Skip to main content

Table 3 Pooled summary estimates of all studies

From: Usefulness of sonication in the microbiological diagnosis of cardiovascular implantable electronic device infections: systematic review, meta-analysis and meta-regression

 

Method

Sonicated cultures

No sonicated cultures

Estimate

CI 95%

Estimate

CI 95%

DOR

Univariate DerSimonian-Laird

14.4

6.8–30.7

5.8

3.3–10.4

Bivariate Zwindermann & Bossuyt

10.6

5.5–18.6

5.5

3–9.3

Bayesian Bivariate model

14.9

5.13–36.9

6.7

2.9–13.8

AUC

Univariate Holling’s proportional hazard

0.889

0.857–0.923

0.723

0.665–0.791

Bivariate diagnostic random-effects model (*)

0.804

NA

0.771

NA

Bayesian Bivariate model (**)

0.87

NA

0.78

NA

Sensitivity

Univariate model

0.799

0.702–0.895

0.459

0.321–0.596

Bivariate diagnostic random-effects model

0.756

0.677–0.820

0.446

0.310–0.590

Bayesian Bivariate model

0.808

0.704–0.903

0.447

0.306–0.599

Specificity

Univariate model

0.744

0.621–0.867

0.872

0.815–0.929

Bivariate diagnostic random-effects model

0.766

0.605–0.874

0.868

0.795–0.917

Bayesian Bivariate model

0.779

0.629–0.895

0.887

0.825–0.937

  1. (*) Partial AUC restricted to observed False Positive Rates and normalized are 0.742 for sonicated and 0.483 for non-sonicated cultures
  2. (**) Partial AUC restricted to observed False Positive Rates and normalized are 0.756 for sonicated and 0.479 for non-sonicated cultures