Abstract
Background: :
Approximately 60% of controlled clinical trials (CCTs) presented at scientific meetings and published as abstracts in conference proceedings are published in full. Systematic reviews are essential to evidence–based practice and guidelines, and rely on access to all trial data. To identify unpublished CCTs in vision research, the US satellite of the Cochrane Eyes and Vision Review Group (CEVG@US) handsearches conference proceedings to find reports of CCTs to include in trials registers [and ultimately, systematic reviews].
Purpose: :
To test whether abstract authors could contribute to efficient identification of CCT reports published in ARVO proceedings.
Methods: :
With the cooperation of the ARVO conference organizers, we added a definition and check–off box to the 2004 submission form for authors to indicate whether a submitted abstract described a CCT. We asked "Is the research presented in your abstract a human clinical trial? [Yes] [No] (see definition); Definition of Human Clinical Trial: A planned study in humans designed to assess the efficacy and/or safety of one or more test interventions by comparing outcomes in individuals assigned the test intervention(s) with those receiving no intervention or a comparison intervention, and where individuals in all groups are enrolled, treated, and followed concurrently." In 2005, we modified the question to include the word "controlled" so that the question read "Is the research presented in your abstract a human controlled clinical trial?" We compared author classification of ARVO abstracts with handsearch results for 2004 and 2005, and assumed the handsearcher identified the "gold standard" of all CCTs.
Results: :
Handsearchers identified 295 CCTs from 5,610 2004 abstracts. Authors classified 634 as CCTs, of which 229 were correctly classified ("true positives") and 405 were not ("false positives"). Authors failed to classify 66 CCTs correctly. Thus, sensitivity was 229/295 (78%). For the ARVO 2005 meeting, handsearchers classified 331/5,732 abstracts as CCTs. Authors classified 490 as CCTs, of which 235 were true positives and 255 were false positives. Authors failed to classify an additional 96 CCTs (sensitivity = 235/331 [71%]).
Conclusions: :
Inclusion of the modifier "controlled" in the question appeared to lead to fewer false positives, but did not increase the sensitivity of author classification. Although classification of study design by ARVO authors could reduce or eliminate the need for handsearching, current methods are not satisfactory.
Keywords: clinical (human) or epidemiologic studies: treatment/prevention assessment/controlled clinical trials • clinical (human) or epidemiologic studies: biostatistics/epidemiology methodology