Science / Microbe of the Month: One wedding and a chicken oscar: Salmonella bacteria multiply prodigiously, but it is the initial dosage that is critical, says Bernard Dixon

Take 16mg of strychnine and you're a gonner. Take less and you stand a chance of surviving the poison.

The relevance of the dose of a chemical to its harmful effects is well known. We are all, day by day, exposed to minute traces of substances which, in much larger quantities, would be hazardous. Far less widely recognised is the fact that the dose of a potentially harmful microbe also has a strong bearing upon whether we become ill following infection.

Details of an epidemic of salmonella food poisoning in Connecticut, published this month in Epidemiology and Infection, demonstrate dramatically how the dose of this bacterium affects not only the severity of the disease, but also the incubation period and even symptoms suffered. The story carries an important lesson for food safety, particularly in the warm summer months. For example, the longer a piece of chicken is left unrefrigerated, the greater the opportunity for food-poisoning bacteria in the meat to proliferate and reach a level at which they can trigger a serious illness.

Surprisingly, perhaps, the importance of the dose of salmonella bacteria has not been thoroughly studied in the past. Indeed, investigations of food-poisoning outbreaks, based on the pinpointing of particular items eaten, tend to reinforce the view that infection is an all-or-nothing affair. Eat a contaminated pie and you become ill; don't and you don't.

The Connecticut incident came to light when a nurse reported 10 patients with food poisoning, on a single day, to a local health department. All had attended a wedding reception the previous day, and the only meal they reported in common was dinner, which had been served to 180 people. It consisted of an appetiser buffet, salads, vegetables, ice-cream, wedding cake and chicken oscar topped with hollandaise sauce.

The investigation by a team from the Centers for Disease Control and Prevention (CDC) in Atlanta, Georgia, was in part conventional. Questionnaires were given to all of the guests to determine how many had become unwell. Of the 178 who returned them, 168 had been ill with diarrhoea following the reception. One other person did not have diarrhoea but reported abdominal cramps - another feature of salmonella food poisoning. Every one of these 169 guests reported eating the chicken oscar, as did five of nine respondents who had not succumbed. None of the other food items were clearly linked with illness in this way.

Soon, suspicion fell on the hollandaise sauce, rather than the chicken itself. In addition to its strong association with the guests' food poisoning, it was the only item of food that was sampled by two chefs in the kitchen, and they also proved to have Salmonella enteritidis in their stools. The sauce had been prepared, using 150 eggs, at temperatures that do not kill salmonellae. And tests on hens on the farm that produced the eggs showed that one of them was carrying the same strain of S. enteriditis in her ovaries.

As the Edwina Currie affair reminded us in 1988, salmonella infection is not uncommon in hens' eggs. That being so, it is hardly surprising that hollandaise and bearnaise sauces, prepared from eggs with only very gentle heating, are potential sources of the infection. On this occasion, hollandaise sauce made from contaminated eggs was established beyond reasonable doubt as the origin of the epidemic.

The CDC investigators did not stop there, however. They were also keen to examine whether the guests' illnesses reflected in any way the actual dose of bacteria they had received - based on the reasonable supposition that, following thorough mixing of the sauce, every portion would contain roughly equal numbers of salmonellae. So the team used further questionnaires in which they asked how much of each food item the guests had consumed ('none', 'some', 'most' or 'all') and sought detailed information about their food poisoning. Questions included the precise symptoms, the severity of the condition on a scale from one to 10, the maximum number of loose stools in any 24-hour period and a day-by-

day calendar of their health following the wedding.

The results were striking. First, as the dosage increased, the incubation period of the disease (assessed from the daily calendar) declined. Whereas the guests who had consumed the smallest amounts of sauce became ill on average about 37 hours afterwards, those who had eaten the most developed symptoms in 17.5 hours or less. At the same time, greater proportions reported vomiting and bodily aches. Likewise, the higher the dose of S. enteritidis, the greater the maximum daily number of stools, the greater the loss in body weight, the worse the patients' estimated severity of their condition and the more days they were confined to bed.

There were, as expected, variations in each case. Many factors, other than the virulence of a microbe, help to determine the outcome when an individual is infected. They range from age and nutritional well-being to the state of various bodily defences. Nevertheless, these dose-response trends were all highly significant statistically, and similar to those one would expect if the food had contained a chemical poison rather than a disease-causing microbe.

S. enteritidis is closely related to S. typhi, the agent of the potentially fatal disease typhoid fever. Food poisoning caused by S. enteritidis is by no means a trivial condition and, can, as in the Connecticut outbreak, have serious consequences. In this case, 52 of the individuals became so dehydrated that they had to receive intravenous fluids, and 20 were sufficiently ill to be admitted to hospital.

The CDC study holds important lessons for future investigators - for example, in showing that the pattern of symptoms seen in an epidemic can suggest when the contaminated food was consumed. But its most compelling message is for the kitchen, shop and restaurant. Left at room or body temperature, food so lightly contaminated that it poses no real risk may, after just a few hours of prodigious multiplication, contain a population of bacteria capable of provoking life-threatening disease.