Michael Nelson, Biostatistics trainee
In April 1940, Dr AM Rubin, epidemiologist-in-training in New York, was assigned to conduct an investigation into the occurrence of an outbreak of acute gastrointestinal illness in the village of Lycoming, Oswego County.[1]
When Dr. Rubin arrived he learned that all persons known to be ill had attended a church supper held on the previous evening. Family members who did not attend the church supper did not become ill. Dr Rubin focused the investigation on the supper. He completed interviews with 75 of the 80 persons who attended, collecting information about onset of symptoms and foods consumed. Of the 75 persons interviewed, 46 persons reported gastrointestinal illness.
The illness was characterized by acute nausea, vomiting, diarrhoea and abdominal pain. None of the ill persons reported having an elevated temperature and all recovered within 24 to 30 hours. No faecal specimens were obtained for bacteriologic examination.
The supper was held in the basement of the village church. Foods were contributed by numerous members of the congregation. Food was spread out on a table and consumed over several hours. Data regarding onset of illness and food eaten or water drunk by each of the 75 persons interviewed has been collated. Of the various food and drink available at supper, the most commonly eaten foods amongst the 46 persons who reported gastrointestinal illness were vanilla ice cream, baked ham, cakes, spinach and chocolate ice cream.
Forty three of 54 persons who ate vanilla ice cream reported gastro while 3 of 21 who did not eat the vanilla ice cream reported gastro.
What is the best measure of the effect of eating vanilla ice cream with respect to suffering from gastro? Calculate it.[2]
Answer
The Relative Risk (RR) is the best measures of effect.
A RR is the ratio of the probability of the event in the exposed group and the probability of the event in the unexposed group.
The exposed group are those who ate vanilla ice cream. The probability of reporting gastro in this group was 43/54=0.7963.
The unexposed group are those who did not eat vanilla ice cream. The probability of reporting gastro in this group was 3/21=0.1429.
The relative risk is therefore 0.7963/0.1429=5.6.
People who ate vanilla ice cream were 5.6 times more likely to get gastro compared with those who did not eat vanilla ice-cream.
