Abstract
Purpose: :
Starting in 2007, ARVO required that abstracts reporting controlled clinical trials note the place of trial registration and the trial registration number. Our purpose was to determine (1) whether randomized controlled trials (RCTs) accepted for presentation at the 2007 ARVO meeting had been registered in a clinical trial registry, and (2) if all abstracts that included trial registration information were indeed controlled clinical trials.
Methods: :
We hand searched the 2007 ARVO program abstracts (n = 6,044) to identify all those describing RCTs. We also reviewed all 2007 abstracts where the author had entered information in the space allocated for trial registration (n=258). We determined the proportion of abstracts describing RCTs that had been registered prior to submission, and abstracted trial registration information. We also examined all abstracts where authors included registration information, and classified the study design, some of them RCTs and some of them non-RCTs.
Results: :
We classified 2.9% (173/6044) of all abstracts as describing an RCT. This proportion was lower than the proportion of abstracts classified as an RCT for abstracts submitted to ARVO in the previous three years (2006, 3.6% (219/5920); 2005, 3.6% (212/5732), and 2004, 3.4%% (189/5610)). Only 62% (107/173) of RCTs had been registered before submission, with the majority (88/105) registered at clinicaltrials.gov, the International Standard Randomized Controlled Trials Number, or ACTR. Other entries included the EudraCT trials register (n = 2), regulatory agencies (n=3), ethics committees (n=3), and other (n = 11). Authors of 39% (66/173) of RCTs did not provide any trial registration information.Authors of an additional 151 non-RCT abstracts provided a response in the space allocated for trial registration information. Three of these entries were a statement that registration was not required due to study design. Of the remaining 148 abstracts, 18 described a controlled clinical trial, 96 described an uncontrolled clinical trial, 22 described a case control study, and 12 we classified as other, including animal studies (n=5), laboratory experiments (n=4), image evaluation with no human involvement (n=2), and a simulation experiment (n=1).
Conclusions: :
The proportion of accepted abstracts describing RCTs decreased in 2007 compared with that of previous years, perhaps indicating that registration was a barrier for submission or acceptance. Nearly 40% of authors did not adhere to the requirement to register RCTs before submission to the 2007 ARVO meeting.
Keywords: clinical research methodology • clinical (human) or epidemiologic studies: biostatistics/epidemiology methodology