Smittskyddsinstitutet (Swedish Institute for Infectious Disease Control, SMI), Solna, Sweden
Salmonellosis is a common infection estimated to affect three billion people and to cause 200,000 deaths every year . Salmonella enterica
is, beside Escherichia coli
, the best studied bacterium. A number of virulence factors and virulence mechanisms are suggested, among those, type III secretion system, lipopolysaccharide, intracellular survival and pathogenicity islands (SPI). However, the relevance of these virulence factors in human infections and especially infections with other serovars than Typhimurium is not fully understood .
Salmonella enterica is divided into more than 2,500 serovars . It is common knowledge that infections with some serovars, e.g. serovar Dublin, result in more severe infections than other non-typhoidal serovars [1,3]. In a study in the United States, of 540 cases of invasive Salmonella infection, 71% required hospitalisation and 5% were fatal, with 45% of deaths occurring in the age group >=60 years. These results suggest that invasive capacity is one of the most important features for causing severe disease .
There are, however, few reports regarding the virulence of different Salmonella serovars. In the study presented here, the invasive capacity of different serovars of Salmonella is analysed using the Enter-net database .
Enter-net, an international surveillance network for human gastrointestinal infections, has a database containing information on about 2 million Salmonella enterica isolates . As it includes few variables that estimate morbidity and mortality, in this study we sought to determine invasive capacity by way of comparing the ratio of blood isolates (B) to the total (T) number of faecal and blood isolates for each Salmonella serovar (B/T ratio).
The material analysed includes the strains reported to the Enter-net database between 1994 and 2004. Of these strains, 956,786 fulfilled the inclusion criteria by virtue of being human isolates and including information on both serovar and site of isolation.
The results are listed in the Table. All serovars with less than 100 isolates, except Paratyphi C, have been removed from the Table. For the majority of the excluded serovars there were no blood isolates. Serovar Newington does not exist in the Kauffmann-White scheme anymore and has been replaced by Anatum var 15+. However, even if these isolates had been included in Anatum, it would not have changed the B/T ratio calculated for Anatum.
There are different sampling and reporting systems in Europe. For example, during the period studied, Sweden reported isolates rather than cases, meaning that one case could be reported twice, if positive in both blood and faeces. This represents a selection bias and could be a potential source of error in this study.
Some conclusions, however, can be made on the basis of existing data. As expected, serovars known to cause more severe disease gave high B/T ratios, serovar Choleraesuis (33%), Dublin (41%), Paratyphi A (55%), Paratyphi C (54%) and Typhi (60%). This supports previous findings showing that the invasive capacity of a serovar is an important quality for the ability to cause disease .
We also found a difference between d-tartrate positive and d-tartrate negative Paratyphi B serovars with B/T ratios of 2% versus 12%, respectively. Approximately the same ratios were reported in a previous study .
Most of the remaining serovars have B/T ratios between 0% and 4%. It is difficult to interpret the results for these serovars without epidemiological data, especially for serovars with low number of reported isolates. For example, a Salmonella outbreak occurring in an institutional setting (i.e. hospital or nursing home) and generating hundreds of cases would result in many positive blood cultures. Other serovars, not affecting vulnerable groups, would have a comparatively lower B/T ratio. The resulting high B/T ratio would therefore potentially falsely convey that the cultured serovar had become more invasive when it was rather due to the temporal association and common methodology used to respond to the outbreak.
Still, some general observations can be made. A high number of isolates with a low ratio indicates that a given serovar possesses low invasive capacity. For the Salmonella serovars Emek and Isangi with 580 isolates and 486 isolates, respectively, no blood isolates were reported to suggest low invasive capacity for both species.
As discussed above, a high ratio does not necessary indicate a high virulence capacity, it could be due for example to outbreaks occurring among vulnerable groups. However, if the number of isolates is high and the difference in B/T ratio is large, it is more likely that these reflect a true difference in invasive capacity. Looking at two serovars with the most extreme ratios and highest numbers of isolates, Salmonella Braenderup (B/T=0.4) and Brandenburg (B/T=3.9), we see a significant difference with resulting p-value of 10–16. It is therefore likely that in both serovars we have detected a true difference in invasive capacity rather than differences in the affected populations.
In the Table, the initially missing values of 95% CI for eight serovars have been completed. The correction was made on 3 October 2007.