Barbosa Fabiano Timbó, de Souza Diego Agra
Universidade Estadual de Ciências da Saúde de Alagoas.
Rev Bras Anestesiol. 2010 Sep-Oct;60(5):528-36. doi: 10.1016/S0034-7094(10)70064-7.
Statistical analysis is necessary for adequate evaluation of the original article by the reader allowing him/her to better visualize and comprehend the results. The objective of the present study was to determine the frequency of the adequate use of statistical tests in original articles published in the Revista Brasileira de Anestesiologia from January 2008 to December 2009.
Original articles published in the Revista Brasileira de Anestesiologia between January 2008 and December 2009 were selected. The use of statistical tests was deemed appropriate when the selection of the tests was adequate for continuous and categorical variables and for parametric and non-parametric tests, the correction factor was described when the use of multiple comparisons was reported, and the specific use of a statistical test for analysis of one variable was mentioned.
Seventy-six original articles from a total of 179 statistical tests were selected. The frequency of the statistical tests used more often was: Chi-square 20.11%, Student t test 19.55%, ANOVA 10.05%, and Fisher exact test 9.49%. The frequency of the adequate use of statistical tests was 56.42% (95% CI 49.16% to 63.68%), erroneous use in 13.41% (95% CI 8.42% to 18.40%), and an inconclusive result in 30.16% (95% CI 23.44% to 36.88%).
The frequency of inadequate use of statistical tests in the articles published by the Revista Brasileira de Anestesiologia between January 2008 and December 2009 was 56.42%.
