Measures of Association
Epidemiology is all about identifying the link between exposure and disease. Are smokers more likely to develop lung cancer? Do helmets reduce the risk of head injury? To answer these questions, researchers rely on measures of association.
This blog post summarizes everything from your video lecture, with added detail and examples, so you can revise thoroughly.
Today’s lecture takes you from
- Understanding what association means in epidemiology,
- The study designs used for measurement of association,
- Tools or formula used to measure it and its interpretation, &
- Its public health implications and finally applying these concepts in case scenarios
🔎 What Do We Mean by “Association”?
In epidemiology, association means “a statistical relationship where an exposure and an outcome (disease) occur together more often than expected by chance.”
👉 Example: If lung cancer occurs much more frequently in smokers than in non-smokers, smoking and lung cancer are said to be associated.
🧪 Types of Observational Studies used to measure the association
To measure association, we first look at the types of observational studies available. we choose from three main observational study designs. Cross Sectional, Case-control and Cohort Study.
Cross-Sectional Study:
Cross-sectional studies give us a quick snapshot—exposure and disease are measured at the same time. They’re good for identifying correlations but not strong for cause-and-effect. in short:
Its a Snapshot in time.
Where, Exposure and outcome measured simultaneously.
They’re good for identifying correlations but not strong for cause-and-effect association.
Case-Control Study:
Case-control studies look backward in time, starting with people who already have the disease and comparing them with those who don’t, to see if past exposures differ. In short:
It works backward: start with people who have the disease (cases) and those who don’t (controls).
Look at past exposures
Cohort Study:
Cohort studies move forward in time, following exposed and non-exposed groups to see who develops the disease. These follow groups of exposed and non-exposed individuals over time to see who develops the disease, making them strong for showing cause and effect. In short
It works forward: start with cohort and followed for exposure and outcomes among them over time.
See who develops the disease.
“Because time frame is built into case-control and cohort designs, these studies are especially powerful for figuring out whether an exposure is truly associated with an outcome, so association can be measured. While in cross-sectional studies, exposure and disease are measured at the same time, limited in proving cause-and-effect since the sequence of events cannot be established”.
Once we have identified the study design, the next step is to know the tools for the measure of association.
🧮 Tools for Measures of Association in Case-Control and Cohort Studies
In case-control studies, the key measure of association is the Odds Ratio (OR), which compares the odds of exposure between cases and controls.
And in cohort studies, we’ll explore several important measures of association.
- Relative Risk, which compares the risk of disease in exposed versus non-exposed groups.
- Attributable Risk, which tells us how much of the disease among the exposed can actually be linked to the exposure.
- Attributable Risk Ratio, which expresses that excess risk as a percentage. And
- Population Attributable Risk, which reflects the overall impact of an exposure on the entire community.
📝 Approach to Calculation of Measures of Association
Now that we know which measures of association are used in case-control and cohort studies, let’s dive into the calculation and interpretation. The approach is simple and follows three steps:
- First, we summarize the data in a 2by2 contingency table.
- Next, apply the appropriate formula to calculate the measure of association.
- And finally, focus on the interpretation, to understand what the number really means in terms of risk or protection.
Now it’s time to discuss one of the most important tools in epidemiology—the 2by2 contingency table.
📐 1. Understanding the Contingency Tables
This table above is 2by2 Contingency table which helps us organize study data by dividing participants into four groups based on exposure status and disease outcome. From here, we can easily calculate measures of association like the Odds Ratio in case-control studies, or Relative Risk and Attributable Risk in cohort studies. In short, the 2by2 table is the foundation for turning raw data into meaningful epidemiological insights.
Now that we’ve built our 2by2 contingency table, it’s time to discuss the formulas for calculating measures of association.
📐 2. Calculating Measures of Association: Key Formulas
a. For Case-Control Study
For case-control studies, we’ll use the formula for the Odds Ratio, which compares the odds of exposure among cases to the odds of exposure among controls.
In the table below, the columns separate those with the disease from those without the disease. In the top row we have people who were exposed, and in the bottom row those who were not exposed. So, box ‘a’ is exposed with disease, ‘b’ is exposed without disease, ‘c’ is unexposed with disease, and ‘d’ is unexposed without disease. From this simple setup, we can calculate the Odds Ratio using the formula: ad divided by bc.
b. For Cohort Study
For cohort studies, we’ll calculate the
- Relative Risk, which compares the risk of disease in exposed versus non-exposed groups.
- From the same table, we can also derive the Attributable Risk,
- The Attributable Risk Ratio, and
- The Population Attributable Risk
each giving us deeper insight into how exposure influences disease, both at the individual and population level.
Here we have the 2by2 table for a cohort study. From this table, we can calculate several key measures of association:
- Incidence in the exposed group is simply a over a plus b.
- Incidence in the non-exposed group is c over c plus d.
- Relative Risk is the ratio of these two incidences—Ie divided by Io.
- Attributable Risk (AR) is the difference—ie minus Io.
- Attributable Risk Percent (AR% or Risk Ratio) shows the percentage of disease among the exposed that is actually due to the exposure, calculated as AR over Ie times 100.
- Finally, Population Attributable Risk (PAR) looks at the overall population by subtracting Io from the total incidence, while PAR% expresses this as a percentage of the total risk.
🧾 Examples of Measures of Association
These are the examples for you to work on. Try identifying the study design, choose the correct measure of association. The approach is 3 step simple method. As of now try to summarize your data in 2by2 table and apply the formula for the calculation of measures of association in two study set up.
For the calculation and interpretation, we will discuss everything step by step in the next session. This brings us to the end of today’s session.
References:
- Park’s Textbook of Preventive and Social Medicine (28th Edition)
- GORDIS EPIDEMIOLOGY (6th Edition)
- Medical Biostatistics by Abhaya Indrayan (4th Edition)