Gill disease surveillance

Gill disease affects fresh and marine aquaculture worldwide. In marine cultured Atlantic salmon, gill disease has been established as one of the most significant health challenges in the last decade. Amoebic gill disease (AGD) caused by the amoeba Neoparamoeba perurans is the most established, easiest recognized and only treatable type of marine gill disease. Other types are parasitic gill disease, viral gill disease, bacterial gill disease, zooplankton-associated gill disease (caused often by cnidarian nematocysts), harmful algal gill disease and chemical/toxin-associated gill disease. Especially zooplankton-associated gill disease in particular has caused severe mortality in Scottish salmon aquaculture in 2022 and 2023. When principal pathological changes in the gill are non-specific, either in combination with, or in the absence of, one or more of the seven distinctive types of gill disease (including AGD), the type of gill disease is referred to as complex gill disease (CGD). This talk uses surveillance for AGD and CGD as an example to discuss surveillance of gill disease for specific and non-specific conditions.

Disease surveillance consists of many components. Diagnostic test evaluation and sample size calculations ensure accuracy, reliability, and representativeness of collected data. Together with data collection, management, analysis, dissemination, information technology, and collaboration, they form a comprehensive framework that can enable effective disease monitoring and decision-making. A variety of diagnostic tests is used for surveillance of AGD and CGD. Common tests for AGD are gross gill scores, histopathology, molecular techniques, and immunofluorescence antibody tests. For CGD there are fewer options because the case definition includes a variety of infectious agents or no primary pathogens in the case of environmental insults. Commonly used tests are therefore less pathogen-specific gill scores and histopathology. These tests have different characteristics, including expense, whether the test is lethal, type of information they provide, level of expertise needed to carry out the test and test performance.

Which of these test(s) to include in a surveillance strategy is a complex process, which often starts with understanding how well tests perform, i.e. determining diagnostic sensitivity and specificity. Reason is that a test with a poor performance may result in numerous false positives and negatives – thereby confusing interpretation of results, and potentially leading to suboptimal mitigation strategies which can turn out to be very costly in the long term. Gross gill scores are medium to good in designating truly diseased fish and excellent in designating non-diseased fish. These non-lethal tests for AGD and CGD are frequently, usually weekly, performed on all salmon sites in Scotland at the same time as mandatory lice counts to minimize discomfort to fish. The tests are quick and easily performed by lifting the operculum of a sedated fish and visually observing all 8 gill arches, then giving one score between zero and five for the entire fish based on a standardized scoring criterion. Because of these characteristics, gross gill scores are a valuable tool for general field surveillance, and that can be augmented by other tests when there is a need for more information or increased test performance. Histopathology is a lethal test that requires a high level of technical expertise that is often not available in-house by salmon producers and therefore outsourced to diagnostic companies. Histopathology is often considered the gold standard (reference standard) for any type of gill disease because it is accurate and provides an understanding of the state of the fish. For example, if lesions are focal or diffuse, whether there is an acute inflammation or a more chronic condition, and which type of pathogens may be involved. It is difficult to accurately estimate diagnostic sensitivity and specificity of histopathology, because in scientific studies usually the second gill arch on the right is used. This leads to missed abnormalities when they occur elsewhere, leading to false negatives. Molecular PCR tests are very valuable non-lethal tests that can be used when a specific pathogen is targeted, such as N. perurans in the case of AGD, but not when no specific pathogens are always involved, such as for CGD. Generally, it is important to develop understanding of diagnostic sensitivity and specificity of tests for test selection. However, these test characteristics can also be used to estimate true prevalences and predictive values, which can help enrich surveillance strategies.

Another important component of surveillance is the sample size. Mostly it is much more convenient and economical to examine a sample of a population rather than conduct a census, therefore we sample a subset of a population. In a census, the only error is in the measurement itself; with samples, error can arise from both measurement and sampling error. By designing a well-planned sampling strategy, the information obtained can be almost as good as information obtained from a census, when the sample size is correct. In aquaculture, a sample size consideration is rarely as simple as “how many fish?”, because fish are grouped in pens and pens in sites, and individual fish characteristics within a pen are more similar to each other than those of fish in different pens. In addition, for gross gill scores, the variation between fish is not similar at all levels of gill scores; at higher gill scores there is more variability compared to that at lower gill scores. Using intra-correlation coefficients, we developed a decision support tool that considers the variance-covariance components so that sample sizes can be estimated that take the structure of the system into account (https://epidemiology.sruc.ac.uk/shiny/apps/gillhealth/). Evidence based sample size calculators, such as the one described, are not meant to be used on a day-to-day basis but are designed to assist when a sampling protocol is developed, or an existing protocol reviewed. They ensure that a representative and statistically meaningful amount of data is collected, increase the precision of estimates, and reduce the likelihood of errors and biases. This ensures that the surveillance findings are robust. They also help optimize resource allocation because they enable cost-effective data collection, while minimizing wastage of resources.

Other essential components of surveillance include data collection methods, data management pipelines, analysis, and dissemination. Data collection methods must be standardized and consistent to ensure data quality. This is often challenging in the context of salmon aquaculture because companies own multiple farms and thus time and effort is required to ensure consistency between personnel at different locations. Online reoccurring courses can help improve standardization. Data management involves secure storage, timely entry, and rigorous quality control. Analysis of collected data helps identify trends and potential outbreaks, supporting decision-making. When such analysis includes multiple producers, it is crucial to consider case definitions and ensure that models include structure, e.g. through random effect components, so that between-company variability is considered. Finally, sharing surveillance findings with relevant authorities and the public helps dissemination, and promotes transparency and awareness where deemed appropriate.