Evaluation of the surveillance and control programme for VHS and IHN

The National Veterinary Institute has evaluated the surveillance and control programme for viral haemorrhagic septicemia (VHS) and infectious haematopoietic necrosis (IHN). The methodology used is a quantitative analysis of multiple complex data sources, and is based on scenario tree analysis and stochastic simulation. By evaluating the surveillance programme according to this method, we will have quantitative estimates for the probability of detecting disease (the system surveillance component sensitivity, SSCSe) for the various surveillance strategies. The most cost-effective surveillance strategy is the strategy that yields at least 95% SSCSe with lowest cost.

The present model shows that if surveillance is risk-based we have a high probability of detecting disease. However, the surveillance system component sensitivity is dependent on the number of samples taken within farms, and the design prevalence (i.e. the hypothetical prevalence of disease that the surveillance program is assessed against).

If the surveillance is targeted towards farms with rainbow trout, a minimum of 20 samples per farm from fish with disease signs will be needed for the detection of VHS given the farm is infected with a within-farm prevalence of 5 %. Furthermore, a number of 487 farms (a total of 9740 samples) will be needed to achieve a 95 % certainty that the programme will detect a VHS-infected farms assuming that there is at least 2 infected farms in the whole salmon farm population (design prevalence of 0.2 %). Because there are not that many farms with rainbow trout, 20% of the salmon farms have to be added in the strategy to achieve at least 95% sensitivity.

Cost considerations indicate that running PCR for VHS virus on few, large batches of samples will be more cost effective than the present programme using cell culture. However, rapid reply is not possible if samples are collected over a period of time, so the need for early detection must be considered in relation to response time.

The model presented is targeted towards VHS. Historically, IHN has been included as part of the same sampling regime as VHS. The results and conclusions in this model may therefore be relevant also for IHN but needs to be validated.