Table D-2Summary results from all meta-analysis methods (for the example in Table D1)

ModelEstimation (within-study likelihood)Logit-sensitivity (95 CI or CrI*)Logit-specificity (95% CI or CrI*)
Univariate, FEIV (normal)1.708 (1.646, 1.770)1.855 (1.797, 1.912)
Univariate, REDL (normal)1.803 (1.561, 2.044)2.354 (2.049, 2.659)
Univariate, REREML (normal)1.799 (1.569, 2.029)2.422 (2.017, 2.828)
Univariate, REML (binomial)1.840 (1.607, 2.074)2.556 (2.110, 3.002)
Bivariate, REmultivariate DL (normal)1.805 (1.564, 2.047)2.352 (2.048, 2.657)
Bivariate, REREML (normal)1.801 (1.569, 2.033)2.416 (2.005, 2.827)
Bivariate, REML (binomial)1.839 (1.605, 2.072)2.547 (2.104, 2.990)
Bivariate, REFully Bayesian1.840 (1.589, 2.105)2.558 (2.080, 3.076)

CI = confidence interval; CrI = credibility interval; DL = DerSimonian-Laird; FE = fixed effect; IV = inverse variance; ML = maximum likelihood; RE = random effects; REML = restricted maximum likelihood;

*

Credibility intervals are presented for Bayesian analyses.

Credibility intervals are presented for Bayesian analyses.

From: Appendix D, Worked Meta-Analysis Example

Cover of An Empirical Assessment of Bivariate Methods for Meta-Analysis of Test Accuracy
An Empirical Assessment of Bivariate Methods for Meta-Analysis of Test Accuracy [Internet].
Dahabreh IJ, Trikalinos TA, Lau J, et al.

NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.