Accuracy of data in abstracts of veterinary ophthalmology research articles published in peer-reviewed journals

Assess the accuracy of abstracts in published veterinary ophthalmology articles.

Abstracts and contents of 204 original research articles in veterinary ophthalmology published in seven peer-reviewed journals between 2016-2020 were reviewed. Abstracts were considered inconsistent if they contained data that were either missing from or inconsistent with corresponding data in the article's body. Each abstract was graded between 0 (inaccurate) to 3 (accurate), and each inconsistency was subjectively classified as minor or major. The influence of selected variables was assessed: journal, impact factor, year of publication, number of words in abstract, study type (prospective/retrospective), and characteristics of the corresponding author [institution (academia/private practice), country of domicile (native/non-native English), number of publications].

Most abstracts were accurate, with 1%, 4%, 9% and 86% receiving a score of 0, 1, 2 and 3, respectively. When detected, most inconsistencies were considered minor (77%). Although not statistically significant (p ≥ .130), the proportion of articles with a perfect score (=3) was higher in prospective (88%) vs. retrospective (81%) studies, academia (88%) vs. private practice (78%), and studies from corresponding authors domiciled in English (89%) vs. non-English (83%) speaking countries. A significant but very weak (r = -0.15 to -0.19; p ≤ .034) negative correlation was found between accuracy score and number of words, as well as 1-year and 5-year impact factors.

Although relatively uncommon, data in abstracts that are inconsistent or missing from the article's body do occur in veterinary ophthalmology articles, and could adversely influence a reader's interpretation of study findings.