- Jean-Claude Desenclos, medical epidemiologista,
- Philippe Bouvet, deputy directorb,
- Elisabeth Benz-Lemoine, public health physicianc,
- Francine Grimont, deputy directord,
- Helene Desqueyroux, residenta,
- Isabelle Rebiere, medical epidemiologista,
- Patrick A Grimont, directorb
- a Reseau National de Sante Publique, 94415 Saint-Maurice Cedex, France
- b Centre National de Reference des Salmonella-Shigella, Institut Pasteur, Paris
- c Direction Departementale des Affaires Sanitaires et Sociales des Deux-Sevres, Niort, France
- d Centre National de Reference de Typage Moleculaire Enterique, Institut Pasteur, Paris
- Correspondence to: Dr Desenclos.
- Accepted 2 November 1995
Objective: To assess the magnitude of a nationwide outbreak of infection with Salmonella enterica serotype paratyphi B and identify the vehicle and source of infection.
Design: A case finding study of S paratyphi B infection between 15 August and 30 November 1993; a pair matched case-control study; an environmental investigation at a processing plant that produced a raw goats' milk cheese incriminated in the outbreak; phage typing and genotyping of food and human S paratyphi B isolates.
Setting: France, 15 August to 30 November 1993.
Subjects: 273 patients with S paratyphi B infection; 59 pairs of cases and controls matched for age, sex, and city of residence.
Main outcome measures: Numbers of cases and incidence rates by region of residence and age; matched odds ratios for dairy food preferences.
Results: Among the 273 cases there was one death; 203 (78%) strains belonged to phage type 1 var 3. The incidence of infection was greatest in the region where goats' milk cheese is commonly produced. Comparison of cases and controls showed a 12-fold greater risk of illness (95% confidence interval 1.6 to 92.3) from eating brand A unpasteurised goats' milk cheese. S paratyphi B isolates of phage type 1 var 3 were recovered from cheese A, goats' milk at the plant processing cheese A, and goats' milk supplied to the plant by a single farm. Genotypic IS 200 typing of food and human 1 var 3 phage type isolates showed a common IS 200 pattern.
Conclusion: This outbreak emphasises the potential health hazards of widely distributed unpasteurised milk products in France and the need for their close bacterial monitoring.
Though the cheese was probably contaminated for more than two months, the outbreak continued undetected for a further two months
The source of the infection was goats' milk from one of the 40 farms that supplied the cheese processing plant
Internal microbiological monitoring at the plant was not sensitive enough to detect the salmonella contamination initially
Any batch of unpasteurised cheese or milk product should be closely monitored for Salmonella and should not be distributed until known to be clear
Salmonella enterica serotype paratyphi B causes sporadic gastroenteritis and, less frequently, paratyphoid fever.1 Few foodborne outbreaks of S paratyphi B infection have been reported. In France two or more outbreaks have occurred in the past 10 years.2 3 An outbreak in 1990 (277 cases) was possibly related to contaminated goats' milk cheese.3 Unpasteurised dairy products have caused outbreaks of salmonellosis, campylobacteriosis, listeriosis, and the haemolyticuraemic syndrome.1 4 5 6 7 8 9 10 In France large amounts of many different types of raw milk cheeses are consumed, yet raw milk cheese has only rarely been incriminated in foodborne outbreaks.3 10
We describe a large nationwide outbreak of S paratyphi B infection caused by unpasteurised goats' milk cheese.
Subjects and methods SALMONELLA SURVEILLANCE
In France surveillance for salmonellosis is carried out by the National Reference Centre for Salmonella and Shigella, which receives isolates for serotyping from one third of the 4000 microbiology laboratories. For the past 12 years monthly trends have been computed for each serotype of Salmonella isolated. During the third week of October 1993 a nationwide increase in the number of S paratyphi B isolates submitted for typing was observed.
A case was defined as a resident of France from whom a specimen (stools, blood, or other body tissue) had been culture positive for S paratyphi B between 1 August and 30 November 1993. Cases were identified by reviewing isolates received by the National Reference Centre. Multiple isolates from the same patient were excluded. Additional cases were sought by district public health officers from local laboratories. For each case identified, the patient's sex and age (<1, 1-5, 6-14, 15-64, and >/=65 years) and the date, site, region, and laboratory were recorded. Missing data were obtained by contacting the relevant laboratory.
After the outbreak was recognised a food questionnaire given to a few patients indicated that most had consumed brand A medium size round goats' milk cheese. Because a relation with a similar cheese had been suggested in the 1990 outbreak3 we hypothesised that this cheese was the vehicle of infection in the present outbreak. To test this hypothesis we conducted a case-control study. Patients were included as cases if they had S paratyphi B gastroenteritis (more than three loose stools daily) or septicaemia. For each case investigated a community control matched for age (within the ranges <1, 1-4, 5-14, 15-34, 35-44, 45-54, 55-64, and >/=65 years), sex (for cases aged >5 only), and city of residence was sought from the telephone directory. People whose names (different spelling from the case name) came after the case name in the directory were called alphabetically until one was located who met the matching criteria. Potential controls who reported diarrhoea (more than three loose stools daily) in the previous three months were excluded.
Cases and matched controls (or their mothers if under 18) were interviewed by telephone by district public health physicians or a medical epidemiologist from the National Public Health Network using a standardised questionnaire. This was mainly targeted at milk products, particularly cow and goats' milk cheeses, and included questions on names and types of cheeses. Cases and controls were interviewed two to 12 weeks after the illness. Hence rather than ask them to try to recall the foods actually eaten during the three days before the illness we aimed at ascertaining their food preferences. Interviewers were not blinded to the subjects' case or control status.
In early November the processing plant (plant A) that produced the suspect cheese was inspected by a district veterinarian from the Ministry of Agriculture and one of us (EBL). Goats' milk sources, cheese production and storage, and the microbiological monitoring of milk and cheese produced at the plant between June and November 1993 were reviewed. In addition, stool specimens were obtained from goats, cows, dogs, a cat, and workers from a farm that had supplied plant A with milk found to be contaminated by Salmonella.
Salmonella serotype paratyphi B isolates (human, milk, and cheese) were phage typed at the National Reference Centre for Enteric Molecular Typing by using Felix and Callow's international system.11 Human and food isolates were also subjected to genotypic IS 200 typing.12
Two hundred and seventy three cases (4.3/million residents) were recorded (259 by the National Reference Centre, 14 by other laboratories). In 240 cases (88%) S paratyphi B was isolated from stools, in 15 (5.5%) it was isolated from blood, and in 14 (5%) it was isolated from other tissue (site unknown for four isolates). Clinical details were obtained for 97 (36%) patients by telephone interview. Thirty six (37%) had been admitted to hospital and one died. Gastroenteritis was characterised by diarrhoea (three to over 20 stools daily, median 6) that lasted from three to 27 days (median 5 days), fever (>38°C; 80 cases), abdominal cramp (76 cases), nausea (33 cases), and vomiting (30 cases). Of the 259 isolates phage typed, 203 (78%) belonged to phage type 1 var 3.
The outbreak began during the second week of August 1993 and continued till the second week of November (fig 1). Most of the infections were due to phage type 1 var 3. Cases were distributed nationally; however, the incidence was greatest in Poitou-Charentes (a traditional area of goats' milk cheese production) and the surrounding regions (fig 2). The isolation rate was highest among infants and children aged 1 to 5 years (table 1).
Food questionnaires were completed for 72 pairs of cases and controls. Because most of the infections in the outbreak were attributed to phage type 1 var 3 the analysis was restricted to the 59 (82%) pairs in which that phage type was isolated. The 59 cases did not differ from the total series of 203 patients infected with phage type 1 var 3 in age (P=0.9), sex (P=0.9), or region of isolation (P=0.2). There was a trend towards an increased risk in the presence of an underlying illness (for example, diabetes, malignancy, treatment with corticosteroids, chemotherapy; odds ratio 2.0 (95% confidence interval 0.6 to 6.7)). Analysis of the consumption of milk products showed a 3.8-fold greater risk of illness among people who ate goats' milk cheese (table 2). For the subsequent analysis cheeses were categorised as brand A goats' milk cheese (a medium size round cheese), medium size round goats' milk cheese of unspecified brand, other types of goats' milk cheese, and non-goats' milk cheese. Compared with the risk of illness among people who did not eat goats' milk cheese, there was a 12-fold greater risk among those who ate brand A and a sixfold greater risk among those who ate medium size round goats' milk cheese of unspecified brand. There was no substantially increased risk of illness among people who ate other goats' milk cheeses (table 2).
ENVIRONMENTAL AND MICROBIOLOGICAL INVESTIGATIONS
Cheese A is made from raw goats' milk at a single plant and distributed to food stores and supermarkets nationally. Every other day two batches of cheese A are made (11000 to 15000 cheeses (200 g each) per batch), each corresponding to a pool of 40 farms supplying goats' milk. Cheeses are stored for 11 days at the plant for maturation before distribution. The “use by” date is 45 days after the cheeses leave the plant.
Before October 1993 internal control for Salmonella at the plant consisted of a weekly culture on five cheeses picked from a single batch. Then, on 6 October, one brand A cheese grew Salmonella, later typed as paratyphi B. Subsequently, from 7 October, all batches of cheese stored at the plant, milk pools, and milk from all the farms that supplied each pool were sampled daily for Salmonella. The district public health authorities remained unaware of these matters until 8 November, when the district public health physician contacted the district veterinarian's office about a possible link of the outbreak with cheese A. The milk pool corresponding to the batch that grew S paratyphi B on 6 October was also positive for S paratyphi B on 9 October but negative on the 7th, 11th, and 13th.
S paratyphi B was recovered from the milk of only one of the 40 suppliers. No salmonella was found in stool specimens from workers, cows, goats, and pets at the farm. Cheese A and goats' milk isolates belonged to the epidemic phage type (1 var 3). IS 200 genotypic typing was done on three human and four cheese A 1 var 3 isolates. All seven strains exhibited a common IS 200 pattern (profile 2.7).
Around 30 tonnes of cheese, corresponding to the batches stored at the plant between 21 September and 6 October, were destroyed after the isolation of S paratyphi B from cheese A, and cheese production from the relevant milk pool was pasteurised until daily Salmonella control of each batch was implemented. Subsequently all batches produced have been tested for Salmonella on day 1 (milk pool), on days 2 and 6 of the maturation process, and on days 9 and 12 (packaging and distribution, respectively).
This large nationwide outbreak of salmonellosis was caused by unpasteurised goats' milk cheese made in a single plant. Evidence comes from the results of the case-control study and the isolation from cheese A and goats' milk of an S paratyphi B strain of the same phage type and IS 200 pattern as the epidemic strain. The increased risk suggested in the case-control study for medium size round goats' milk cheese of unspecified brand may reflect consumption of cheese A, because this exposure category may have included people who ate cheese A but could not recall the brand name.
Contamination of the milk pool from which one of the two batches was made originated from a single farm. However, the precise source of infection (human, animal, or environmental) was not identified. The duration of the outbreak (three months) indicated that cheese A had been contaminated for a similar period, probably from mid-June (date of onset of the epidemic minus three days for the incubation period, 11 days for cheese maturation, and 45 days before the use by date). The outbreak was detected during the third week of October, when contamination had gone unnoticed for almost three months. Detection of contamination at the plant in early October was not notified to the authorities and so did not contribute to faster recognition of the outbreak. (Routine microbiological control programmes in food processing plants are carried out voluntarily by producers to reduce the risk of foodborne infections; however, the results are not required to be notified to public health authorities.) Routine daily control of each batch for Salmonella would have detected the contamination much earlier.
In France only one third of laboratories send isolates of Salmonella for typing. Furthermore, only about 6% of patients with diarrhoea have a stool culture15 and some patients with diarrhoea do not see a doctor at all. Hence the true size of the epidemic was probably much underestimated. Several thousands of cases may have occurred because of contamination of the cheese.
Despite the amount of raw milk cheese consumed daily in France outbreaks of infection remain comparatively rare.3 10 In France pasteurisation of raw milk cheeses is not feasible for cultural, social, and economic reasons. Strategies for preventing infection by raw milk cheeses should therefore be aimed at both producers and consumers. Strict and carefully planned hazard analysis critical control point procedures should be developed and implemented for unpasteurised dairy products.16 As part of this procedure any batch of cheese made from raw milk should be closely monitored for Salmonella and not be distributed until known to be clear. Producers should also report positive results of end production internal sampling to public health authorities. Consumers—particularly those susceptible to infectious diseases (for example, infants, elderly people, immunocompromised patients)—should also be warned that a nil risk cannot be warranted for raw milk products.
We thank the following public health physicians who participated: M Andrillon, A Armangaud, C Barbier, F Belingard, D Bousquet, E Boutin, V Bleuze, B Cabo, C Cazenave, F Charlet, M Charron, M J Communal, M Cunnac, M Desvaux, M C Dubois, M L Ferial, M Feltin, J Y Goarant, L Gossel, A Hetru, M Juge, A Meunier, J Patureau, C Piau, E Pons, F Quittancon, and P Rogez (Directions Departementales de l'Action Sanitaire et Sociale); V Goulet, B Hubert, N Lacan, E Laurent, I Mehl-Augier, R Pinget, and E Sariot (Reseau National de Sante Publique); A Lepoutre and C Moyse (Direction Generale de la Sante). We also thank public health veterinarian F Peyre (Direction des Services Veterinaires des Deux-Sevres).
Funding This study was conducted as part of routine activities in the institutions concerned. The Reseau National de Sante Publique is funded by the Ministere de la Sante et de l'Assurance Maladie; the Centre National de Reference des Salmonella-Shigella and the Centre National de Reference de Typage Moleculaire Enterique are funded by the Ministere de la Sante et de l'Assurance Maladie and the Pasteur Institute; the Direction Departementale des Affaires Sanitaires et Sociales des Deux-Sevres is funded by the Ministere de la Sante et de l'Assurance Maladie.
Conflict of interest None.