The aerobic colony counts, coliforms and E. coli counts were significantly higher (p<0.05) in the main dishes than in the breakfasts (I, Table 2). Although the means of total counts and coliforms or Enterobacteriaceae were rather low in the hot meals, high maximum values were detected. Total counts higher than 10 6 cfu/g, which is the AEA limit for food items handled after heat treatment, were found in 9.2% of hot meal samples. Many of the hot meals 33 (8.2%) also exceeded the microbiological standards given by the AEA (1996) for E. coli . The number of the hot meal samples not meeting the standard were for S. aureus 6 (0.6%), B. cereus 7 (0.7%) and C. perfringens 2 (0.7%) (I, Table 4). The maximum counts were 4.0 x 10 3 cfu/g for S. aureus , 3.0 x 10 4 cfu/g for B. cereus and 1.0 x 10 3 cfu/g for C. perfringens .
There were statistically significant differences between preparing countries regarding the means of total count and E. coli in hot meals (I, Table 5). The two countries showing the highest means of total bacteria also showed highest means of E. coli .
Appetisers and salads showed significantly higher aerobic colony counts and E. coli counts (p<0.05) than desserts (II, Table 2). E. coli counts were over 10 cfu/g in 33 (18%) of the appetizers, in 23 (16%) of the salads and in eight (6%) of the desserts, and totally in 64 (14%) of the cold meals (II, Table 3). S. aureus, B. cereus or C. perfringens were found in 50 (7%) of the cold meal samples (II, Table 4). The requirements set for S. aureus and B. cereus by the AEA standard were not met in 23 (7%) and 13 (3%) of the samples, respectively. In two poultry appetisers and one fresh salad prepared in the same country E. coli counts as high as 1.0 x 10 6 cfu/g were found (II, Table 2). The maximum value of S. aureus , 3.3 x 10 3 cfu/g, was found in an appetiser containing meat as the main ingredient. For B. cereus , the highest value was 5.0 x 10 4 cfu/g, which was found in an appetiser containing pâté.
The means of total counts and E. coli counts showed significant differences between preparing countries in cold meals as well as in hot meals. In addition, the means of B. cereus also revealed differences (II, Table 5).
In the microbiological survey including Salmonella , Salmonella was detected in 3 (0.3%) out of 1011 hot dishes (I, Table 3), but not in any of the 653 cold dishes (II, Table 4). In the specific Salmonella survey, out of the 1288 hot dishes five (0.4%) and out of the 923 cold dishes one (0.1%) were found to be Salmonella positive (III, Table 1). Altogether, Salmonella was detected in seven (0.3%) out of 2299 hot meals and in one (0.1%) out of 1576 cold meals.
The serotypes found in hot dishes were, Salmonella enterica serovar Manchester (hereafter S . Manchester), Salmonella enterica serovar Morbificans (hereafter S. Morbificans), Salmonella enterica serovar Hadar (hereafter S . Hadar) and Salmonella enterica serovar Braenderup (hereafter S . Braenderup). S . Manchester was isolated in a main dish containing beef, potatoes and cooked vegetables prepared in Kenya, S . Morbificans in a main dish containing chicken and boiled vegetable, S . Hadar in a breakfast containing omelette and cheese. Four hot dishes prepared in China, which contained chicken, potatoes and beans, beef, potatoes and cooked vegetable, fish, rice and cooked vegetable and snack crepes, and which were sampled during the same week, were shown to be contaminated by the same serotype, S . Braenderup.
The serotype found in the cold dish was Salmonella enterica serovar Ohio (hereafter S . Ohio), and it was isolated in an appetiser prepared in Thailand. It contained ham, Edam-type cheese, boiled egg and cooked and marinated vegetables.
An outbreak by Salmonella enterica serovar Infantis (hereafter S . Infantis), infecting a total of 226 people, occurred in Finland at the beginning of August 1986. It was found that three clusters of people were infected; railway passengers, charter flight passengers and employees of a Finnish catering establishment. The source of infection was traced to foods prepared in the same catering establishment. In spite of operating as a flight kitchen, the catering had other food business, such as supplying sandwiches for trains. The number of culture-confirmed persons was 226. The number of persons at risk and those infected among railway and airline passengers and catering establishment staff was 107/600 (18%), 91/350 (26%) and 28/162 (17%), respectively. The test results of catering staff (162) revealed that the infection had spread to almost every group of employees. Out of the 118 food handlers, 23 (19%) became infected. Of those infected, 17 (74%) were symptom-free carriers and only 6 (26%) had symptoms (IV, Table 3).
Through questionnaires and other investigations, the source of infection was traced to the following foods: egg sandwiches served on trains, the meal served on aircraft consisting of Viennese goulash, fresh salad and Swiss roll, and cold cuts served to the catering establishment staff during breakfast. A statistical significant association was shown between the infection of catering staff and the cold cuts from the cold kitchen that were served during breakfast (p<0.05), whereas there was no statistical association between eating lunch and illness. No single dish served to air passengers revealed a statistically significant association with illness.
S . Infantis was detected in the routine control from one hot meal sample taken from a batch of 1200 Viennese goulash portions, from which 350 portions had been sent to a charter flight to Rhodes. Salmonella was not detected in any other food samples (152) tested during investigation of the outbreak.
Inspection of the catering establishment revealed several structural and functional shortcomings. The transport routes of raw and cooked foods were not separated, thus causing a risk of cross-contamination. There was no facility for fast chilling hot food. A lack of cold storage was also reported. Of great importance was that the food handlers suffering from diarrhoea were not excluded from work. Neither was any food hygiene training included in their education. The spread of the outbreak was further influenced by a heat wave at that time in Finland.
The prevalence of S. aureus was much higher on the basis of nasal sampling compared with hand sampling, 32 of 111 (29%) and 10 of 117 (9%) respectively. Seven persons showed growth both in nasal and hand samples, and 20 persons only revealed S. aureus from nasal culture. Almost all hand carriers showed growth also in nasal samples (V, Table 1). Enterotoxigenic S. aureus types were found in 13 out of 111 (12%) and 7 out of 117 (6%) food handlers according to nasal and hand sampling, respectively.
PFGE macrorestriction profiles revealed a total of 32 different types associated with the 35 employees carrying S. aureus . Eight PFGE types were obtained from the hands and 30 types from nasal samples. In 4 cases out of 7 the same type colonised both hand and nose. PFGE type 6 was the most common type, colonising 5 persons. PFGE types tested for enterotoxin production showed that 12 of 32 (38%) types produced enterotoxin. The most common PFGE type 6 produced enterotoxin B.