Winter outbreaks of diarrhoea occur in United Kingdom too

EDITOR—Letrilliart et al conclude that the winter epidemic of diarrhoea in France in 1995-6 was not associated with consumption of tap water or shellfish, that person to person spread was implicated, and that the epidemic was of viral aetiology.1 Despite considerable reservations about the validity of their evidence we agree that the epidemic may have been largely attributable to viral diarrhoea.

Mean weekly incidence of infectious intestinal disease seen in general practice, by four week period and age group (from Royal College of General Practitioners' weekly return service, 1992-6)

• Download figure
• Open in new tab
• Download powerpoint

Winter outbreaks of viral diarrhoea associated with shellfish have been described in the United Kingdom,2 but the French study's failure to show an association may have been because of the methods used. Infection with small round structured virus, the most frequently identified cause of foodborne viral gastroenteritis, induces short term immunity,3 which could produce perverse effects in a study of this design—for example, people who eat oysters frequently might be differentially immune owing to higher exposure to risk inducing an apparent protective effect for frequent consumption. Separate analysis of primary and secondary cases would be more appropriate to test the primary hypothesis that food and water consumed were responsible for the epidemic.

The descriptive epidemiology and the design of the study are inadequate to explain the hypotheses selected. The apparent epidemic peaks in the incidence of infectious intestinal disease in January of most years are not clear,1 and figure 1 suggests that a high incidence occurred over the whole winter. A high intake of shellfish at Christmas and New Year would be expected to cause diarrhoea only in weeks 52, 1, and 2.

Disappointingly, the authors did not address age distribution; age is a crucial explanatory factor in the epidemiology of diarrhoeal illness. The incidence of diarrhoea is much higher in children than adults. During 1996 in the United Kingdom 17 140 episodes of infectious intestinal disease were reported in the weekly return service of the Royal College of General Practitioners; of these, 9283 were in children aged 0-4 and 5267 in children aged 5-14.4 We estimated that rotavirus was causing roughly three tenths of episodes of diarrhoea in children under 5 who consulted in general practice in the United Kingdom, and these cases showed a distinctive winter peak (figure).5 No seasonal pattern was apparent in older age groups.

To explain the aetiology of winter diarrhoea, detailed descriptive epidemiology combined with adequate bacteriological and virological investigation of the cases would be a more appropriate approach to assessing the potential risk factors. Such an approach is currently being taken by the Royal College of General Practitioners in collaboration with the Public Health Laboratory Service.

Authors did not rule out shellfish as a factor

EDITOR—Letrilliart et al claim to have excluded consumption of shellfish from the factors accounting for the winter outbreaks of diarrhoea in France.1 We did not find evidence in their paper to support this claim.

Firstly, the selection of cases was biased. The authors should have attempted to include all cases who consulted the doctors throughout the one month study rather than selecting only the first three cases who consulted each doctor.2 The first three patients with diarrhoea seen by a doctor during an epidemic of diarrhoea are unlikely to be representative of all the patients with diarrhoea seen by the same doctor during one month of the epidemic. Such a sample of cases is even less likely to be representative of cases in the community.

Secondly, the interpretation of the results given in the paper does not support the claim that consumption of shellfish was excluded as the source of the winter outbreaks. The finding that the cases with diarrhoea were more likely to have had contacts with other people with diarrhoea suggests that these cases were mostly secondary cases. Nothing was mentioned about the index cases, who are the key to the “cause” of the outbreak.

Index cases may have consumed raw shellfish and got infected with, for example, a Norwalk-like virus (small round structured virus).3 Transmission from person to person may have led to the secondary cases. Norwalk-like viruses can be transmitted effectively from person to person as well as through a contaminated food source (such as raw oysters) or water.3 The authors did not report other clinical symptoms that accompanied diarrhoea in their patients. Vomiting is one of the common symptoms of gastroenteritis due to Norwalk-like virus and the source of the name “winter vomiting disease.”

Finally, seasonality of gastroenteritis in the community is not limited to France. In England and Wales, general outbreaks of infectious intestinal diseases (community and institutional outbreaks combined) during 1992-4 peaked during summer.4 Salmonella infections, the commonest cause of these outbreaks, had peaks in July, while infections with Norwalk-like virus, the second commonest cause, had peaks in October.4

Authors' reply

EDITOR—We do not understand what Fleming et al mean by the statement that “the apparent epidemic peaks in the incidence of infectious intestinal disease in January of most years are not clear”; these peaks are clearly shown in figure 1 of our paper, in which the visual impression is supported by state of the art time-series analysis.1 We agree with Fleming et al that breaking down this figure by age group is common in epidemiology, as the Royal College of General Practitioners does routinely. We do not, however, believe that it was necessary in this paper, which was based on a case-control design; those interested can find this additional approach illustrated in Valleron's paper.2 That shows that, contrary to Fleming et al's assertion, diarrhoea in each age group, and not only young children, has a seasonal pattern, despite the differences in incidence. Osika and Muganwa-Kamya comment on the representativeness of our patients. Our efforts focused on the appropriate selection of controls so that biases were avoided and the relative risks were estimated as accurately as possible. Nevertheless, the cases that we included in the study were comparable for age, fever, history of contact, and geographical distribution with the cases reported in the Sentinelle system during the same period. The predictive value of vomiting in diarrhoea due to Norwalk-like viruses seems to us debatable.3 The hypothesis that some cases may be attributable to person to person transmission by index cases contaminated by infected shellfish does not contradict our results. We stated only that most cases of acute diarrhoea are not due to consumption of shellfish. From our results we calculated the maximum proportion of cases of acute diarrhoea attributable to consumption of raw shellfish to be 3.4% (95% confidence interval 0% to 10.8%), which is not significant.