FIGURE 7-5: Example of a typical case-control study 2x2 table.

As shown in Figure 7-5, what is important is not the 2x2 table itself, but the order in which the counts

are filled in. Notice that at the beginning of the study, you already knew the case total was 30, and you

had determined that your control total would be 30 (although you are allowed to sample more controls

if you want in a case-control study).

The correct measure of relative risk to present for a case-control study is the OR (as

described in Chapter 14). It is important to acknowledge that when you present an OR from a

case-control study, you interpret it as an exposure OR, not an outcome or disease OR. (It is also

acceptable to present an OR in a cross-sectional study, but in that case, you are presenting an

outcome or disease OR.)

In a case-control study, because the condition is rare, you are sampling on the outcome and

calculating the likelihood that the cases compared to controls were exposed. This study design is

seen as extremely biased, which is why cohort studies are preferred, and are at a higher level of

evidence. However, case-control study designs are necessary for rare diseases.

Following a cohort over time

In the previous section, we pointed out that case-control studies are used for rare diseases. Therefore,

case-control studies do not have large sample sizes, which is evident in Figure 7-5 where the total

sample size is 60. In contrast, cohort studies are used for studying common conditions, such as HTN,

so they include very large sample sizes — but they still use the same 2x2 table for interpretation.

Figure 7-6 offers an example of what a 2x2 table for a cohort study might look like where the exposure

is high alcohol intake and the outcome is HTN.

© John Wiley & Sons, Inc.

FIGURE 7-6: Example of a typical cohort study 2x2 table.

As shown in Figure 7-6, the total number of participants is large. This cohort of 600 participants could

have been naturally sampled from the population, or they could be stratified by exposure, meaning that

the study design could require a certain number of participants to be exposed and to be unexposed.

Imagine that you insisted that 300 of your participants have high alcohol intake, and 300 have low

alcohol intake. It may be harder to recruit for the study, but you would be sure to have enough exposed