Abstract
Purpose:
To identify the most commonly used statistical analyses in the ophthalmic literature and to determine the likely gain in comprehension of the literature that readers could expect if they were to sequentially add knowledge of more advanced techniques to their statistical repertoire.
Methods:
All articles published from January 2012 to December 2012 on Ophthalmology, American Journal of Ophthalmology and Archives of Ophthalmology were reviewed. A total of 780 peer-reviewed articles were included in this cross sectional study. Two reviewers examined each article and assigned categories to each one depending on the type of statistical analyses used. Discrepancies between reviewers were resolved by consensus. Total number and percentage of articles containing each category of statistical analysis were obtained. Additionally we estimated the accumulated number and percentage of articles that a reader would be expected to be able to interpret depending on their statistical repertoire.
Results:
Readers with little or no statistical knowledge would be expected to be able to interpret the statistical methods presented in only 20.8% of articles. In order to understand more than half (51.4%) of the articles published, readers were expected to be familiar with at least 15 different statistical methods. Knowledge of 21 categories of statistical methods was necessary to comprehend 70.9% of articles, while knowledge of more than 29 categories was necessary to comprehend more than 90% of articles. Retina and glaucoma showed a tendency for using more complex analysis when compared to cornea.
Conclusions:
Readers of clinical journals in ophthalmology need to have substantial knowledge of statistical methodology to understand all the results of published studies in the literature. The results of this study could provide guidance to direct the statistical learning of clinical ophthalmologists, researchers and educators involved in the design of courses for residents and medical students.
Keywords: 468 clinical research methodology