Purchase this article with an account.
Pooja Mehta, Neil M Bressler; Identification and Recognition of Limitations of Retrospective Studies in Select Peer-reviewed Journal. Invest. Ophthalmol. Vis. Sci. 2016;57(12):5556. doi: https://doi.org/.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Since retrospective designs can introduce bias or confounding on results, and as such, should be noted by authors. This study aimed to determine how often clinical research publications in select ophthalmology journals that are retrospective in design state “retrospective” design in the title or methods, and how often limitations of that design are mentioned or commented upon in the discussion section.
Original investigations from three ophthalmology journals with relatively high impact factors as well as a similar specialty journal in otolaryngology and a subspecialty journal in retina were reviewed in 2004 to 2005 and in 2014 to 2015. All original investigations, excluding brief reports; case series; case reports; or observations, were reviewed to determine if the study design was retrospective. Retrospective designs were defined as studies performed after the outcomes of interest had already occurred; most commonly case control studies, but also retrospective cohort or case series.
Journals reviewed included the American Journal of Ophthalmology, JAMA Ophthalmology, JAMA Otolaryngology, Ophthalmology, and RETINA. Preliminary results suggest that fewer than one-third of all of the articles mentioned in the Discussion section that the retrospective design was a limitation to the interpretation of the results in 2004 to 2005; this increased relatively by 50% to 100% by 2014 to 2015. However, in fewer than 5% of the discussions did the authors mention how the retrospective design might affect the interpretation of the results both last decade and this decade.
Over the past decade, authors in the journals evaluated are more likely to indicate that a retrospective study has limitations that can affect the interpretation of the results, but few authors indicate exactly how those limitations might affect the interpretation of the results.
This is an abstract that was submitted for the 2016 ARVO Annual Meeting, held in Seattle, Wash., May 1-5, 2016.
This PDF is available to Subscribers Only