Outbreak of Salmonella enterica serotype Poona in infants linked to persistent Salmonella contamination in an infant formula manufacturing facility, France, August 2018 to February 2019

We describe a Salmonella Poona outbreak involving 31 infant cases in France. Following outbreak detection on 18 January 2019, consumption of rice-based infant formula manufactured at a facility in Spain was identified as the probable cause, leading to a recall on 24 January. Whole genome sequencing analysis linked present outbreak isolates to a 2010–11 S. Poona outbreak in Spain associated with formula manufactured in the same facility, indicating a persistent source of contamination.

We describe a Salmonella Poona outbreak involving 31 infant cases in France. Following outbreak detection on 18 January 2019, consumption of rice-based infant formula manufactured at a facility in Spain was identified as the probable cause, leading to a recall on 24 January. Whole genome sequencing analysis linked present outbreak isolates to a 2010-11 S. Poona outbreak in Spain associated with formula manufactured in the same facility, indicating a persistent source of contamination.

Outbreak alert
On 18 January 2019, a cluster of four isolates of Salmonella enterica serotype Poona (S. Poona) was identified in infants under 1 year of age. The isolates were received from 20-24 December 2018 at the French National Reference Centre for Escherichia coli, Shigella and Salmonella (NRC-ESS, Institut Pasteur). Initial investigations identified consumption of the same brand of rice-based powdered infant formula (Brand A formula).
On 21 January 2019, the Directorate of the Ministry of Economy in charge of consumers' affairs (DGCCRF) identified that the Brand A formula had been manufactured at a single production facility in Spain, Facility X. After additional isolates of S. Poona in infants consuming Brand A formula were identified, the French distributer initiated a recall of all Brand A formula manufactured at Facility X on 24 January ( Figure 1) [1].

Epidemiological investigations
For the investigation, cases were defined by Santé Publique France (SpFrance) according to the case definition in the box.
On 18 January 2019, after the initial alert, the families of three infants were interviewed by telephone by SpFrance. A standardised Salmonella questionnaire was used, which included questions relating to the consumption of foods and drinks e.g. infant formula, as well as methods of bottle preparation and exposure to reptiles, a known risk factor for S. Poona [2]. Two of the three infants had consumed different products of Brand A formula. Between 18 and 23 January, the NRC-ESS identified a total of nine S. Poona isolates in infants: four belonging to the same genomic cluster by single nucleotide polymorphism (SNP) analysis and core genome MLST (cgMLST) analysis and five infants with S. Poona isolates undergoing cluster analyses. Interviews of parents/guardians for seven of these infants (two families were unreachable) confirmed consumption of Brand A formula.
As at 28 March 2019, 30 confirmed cases and one possible case have been identified across 11 regions in France. S. Poona infection from suspected person-to-person transmission was identified in an older sibling (> 36 months) of a confirmed case but the child was excluded from the analysis as they did not fulfil the case definition. The median age of the 31 cases was 10 months (range: 2-28 months) including 13 girls and 18 boys. Week of Salmonella isolation ranged from week 34 2018 to week 8 2019. From 18 January to 13 March, the parents/guardians of 29 cases were interviewed. Symptom onset ranged from week 34 2018 to week 6 2019 (one case consumed formula after the recall as caretakers did not immediately receive the information) ( Figure 2). No cases reported symptom onset after 8 February 2019 and the outbreak is considered over. All cases had diarrhoea, with 13 of 29 cases reporting bloody diarrhoea; 27 cases had fever (> 38°C). Thirteen cases were hospitalised, with two cases having underlying medical conditions. All infants were recovering or had recovered at the time of interviews.
Parents/guardians all reported using Brand A formula during interviews. Rice based formula is usually recommended for lactose intolerant infants. Brand and batch numbers for formula consumed or purchased in the week before symptom onset were collected during interviews for trace back and microbiological investigations. Three different Brand A formula products were identified: first age formula for infants aged 0-6 months (two cases), second age formula for infants aged 6-12 months (standard format; 16 cases) and 6-36 months (anti-regurgitation format; 11 cases). No other common food items, including bottled water, were identified. Practices for bottle preparation were according to recommendations and did not favour bacterial proliferation.
Sixteen countries responded to an urgent enquiry in the European Centre for Disease Prevention and Control (ECDC) Epidemic Intelligence Information System for Food-and Waterborne Diseases and Zoonoses (EPIS-FWD) posted on 21 January and two additional confirmed cases were identified, one in Belgium and one in Luxembourg (ages 9 and 27 months, respectively). Both cases had consumed Brand A formula, the former  Case interviews were conducted with the parents/guardians on behalf of the infant cases.
purchased at a pharmacy in France and the latter on the Internet. alleles (HC5). Two other human isolates submitted by the NRC in Belgium and the National Health Laboratory in Luxemburg also belonged to the same SNP and cgMLST HC5-164707 cluster as the French outbreak isolates ( Figure 3). All studied genomes were deposited into EnteroBase and raw reads of two representative outbreak isolates (201811387 and 201900647) were also deposited to the European Nt Archive, under study accession number PRJEB31267.

Control measures and product investigations
Isolates from five infant and infant formula samples collected during the 2010-11 S. Poona outbreak in Spain were sequenced by the Spanish National Reference Laboratory (Instituto de Salud Carlos III, Madrid) and submitted to Enterobase for comparison with the French outbreak isolates. Two S. Poona isolates from infants in Spain collected in 2018 were also available for comparison. The cgMLST (HC20-44730) and SNP (Figure 3) analyses confirmed that the 2010-11 Spanish isolates were related to the French outbreak isolates, while the 2018 isolates were unrelated.

Discussion
A total of 33 S. Poona cases were identified in this outbreak. While half of the cases were hospitalised for their salmonellosis, fortunately no serious illnesses were reported. Early detection of S. Poona isolates alerted SpFrance to a potential outbreak, which resulted in an immediate investigation of subsequent cases and rapid implementation of control measures, with a recall of the suspected product 6 days after outbreak detection. This outbreak illustrates the sensitivity of the French Salmonella surveillance system, which is based on routine WGS of human isolates received at the NRC-ESS.
In 2008, an outbreak of S. Give in infants in France was also linked to infant formula manufactured at Facility X [8,9]. More recently, in 2010-11, an outbreak of S. Poona in Spain, affecting more than 280 infants, was linked to infant formula manufactured at the same facility [10]. The brands of infant formula identified in these two outbreaks were different from Brand A. While no Salmonella contamination has been confirmed in Brand A formula or in environmental samples from Facility X in 2019, the results of the epidemiological investigations support the hypothesis that Brand A formula is the origin of the outbreak. Furthermore, genomic analyses confirm that the current outbreak and the 2010-11 S. Poona outbreak isolates are related.  Investigations are ongoing at Facility X in an effort to identify the source of contamination.
Similar outbreaks linked to contamination of powdered infant formula (PIF) by a variety of Salmonella serotypes have been reported in several countries [10][11][12][13][14][15]. This is the fourth outbreak of Salmonella linked to PIF in France and the second in less than a year for which investigations strongly suggest a persistent source of Salmonella contamination inside a manufacturing facility [8,16,17].
The current outbreak and the S. Agona outbreak in 2017 in France highlight the risk of persistent Salmonellacontamination in PIF manufacturing facilities and demonstrate the importance of WGS in identifying recurrent outbreaks. Persistent contamination in a production facility has also been reported in an outbreak linked to dry cereal [18]. PIF is not a sterile product and Salmonella contamination, in particular at low levels, can occur [19]. A 2008 report by the French Agency for Food, Environmental and Occupational Health and Safety (ANSES) evaluated the contamination risk for PIF and the difficulties associated with detection [20]. Low level or point source contamination, uneven distribution of Salmonella contamination in PIF and the efficacy of sampling plans can all impact detection [19,20]. Additionally, factors related to environmental persistence are still not fully understood. ANSES recommended that product sampling be coupled with environmental sampling as testing the product alone is not sufficient to detect potential contamination.
While in previous outbreaks in France, S. Agona in 2005 [17] and S. Give in 2008 [8,9], Salmonella was detected in PIF, this most recent outbreak illustrates the continuing challenge of detecting low level Salmonella contamination. Despite testing PIF consumed by cases as well as environmental and additional PIF samples, no Salmonellacontamination has been detected. Understanding the limits of microbiological testing for the detection of product and environmental Salmonella contamination is crucial for describing the risk of persistent Salmonella contamin ation in dry food production facilities and improving detection and prevention methods. L Herrera-Leon and S Herrera-Leon carried out analyses on bacterial isolates in Spain, and carried out WGS and cgMLST analyses. C Varela coordinated epidemiological investigations in Spain.
J Mossong was responsible for microbiological investigations in Luxembourg.
W Mattheus was responsible for microbiological investigations in Belgium. C Huard carried out epidemiological investigations of cases in Belgium. All authors revised the manuscript.