A 2-stage hierarchical interrupted time-series analysis to quantify the long-term effect of subclinical bacterial kidney disease on performance of farmed Atlantic salmon (Salmo salar L.).
Prev Vet Med. 2019 Sep 20;172:104776
Authors: Boerlage AS, Stryhn H, Armstrong B, Hammell KL
Bacterial Kidney Disease (BKD) is an economically significant disease in salmonid aquaculture and commonly requires antibiotic treatments to reduce its impact. Once a pen of fish is diagnosed with BKD, fish are considered chronically infected, potentially until harvest. Although there appears to be little or no evidence to support it, it is often assumed that subclinical infections affect productivity over the long term. We used a 2-stage hierarchical interrupted time series (ITS) analysis in an attempt to quantify the effect of subclinical BKD on mortality, growth, and food conversion ratio (FCR) of Atlantic salmon cultured in marine farms in Atlantic Canada. For all three outcomes, BKD had for some site cycles a positive effect, and for others a negative effect. Overall, the effect of BKD on mortality and growth could not be detected (effect -0.08 ((95% ci: -0.51, 0.35) and 0.00 (-0.02, 0.02)), while a very small effect showing an increase in FCR was detected (0.07 (-0.01, 0.15)). We hypothesized that minimal interference with fish performance may be compatible with the ecology of Renibacterium salmoninarum, the causative agent of BKD. For this organism, vertical transmission is a primary mode of propagation in low-density host populations as found in the wild. Since farms are always adapting and optimizing their farm management of BKD, these constant adjustments may also have negated our ability to detect the effect of many factors contributing to BKD productivity impacts. Hierarchical ITS analysis is considered an appropriate methodology to investigate the complex relationships with productivity measures over time under farming conditions. In the highly innovative salmon aquaculture industry, health records generating data available for time-series analysis is expected to become more accurate and abundant in the future, providing more opportunities for time-series regression studies.