Descriptive epidemiology

The total number of confirmed cases reached 100 and was distributed throughout 18 of Sweden’s 21 regions. The largest peak in disease onset of cases was during week 34 in August. Subsequently, smaller peaks followed in weeks 37 and 40 (Figure 1). The age and sex distribution indicated that a higher proportion of confirmed cases were female (65%), and that the highest proportion of cases was observed among individuals aged 0–10 years (17%) and 41–50 years (20%) (Figure 2).

Figure 1. Confirmed cases of Salmonella Typhimurium sequence type 36 in an outbreak linked to alfalfa sprouts, by week of symptom onset and cluster, and timeline of outbreak events, Sweden, August–November 2024.

Figure 2. Confirmed cases of Salmonella Typhimurium sequence type 36 in an outbreak linked to alfalfa sprouts, by age group and sex, August–November 2024.

Matched case–control study

In total, 11 food items were included in the matched case–control study. Seven of the items were reported by >50% of confirmed cases in the trawling questionnaire (tomatoes, cucumber, carrots, apples, minced beef, yellow onions, peppers) and four items were reported by >30% of confirmed cases and had a significantly higher odds ratio in the case–case comparison (sprouts and shoots, salami, goat cheese, iceberg lettuce) (Supplementary Table S1).

A total of 32 confirmed cases and 466 controls responded to the case–control questionnaire, giving a response rate of 48% and 62%, respectively. A total of 421 controls were included after excluding controls reporting travel abroad or gastrointestinal symptoms. Characteristics of the study population and results prior to re-categorization can be found in Supplementary Tables S2 and S3, respectively. Results from conditional logistic regression models by each exposure item, where the confirmed cases and controls were matched on age, sex, and region, showed that alfalfa sprouts and pea shoots had higher odds of consumption in cases compared to matched controls (aOR = 7.66, 95% CI = 2.75–21.3 and aOR = 10, 95% CI 2.96–34.1, respectively) (Table 1). We performed multivariable analysis including alfalfa sprouts, pea shoots, cress, alfalfa seeds, bean sprouts, apples, and carrots. One case and 174 controls were excluded from the regression analyses due to the absence of a matching case or control. After backwards stepwise regression, only alfalfa sprouts and carrots remained in the conditional regression model, with alfalfa sprouts having an adjusted OR >1, both in the main analysis (aOR = 8.94, 95% CI = 2.96–27.1), and the sub-analysis excluding children (aOR = 11.3, 95% CI = 3.28–39.2).

Table 1. Number and percentage of exposed cases and controls, and odds ratios in univariable and multivariable models in an investigation of an outbreak of Salmonella Typhimurium sequence type 36 linked to alfalfa sprouts, Sweden, September–November 2024

^* Conditional regression models are adjusted for age group, region, and sex.

^† Adjusted odds ratio.

Microbiological investigation

Isolates from the 100 cases were distributed over 14 genetically related strains of S. Typhimurium ST36, in nine clusters and five individual strains, with an overall maximum SNP difference of 53 SNP (Figure 3). We observed temporal differences in the symptom onset date for cases with isolates belonging to the different microbiological clusters (Figure 1). The 11 cases from the case–control study who reported sprout/shoot consumption were distributed across clusters: 1 (n = 1), 3 (n = 2), 4 (n = 3), 5 (n = 1), 6 (n = 2), 7 (n = 2), and were observed during weeks 36, 37, 38, 40, 41, 42, 44, and 45, in seven different regions.

Figure 3. Minimum spanning tree of Salmonella Typhimurium sequence type 36 isolates from outbreak cases, Sweden, August–November 2024 (n = 100). Single-nucleotide polymorphisms (SNPs) are shown next to the branches (branch length is not proportional to genetic distance). A representative sequence from the outbreak was used as a reference. Recombinations are marked “R.” Colors demonstrate clusters and individual strains.

Traceback and environmental investigation

The Swedish fresh sprout market is dominated by two cooperating, geographically adjacent companies. Traceback analysis from two restaurants linked to four cases, in addition to a receipt from one case, confirmed that the cases had consumed sprouts produced by both of these major companies. Both producers routinely use the same batches of seeds, which undergo decontamination through hot water treatment (78 °C for 15 s) before sprouting. Inspection of the pasteurization process did not identify any deviations. Cultivation takes 7–8 days, and the resulting sprouts have a shelf life of 8 days. In May 2024, a 10-t batch of alfalfa seeds from an Italian supplier was delivered to one of the two companies. Sprouting of the alfalfa seeds from this batch began in the latter half of July 2024 and occurred continuously throughout the summer and autumn. At the time of sampling, two out of 10 tonnes of alfalfa seeds remained from the original batch that had not yet been sprouted or distributed. Analysis of Salmonella and STEC was performed without detection of either pathogen according to the requirements in Commission Regulation (EC) 2073/2005 on microbiological criteria for foodstuffs.

Outbreak control measures

PHAS first published information about the outbreak on its website on 12 September 2024. It was not observed that the outbreak was continuing throughout the autumn until late October, when additional genetically-related clusters were detected. On 8 November 2024, PHAS communicated that the outbreak was ongoing. Findings indicating that alfalfa sprouts were the likely source of the outbreak were communicated to the SFA, who then contacted the competent authority, the CAB, for the two major seed-producing companies in Sweden. After the inspection performed by the CAB on 21 November 2024, the two companies recalled the remaining alfalfa sprout products from the implicated batch on 26 November 2024. After testing of Salmonella and STEC with negative results, the two companies began distributing alfalfa sprouts again, using a new batch of seeds from a different supplier on 18 December 2024. Since the recall, no additional cases from this ST36 cluster have been reported, and the outbreak was officially declared over in Sweden on 23 December 2024.

Discussion

We describe an outbreak involving multiple distinct but genetically related clusters of S. Typhimurium ST36 that occurred August–November 2024 in Sweden and involved 100 confirmed cases. Epidemiological, microbiological, and traceback investigations support that contaminated sprouted alfalfa seeds were the source of the outbreak, despite the absence of detection of Salmonella from sprouted seed samples in Sweden. The outbreak ceased after the recall of sprouts produced from the implicated seed batch and the switch to a different seed supplier. Our findings are further corroborated by a Norwegian multi-serovar outbreak investigation identifying the same source [Reference Rakover21]. Further, multiple outbreaks and findings of Salmonella of additional serovars and strains in Finland, Spain, Germany, the Netherlands, and Italy have been linked to the same Italian region, according to reviewed EpiPulse and RASFF notifications, and the ROA published by ECDC and EFSA [20]. The multinational scope of these findings, combined with the involvement of multiple Salmonella serovars and strains, suggests widespread and persistent contamination during seed production. Communication and sharing of independent findings with Norwegian colleagues and others were essential for strengthening our conclusion as to the source of the outbreak and for identifying the wide scope of this contamination and accompanying outbreaks.

We observed temporal trends both in terms of the timing of the outbreak in relation to the delivery of the implicated batch, as the Swedish outbreak began shortly after cultivation started, and in terms of the sequential pattern of separate ST36 clusters detected over time. We hypothesize that this pattern may be due to seed production and widespread contamination at the seed supplier. The implicated batch was delivered to Sweden in May 2024 in multiple bags, from which the seeds were sprouted sequentially, beginning at the end of July and continuing throughout the following months. The separate bags in which the seeds were stored may have contained different strains of ST36, and their usage may align with the temporal trends in cluster detection. However, the majority of the seeds and bags did not remain at the time of the study, so this hypothesis cannot be confirmed. The lack of findings of Salmonella from sprouted seed samples may be explained simply due to the large size of the batch and a potentially uneven or sporadic distribution of bacteria.

The involvement of alfalfa sprouts in particular created limitations for the epidemiological investigation. Sprouts are typically used as a garnish, likely leading to poor recall among cases, or in some instances, cases may have been unaware that they consumed sprouts at all. Recall may have been especially difficult for parents and/or guardians reporting consumption on behalf of their child, as they may not be aware of what their child ate while not at home, and garnishes and/or side salads are often not written on printed school menus. Exposures may have been misclassified as a result. As such, approximately one-third of cases reported sprout or shoot consumption in the trawling questionnaire. The low recall observed is somewhat expected and aligns with past sprout-associated outbreaks [Reference Werner13, Reference Rimhanen-Finne14, Reference Harfield22]. Nevertheless, the regional CMOs reported that sprout consumption among cases was unusually high. This was further corroborated by PHAS in a case–case comparison with other sporadic Salmonella cases observed in the same time period. Recall bias due to the timing of the study is also a limitation that may have led to underestimated odds ratios, as cases were asked, in some instances, to recall food consumption several weeks in the past. To mitigate recall bias, we limited inclusion to cases reported after 1 September 2024. This decision was influenced by the need to compare cases and controls from the same population and season, ensuring that dietary and other relevant exposures were comparable. However, we acknowledge that this exclusion criterion may have introduced selection bias among the cases due to seasonal differences. In the case–control study, reported sprout consumption was distributed across time and did not indicate any clear cluster-specific pattern. Additionally, results indicate an association with carrot consumption. While it is possible that carrot consumption may be linked to an unmeasured variable that could explain this, we suspect that this may be a spurious association, as there is no clear mechanism or prior evidence suggesting why carrots would have a protective effect in this context. Furthermore, the use of a web panel for control data selection may have induced selection bias, as respondents may be more prone to participating in surveys than the population that gave rise to the cases, and re-coding of ambiguous and missing responses may have introduced misclassification bias. Together, these limitations may affect both the internal and external validity of the study.

The findings from this outbreak have implications for future food safety measures and outbreak investigations. There are multiple possible points of contamination along the sprout production chain, for example, from contaminated water, soil, storage units, or improper handling at the sprouting facility [6]. Our findings suggest widespread contamination at or before the seed supplier’s facility. The diversity of genetically similar yet distinct strains of ST36 suggests that this contamination also had time to evolve. Evolution during seed storage is less likely due to the water-stressed conditions present in dried seeds that may inhibit bacterial growth [Reference Morasi23]. Instead, the number of different serovars identified in other countries suggests that the contamination originates from several sources or, alternatively, a single source with a mixture of contaminations, such as sewage. We recommend that future outbreak investigations consider multiple serovars or pathogens from the same source, particularly in open-environment food production systems, as it could enhance root cause analyses and lead to more effective prevention and control measures. Furthermore, our results point to difficulties in decontaminating seeds as Salmonella repeatedly remained in finished sprouts despite inspection of the pasteurization process not revealing any deficiencies. Previous studies have highlighted this challenge, particularly in regards to Salmonella, as it is resilient and can survive in products with low water activity [Reference Waldner24, Reference Yao, LiBrizzi and Chen25]. This, together with the other sprout-related outbreaks mentioned above, underlines the importance of avoiding contamination at the primary seed producer and highlights the persistence of sprouts as an international food safety concern [Reference Taormina, Beuchat and Slutsker26]. We emphasize the need for primary producers of alfalfa seeds and sprouts to identify and prevent possible contamination points.