In 1995 there was an outbreak in Sydney’s south involving 70 patients diagnosed with laboratory confirmed cryptosporidiosis. There was no obvious link between the patients. Public health investigators decided to do an epidemiological study to try to identify the source of the outbreak. With their GP’s permission, investigators contacted the patients and interviewed them about risk factors.
The reported exposures of the 35 most recently diagnosed patients were then compared with a comparison group. The comparison group included 35 people who were recruited by calling people selected from the electronic white pages who lived in the same neighbourhoods and who were about the same ages as the patients, but denied illness. The comparison group was asked about the same exposures as the cases, over a comparable 2 week period.
The distribution of reported exposures of the patients and the comparison group were similar for potential sources of infection, including consumption of treated and untreated water and attendance at childcare. However 17 patients reported swimming at a particular pool (pool A) in the 2 weeks before the onset of their illness, compared with 7 people in the comparison group.
What is the epidemiological measure you could use to compare these exposures, and how would you calculate it?
Answer
You can do a case control study and calculate an odds ratio. If the case control study involved matching a control to each case (eg, based on age and location of residence), then you should calculate a matched odds ratio. But we will assume the study was unmatched. Using the cell labels in the table below, the odds ratio (OR) is calculated as:
OR = (AxD)/ (BxC)
That is, OR = (17x28)/(7x18) = 476/126 = 3.78
Reference
