Evaluation of diagnostic tests for infectious salmon anaemia (ISA)

Reverse transcriptase polymerase chain reaction (RT-PCR), virus isolation (VI) and indirect fluorescent antibody tests (IFAT) are three assays currently used by the salmon industry to identify fish infected with the infectious salmon anaemia virus (ISAv). It is important to know the accuracy (sensitivity and specificity) and precision (repeatability and reproducibility) of these diagnostic assays currently used for disease control and research programs by the aquaculture industry in Canada. It is also important to know the effect of freezing on RT-PCR and VI as samples are often stored at -20°C (RT-PCR) or -80°C (VI) before they are processed for diagnostic and research purposes (-80°C).; In order to evaluate these test characteristics, 5 laboratories participated in a blinded study. To ensure that salmon from all states of infection were included in the study and that populations with different infection prevalence were included, salmon were selected from 4 groups assumed to have a different prevalence of infection. A total of 403 fish (100 per group except for one group with 103) were sampled. Each fish had multiple samples taken, which could be submitted to multiple laboratories for multiple tests. Assay methods evaluated included IFAT (1 lab), RT-PCR (3 labs), and VI (2 labs). Each laboratory used its own testing protocols.; Results indicated that freezing did not affect VI but improved the sensitivity of RT-PCR. The repeatability and reproducibility of VI was almost perfect. There were substantial differences in repeatability of RT-PCR among the three laboratories (kappa ranging from 0.5 to 0.96) and consequently there was only a moderate reproducibility between these laboratories (it ranged from serious disagreement to substantial disagreement). The repeatability of IFAT was moderate (kappa=0.68) when the IFAT results were analyzed using 1+ and above as positive result.; Due to the absence of a gold standard we used latent class models to evaluate the sensitivity and specificity. To use the method, three main assumptions needed to be met. We formally assessed those assumptions using pseudogold standards. Our results suggested that there was a conditional dependence between IFAT and VI and between IFAT and PCR. We accounted for the conditional dependence by incorporating into our analysis covariance between tests using maximum likelihood and Bayesian methods. The estimates obtained from this study suggested that RT-PCR was the most sensitive assay (range from 77-99%), followed by VI (range from 80-89%). IFAT was the least sensitive method (62%). Except for RT-PCR performed by one laboratory, the three tests had similar high specificities (range from 98-100%).; Different testing strategies (single testing, multiple testing interpreted in series or in parallel) were evaluated using the estimates of the sensitivity and specificity previously computed. Our analyses showed that the best testing strategy depends on the production phase. If sampling is to be carried out in a freshwater facility, then broodstock should be tested by VI alone, while pre-smolts should be tested with IFAT and VI used in series. For fish reared in saltwater, parallel testing with VI and RT-PCR or testing with VI alone, are appropriate testing strategies for broodstock. For market-fish. PCR by itself is a good screening option. If one assumes the prevalence in moribund fish is at least 50%, then a maximum of 5 fish (at a cut-point of (1)need to be tested at a cost of {dollar}220 to detect ISA in a cage. If one desired to have a perfect GSp (i.e. no false positive cage designations), serial testing with IFAT and VI is a better option. However, now a maximum of 9 fish (at a cut-point of 1) need to be tested at a cost of {dollar}472.