Suppose that each subject in a large study, at a particular time, is classified
as positive or negative according to some risk factor, and as having or not
having a certain disease under investigation. For any such categorization the
population may be enumerated in a 22 table (Table 1.11). The entriesA,B,
CandDin the table are sizes of the four combinations of disease presence/
absence and factor presence/absence, and the numberNat the lower right cor-
ner of the table is the total population size. The relative risk is
RR¼
A
AþB
o
C
CþD
¼
AðCþDÞ
CðAþBÞ
In many situations, the number of subjects classified as disease positive is
small compared to the number classified as disease negative; that is,
CþDFD
AþBFB
and therefore the relative risk can be approximated as follows:
RRF
AD
BC
¼
A=B
C=D
¼
A=C
B=D
where the slash denotes division. The resulting ratio,AD=BC, is an approxi-
mate relative risk, but it is often referred to as anodds ratiobecause:
1.A=BandC=Dare the odds in favor of having disease from groups with
or without the factor.
TABLE 1.11
Disease
Factor þ Total
þ AB AþB
CD CþD
Total AþCBþDN¼AþBþCþD
20 DESCRIPTIVE METHODS FOR CATEGORICAL DATA