Keep both exposure groups
In a hypothetical meal investigation, 30 of 50 people who ate a salad became ill, compared with 10 of 50 who did not.
| Group | Ill | Total | Attack rate |
|---|---|---|---|
| Ate salad | 30 | 50 | 60% |
| Did not eat salad | 10 | 50 | 20% |
The risk ratio is 60% divided by 20%, or 3. The risk difference is 40 percentage points.
The arithmetic does not identify the ingredient
People who ate the salad may also have eaten another food. Recall may be imperfect, and the case definition or attendance list may be incomplete. The observed association needs investigation.
A high attack rate among eaters alone is less informative without the comparison group. If almost everyone ate a food, few non-eaters may remain for a precise contrast.
Align the definitions
Use a consistent illness definition and follow-up period. Distinguish “did not eat” from “does not remember.” Examine other exposures and supporting laboratory or environmental evidence where appropriate.
Food-specific rates organize a hypothesis. They do not justify accusing a source solely because one item has the largest number in a table, particularly when many foods were compared and consumption patterns overlap.
