Field evaluation of diagnostic sensitivity (DSe) and specificity (DSp) of common tests for amoebic gill disease (AGD) and complex gill disease (CGD) in cultured Atlantic salmon (Salmo salar) in Scotland using Bayesian latent class models

AS Boerlage*, Angela Ashby, JI Eze, GJ Gunn, A Reeves

*Corresponding author for this work

    Research output: Contribution to journalArticlepeer-review

    4 Downloads (Pure)

    Abstract

    Amoebic gill disease (AGD) and complex gill disease (CGD) are the most significant marine gill diseases in salmon aquaculture in Scotland. Little is published about diagnostic performance of tests to detect these diseases, making it difficult to interpret test results. We estimated diagnostic sensitivity (DSe) and specificity (DSp) of common tests for AGD (gross AGD score, qPCR for Neoparamoeba perurans, histopathology) and CGD (gross proliferative gill disease (PGD) score, gross total gill score, histopathology). Because specifications in our sampling protocol implemented to encourage consistency across the farms might affect diagnostic performance of histopathology (historically the reference standard for gill diseases), we used Bayesian latent class models without reference standard. Cases and non-cases were based on less, medium, and severe stringent case definitions, representing different cut-off levels for the different tests. Gross gill scores for both diseases were excellent in designating non-diseased fish, DSps were generally around 1. To detect CGD, DSe of gross total gill score and gross PGD score were between respectively 0.81 (0.73 – 0.91 lower to upper 95% credible interval) and 0.53 (0.46 – 0.64) for medium stringent case definitions, and to detect AGD the DSe for the gross AGD score was between 0.53 (0.48–0.57) and 0.14 (0.07 – 0.22) for respectively the less and severe stringent case definition. Thus, gross gill scores were medium to good in designating truly diseased fish, implying some false negatives are expected. For CGD the DSe for gross total gill scores were the highest, for AGD it was the qPCR test at a DSe of 0.92 (0.86 – 0.99). For both diseases, DSe was lowest for histopathology, e.g. 0.23 (0.16 – 0.30) for AGD and 0.1 (0.07 – 0.14) for CGD under medium stringent case definitions, perhaps due to collecting the second gill arch on the right rather than the worst affected arch, whilst PCR sampling and gross gill scoring included multiple (PCR) or all (gross scoring) gill arches. The diagnostic goals of these tests differ; gross gill scoring provides a low-cost presumptive diagnosis, PCR a non-lethal confirmation of the presence of a specific pathogen and histopathology provides information on the underlying aetiology of gill damage as well as the extent, severity, and chronology of gill disease. An effective gill health surveillance strategy is likely to incorporate multiple diagnostic tools used in a complementary manner.

    Original languageEnglish
    Article number105654
    JournalPreventive Veterinary Medicine
    Volume204
    Early online date22 Apr 2022
    DOIs
    Publication statusPrint publication - Jul 2022

    Keywords

    • Gross gill score
    • Histopathology
    • Latent class models
    • Marine gill diseases
    • Sensitivity
    • Specificity

    Fingerprint

    Dive into the research topics of 'Field evaluation of diagnostic sensitivity (DSe) and specificity (DSp) of common tests for amoebic gill disease (AGD) and complex gill disease (CGD) in cultured Atlantic salmon (Salmo salar) in Scotland using Bayesian latent class models'. Together they form a unique fingerprint.

    Cite this