Salmonella enterica bacteraemia: a multi-national population-based cohort study

Salmonella enterica is a major cause of invasive infections worldwide [1‐7]. Although a wide range of serotypes may cause human disease, they may be broadly grouped into the typhoidal species that are specific human pathogens and includes serotypes Typhi and Paratyphi, and other serotypes that are primarily spread to humans from animal sources (non-typhoidal). In high income countries, a major risk factor for acquiring typhoidal salmonella bacteremia is travel to an endemic region [8, 9]. Foreign travel is also a risk factor for acquiring non-typhoidal salmonella infections, and several reports have highlighted the spread of resistant species globally [10, 11]. However, non-typhoidal salmonella may frequently cause human disease in high income countries and in these cases risk factors include exposure to contaminated food, extremes of age, and the presence of a number of co-morbid illnesses [12‐16].

In order to best establish the distribution and determinants of an infectious disease, population-based studies are optimal. This is because in these designs, selection bias is minimized by inclusion of all cases of disease fulfilling a case definition occurring among residents of a defined population. In addition, because the population at risk can be defined, incidence rates can be calculated that facilitate comparison among different populations and time periods. Only a few, relatively small studies have investigated the epidemiology of S. enterica bacteremia at the population level in single regions [17‐19]. The objective of this study was to define the occurrence of S. enterica bacteremia in a large international population and to evaluate temporal and regional differences.

This study reports the first novel results of the evolving International Bacteremia Surveillance Collaborative and provides new population-based information on the epidemiology of salmonella bacteremia. Our collaborative was initially formed in 2008 to address the lack of coordinated population based study of invasive bacterial diseases internationally [20]. Our present results highlight the value of such an initiative. The more than 60 million combined patient years of observation enabled us to study a large number of cases in low incidence areas. In addition, we were able to directly calculate the incidence of disease because in our surveillance areas all cases were included (sampling was not performed) and we had well-defined populations at risk for denominator data. This contrasts with non-population based studies such as hospital-based case series where sampling bias is a potentially major concern and the population at risk is rarely definable for community-onset disease [25]. Furthermore, by age and gender standardization to the EU27 population, we were able to meaningfully compare rates of disease in our surveillance regions. We observed that salmonella bacteremia is a relatively uncommon cause of bacteremia in our populations, but that considerable demographic and regional differences in occurrence exist.

The observations of significant differences among regions and by species group are important and novel. While we believe these represent true differences in incidence, a number of other factors could potentially at least in part explain these observations. Since the risk for salmonella bacteremia is influenced by age and gender, where populations differ in demographic structure then crude incidence rates may be expected to differ. However, we controlled for this possibility by age and gender standardizing our overall incidence rates. A second potential issue is that there may have been different rates of case ascertainment (identifying all culture-positive cases among residents) among our surveillance regions. While in some cases ascertainment was expected to approach or equal 100% such as with the national mandatory reporting system in Finland, this was not strictly the case in every other region [20]. However, all the other surveillance areas in this study are highly captive and estimated to be 95% or higher, such that the 2-3 fold differences in rates could not be explained by these small potential differential rates of ascertainment. A third possibility is that there may have been differences in the culturing practices among regions. This arises if the decision to sample blood for culture is significantly different among regions, and it may be expected that areas that sample more frequently may observe higher numbers of culture-positive cases. It is important to note that in all of our study regions the cost of blood culturing is fully publicly funded and available to all residents irrespective of financial means. While we cannot directly assess the issue of differential blood culturing pratices among our regions as we do not have data on negative cultures, it is notable that in previous individual publications on other common species causing bloodstream infections such as Escherichia coli and Staphylococcus aureus, similar crude incidence rates were reported [22‐24, 26, 27]. Although we may speculate as to why the incidence of salmonella bacteremia differs so greatly among our regions, further studies are needed to explore factors such as travel, food preparation practices, agricultural, and other factors [8, 28].

Our observation of increased risk for salmonella bacteremia during the summer has been previously well documented [17, 29, 30], and has been proposed to be due to increased travel abroad and lax food preparation related to outdoor cooking in that season. Several previous studies have documented increased colonization rates with salmonella species among food animals during the summer months [31‐33]. Our observation of an increased risk for non-typhoidal salmonella bacteremia with advancing age is likely at least in part due to a higher prevalence of co-morbid illnesses in the older age group that predispose to bacteremia risk [25, 34]. The excess population risk in males for non-typhoidal salmonella bacteremia is a novel finding. We speculate that this may be due to agricultural or diet exposure in males. Typhoidal salmonella bacteremia occurred in younger individuals and no gender-associated risk was evident. We suspect that this may reflect a greater rate of foreign travel to exotic typhoidal salmonella endemic regions in the younger age groups or possibly may reflect younger immigrant populations. However, while we speculate reasons as to why age and gender may modify risk, we do not have empiric data to support these suspicions. The determinants of increased risk for salmonella bacteremia in our surveillance regions merits further prospective investigation.

While this study benefits from the large, multi-national, population-based design, there are some limitations that merit discussion. The surveillance regions involved in this study are high income Western countries with largely publicly funded health systems. These results should therefore not be generalized to other dissimilar jurisdictions especially low income countries and tropical areas. In low income countries salmonella typically is among the most common bacterial pathogens and may be responsible for as many as one-half of all bacteremias [6, 35, 36] whereas they typically represent ≤1% in high income countries [18, 37]. Incidence rates for typhoid in the developing world, especially the Indian subcontinent, may be several orders of magnitude higher than that reported here [7, 36]. Second, the information that we collected was retrospective. While we used common criteria for inclusion procedures and data were abstracted using a standardized template, the routine laboratory testing procedures and methods of clinical data collection varied somewhat site-by-site. For example, only basic information was available from Finnish cases, and microbiology and clinical details from the other centers were determined using the local data sets and procedures. These included prospective review in Denmark and electronic records review and/or database extraction in other sites. In addition, ideally isolates would have been serotyped and tested for antimicrobial susceptibility in a single centralized laboratory. Finally, it would have been useful to have added information on other potential risk factors such as food and animal exposures, international travel, and co-morbid illness.

In summary, in this multi-national collaborative study we report the epidemiology of salmonella bacteremia and find regional and temporal differences in occurrence. Further studies designed to define the determinants of risk for acquiring salmonella bacteremia are warranted. This report highlights the benefit of international co-operation in bacteremia surveillance and sets the stage for future collaborative studies.