Hatchery-level predictive values for infectious pancreatic necrosis virus and Aeromonas salmonicida in Ontario, Canada

The probability and uncertainty of correctly classifying the IPNV and Aeromonas salmonicida status of fish-rearing and natural sites in Ontario were estimated through Monte Carlo simulations. Propagating several uncertain inputs showed the extent to which natural variability and our present lack of knowledge affect the probability of site misclassification. For the scenarios investigated, the site-level negative predictive values (SNPVs) were high and fairly constant. The site-level positive predictive values (SPPVs) - given a test specificity ranging between 0.999 and 1.0 - were much lower, more variable, and highly affected by cut-off point and sample size. Substantial uncertainty resides in classifying the pathogen status of test-positive sites, whereas much less uncertainty resides in classifying pathogen status of test-negative sites.