Abstract

CORTES-REYES, Édgar; RUBIO-ROMERO, Jorge Andrés  and  GAITAN-DUARTE, Hernando. Statistical methods for evaluating diagnostic test agreement and reproducibility. Rev Colomb Obstet Ginecol [online]. 2010, vol.61, n.3, pp.247-255. ISSN 0034-7434.

Introduction: when evaluating a diagnostic test’s usefulness, one often has to assess the results’ repeatability or their degree of agreement when compared to another test which is not used as gold standard for the entity in question. This paper was aimed at presenting the statistical methods used for evaluating clinical and laboratory observations’ repeatability or reproducibility and agreement, their theoretical basis and showing some examples of how they have been applied. Methodology: the theoretical bases for evaluating agreement and the repeatability of results were reviewed and examples of their use were taken from pertinent obstetrics- and gynecology-related literature. Results: the Kappa coefficient is usually used for evaluating the degree of agreement or concordance for dichotomic or categorical variables. The use of the intraclass correlation coefficient (ICC) or Lin’s concordance correlation coefficient should be preferred over Pearson’s correlation coefficient or paired Student’s t-test for assessing continuous variables’ concordance. These methods must be interpreted according to the clinical context in which they were used. Conclusions: the selection of statistical methods for evaluating agreement and reproducibility depends on the type of variable being measured and on the parameters being evaluated for assessing either reproducibility or validity.

Keywords : reproducibility of results; correlation; agreement; concordance.

        · abstract in Spanish     · text in Spanish     · Spanish ( pdf )