Exploring Options for Syndromic Surveillance in Aquaculture: Outbreak Detection of Salmon Pancreas Disease Using Production Data from Norwegian Farms

Table 5

Performance of the proposed syndromic surveillance system using mortality data in detecting salmon pancreas disease outbreaks monthly and at cohort level.

Area	Monthly performance^†				Cohort performance^†
	3	4	5	6	3	4	5	6
TP	500	582	147	629	74	76	18	89
FP	993	569	306	917	69	37	33	51
TN	1,123	688	281	808	80	38	9	48
FN	72	107	34	152	12	19	5	20
Se	87.4%	84.5%	81.2%	80.5%	86.0%	80.0%	78.3%	81.7%
FAR	46.9%	45.3%	52.1%	53.2%	46.3%	49.3%	78.6%	51.5%
PPV	33.5%	50.6%	32.5%	40.7%	51.7%	67.3%	35.3%	63.6%
NPV	94.0%	86.5%	89.2%	84.2%	87.0%	66.7%	64.3%	70.6%
Time^‡	—	—	—	—	1 (−3 to 4)	1 (−2 to 3)	−1 (−3 to 1)	0 (−4 to 4)

The values in this table refer to the system implemented using the parameters presented in Table 4. TP = true positive, FP = false positive, TN = true negative, FN = false negative, Se = sensitivity, FAR = false alarm rate, PPV = positive predictive value, NPV = negative predictive value, and Time = timeliness. ^†Sum of TP, FP, TN, and FN is not always equal to the numbers given in Table 3 due to a minimum number of months with aberration detection required to trigger outbreak alarms. ^‡Time is given as the median (interquartile range) of the difference between the month of PD detection by the SyS system and the current surveillance program.