Sixty-Four Diarrhoea Cases After a Wedding Lunch: Attack Rates and a Red Herring
On Monday morning a district health office receives a call from a wedding hall. Of 180 guests who attended a lunch on Saturday at 13:00, 64 have since had watery diarrhoea, and 14 report fever above 38 °C. The median time from the meal to first symptom is 28 hours, the median illness duration is three days, and no one has been hospitalised. A rapid telephone survey of guests gives: chicken karahi eaten by 90 guests, of whom 55 are ill; rice eaten by 130, of whom 48 are ill; salad eaten by 70, of whom 30 are ill; and of the 90 guests who did not eat chicken, 9 are ill. The kitchen has a single hand-washing sink, one food handler has a discharging finger wound, and leftovers were kept in a warm room for four hours before refrigeration. Stool samples from five patients are being sent to the laboratory.
1. Calculate the overall attack rate and the attack rate ratio for each food item. State which item is implicated and how confident you are, indicating the analysis that would firm up the conclusion.
Read the worked answer
Overall attack rate = 64/180 = 35.6%.
Chicken karahi: 55/90 = 61.1% among those who ate it versus 9/90 = 10.0% among those who did not. Attack rate ratio = 6.1, with a risk difference of 51 percentage points.
Rice: 48/130 = 36.9% versus 16/50 = 32.0%, ratio 1.15.
Salad: 30/70 = 42.9% versus 34/110 = 30.9%, ratio 1.39.
Interpretation. Chicken is the only item with a large, specific association, and the 28-hour median incubation fits a food-borne infection rather than a preformed toxin. Rice and salad show weak associations of the size expected from confounding: nearly everyone ate chicken, so the unexposed reference groups are small and contaminated hands or surfaces used for the salad may simply reflect the same source.
How to firm it up. Interview all 180 guests - not a convenience sample - and stratify by chicken consumption: if the salad ratio falls towards 1 within both chicken strata, salad is confounded. Also examine dose-response (amount of chicken eaten), check whether the salad and chicken were handled by the same unwashed hands, and confirm with stool or food cultures. Withholding judgement until the stratified analysis is the discipline that distinguishes a field investigation from a guess.
2. The caterer insists the food handler's finger wound proves this is staphylococcal food poisoning. Evaluate that claim using the incubation period and case features, and name the more likely agents.
Read the worked answer
Reject it on the epidemiology. Staphylococcus aureus food poisoning is caused by preformed enterotoxin in food held at room temperature. It has an incubation of 1-6 hours (usually 2-4), a duration under 24 hours, vomiting as the dominant symptom, and little or no fever. Here the median incubation is 28 hours, diarrhoea dominates, 14 of 64 cases (22%) have fever, and the median duration is three days. Four independent features are incompatible with staphylococcal intoxication.
More likely agents are those causing infection or toxin-mediated enteritis after 18-48 hours: non-typhoidal Salmonella, enterotoxigenic Escherichia coli, Campylobacter jejuni or Shigella species. Four hours of warm holding of leftovers is a textbook amplification step, and the low infective dose of Salmonella and Shigella fits a large single-source outbreak.
The wound is a red herring but not irrelevant. A finger wound matters chiefly for staphylococci and, if faecally contaminated, for Salmonella typhi or paratyphi and hepatitis A. Manage it as a hygiene breach - swab it, exclude the handler from food work until healed, and review hand-washing - but do not let it dictate the aetiology. State plainly that stool and food culture or PCR will settle the question, and that the investigation should not be steered by the caterer's liability concerns.
3. List your immediate public health actions, including which specimens to collect, and state what you would tell the public and the caterer about antibiotics.
Read the worked answer
1. Notify. A suspected food-borne outbreak is notifiable; inform the public health authority the same day and preserve all remaining food, leftovers, packaging and kitchen records.
2. Case finding. Build a line list of all 180 guests with exposure, onset time and symptoms; interview by telephone. Collect stool or rectal swabs from at least 10 ill guests, from all food handlers including the one with the wound, and swab the wound itself. Send implicated food and water samples for culture and, where available, multiplex PCR.
3. Environmental investigation. Inspect the kitchen against hand-washing, refrigeration, separation of raw and cooked food, and holding-temperature standards. Suspend the hall from catering until corrected; exclude the handler until asymptomatic for 48 hours and the wound is healed.
4. Control measures. Safe water, thorough reheating, no warm holding beyond two hours, staff training, and written records.
5. Antibiotics. Tell the public and caterer that this illness is usually self-limiting in healthy adults and that routine antibiotics are not indicated: they do not shorten most cases, they prolong convalescent carriage, and in some Salmonella infections they increase the risk of relapse and complications. The main treatment is oral rehydration. Reserve antibiotics for severe disease, invasive infection, or high-risk patients, and only on clinical advice.
6. Communicate a single agreed spokesperson, and commit to a written final report with the analytical results.