Based on date of laboratory confirmation of diagnosis, the current bluetongue serotype 8 epidemic is now into its 10^th week, with the case total having risen to 1196 herds. The vast majority of cases having been reported from Belgium, Germany and the Netherlands, and over half of them within the last 3 weeks in October. The spatial spread has continued, but visual analysis suggests the appearance of three distinct clusters, which include the original focus in the area where Belgium, Germany and The Netherlands share borders, the west of Belgium, and an area around the Bergisches Land in Germany. It needs to be emphasized that this observation has to be interpreted cautiously as this observed clustering may be the result of spatial variation of the population at risk. 

Until very recently, bluetongue virus (BTV) had not been reported from the more northern parts of Europe, and serotype 8 had not previously been reported in the EU. The epidemiological mechanisms underlying the current bluetongue epidemic are unclear, including its origin and the geographic, climatic, and husbandry factors influencing the establishment and the spread of this disease. EFSA has been requested by the European Commission to investigate these issues through an epidemiological analysis of the surveillance data for the region affected by the epidemic. One aspect of the epidemiological analysis is a descriptive analysis of the spatial and temporal disease pattern during the progression of the epidemic. The findings are reported in this Bulletin, which is to be updated at weekly intervals in the light of the epidemic.

It should be noted that at this stage the analyses are of a descriptive nature, and it is not yet possible to draw causal inferences, i.e. to determine which factors affect for example the spread of the disease. Due to the impact of several biases in the data, which are further discussed below, caution has to be applied when using these findings for developing causal hypotheses.

The results indicate that the disease was first confirmed in mid August 2006 in an area where Belgium, Germany, and the Netherlands are sharing borders. Since the disease was previously absent in this part of Europe, farmers’ and veterinarians’ awareness of its clinical symptoms was likely to be low in this initial period, resulting in an unknown level of underreporting bias.

• epidemic curves which show the daily number of cases, by country (Figure 1a) and overall (Figure 1b) for cattle and small ruminants. This is not shown for France as it only had five confirmed cases by October 24, 2006;
• an animated point map with the weekly number of cumulative and new cases (Figures 2a through 2j); and
• maps showing the monthly cumulative number of cases for all species (Figure 3a) and by animal species (Figures 3b and 3c).

Spatial pattern

The temporal pattern of the epidemic has been accompanied by geographical spread beyond the region originally affected, with new cases being reported over a steadily widening geographical area (see Figure 2a-j and Figure 3a). From the original cluster of cases in the area where , the and share borders, disease seems to have extensively spread along an east-west axis with distinct new clusters forming in the west of and around the Bergisches Land region of . Limited spread has occurred in a northern direction, and very little towards the south from the original focus. In addition, five cases were identified in the north of . It is notable that the original focus of infection continues to produce cases, although at a smaller rate. The spatial patterns for large and small ruminants appear to largely mirror each other (see Figures 3b-c).

It needs to be emphasized that any apparent spread or lack thereof deduced from these data will be affected by an unknown level of underreporting bias, as well as varying temporal lags between occurrences of infection, its detection, and its subsequent confirmation. It is also to be noted that observed clusters of high or low numbers of cases may well be a reflection of variation in the population at risk for the animal species affected.

The following specific sources of underreporting bias have to be considered:

• Animals, especially cattle, may not show clinical signs.
• Animals with clinical signs may recover rather quickly and are therefore not detected by the farmer. Furthermore, in many cases sheep are kept under extensive production systems and therefore not observed daily by farmers.
• Clinical cases may not be reported in order to avoid movement restrictions that are imposed in affected herds.

Sub-clinical cases can be detected through serological screening programmes. This is illustrated by the reports from France where 3 of the 5 case herds were identified through serology at a time when none of the animals in these herds showed clinical symptoms. Similarly, animals that were exposed to the virus can also be detected when tested serologically for certification prior to trade or export.

The Working Group is currently organising access to additional data sources such as livestock population denominator data, so that more detailed epidemiological analyses can be conducted and therefore the effect of some of the aforementioned biases can be reduced.

The bluetongue disease data used in this analysis are obtained as follows:

• Generally, suspicious cases are identified by the veterinary practitioner who has been consulted by the animal owner.
• The cases¹ of bluetongue that are presented in this bulletin represent herds in which one or more clinically-suspected animals were confirmed positive using a laboratory test (Polymerase chain reaction (PCR) or serology) and that were notified to the veterinary authorities.
• Subsequently, the EU Member States’ veterinary authorities notify the Commission and the Member States of disease outbreaks, in accordance with Directive 82/894/EEC.
• These notifications are entered in the European Commission’s Animal Disease Notification System (ADNS) (http:/ec.europa.eu/food/animal/diseases/adns/index_en.htm).

From the ADNS database the following data fields were used for each case: the country name, the date of confirmation, the geographical coordinates, and the animal species. The data for these four fields were available for this bulletin up until 24 October, 2006 (extracted 25 October, 2006).

The geographical references were verified prior to their use by checking for obvious inaccuracies. Missing or implausible geo-references were excluded from the analysis, except for data missing only at the highest level of precision (seconds). Cases with incomplete or implausible information are brought to the attention of the European Commission which is seeking clarification from the Member States for these data.

Justification for using confirmation date for this analysis

Different dates² can be recorded in the ADNS system:

• the date of first infection,
• the date of clinical suspicion - which is normally the date of sampling for laboratory analysis, and
• the date of confirmation of the laboratory diagnosis.

Whereas the actual date of first infection is unknown, the date of first clinical signs observed by the farmer could potentially be used, if obtained as part of a formal outbreak investigation. However, these data are not always available when the notification to the ADNS is made. Since the date of confirmation is always provided, it was therefore decided to use it for the purpose of the weekly reporting in this Bulletin.  

The data utilised by the EFSA Bluetongue Epidemiology Working Group are the result of the work of a large number of dedicated individuals, including animal owners, veterinary practitioners, experts in diagnostic and other aspects of bluetongue disease, and national and European-level veterinary services.

The EFSA Bluetongue Epidemiology Working Group consists of a group of experts from all Member States which have been invited to join. It is chaired by D. Pfeiffer (RVC, UK).

• 25 October 2006