How often do articles depend on suppression effects for their findings? How often do they disclose this fact? By suppression effects, we mean control-variable-induced increases in estimated effect sizes. Researchers generally scrutinize suppression effects as they want reassurance that authors have a strong explanation for them, especially when the statistical significance of the key finding depends on them. In a reanalysis of observational studies from a leading journal, we find that over 30% of articles depend on suppression effects for statistical significance. Although increases in key effect estimates from including control variables are of course potentially justifiable, none of the articles justify or disclose them. These findings may point to a hole in the review process: journals are accepting articles that depend on suppression effects without readers, reviewers, or editors being made aware.

In 2003/2004 a field trial was conducted in Northern Ireland to assess the diagnostic accuracy of six serological tests for bovine brucellosis caused by Brucella abortus. Whereas between-test comparisons have been used to calculate test performances so far, the present study used a latent class approach to estimate diagnostic test accuracy parameters in the absence of a gold standard for these six tests simultaneously and to estimate the true prevalence, while accounting for clustering in the study population and risk factors for true prevalence. Results obtained in this study with regard to prevalence, sensitivity and specificity were largely in accordance with previous findings. Screening tests (SAT and EDTA) appeared to be the most sensitive; however, at low prevalences the EDTA and CFT showed the highest positive predictive values of all investigated tests. The specificities and negative predictive values of all diagnostic tests were found to be very high. Differences of prevalence between three groups of the study population with different risk of exposure could be attributed to the mode of sampling indicating that a more risk-based sampling will result in a higher prevalence than a cross-sectional sampling mode. Age, dairy status and history of abortion were shown to influence the prediction of the latent true infection status.