Table 2.

Performance in outbreak detection according to germ criterion and a upper control limit defined by 90% confidence interval, with and without clustering.

Parameter	Without clustering	With clustering (d=0)g
Sensitivity (%)a	90.0 (27/30)	100 (30/30)
Specificity (%)b	84.0 (630/750)	86.7 (650/750)
PPV (%)c	18.4 (27/147)	23.1 (30/130)
NPV (%)d	99.5 (630/633)	100 (650/650)
AUCe (95% CIf)	0.87 (0.81-0.94)	0.93 (0.91-0.95)

^a Sensitivity = TP/(TP+FN) (where TP (true positive) is an outbreak correctly identified as an outbreak, and FN (false negative) is an outbreak wrongly identified as a non-outbreak).

^b Specificity = TN/(TN+FP) (where TN (true negative) is a non-outbreak correctly identified as a non-outbreak, and FP (false positive) is a non-outbreak wrongly identified as an outbreak).

^c Positive predictive value (PPV) = TP/(TP+FP) (variables defined above).

^d Negative predictive value (NPV) = TN/(TN+FN) (variables defined above).

^e AUC: area under receiver operating characteristic curve.

^f CI: confidence interval.

^g d: cutting Euclidean distance.