Skip to main content
. 2016 Mar 25;54(4):1017–1024. doi: 10.1128/JCM.03104-15

TABLE 3.

Diagnostic performance of selected qPCR assays using cutoffs determined by ROC analysis

Organism or model AUC (95% CI) Cutoffa Sensitivity (%) (95% CI) Specificity (%) (95% CI) Likelihood ratio (95% CI)
G. vaginalis 0.90 (0.87–0.94) 1.10E5 83.6 (77.2–88.8) 84.1 (78.5–88.7) 5.26 (3.84–7.21)
A. vaginae 0.92 (0.89–0.95) 1.10E5 87.1 (81.2–91.8) 90.7 (85.9–94.2) 9.32 (6.12–14.2)
BV 0.87 (0.84–0.91) 1.00E3 78.9 (72.1–84.8) 92.5 (88.1–95.7) 10.6 (6.55–17.0)
G. vaginalis + A. vaginae 0.96 (0.94–0.98) 1.26E6 87.1 (81.2–91.6) 93.9 (89.8–96.7) 14.3 (8.44–24.4)
G. vaginalis + BVAB2 0.93 (0.91–0.96) 6.41E5 85.4 (79.2–90.3) 90.7 (85.9–94.2) 9.14 (5.99–13.9)
A. vaginae + BVAB2 0.95 (0.93–0.98) 1.53E5 92.4 (87.4–95.9) 90.2 (85.4–93.8) 9.42 (6.26–14.2)
G. vaginalis + A. vaginae BVAB2 0.96 (0.94–0.98) 1.17E6 91.2 (85.9–95.0) 93.0 (88.7–96.0) 13.0 (7.97–21.2)
BICb model 0.98 (0.97–0.99) 0.20 91.8 (86.6–95.5) 95.3 (91.6–97.7) 19.7 (10.7–36.1)
a

Cutoff is copies/reaction except for BIC, for which the cutoff is the predicted probability of the logistic regression model, where 0 = healthy and 1 = BV.

b

BIC, Bayesian information criterion.