Reporting of results from clinical investigations is an important component of the ethical conduct of research. We conducted a review of the publication record of all protocols submitted to our Research Ethics Board (REB) for the periods 1995-1996.
Protocols for studies closed by the principal investigator (PI) were identified from the REB database. We identified the local PI, the sponsor, and the phase of clinical investigation. For each publication, we identified publication authors, publication title, journal, and whether a randomised clinical trial or not. For data analysis Systat V10 was utilised. p<0.05 was taken as statistically significant.
We identified 110 unpublished studies and 85 publications (44%). The median time to publication was 32.5 mos (975 days)(mean 33.1±17.8 mos) and the median number of subjects enrolled per study was 329. For unpublished studies, subject recruitment was reported for 45 studies. In these studies there was a median of 2 subjects recruited by local investigators per study. Physicians authored 79/85 publications (93%) and were the principal investigators in 89% (173/195) of studies. Publication by sponsor is given in the Table. Of the 85 published studies, results could be described as positive findings in 72 (85%). Of 13 negative published trials, 11 were sponsored by the pharmaceutical industry, one by the health authority, and one by a federal granting agency.
Publication of research results was unacceptable, less likely to occur if sponsored by the pharmaceutical industry, and more likely to occur if reporting positive findings. Calls for public registration of clinical trials appear warranted.
Publication of research results is an ethical obligation of researchers and failure to publish results is considered by some to represent scientific misconduct. In addition failure to publish scientific results violates the social contract research subjects undertake when they participate in research and misinforms public policy through failure to provide adequate information upon which to base guidelines and perform meta-analysis. Our results suggest much further education around these issues is required.
TABLE.SponsorYesPublication No% SuccessPharmaceutical Company607246Local Health Authority113424*Federal Granting Agency11191*Charity30100Other030*
p<0.05 vs Pharmaceutical Company
Richard Hall, None.