relationship is and in what direction it goes. A simple solution to this is to present the test statistic and

p value to represent how strong the relationship is, and to present the actual results as row or column

percentages to indicate the direction. For the data in Figure 13-2, you could say that being obese was

associated with having HTN, because 14/21 = 66 percent of obese participants also had HTN, while

only 12/39 = 31 percent of non-obese participants had HTN.

Quantifying associations

How strongly is an exposure associated with an outcome? If you are considering this question with

respect to exposures and outcomes that are continuous variables, you would try to answer it with a

scatter plot and start looking for correlation and linear relationships, as discussed in Chapter 15. But

in our case, with a fourfold table, you are essentially asking: how strongly are the two levels of the

exposure represented in the rows associated with the two levels of the outcome represented in the

columns? In the case of a cohort study — where the exposure is measured in participants without the

outcome who are followed longitudinally to see if they get the outcome — you can ask if the exposure

was associated with risk of the outcome or protection from the outcome. In a cohort design, you could

ask, “How much does being obese increase the likelihood of getting HTN?” You can calculate two

indices from the fourfold table that describe this increase, as you discover in the following sections.

Relative risk and the risk ratio

In a cohort study, you seek to quantify the amount of risk (or probability) for the outcome that is

conferred by having the exposure. The risk of getting a negative outcome is estimated as the fraction of

participants who experience the outcome during follow-up (because in a cohort design, all participants

do not have the outcome when they enter the study). Another term for risk is cumulative incidence rate

(CIR). You can calculate the CIR for the whole study, as well as separately for each stratum of the

exposure (in our case, obese and nonobese). Using the notation in Figure 13-1, the CIR for participants

with the exposure is

. For the example from Figure 13-2, it’s

, which is 0.667 (66.7

percent). And for those without the exposure, the CIR is represented by

. For this example, the

CIR is calculated as

, which is 0.308 (30.8 percent).

The term exposure specifies a hypothesized cause of an outcome. If it is found that a certain

exposure typically causes risk for an outcome, it is called a risk factor, and if it is found to confer

protection, it is called a protective factor. Higher education has been found to be a protective

factor against many negative outcomes (such as most injuries), and obesity has been found to be a

risk factor for many negative outcomes (such as HTN and Type II diabetes).

The term relative risk refers to the amount of risk one group has relative to another. This chapter

discusses different measures of relative risk that are to be used with different study designs. It is

important to acknowledge here that technically, the term risk can only apply to cohort studies because

you can only be at risk if you possess the exposure but not the outcome for some period of time in a

study, and only cohort studies have this design feature. However, the other study designs — including

cross-sectional and case-control — intend to estimate the relative risk you would get if you had done a

cohort study, so they produce estimates of relative risk (see Chapter 7).

In a cohort study, the measure of relative risk used is called the risk ratio (also called cumulative

incidence ratio). To calculate the risk ratio, first calculate the CIR in the exposed, calculate the CIR in