product. Two expert organizations international reputations systematic reviews given BMP-2 task, resources, timeline.7,8 Nevertheless, final reports, methods identical ways. Consistent meta-analyses topics using do show concordant results.9 The current reviewbyEbrahimet al, aswell cited work, suggests several next steps needed ensure transparencyandopenaccess inRCTs.First, allRCTs, prespecified protocols plans, should be registered community, fulfilling ethical promise made enrolled Such step contribute improvements standardization registration reportingof results,which remainsvarieddespitegovernmental regulation policies.10-12 Full availability essential allowpeer reviewers journals tomonitor andanalyticplans toensureconsistencyand thereby reduce someof thevariation thatmayoccur reporting results, particularly respect primary, secondary, exploratory outcomes. Second, rawdata andmetadata (all data) ideally bemade available seek verification markedly increase community’s confidence Evenwhen importantly, open dialogue promote deeper understanding its interpretation. demonstrates Related page 1024 Opinion