Randomized, controlled trials of patients with rheumatoid arthritis are published sooner and more likely to be published if the outcome is positive, indicating publication bias, a recent study found.
The study surveyed ClinicalTrials.gov for randomized, controlled trials (RCTs) of patients with rheumatoid arthritis (RA), finding 143 eligible studies registered before 2012 and completed before 2010. The outcome of 16 trials could not be ascertained. Among the 127 RCTs whose outcomes were known, 74 had a positive result, and 53 were negative, according to Dr. Nasim A. Khan of the University of Arkansas, Little Rock, and his associates (Arthritis Rheumatol. 2014;66:2664-74). The researchers also found that trial registration at ClinicalTrials.gov increased dramatically after July 2005, when the International Committee of Medical Journal Editors made trial registration in a public registry a requirement for publication, with 37 trials being registered before 2006, and 106 registered during 2006-2009.
Of the 143 trials, 95 studies had been published, with 64 of these having positive results and 31 having negative results. Only 10 trials with positive outcomes had not been published, compared with 22 RCTs with negative outcomes, Dr. Khan and his associates reported. The median time to publication for studies with a positive outcome was 30 months, while studies with a negative outcome had a median time to publication of 49 months.
The rate of publication was slightly higher for studies completed after 2005, with 65% of studies completed before 2006 being published, 78% of studies completed from 2006-2007 being published, and 59% of studies completed from 2008-2009 being published. Some of those later trials may still be published, Dr. Khan and his fellow investigators conceded. The time to publication was much slower for studies completed before 2006, taking a median time of 55 months. Studies completed from 2006-2007 took 30 months to publication, and studies completed from 2008-2009 took 34 months.
The authors noted that while progress has been made to increase transparency, there are still issues within the system, and that “efforts must be made to increase awareness of the existence and drawbacks of publication bias among current and future physicians, scientists, individuals working in the pharmaceutical industry and insurances companies, and patients.”
The authors had no disclosures to report.