Quality of reporting of prospective in vivo and ex vivo studies published in the Journal of Veterinary Emergency and Critical Care over a 10-year period (2009-2019)

Abstract

Objective: To evaluate the reporting of key items associated with risk of bias and weak study design over a 10-year period.

Design: Literature survey.

Setting: Not applicable.

Animals: Not applicable.

Interventions: Papers published in the Journal of Veterinary Emergency and Critical Care between 2009 and 2019 were screened for inclusion. Inclusion criteria consisted of prospective experimental studies describing in vivo or ex vivo research (or both), containing at least 2 comparison groups. Identified papers had identifying information (publication date, volume and issue, authors, affiliations) redacted by an individual not involved with paper selection or review. Two reviewers independently reviewed all papers and applied an operationalized checklist to categorize item reporting as fully reported, partially reported, not reported, or not applicable. Items assessed included randomization, blinding, data handling (inclusions and exclusions), and sample size estimation. Differences in assessment between reviewers were resolved by consensus with a third reviewer. A secondary aim was to document availability of data used to generate study results. Papers were screened for links to access data in the text and supporting information.

Measurements and main results: After screening, 109 papers were included. Eleven papers were excluded during full-text review, with 98 papers included in the final analysis. Randomization was fully reported in 31.6% of papers (31/98). Blinding was fully reported in 31.6% of papers (31/98). Inclusion criteria were fully reported in all papers. Exclusion criteria were fully reported in 60.2% of papers (59/98). Sample size estimation was fully reported in 8.0% of papers (6/75). No papers (0/99) made data freely available without a requirement to contact study authors.

Conclusions: There is substantial room for improvement in reporting of randomization, blinding, data exclusions, and sample size estimations. Evaluation of study quality by readers is limited by the low reporting levels identified, and the risk of bias present indicates a potential for inflated effect sizes.

Keywords: bias; literature; quality; randomization.