[Use of statistics and the accessibility of original articles published in 3 anesthesiology journals]

To determine the use of various statistical analyses and the degree of analysis-dependent and article-dependent accessibility to the reader of publications in anesthesiology journals with impact factors (included in the Science Citation Index) in comparison with Revista Española de Anestesiología y Reanimación.

We analyzed original articles published in 2004 by 3 journals: Anesthesiology, the British Journal of Anaesthesia, and the Revista Españlola de Anestesiología y Reanimación. We found 1214 instances of use of statistics in 386 of the 412 articles analyzed. Twenty-six original articles did not use statistical analyses.

Statistics were used in 93.7% of the articles. The mean (SD) number of statistical tests used per article was 3.16 (132). Readers able to understand only descriptive statistics would find only a quarter of the articles' analyses comprehensible in all 3 of the journals studied. A reader familiar with second-level statistics (after the classification of Emerson and Colditz), would have an overall understanding of the analyses in 61.94% of the articles. Such a reader would be able to understand a higher percentage (76.67%) of articles in the Spanish anesthesiology journal.

As a result of marked advances in the use of statistics in anesthesiology journals, readers are seeing more applications of higher-level statistics. Readers must therefore acquire greater knowledge of statistics in order to understand the methods used in original research publications. The results we report for analysis-dependent accessibility show that Revista Española de Anestesiología y Reanimación is easier to understand for readers with knowledge of middle-level statistics. Both Anesthestiology and the British Journal of Anaesthesia publish articles that apply more complex statistical analyses.