Foodborne Outbreak at a Village Wedding
A wedding reception in a village serves 120 guests. Within 48 hours, 45 guests develop diarrhea, abdominal cramps, and fever. The local health team interviews all guests about foods eaten. Among 80 guests who ate rice, 40 became ill; among 40 who did not eat rice, 5 became ill. Among 60 who ate chicken, 35 became ill; among 60 who did not, 10 became ill. Among 50 who ate salad, 20 became ill; among 70 who did not, 25 became ill. The rice was cooked in the morning and kept at room temperature until the evening. The team wants to know which food was the likely vehicle and whether a cohort study is needed. Several guests who did not eat rice ate chicken and also became ill.
1. Calculate the attack rate for rice eaters and non-rice eaters, and the relative risk and attributable risk for rice. Show your work.
Read the worked answer
Rice eaters: 40 ÷ 80 = 0.50, or 50%. Non-rice eaters: 5 ÷ 40 = 0.125, or 12.5%. Relative risk = 0.50 ÷ 0.125 = 4.0. Attributable risk = 50% − 12.5% = 37.5 percentage points. This means rice eaters had four times the risk of illness, and about 37.5 cases per 100 rice eaters could be attributed to rice if the association is causal.
2. Which food is the most likely vehicle? Use the attack rates for chicken and salad to justify your answer.
Read the worked answer
Chicken: 35 ÷ 60 = 58.3% among eaters versus 10 ÷ 60 = 16.7% among non-eaters, RR = 3.5. Salad: 20 ÷ 50 = 40% versus 25 ÷ 70 = 35.7%, RR about 1.1, so salad is not a strong candidate. Rice has the highest RR (4.0) and a biologically plausible vehicle: cooked rice held at room temperature can grow Bacillus cereus. However, chicken also has a high RR, and guests often ate both foods. The most likely single vehicle is rice, but a stratified or multivariable analysis is needed to separate rice from chicken.
3. What study design is this, and what confounding or bias could explain the association with rice?
Read the worked answer
This is a retrospective cohort study because exposure and illness were assessed after the outbreak in a defined group of guests. Confounding could occur because rice eaters were more likely to eat chicken, so the rice association may partly reflect chicken. Recall bias is possible if ill guests remember foods differently. Food-specific attack rates do not adjust for other foods. Stratify by chicken consumption or use a multivariable model, and inspect kitchen storage and cooking times. A case-control design would be less efficient here because the cohort is known.