Background
In mid-2002, the Nationale Referenzzentrale für Salmonellen (NRCS) noticed a regional cluster of human Salmonella enterica subsp. enterica ser. Enteritidis phage type 5 (S. Enteritidis PT5) infections (1) in a region consisting of two neighbouring districts (combined population of the two
districts: 235 000; population of Austria: 8 031 000). From November 2001 to August 2002, 61 human isolates of S. Enteritidis PT5 were ascertained in these two districts (figure 1).
There had been a small outbreak of S. Enteritidis PT5 in this region in 1999. A least 8 people had been infected after eating cake that had been left unrefrigerated for about five days. The eggs used in the preparation of this cake had been traced back to a local egg producer. The reappearance of S. Enteritidis PT5 in 2002 in the same two districts was a reason to reinvestigate the farm involved in the 1999 outbreak.

Material and methods
The NRCS receives almost all human and non-human Salmonella isolates from Austrian laboratories. No preselection is done by the dispatching laboratories. The strains are accompanied by basic epidemiological data which includes date of receipt by the primary laboratory, patient's place of residence, etc. All S. Enteritidis strains are phage typed according to international standards (2).
In Austria human outbreaks are investigated by local health authorities. In case of outbreaks of major interest the NRCS collaborates actively with local investigators.
In September 2002 a questionnaire asking about symptoms and a basic food history in the three days prior to infection was sent to the patients. At that time, the only available information about the S. Enteritidis PT5 outbreak in 1999 in the same region was anecdotal. Therefore the Österreichische Qualitätsgeflügelvereinigung (QGV) - Austrian Poultry Health Service - APHS) was contacted to locate the farm involved in the 1999 outbreak and to gather information about both the 1999 and the current outbreaks. When the farm had been located, voluntary bacteriological examinations of the farm started.

To exclude the possibility that the farm had been contaminated with S. Enteritidis PT5 by the recently introduced hens, the APHS also gathered all available information about the status of the rearing flocks of the poultry breeders who had supplied the flocks to the farm.
When the evidence concerning the current outbreak was sufficient to incriminate this farm, the farmer immediately stopped selling table eggs from the farm.
To assess whether or not there had been a real increase of human Salmonella infections in the region, the number of isolates ascertained during the peak of the outbreak (calculated by adding together all isolates identified during the months of July, August and September) were statistically compared with the retrospective numbers from July to September in previous years (1998 - 2001), assuming a Poisson distribution.

Results
Analysis of the returned questionnaires did not give conclusive results about the origin of the outbreak. Some human infections of S. Enteritidis PT5 reported outside the two districts were related to travel in the outbreak region. The returned questionnaires also revealed that one person had died.
The farmer had destroyed his flocks following the 1999 outbreak, and the poultry sheds had remained unoccupied for several weeks afterwards. During this time, the poultry sheds were disinfected thoroughly using formalin gas. During the years that followed, only vaccinated flocks were introduced in the farm. An inactivated vaccine against S. Enteritidis was used (TALOVAC SE, Lohmann Animal Health, Germany). In the summer of 2001, however, the farmer decided to stop vaccinating his flocks. The first flock not to be vaccinated against S. Enteritidis was introduced into the farm in August 2001, and a second flock was introduced at the end of April 2002. Both of these flocks comprised about 6 000 hens. The output of table eggs from the farm was about 1.7 million eggs per year (annual egg consumption in Austria is around 1 870 million eggs). The hens were kept in cages. The eggs were sold locally by the Austrian system of direct marketing (for example, in local market places, and to restaurants).

The untreated table eggs were withdrawn from sale at the end of September 2002 after S. Enteritidis PT5 was found in eggs from the farm. All eggs from this farm were pasteurised thereafter before being used in industrial food production.
In September 2002, in collaboration with the owner of the farm, bacteriological samples were taken from the farm. S. Enteritidis PT5 was found in pooled faecal samples, on eggshells, and in swabs from the room where eggs were stored before leaving the farm.
There was no evidence for S. Enteritidis PT5 infection in hens from the same rearing flocks as the outbreak flocks that were sold to other farms. Most of these flocks were tested (independently from the described outbreak) for Salmonella, but none of these samples tested positive. The two rearing flocks were repeatedly tested for Salmonella (meconium after hatching, pooled faeces in the tenth and fourteenth weeks of rearing, swabs taken from the devices used for the transport of hens). These bacteriological examinations also showed no growth of Samonella.

Retrospectively, the first sporadic cases of S. Enteritidis PT5 were apparent by the end of 2001. A higher number of isolates from the two districts were send to the NRCS from July 2002 onwards. From January to the end of October 2002, 70 human isolates of S. Enteritidis PT5 were recognised in the two districts. Sixty eight isolates were derived from faecal samples, one from a blood culture and one from an abdominal wound swab. Thirty two patients were female, and 38 male. All age groups were represented; the median age was 38 years. One patient is known to have died from S. Enteritidis PT5 infection.
There had also been human infections with other types of Salmonella in the two districts. In figure 2, S. Enteritidis PT5 infections are compared to non-S. Enteritidis PT5 infections. When the total number of isolates (S. Enteritis PT5 and non S. Enteritidis PT5) in the major outbreak months (July-September 2002) was compared with the preceding years, a statistically significant increase (p < 0.001) was observed. The expected number of cases was 58.5, compared with the 132 actually observed.

Untreated table eggs from the farm were withdrawn from sale at the end of September 2002. In October 2002 only one isolate of S. Enteritidis PT5 was reported in the two districts, although the overall number of Salmonella isolates in Austria was high and comparable to the three preceding months.

Discussion
The problem of re-infection of new flocks placed in previously infected chicken sheds is well known. The 2001 annual zoonoses report from Denmark mentions nine egg producers with repeated infections after cleaning and disinfection (3). After excluding the possibility that S. Enteritidis PT5 had been introduced by primarily infected hens (S. Enteritidis PT5 was not found in the farms where the hens had been reared originally from the same rearing flocks) it is most likely that the infection of the flocks in 2002 resulted from S. Enteritidis PT5 contamination in the environment of the farm. Presumably there had been niches in the farm (poultry litter conveyer belts, rodents, etc.) where S. Enteritidis PT5 had survived. As long as the flocks were vaccinated against S. Enteritidis, however, the hens (and their eggs) were protected.
Vaccination of poultry may reduce the contamination of the environment of a farm with Salmonella (4). It is tempting to believe that this effect may eventually contribute to the eradication of Salmonella from a farm. This outbreak demonstrates the consequences of the persistence of Salmonella.

Nevertheless, the near total disappearance in 2000 and 2001 of human cases of S. Enteritidis PT5 in the region after the implementation of vaccination is an example of the effectiveness of vaccination in preventing human infections under field conditions (figure 1).

Laboratory confirmed Salmonella isolates represent only a small proportion of all human infections. A recent report from the Netherlands states that between 1996 and 2000, only 3729 (6.9%) out of 53 500 estimated cases of human gastrointestinal salmonellosis were laboratory confirmed (5). In the United States the degree of underreporting has been estimated at around 38-fold (6). It is therefore realistic to assume that several hundred infections occurred in the course of this outbreak in the space of only a few months. This is noteworthy, because of the relatively small amount of table eggs sold by this producer. It emphasises the importance of eggs as vehicles for human infections.
Occurrence of 'extra' cases of S. Enteritidis PT5 infections, supported by the statistical comparison of the total number of isolates in the region from July to September 2002 with the corresponding figures of the years 1998 - 2001, would probably have been preventable if the farmer had continued to vaccinate his flocks or if Salmonella had been eradicated from the farm after earlier clean up.