ITTIOPATOLOGIA, 2008, 5: 229-230

WORKSHOP “ACQUACOLTURA MEDITERRANEA: ASPETTI NORMATIVI E SANITARI A CONFRONTO” XV CONVEGNO NAZIONALE S.I.P.I. – ERICE (TP) 24 ottobre 2008

Epidemiology in the control of fish disease L’epidemiologia nel controllo delle malattie dei pesci Ignacio de Blas Department of Animal Pathology, Faculty of Veterinary, University of Zaragoza (Spain) c/ Miguel Servet 177 - 50013 Zaragoza (Spain)

_________________________________

Epidemiology is an important tool for the control and eradication of animal diseases, and it gets more importance since approval of European Directive 88/2006, in which a risk-based epidemiological surveillance system was proposed for disease investigation for diseases of aquatic animals. Epidemiology provide theoretical basis for disease study like sampling strategy, sample size calculation, reliability of diagnostic tests, calculation of prevalence and incidence and estimation of risk factors, among others. Control of fish diseases has two basic approaches: detection of disease (as key part of surveillance and eradication programmes) and calculation of prevalence (to evaluate efficacy of a control programme). In first case we need to know if diseases (or infection) are present in a territory or a population. So we must design a surveillance system taking into account different epidemiological tools. First of all is to design an efficient sampling methodology, and in this case it is recommended a non probabilistic sampling method, since animals with more probability of infection are selected. In order to get this aim, we use epidemiological data, as susceptibility and risk factors, to increment probability of detection of a diseased/infected animal, and in consequence we establish some objective criteria for sampling, usually related with features of animal (specie, age, sex…) or environment (temperature, season…). Another important element to determine is the sample size (n), that in this case it is related with the expected minimum prevalence (normally, 2% or 5%) (really, it is the minimum expected diseases animals: d), population size (N) and confidence level (CL). It is relevant to indicate that with great populations (as occurs with fish) the sample size is more or less constant, and it is frequent to work with samples sizes of 150, 60 and 30 fish, that correspond respectively with 2%, 5% al 10% of minimum expected prevalence with a 95% of confidence level. Used formula is: 1 ⎛ ⎞ ⎛ d − 1⎞ n = ⎜ 1 − (1 − CL ) d ⎟ ⋅ ⎜ N − ⎟ ⎜ ⎟ ⎝ 2 ⎠ ⎝ ⎠

229

ITTIOPATOLOGIA, 2008, 5: 229-230

The last factor to take into account is diagnostic accuracy since we need to use the more sensitive diagnostic test (to minimize false negatives). False negatives are infected animals that we cannot to detect, and in consequence our final classification of a population could be wrong. Furthermore, we also must take into account other factors like imperfect specificity or use of pools for diagnostic. The second case to discuss is the estimation of prevalence, when we know that a disease is present in a population, we need to know the prevalence, to decide measures to implement, and its evolution, as an indicator of efficacy of disease control measures. This kind of surveys is significatively different of previous ones. Related with sampling methodology, we must use a probabilistic sampling method, in other words, all the fish in a population have the same probability to be select to take part of the sample. So the sampling strategy is usually complex, and different approaches are possible: pure random, systematic, stratification, cluster selection, multi-step sampling... It is necessary to know the characteristics of the population to select the more adequate. The second question related with sampling is the calculation of sample size that depends on expected prevalence (P), expected precision (E) and confidence level (Z, which corresponds to value of desired confidence level in a normal distribution: 1.96 for a 95% confidence level). The formula to be applied in this case is the next one:

n = Z2 ⋅

P ⋅ (1 − P ) E2

Based in previous formulae, it is possible to see that maximum sample size corresponds to a prevalence of 50% (0.5), and in consequence this is the value to use when we work with populations where prevalence is unknown. Finally in this case we need to consider also sensitivity and specificity of diagnostic test in order to understand the role of false negatives and false positives in the calculation of true prevalence. In other case, we only know the apparent prevalence, calculated as percentage of positive results in a sample. So if we use different diagnostic protocols, could be difficult to compare results. Formulae to calculate true prevalence is:

True Prevalence =

Apparent Prevalence + Specificit y − 1 Sensitivit y + Specificit y − 1

Unfortunately we don’t know currently sensitivity and specificity for most of diagnostic protocols used in fish health, and more efforts are needed in this direction. However, there are more applications of epidemiology for control of fish disease, and for example it is possible to design epidemiological surveys to detect risk factors that can be used to improve the epidemiological surveillance or to design new strategies in order to control the disease or to minimize the health impact.

230