92 PART 2 Examining Tools and Processes

Making statements about the population

Ecologic studies (also called correlational studies) and cross-sectional studies

appear in the next level up from expert opinion and case studies and case series.

Though ecologic studies are still descriptive designs that provide weak evidence,

they have the advantage of having potentially very large samples.

In ecologic studies, the experimental units are often entire populations (such of a

region or country). For example, in a study presented in Chapter 16 of Epidemiology

For Dummies by Amal K. Mitra (Wiley), the experimental unit is a country, and

15 countries were included in the analysis. The exposure being investigated is fat

intake from diet (which was operationalized as average saturated fat intake as a

percentage of energy in the diet). The outcome was deaths from coronary heart

disease (CHD), operationalized as 50-year CHD deaths per 1,000 person-years

(see Chapter 15 for more about rates in person-years). Figure 7-3 presents the

results in the form of a scatter plot.

As shown in Figure 7-3, the country’s average value of the outcome (rate of CHD

deaths) is plotted on the y-axis because it’s the outcome. The exposure, average

dietary fat intake for the country, is plotted on the x-axis. The 15 countries in the

study are plotted according to their x-y coordinates. Notice that the United States

is in the upper-right quadrant of the scatter plot because it has high rates of both

the exposure and outcome. The strong, positive value of correlation coefficient r

(which is 0.92) indicates that there is a strong positive bivariate association

between the exposure and outcome, which is weak evidence for causality (flip to

Chapter 15 for more on correlation).

FIGURE 7-3:

Ecologic study

results.

© John Wiley & Sons, Inc.