Unnatural Selection? An Analysis of the Factors Associated with Unpublished Prospective Clinical Research in Radiation Oncology

Abstract

<h3>Purpose/Objective(s)</h3> Factors associated with successful publication of clinical trials in radiation oncology have not yet been identified. We investigated whether publication bias was associated with factors such as funding source, size, phase of trial, study length, randomized design, and trial origin. <h3>Materials/Methods</h3> Clinical trials between 2000-2005 were extracted by filtering for "radiation oncology" using the National Institutes of Health database, clinicaltrials.gov. The dates were chosen to select for studies that had adequate time to accrue. Studies were excluded if they were incomplete, observational, Phase 4, or lacked sufficient descriptions of their methods. Included studies underwent independent samples <b>t-</b>tests and bivariate analyses to evaluate for publication status using the Pearson Chi-Square test through JMP. <h3>Results</h3> 538 studies were candidates for analysis. Multi-center trial sites, government funding, Phase III status, and randomized allocation were factors independently associated with increased likelihood of publication. The average number of study arms, lengths of the study in years, and number of participants were all significantly greater in trials that achieved publication. <h3>Conclusion</h3> The results of the review demonstrate publication bias. Less than half of the completed clinical trials in radiation oncology from 2000-2005 were published; Phase III status, multi-center sites, non-industry funding, randomized allocation larger participant pools, and a longer study duration all contributed to successful publication of those trials. Ascertaining facilitative and inhibitory factors to publication may facilitate transparency in medical literature and thereby improve clinical advancements and the scientific discovery process.