Estimating diagnostic test accuracy for infectious salmon anaemia virus in Maine, USA

Abstract

Infectious salmon anaemia virus (ISAV) is a pathogen of consequence to farmed Atlantic salmon, Salmo salar L. ISA control centres on active surveillance for early detection by reverse transcription polymerase chain reaction (RT-PCR), indirect fluorescent antibody assay (IFAT) and virus isolation. Because diagnostic test performance varies among populations and laboratories, the Office International des Epizooties (OIE) recommends an evaluation of test accuracy in each region of use. This is complicated in Maine, USA by the co-existence of ISAV genotypes homologous to North American (NA) and European (EU) isolates. While NA ISAV genotypes isolated in Maine are characterized by high morbidity and mortality, the single EU genotype in Maine has not yet been linked to disease or isolated by cell culture. Consequently, distinguishing among genotypes is critical to infection response. Accuracy in NA genotype detection was estimated from ISA surveillance data using latent class models. Results suggested that RT-PCR is an excellent screening test for NA ISAV genotypes in Maine, although probably with reduced specificity in the presence of EU genotypes. IFAT, in contrast, was a poor screening test for detection of ISAV in Maine, although it may be useful in confirmation of NA genotypes during disease outbreaks.

Quantitative Polymerase Chain Reaction (PCR) for Detection of Aquatic Animal Pathogens in a Diagnostic Laboratory Setting

“…The final approach is latent class modeling, which can be used when there is no appropriate reference test available (Hui and Walter 1980). Latent class modeling allows for estimations of DSe and DSp as well as the prevalence in a population by employing either maximum likelihood or Bayesian estimation procedures (Nerette et al 2005;Gustafson et al 2008;Nerette et al 2008). There are three assumptions that must be satisfied to use latent class models: (1) at least two populations must be tested and prevalence must vary between these two populations, (2) DSe and DSp should be constant across the populations, and (3) the tests should be conditionally independent given the infection status (although methods to account for conditional dependence of tests in latent class models were reported by Qu et al 1996).…”

Infectious Salmon Anemia

“…И в этой ситуа-ции изучение экологии ISAV превращается в один из «пробных камней», позволяющий судить о нашей технологической готовности «потерять берег из виду» 6 . …”

“…ISAV уже зарекомендовал себя как се-рьёзная угроза для лососевых рыборазвод-ных ферм (ЛРФ), нанося серьёзный эконо-мический ущерб на территории Шотландии [3], Фарерских о-вов [4], Канады [5], США [6], Чили [7], и количество эпизоотий увели-чивается прямо пропорционально количеству новых крупных ЛРФ. Однако следует иметь в виду, что ISAV -природноочаговый вирус, природный резервуар которого не установ-лен.…”

Recommended reporting standards for test accuracy studies of infectious diseases of finfish, amphibians, molluscs and crustaceans: the STRADAS-aquatic checklist

“…Explanation: Prospective and cross-sectional designs, using standardized procedures for sample collection, transportation and handling, usually allow for better quality and consistency of samples with more detailed descriptive data of populations and animals sampled than would normally occur in a retrospective study using repository samples. The example by Gustafson et al (2008) meets those criteria as do the examples in Items 3b and 4b, where the authors used prospectively generated experimental samples for validation of the TUE. The infection trials described in Items 3b and 4b were conducted for the purpose of generating material of known history of pathogen exposure to facilitate evaluation of test accuracy.…”

“…For instance, if a test was evaluated for screening purposes using a mixed population with apparently healthy and moribund fish, one could use the diagnostic sensitivity and specificity estimates of moribund fish only when testing to confirm clinically suspect (moribund) cases. Other study examples that provide covariatespecific estimates for various clinical categories include Jansen et al (2010) and, for separate agent genotypes, Gustafson et al (2008). (Caraguel et al 2009, p. 14).…”