Diagnostic accuracy meta-analyses typically have two main summary outcomes that are correlated, sensitivity and specificity. Due to this, hierarchical models accounting for this correlation have been developed; however, the proportion of diagnostic test accuracy systematic reviews using these hierarchical models and the impact on diagnostic accuracy estimates is unknown.
To determine if authors of systematic reviews of diagnostic accuracy studies published in imaging journals used recommended methods for meta-analysis, and to evaluate the effect of traditional (non-recommended) methods on summary estimates of sensitivity and specificity.
We searched MEDLINE for systematic reviews that included a meta-analysis of test accuracy data in imaging journals published from January 2005 to May 2015. Two review authors independently extracted study data and classified methods for meta-analysis as traditional (univariate fixed- or random-effects pooling or summary receiver operating characteristic curve) or recommended (bivariate model or hierarchic summary receiver operating characteristic curve). We analyzed the use of methods for variation over time. We recalculated the results from reviews in which study authors used traditional univariate pooling methods with a bivariate model.
Three hundred reviews met the inclusion criteria and in 118 (39%) of these the authors used recommended meta-analysis methods. We observed no change in the method used over time. We reanalyzed 51 univariate random-effects meta-analyses with the bivariate model; the average change in the summary estimate was -1.4% (P < 0.001) for sensitivity and -2.5% (P < 0.001) for specificity.
Recommended methods for meta-analysis of diagnostic accuracy in imaging journals are used in a minority of reviews; this has not changed significantly over time. Traditional (univariate) methods allow overestimation of diagnostic accuracy compared to recommended (bivariate) methods.
Patient or healthcare consumer involvement:
Overestimating diagnostic accuracy may lead to missed diagnoses (false-negative test results) and erroneous diagnoses (false-positive test results). Patients needlessly exposed to false-negative and false-positive test results due to inappropriate pooling methods may suffer negative health outcomes and psychological stress.