Are results from pharmaceutical-company-sponsored studies available to the public?

Purpose: Only 53% and 63% of studies and clinical trials results presented at congresses are published. Company-sponsored trial results are being posted on publicly accessible Web sites. We analyzed the public availability (publication or posting on a Web site) rate, time to publication, and factors predicting public availability of results of studies sponsored by a pharmaceutical company. Methods: This was a retrospective cohort study analyzing all studies conducted by GlaxoSmithKline in Spain between 2001 and 2006. Initiation and completion were defined as first participant/first visit and last participant/last visit (or their equivalents). Papers published up to 31 March 2009 were considered. Logistic regression models were used to identify factors predicting public availability of results. Results: The cohort comprised 143 studies (94 clinical trials; of these, 87 were included in international products clinical development plans). Public availability rate was 80% (114/143) for all studies and 78% (73/94) for clinical trials; publication rates were 68% and 61%, respectively. The median time to publication for all studies and trials was 27.3 and 28.4 months, respectively. Study associated to a cancelled project was the only significant factor associated with lower publication rate for all studies [odds ratio (OR) 0.069; 95% confidence interval (CI) 0.02-024; p < 0.001) and trials (OR 0.075; 95% CI 0.016-0.343; p = 0.001) and a lower public availability rate (OR 0.052; 95% CI 0.007-0.382; p = 0.004) for trial results. Therapy area, sample size, positive trial results, duration of experimental phase, and being a clinical trial did not predict publication or public availability. Conclusions: Eighty percent of studies included in this analysis are publicly available. Web site posting increases public availability rate of clinical trial results from 61% to 78%. Cancellation of projects is the single factor negatively influencing publication and public availability rates. [ABSTRACT FROM AUTHOR]