An international point source outbreak of typhoid fever: a European collaborative investigation*

Bulletin of the World Health Organization, 64 (2): 271-278 (1986) © World Health Organization 1986 An international point source outbreak of typhoid fever: a European collaborative investigation* R. E. STANWELL-SMITH' & L. R. WARD2 A point source outbreak of Salmonella typhi, degraded Vi-strain 22, affecting 32 British visitors to Kos, Greece, in 1983 was attributed by a case-control study to the consumption of a salad at one hotel. This represents the first major outbreak of typhoid fever in which a salad has been identified as the vehicle. The source of the infection was probably a carrier in the hotel staff. The investigation demonstrates the importance of national surveillance, international cooperation, and epidemiological methods in the investigation and control of major outbreaks of infection. An outbreak of typhoid fever, the largest common source outbreak that has affected residents in the United Kingdom since 1964 (1), occurred in the summer of 1983. All those affected had recently returned from holiday on the Greek island of Kos. Thirty-three cases were confirmed in the United Kingdom and a further 27 cases were reported in eight other countries. MATERIALS AND METHODS The outbreak was identified through national surveillance coordinated by the Public Health Laboratory Service (PHLS) Communicable Disease Surveillance Centre (CDSC), London, and the Com- municable Diseases (Scotland) Unit, Glasgow. Public health authorities in other European countries were informed through WHO. A pilot investigation implicated a single hotel in the week 29 June-5 July * This investigation was carried out with the collaboration of Dr J. A. N. Emslie and Dr J. H. Cossar, Communicable Diseases (Scotland) Unit, Glasgow (Scotland); Dr M. Jahkola, National Public Health Institute, Helsinki (Finland); Dr T. Ronne, Depart- ment of Epidemiology, Statens Seruminstitut, Copenhagen (Den- mark); Dr B. Christenson, Epidemiological Department, National Bacteriological Laboratory, Stockholm (Sweden); Dr K. Wold Haug, Infectious Diseases Unit, Division of Maternal and Child Health, Oslo (Norway); Dr H. Bijkerk, Division of Infectious Diseases, Leidschendam (Netherlands); Professor J. A. Papadakis, Athens School of Hygiene, Athens (Greece); Dr. T. H. Stefanou, Public Health Division, Ministry of Health and Welfare, Athens (Greece). ' Senior Registrar in Community Medicine (Epidemiology), Public Health Laboratory Service Communicable Disease Surveillance Centre, 61 Colindale Avenue, London NW9 5EQ, England. Requests for reprints should be sent to this author. 2 Principal Microbiologist, Central Public Health Laboratory Division of Enteric Pathogens and WHO Collaborating Centre for Phage Typing and Drug Resistance of Enterobacteria, London, England. 1983. Analysis of the numbers and nationalities of guests who stayed at the hotel that week indicated that 118 were British residents, including 32 of the 33 cases confirmed in the United Kingdom. There was one case of typhoid in a Greek visitor to the United Kingdom from the implicated hotel in Kos. Residents of eight other countries had also stayed at the hotel during this week (Table 1). Table 1. Country of origin of guests at the implicated hotel, 29 June-6 July 1983 No. of cases Country No. of guests' of typhoid fever United Kingdom 118 32 (27)b Finland 50 10 (20) Sweden 26 7 (27) Norway 16 5 (31) Netherlands 18 2 (11) Denmark 16 1 (6) Belgium 20 1 (5) Switzerland 15 1 (7) Federal Republic of Germany 10 unknown Total 289 59 (20) a Numbers are approximate (information from hotel and tour operators). b Numbers in parentheses are percentages. 4663 -271- R. E. STANWELL-SMITH & L. R. WARD Questionnaire survey The 118 British guests were asked to complete a detailed questionnaire on menus offered at the hotel during the suspect period. Questions were also in- cluded on other activities, such as swimming, drink- ing, eating meals outside the hotel, and any excur- sions made. The questionnaires for the 32 British cases and for one each of their close contacts who had also stayed at the hotel were completed during an interview with a local or CDSC community physi- cian. The remaining 54 British guests at the hotel were sent the same questionnaire within 4 weeks of their holiday with a request to complete it. Each person traced was asked to provide at least one faecal speci- men. All recipients completed and returned the questionnaires, but two respondents combined their answers on a single questionnaire, and they were excluded from the analysis. The CDSC questionnaire was also used in the outbreak investigations carried out in Finland and Denmark. Thirty-eight of the 50 Finnish guests (76%) at the hotel returned questionnaires. In Denmark the investigation was restricted to one case and one control. Data from British, Danish, and Finnish investigations were analysed separately. Statistical methods The British and Finnish cohorts were divided into three groups for the statistical analysis: those with confirmed typhoid fever, those with no illness, and those with illness other than typhoid fever (Table 2). Comparisons were made, inter alia, between people with no illness and those who had illness other than typhoid fever, as well as between the well group and those with confirmed typhoid fever. Only results relevant to the identification of the vehicle of typhoid fever in the outbreak have been included here. Re- spondents' age was compared using an unpaired Student's t-test and for some analyses reclassified into two categories based on the median ages of the British and Finnish cohorts. Categorical variables were analysed by 2 x 2 tables using Fisher's exact test (two-tailed). Cochran's test was used as a summary test of the association between an exposure and illness after stratification for one or more variables. The clinical picture Most patients with typhoid fever presented with fever and gastrointestinal symptoms following a period of headache and malaise, but only few were seriously ill. There were no fatalities. There were 32 confirmed British cases in the survey, and the 51 individuals without any illness were used as controls in the study (Table 2). The mean age of British cases (34.4 years) was significantly lower than that of the controls (45.9 years) (P= 0.004). No significant difference in this or other respects between the sexes was observed. Onset of en- teric fever for those with confirmed typhoid fever occurred between 11 July and 27 July (median, 20 July) (Fig. 1). Ten of the 38 Finnish guests at the hotel who returned questionnaires had confirmed typhoid fever, 18 had no illness, and 10 had an illness other than typhoid fever. The average age of the Finnish guests was lower (mean age, 28.4 years; median age, 32 years) than that of those from the United King- dom, and cases were not significantly younger than controls. The median date of onset of illness (21 July) was one day later than that of the British cases. Table 2. Details of guests from the United Kingdom and Finland at the implicated hotel, 29 June-5 July 1983 Total in study' Cases of typhoid fever Controlsb Sex and age (years) of guests United Kingdom Finland United Kingdom Finland United Kingdom Finland Male 57 16 12 4 28 6 Female 49 22 20 6 23 12 Male + female 116 38 32 1 0 5 1 1 8 Median age 37 32 31 34 50 34 Mean age 39.5 (± 16.8) 28.4 (± 17.2) 34.4c ( ± 17.9) 27.3 (±15.9) 45.9 (±15.5) 28.9 (± 15.5) a Includes those with illness other than typhoid fever (33 among the British guests and 10 among the Finnish). b Guests who had no illness. c P= 0.004 for difference between cases and controls among the British guests. 272 COLLABORATIVE INVESTIGATION OF TYPHOID OUTBREAK MediOn who handled food; 21 urine cultures from staff were 8 ,I20thMy also examined. An assistant waiter and a female ra 6] 2kitchen assistant were later identified as symptomless Ia,|,] i3 oncetin dcof excretors of S. typhi. The kitchen assistant's first4, nM in 3 c sample was negative for S. typhi, which delayed her identification as a possible carrier. She had worked at f<vlF1 the hotel since 1979 and had had a typhoid infection '6 '8 2 liS 20 s 26 when she 13 of Her duties at the hotel July included washing and cutting salads and vegetables as Fig. 1. Date of onset of symptoms of typhoid fever in the well as preparing cold dishes. As soon as she was outbreak investigation of British cases, Kos, Greece, identified as a carrier, she was removed from service July 1983. at the hotel. The possible vehicle The implicated hotel. The hotel had approximately 400 beds, of which 289 were occupied on 29 June-5 July. There were two sources of water. Drinking- water was supplied by a chlorinated mains supply distributed throughout the building, but water from a borehole in the hotel grounds was used for washing surfaces, and, possibly, for other purposes in the kitchen. Samples of the borehole water taken on 31 July and 7 August were found to be contaminated with Escherichia coli, but Salmonella typhi was not isolated. The hotel staff of 59 included 7 kitchen workers and 17 waiters. Between 2 August and 23 August 1983 at least two and, in most cases, three or more faecal samples were collected from the 24 staff members Food exposure. Analysis of items from the dinner menu (Table 3) revealed a strong association between infection with typhoid fever and consumption of a tomato-and-cucumber salad on 4 July (P=0.001). Side salads with the evening meals were not served with mayonnaise or other dressings, and consump- tion of salad on any other evening was not signifi- cantly associated with infection with typhoid fever. All the Finnish cases and all but one of the group with no illness ate salad on 4 July, and hence a statistical assessment of risk could only be made by combining the Finnish and British data (P=0.001). Consump- tion of melon on 7 July was significantly associated with infection of the Finnish guests (P= 0.02), although the numbers were very small (6 out of 7 Table 3. Number of British guests with typhoid fever and the items of food eaten at dinner on Monday 4 July 19830 No. who ate No. who did not eat With With typhoid typhoid Food item Total fever Not ill Total fever Not ill Fish starter 55 25 (45 30 17 4 (24) 13 Fruit juice 16 5 (31) 1 1 54 22 (41) 32 Meat balls 54 20 (37) 34 21 9 (43) 12 Potato 55 21 (38) 34 15 5(33) 10 Chicken 10 5 (50) 5 54 20 (37) 34 Saladc 60 29 (48) 31 13 0 (0) 13 Melon 8 5 (63) 3 49 17 (35) 32 Ice-cream 23 10 (43) 13 38 13 (34) 25 Cake 24 8 (33) 16 32 13 (41) 19 a Persons with an illness other than typhoid fever were excluded. b Numbers in parentheses are percentages. c Fisher's exact test (two-tailed) for association between eating salad and infection for cases and controls, P= 0.001. 273 R. E. STANWELL-SMITH & L. R. WARD typhoid cases compared with 1 out of 7 guests who had not been ill). This association was not observed in the British guests, only 5 of whom reported having eaten melon on this date, and may have arisen be- cause of differences in the holiday dates as well as food preferences of the Finnish and British groups. Whereas 17 of the British guests had left the hotel by 7 July, all the Finnish cases were resident between 3 July and 10 July. There was only one case among the 16 Danish guests at the hotel, a 10-year-old girl. Both she and her unaffected younger brother had eaten the implicated salad on 4 July but were unsure about their consumption of several other food items, such as melon on 7 July. Detailed epidemiological information was not available for the other Danish guests. Water exposure. For the British cases, significant associations were observed between infection with typhoid fever and swimming in the hotel pool (P= 0.01) as well as with use of ice in drinks (P= 0.004) (Table 4). However, both of these associations were strongly linked to age (P= 0.002 for the use of the swimming pool and P= 0.00004 for ice). After stratifying for both age and consumption of salad, use of the swimming pool was found not to be significantly related to infection with typhoid fever. A high proportion of cases both ate salad on 4 July and took ice in drinks (22 out of 29 (76%)). When stratified only for salad consumption on 4 July (Table 5), use of ice was still significantly associated with illness. The 13 guests who did not eat salad did not become ill. When those who had eaten salad were further stratified according to age (Table 5), the association between use of ice and illness was not statistically significant. In contrast, the association between consumption of salad on 4 July and infection with typhoid fever was independent of the effects of either age or use of ice in drinks. There were no significant associations between infection with typhoid fever and consumption of any other food items or exposure to water in the hotel or elsewhere. Consumption of the salad on 4 July was not significantly associated with illness other than typhoid fever. The 26 Swedish hotel guests were investigated separately, and, as in the British study, there was a significant association between consumption of salad on 4 July, although not with swimming or use of ice in drinks (Dr B. Christenson, personal communication, 1984). Effects ofprevious immunization Only four of the 116 British hotel guests surveyed, and none of the Finnish guests, had been immunized against typhoid fever within 3 years of their holiday, although a further 18 of the British guests and 3 of the Finnish had been immunized at some time in the past. The number of individuals who had recently been immunized was too small to allow investigation of the Table 4. Analysis of British guests with typhoid fever and exposure to various sources of watera No. exposed b No. not exposed With With typhoid typhoid Water source Total fever Not ill Total fever Not ill Paddling pool 4 3 (75)c 1 75 28 (37)c 47 Swimming poold 50 25 (50) 25 31 6 (19) 25 Sea 18 9 (50) 9 62 22 (35) 40 Brushed teeth with tap-water 75 30 (40) 45 7 2 (29) 5 Drank tap-water 26 10 (38) 16 53 21 (40) 32 Drank other non-mineral water 45 19 (42) 26 33 11 (33) 22 Ice in drinksd 47 25 (53) 22 35 7 (20) 28 a Persons with an illness other than typhoid fever were excluded. b Exposure from swimming pools, sea, washing teeth, and drinking tap-water ranged from 3 times a week to every day; not exposed = never or less than 3 times; consumption of non-mineral water and ice = yes or no. c Numbers in parentheses are percentages. d Fisher's exact test (two-tailed) for associations between exposure for cases compared with controls: P= 0.01 for use of swim- ming pool; P= 0.004 for consumption of ice. 274 COLLABORATIVE INVESTIGATION OF TYPHOID OUTBREAK Table 5. Association between use of ice in drinks and infection with typhoid fever, stratified by consumption of salad on 4 July and by age 1) Use of ice, stratified by consumption of salad' No. who did not No. who ate the salad eat the salad With With typhoid typhoid Ice in drinks fever Not ill Total fever Not ill Total Yes 22 13 35 0 4 4 No 7 17 24 0 9 9 Total 29 30 59 0 13 13 2) Use of ice in drinks by those who consumed salad, stratified by age groupb Age: < 37 years Age: > 38 years With With typhoid typhoid Ice in drinks fever Not ill Total fever Not ill Total Yes 15 6 21 7 7 14 No 2 3 5 5 14 19 Total 1 7 9 26 1 2 21 33 a Fisher's exact test for the association between use of ice and typhoid among salad eaters (P= 0.02). Cochran's test for the association between consumption of salad and typhoid fever among those who used ice and those who did not (X= 3.0, P= 0.003). b Cochran's test for the association between use of ice and typhoid fever within age groups (X= 1.91, P=0.06 (not signifi- cant)). degree of protection afforded them, but none of them had confirmed typhoid fever. Microbiological investigations Samples isolated from patients underwent widespread cross-reactions with adapted Salmonella typhi Vi-typing phages and were classified as de- graded Vi-strain 22 (DVS-22) (2). Reactions of this type are reproducible and can therefore be used for epidemiological purposes. Two patterns of reactions were observed between DVS-22 and the typing phages. Strains isolated from most patients, in- cluding the symptomless waiter at the hotel, exhibited only one pattern, whereas a few isolates, including one from the female kitchen assistant, exhibited both patterns. Examination of records at the Division of Enteric Pathogens revealed that a similar strain had been isolated in Athens, Greece, in 1972 and also from three patients who had visited the implicated hotel-one from Sweden in 1981, and one each from Finland and Norway in 1982-prior to their illness. Salmonellae of three other serotypes were also isolated during the investigation: one from a male British guest with an illness other than typhoid fever, whose symptoms commenced on the night of 4 July; one from a Finnish guest who became ill on 10 July; and the third from urine and faecal specimens of three members of the kitchen staff as well as from samples from the hotel's sewerage system. DISCUSSION The investigation established that the most probable vehicle of the S. typhi infection was the tomato-and-cucumber salad consumed on 4 July 1983. It was the only variable in the survey questionnaire for which there was strong independent association with infection with typhoid fever. Salads, in common with other raw foods, carry intestinal pathogens (3, 4). In a recent survey of Greek shops, 1. I% of 538 samples of tomatoes, lettuce, peppers, and other fresh vegetables were positive for salmonellae, although S. typhi was not isolated and no salmonellae were isolated from tomatoes (5). Contamination of lettuce (6) and cucumbers (7) with salmonellae or other enteric bacteria has been reported, but, paradoxically, instances of human salmonellosis caused by raw foods are rare (8). The cases described here appear to result from the first outbreak of typhoid fever in which the epidemio- logical evidence directly implicates consumption of a salad. A lettuce-and-tomato salad, inter alia, has been implicated previously in an outbreak of Salmo- nella typhimurium that affected 36 people who had eaten in a restaurant (9); however, egg or potato salads, particularly those containing mayonnaise, are more frequently identified vectors of intestinal pathogens (10, 11). Melon has previously been implicated in the transmission of salmonellosis (12), and this was the only food item significantly associated with infection of the Finnish guests at the hotel. Although the incubation period for the British and Finnish groups and the interval of 16 days between the appearance of symptoms in first and last cases are consistent with a point source outbreak, a recent analysis of several typhoid outbreaks indicates that a spread of 16 days may be the upper limit for a point source infection (13). It is therefore possible that both the salad on 4 July and melon on 7 July were vehicles of S. typhi. The microbiological findings in the investigation of this outbreak provided vital clues to identification of 275 R. E. STANWELL-SMITH & L. R. WARD both the geographical source of infection and the human carrier. The previous cases of S. typhi associated with the implicated hotel strongly indicated the existence of a carrier who had worked there for some time. Of the two reported excretors of S. typhi, only the female kitchen assistant excreted the two subtypes of DVS-22 found in samples from cases. It is therefore highly probable that she was the carrier in the present outbreak, although the means of contaminating the salads was not established. The salads were probably contaminated while being prepared for dinner; however, whether this was by direct contact of the carrier with the food or via contaminated water remains uncertain. It has been suggested that, although fresh vegetables are frequently contaminated, the number of organisms is too small to cause clinical symptoms, unless they have been allowed to multiply (12). No detailed infor- mation was available on food preparation methods or level of hygiene in the kitchen during the first week of July 1983, but it seems possible that, after contami- nation, the salads stood for some hours at ambient temperature, conditions which favour rapid prolif- eration of bacteria. This hypothesis is corroborated by the high attack rate (27% for the British guests and approximately 20% for all guests at the hotel). It has been reported previously that not more than 20% of those exposed to S. typhi become infected with typhoid fever (14). Salmonella of a different serotype was isolated from other members of hotel staff, suggesting that there was a high level of gastrointestinal infection in the hotel, and this is reflected in the proportion of general illness among the British and Finnish guests (22% of those who did not develop typhoid fever), some of whom became ill either on or shortly after 4 July. This may have arisen because of the contami- nation of the salads by sewage from the borehole water supply. The hypothesis that the salads were washed in contaminated water is corroborated by the presence of faecal contamination of the borehole water sampled at the hotel shortly after the outbreak. However, since a detailed investigation of the hotel's water system was not carried out and no S. typhi was isolated from water or sewage effluent, there is no conclusive evidence to support this hypothesis. The occurrence of illness other than typhoid fever was not significantly associated with consumption of salad, and it seems likely that such illnesses were caused by a variety of sources. Only four of the British visitors to Kos who were investigated in this outbreak had been immunized against typhoid fever within 3 years of their holiday. Although at the time of the outbreak typhoid immunization was officially recommended before travelling to Greece and other Mediterranean countries, it has been suggested that the best protection for travellers is scrupulous attention to food and drink (15). "Holiday typhoid" has long been recognized as one of the hazards of travelling (16), and 88% of the 1092 cases reported in England and Wales between 1979 and 1982 were imported (17). Large outbreaks involving travellers have usually been associated with contaminated water, as in Zermatt (1963), where there were at least 437 cases (18). It should be noted that the risk of acquiring typhoid fever in Medi- terranean countries is less than 1 per 100 000 visitors (15), and that visits to other parts of the world may carry a far higher risk. This is emphasized by what is believed to be the largest ever outbreak of typhoid fever: 9000 cases resulting from a common source of contaminated water in India in 1976 (19). An important aspect of the outbreak in Kos was the rapid identification of cases by national and inter- national surveillance. The first cases were detected from routine reports of CDSC followed by rapid phage-typing. Otherwise there could have been con- siderable delays both in identifying those infected and those at risk and in providing information to control the outbreak. The vehicle of infection was identified by epidemiological investigation, in contrast to pre- vious "holiday outbreaks" of typhoid fever, where the vehicle has remained obscure. ACKNOWLEDGEMENTS We would like to thank Dr H. E. Tillett and Dr R. Straube for advice on and assistance with statistical analysis of the data, as well as Dr C. B. Bartlett, Dr N. S. Galbraith, Dr S. E. Young, and Dr B. Rowe for their advice during the outbreak and in preparing this article. We also gratefully acknowledge the help of our colleagues, the medical officers of environmental health, environmental health officers, clinicians, and laboratory staff who cooperated in this investigation. 276 COLLABORATIVE INVESTIGATION OF TYPHOID OUTBREAK 277 RtSUMt UNE SOURCE PONCTUELLE A L'ORIGINE D'UNE FLAMBEE INTERNATIONALE DE FIEVRE TYPHOTDE: ETUDE COLLECTIVE A L'ECHELLE EUROPEENNE La flamb&e de fievre typholde (Salmonella typhi, souche DVS-22) associ6e A l';le grecque de Kos en juillet 1983 a ete la plus importante flamb6e partant d'une source unique qui ait ete observ6e chez des r6sidents britanniques depuis la flamb6e d'Aberdeen en 1964. Trente-deux Britanniques ayant sejourn6 dans un hotel de l'ile et au moins 27 visiteurs de 8 autres pays europ&ens ont contract6 une fievre typholde. Les epid6miologistes de la plupart de ces pays ont collabor6 A 1'etude de cette flamb6e. Un meme questionnaire a e utilise au Royaume-Uni (n = 116), en Finlande (n = 38) et au Danemark (n = 2), et une 6tude cas-temoins a 6t6 faite sur les groupes britanniques et finnois. Les observations 6pidemiologiques ont permis de mettre en cause la consommation d'une salade de tomates et con- combres servie au diner du 4 juillet (P=0,001). Dans le groupe finnois uniquement, la consommation d'un melon le 7 juillet pr6sentait egalement une association significative avec la survenue de la fievre typholde (P= 0,02) bien que quelques sujets seulement aient 0t6 touch6s. Cette asso- ciation n'a pas e observ6e chez les Britanniques, dont tres peu avaient consomme de ce melon. On a observe une association significative, en fonction de I'Age, entre la fievre typhoide et l'exposition A l'eau chez les malades britanniques (qui avaient utilis6 la piscine de l'h6tel et pris de la glace dans leurs boissons), les sujets atteints etant, de fa;on significative, plus jeunes que les sujets indemnes (P= 0,004). Parmi les cas britanniques consid6res isolement ou associes aux cas finnois, la consommation de salade 6tait le seul facteur ind6pendamment associe a la fievre typholde. II s'agit la de la premiere flamb6e impor- tante dans laquelle une salade est identifi&e comme le vehicule de la fievre typholde. La source de l'infection etait probablement un sujet porteur appartenant au personnel de l'hotel. On a retrouve deux sujets excr6tant Salmonella typhi, DVS-22, dont l'un, une aide de cuisine, 6tait probablement le porteur en cause. L'analyse des registres de la Division of Enteric Pathogens a montre qu'une souche analogue avait ete isol6e chez trois sujets atteints de fievre typhoide au cours des deux ann6es precedentes; ces trois personnes avaient sejourne a l'h8tel en cause. Le mecanisme exact ayant conduit a la contamination des salades n'a pas e 6tabli. II semble toutefois probable qu'elles aient 6t6 directement contaminees par le sujet porteur au cours de leur pr6paration. Neanmoins, 1'even- tualite d'une contamination par les eaux us6es de la source d'eau de l'h8tel, un puits for6 dans le sous-sol, ne peut etre exclue. L'eau du puits 6tait parfois utilis6e pour laver des plans de travail dans la cuisine. Le fort taux d'atteinte (27% des touristes britanniques et environ 20% de l'ensemble des clients de l'hotel) suggere que, quel que soit le mode de contamination, les salades ont et6 laiss6es A la temperature ambiante, condition qui favorise la proliferation des bact6ries. L'etude met en 6vidence l'importance de la sur- veillance nationale, de la cooperation internationale et des methodes epidemiologiques dans l'etude des grandes flambees infectieuses et leur prevention. REFERENCES 1. MILNE, D. ETAL. TheAberdeen typhoid outbreak 1964, report of the departmental committee of enquiry. Edinburgh, H.M. Stationery Office, 1964. 2. ANDERSON, E. S. & WILLIAMS, R. E. 0. Bacterio- phage-typing of enteric pathogens and staphylococci and its use in epidemiology. Journal of clinical pathology, 9: 94-127 (1956). 3. HUCKSTEP, R. L. Typhoid fever and other salmonella infections. London, E. & S. Livingstone, 1962, p.19. 4. CVJETANOVIC, B. Epidemiology and control of water- and food-borne infections. In: Hobson, W., ed. Theory and practice ofpublic health, 5th ed. Oxford, Oxford University Press, 1979, pp. 248-264. 5. PAPADAKIS, J. A. & EFSTRATIOU, M. A. Isolation of salmonellae from fresh vegetables with the use of Rappaport-Vassiliadis magnesium chloride-malachite green enrichment medium. Hippocrates, 5: 1-5 (1980). 6. ERCOLANI, G. L. Bacteriological quality assessment of fresh marketed lettuce and fennel. Applied and environmental microbiology, 31: 847-852 (1976). 7. MENELEY, J. C. & STANGHELLINI, M. E. Detection of enteric bacteria within locular tissue of healthy cucumbers. Journal of food science, 39: 1267-1268 (1974). 8. CHRISTIE, A. B. Typhoid and paratyphoid fevers. In; Infectious diseases: epidemiology and clinical practice. 3rd ed. Edinburgh, Churchill Livingstone, 1980, p. 68. 9. BLASER, M. J. ET AL. An outbreak of salmonellosis involving multiple vehicles. American journal of epidemiology, 114: 663-670 (1981). 10. WEISSMAN, J. B. ET AL. Foodborne shigellosis at a country fair. American journal of epidemiology, 100: 178-185 (1974). 11. DAVIES, R. F. & WAHBA, A. H. Salmonella infections of charter flight passengers. Report on a visit to Spain (Canary Islands), 26 February-2 March 1976. Copen- hagen, WHO Regional Office for Europe, 1976, p. 2. 12. GAYLER, G. E. ET AL. An outbreak of salmonellosis traced to water melon. Public health reports, 70: 311-313 (1955). 278 R. E. STANWELL-SMITH & L. R. WARD 13. NAYLOR, G. R. E. Incubation period and other features of food-borne and water-borne outbreaks of typhoid fever in relation to pathogenesis and genetics of resistance. Lancet, 1: 864-866 (1983). 14. CVJETANOVIC, B. ET AL. Epidemiological model of typhoid fever and its use in the planning and evaluation of antityphoid immunization and sanitation pro- grammes. Bulletin of the World Health Organization, 45: 53-75 (1971). 15. DICK, G. ET AL. Typhoid vaccine, for whom? Lancet, 1: 970-971 (1982). 16. Holiday typhoid. British medical journal, 2: 605-606 (1969). 17. PHLS COMMUNICABLE DISEASE SURVEILLANCE CENTRE. Typhoid fever in England and Wales, 1978-82. British medical journal, 287: 1205 (1983). 18. BERNARD, R. P. The Zermatt typhoid outbreak in 1963. Journal of hygiene, 63: 537-563 (1965). 19. SATHE, P. V. ET AL. Investigation report of an epidemic of typhoid fever. Internationatjournal ofepidemiology, 12: 215-219 (1983).