Abstract
Purpose :
Clinical trial registries such as ClinicalTrials.gov are intended to increase clinical research transparency by documenting all trial designs and outcomes. However, Hartung et al (2014) noted discrepancies between registries and peer-reviewed publications in general medical literature. We attempted to quantify discrepancies in the ophthalmic literature.
Methods :
Clinical trials published in 2014 in the American Journal of Ophthalmology, Ophthalmology, or JAMA Ophthalmology were reviewed. Observational, retrospective, uncontrolled, or post hoc studies were excluded to yield a sample size of 106 articles. If possible, articles were matched to registry entries and assessed for consistency in design, results, and/or funding. Each article was independently assessed by two reviewers and adjudicated by a third. Omission, transposition, or addition of interventions and/or methods were considered discrepant in descriptive data. Differences of ±0.1 were considered discrepant in numeric data.
Results :
No matching registry entries were found for 32 studies (30.2%). Six studies were found in international registries. The remaining 68 studies were analyzed. Discrepancies were identified in:
1) Study design: specific interventions, 11.8% inconsistent (I); sample size, 7.4% (I), 75.0% unreported in registry and/or paper (U); study type, 23.5% (I); primary outcome measure (POM), 47.1% (I); analysis methodology, 1.5% (I), 76.5% (U); secondary outcome measure (SOM) design, 48.5% (I), 26.5% (U)
2) Study results: POM results, 14.7% (I), 55.9% (U); SOM results, 26.5% (I), 60.3% (U); serious adverse events 20.6% I, 66.2% (U)
3) Funding sources: 11.8% (I)
Conclusions :
Registry data could not be found for a high proportion of published clinical trials in ophthalmologic literature. For studies with registry data, widespread discrepancies existed between published and registered data at rates similar to those reported for general medical literature. In many cases, discrepancies reflect missing data; in others, explicit deviations in methodology and/or data were found. Although valid reasons may exist for discrepancies between registry and publication, such variations may also represent post hoc modifications in study design and analysis. Registries do appear to provide some transparency in the planning and execution of clinical trials, but also reveal inconsistencies in information between the two sources.
This is an abstract that was submitted for the 2017 ARVO Annual Meeting, held in Baltimore, MD, May 7-11, 2017.