International travel and the risk of hospitalization with non-typhoidal Salmonella bacteremia. A Danish population-based cohort study, 1999-2008

International travel has increased manifold during the last decades as a consequence of increasing prosperity and reduced costs. Overseas destinations have become popular holiday resorts for Europeans, e.g., Thailand was visited by 758,000 Scandinavians in 2007 [1]. As a consequence there is increasing attention to the exposure to parasitic and viral infections including malaria, hepatitis, dengue fever, and HIV-infection [2–5]. Among travel-related bacterial infections, typhoid and paratyphoid fevers have been addressed in a number of studies [6–8]. Less information is available concerning invasive infections with non-typhoidal Salmonella enterica serovars (henceforth Salmonella) although they are common causes of gastrointestinal infections in travellers. In a recent Swedish study the highest risk of gastrointestinal salmonellosis was associated with travelling to East Africa and the Indian subcontinent [9]. In a Danish cohort study from the 1980's, 18% of Salmonella bacteremias were associated with travel [10].

European studies on travel-related risk of Salmonella bacteremia are few and either clinic-based [11, 12] or confined to a particular serovar or travel destination [13]. Therefore, the proportion of patients hospitalized with travel-related bacteremia remains uncertain and knowledge of potential risk factors and outcomes are lacking. Such data would be of value for providers of guidance to travellers and for clinical decision making. The objective of this study was to determine the proportion of hospitalized patients with non-typhoidal Salmonella bacteremia related to international travel, to assess whether the risk of bacteremia was associated with the travel destination, age or sex, and to evaluate the clinical outcome.

In this cohort study, we found that the overall incidence of patients hospitalized with travel-related non-typhoidal Salmonella bacteremia was low (0.5/100,000 person-years). Still, one in four cases hospitalized with Salmonella bacteremia in Denmark was associated with travel. The risk was most pronounced for the elderly and the young adult travellers and for travellers to countries in the regions of Sub-Saharan Africa, the Middle East and South East Asia. However, patients with travel-related infections were less likely to have poor outcomes than patients with domestically acquired infections. Our data also suggested an increasing trend in the proportion of TRBs, which is further supported by the previous Danish population-based study from the 1980s [10]. One crucial factor is probably the major reduction in the incidence of domestically acquired Salmonella infections which followed implementation of a national Salmonella control program in 1997 [19, 20]. During the last three decades the popularity of overseas travel has been increasing and together these two trends are likely to have changed the national 'balance sheet' for invasive Salmonella infections.

Our study has a number of important limitations. First, our study design only allowed for detection of Salmonella infections among hospitalized patients and our results were therefore based on estimates of the risk of being hospitalized with bacteremia and not on the risk of falling ill for which no data were available. Although, we believe that our study included all severe cases of invasive Salmonella infections, patients with a fever and a history of international travel may have been hospitalized at a lower threshold of clinical suspicion than patients with domestically acquired infections. This would lead to a potential overestimation of the proportion of TRB as compared with DABs. Second, we only included patients with bacteremia diagnosed after their return to Denmark. Travellers who were diagnosed and treated for bacteremia abroad therefore did not appear in the data. Admission and treatment at local hospitals may be more likely for travellers staying abroad for extended periods and the risk of bacteremia may therefore be biased towards destinations typically associated with shorter stays. Third, we were not able to validate the travel information given in hospital charts which relied on the clinicians' notes. If bias were introduced, it would most likely be in the form of 'missing data' on foreign travel. This would entail a potential underestimation of the proportion of TRBs. However, when considering the potential for misclassification of the travel information, we still believe that the different risk estimates reflect genuine differences in exposure.

Previous studies have shown the greatest risk of non-typhoidal Salmonella bacteremia at the extremes of age and we expected to find the same age distribution among patients with TRB [21]. Several predisposing comorbidities are acknowledged risk factors for Salmonella bacteremia, e.g. chronic or immunosuppressing conditions, including malignancy, rheumatological disease, liver disease, and diabetes [22–24], and it is likely that the elevated risk of TRB in the elderly reflected an increased prevalence of comorbidities in this age group. We used the Charlson comorbidity index as a summary measure of the patients' comorbidities, which includes 19 major disease categories, several of which are known risk factors for Salmonella bacteremia. Comparison of the burden of comorbidity in patients ≥65 years with that of patients <65 years (data not included in the results) showed that 67% (8/12) of patients ≥65 years had one or more diseases registered in the Charlson index compared with only 22% (14/64) in patients <65 years (RPP for ≥1 comorbid disease: 3.0 [95% CI: 1.7-5.6]).

A previous study suggested that accommodation, travel characteristics, and dietary hygiene are important factors for the evaluation of traveller's risk of diarrhea [25] and that the higher risk of traveller's diarrhea found in young adults may be associated with ingestion of larger volumes of potentially contaminated food and an adventurous lifestyle in this age group [26]. These behavioural factors would be important for the higher risk of Salmonella bacteremia in young adults, although the risk was only marginally higher than that reported for other age groups. A formal analysis was not possible as we had no data on behavioral factors or precise data on travel form.

Comparison of the characteristics between patients hospitalized with TRB and DAB (Table 1) illustrated that the two groups were fundamentally different. There were major differences in clinical outcomes between the groups. Thus, patients with TRB had a shorter length of hospital stay, fewer complications, and lower mortality than patients with DAB. The explanation for the latter observation is likely to be a "healthy traveller effect" as illustrated by the lower prevalence of comorbidities in the group of patients infected abroad. This is consistent with a Swedish study on Salmonella-associated deaths where persons infected domestically had a much more severe prognosis than those infected abroad [27].

Earlier studies have shown that the distribution of serovars may vary much by geographic location and time [28, 29]. Overall, S. Enteritidis was the predominant serovar in Denmark throughout the study period, even though S. Typhimurium was the most frequent serovar in the last year of the study (2008) because of an unprecedented, large nationwide outbreak [14]. S. Enteritidis was also the most common serovar in Europe and dominant in most of the travel regions [9]. This is in accordance with our finding that S. Enteritidis was the most frequent serovar both in patients with TRB and DAB. S. Virchow and S. Dublin were the second most common serovars isolated from patients with TRB or DAB, even though S. Typhimurium were a much more common serovar in most regions of the world. Both S. Virchow and S. Dublin show a higher propensity for causing bacteremia and more severe infections than S. Typhimurium [21, 30]. This suggests that the distribution of serovars was due to both the overall abundance of serovars as well as the exposure to more invasive serovars.

International travel is an acknowledged risk factor for non-typhoidal Salmonella gastroenteritis, however, little is known regarding the risk of potentially more severe invasive Salmonella infections. This study demonstrated that international travel is a notable risk factor for being hospitalized with non-typhoidal Salmonella bacteremia and the risk differs among travellers according to age and travel destination. Still, the overall risk is low and healthy travellers with bacteremia are less likely to have poor outcomes compared to patients with domestically acquired bacteremia. Our findings may be useful for the clinical evaluation of travellers presenting with fever and may also support pre-travel counselling.