Short definitions for popular epidemiologic study designs

Epidemiologic study designs allow researchers to investigate the cause of an outcome (disease, illness, or death), as well as exposures and risk factors that lead to the outcome.

Epidemiologic studies, when done efficaciously allow us to estimate the risk or odds of getting a particular disease, given a set of exposures or risk factors. This knowledge allows researchers to apply their finding to the group under study or to a larger population.

I recently enrolled in an epidemiology course to sharpen my understanding about various population-level study designs and I am starting to think more broadly about causal inference and similar topics. This post focuses on the study designs that have been discussed so far.

I have also posted about each separately to attempt to distill the details about each one.


What’s the difference between case-control studies and cohort studies ?

I have written previously about case-control and cohort studies. This post compares them to each other.

There are some important similarities and differences, which will be useful to discuss.

Case-Control Studies

In case-control studies, you start with people who already have the disease or a particular condition (these are your cases). You add these people to a group and then you create a similar group. These would be people who do not have the condition of interest but are in every other way, similar to our cases (these are our controls).

With both groups we look back (retrospectively) at different exposures that could have caused the outcome or diease/illness of interest. We then compare these two groups and try to determine if there are any differences in the presence or magnitude of the exposure.

This is the reverse of a cohort study in that in this approach we are starting with people who already have the disease (our cases )and people who do not have the disease (our controles) and looking back in time at their exposures to determine what exposure(s) where different in both groups.

Cohort Studies

In cohort studies, you start with a rare exposure of interest. The cohort can consist of disease free individuals who you follow over time to determine the risk of a specific outcome or disease occurring. Then you find a second group of people (controls) who are in every way similar to the first group except in the exposure of interest. You have to make sure that they are the same because the goal is to compare these two groups. You follow these groups over time and you examine what outcomes emerge.

Cohort studies can either be prospective (starting before the outcome) or retrospective (starting after the exposure is known but the outcome is still not known).


What is an experimental study?

An experimental study, sometimes called an intervention study, is a non-observational study where the investigator assigns the exposure (new drug, diet, programs, or intervention) to a group and follows the group over time to determine the relationship between the exposure and the outcome.

In experimental studies, if the exposure assignment is done randomly, this is a randomized control trial (the gold standard of study designs).

In this case, participants have an equal chance of being assigned to the experimental group or the control group.

To conduct experimental studies, you must have clinical equipoise. That is, there must be uncertainty about the exposure or treatment. Also, based on the Declaration of Helsinki, all study participants, including the controls, must have the best available treatment.


What is a case-control study?

A case-control study is an observational study where the outcome (disease/illness) is measured before the exposure is observed. In other words, we know what the outcome is, but we do not know what caused it (the exposure).

Unlike cohort studies which are suitable for studying rare exposures, case-control studies are useful for studying rare outcomes. In fact, the hardest part of a case-control study is ensuring that those in your control group (your controls) are similar in every aspect to your cases except on the outcome.


What is a cross-sectional study?

A cross-sectional study, sometimes called a prevalence study, is an observational study that collects data on the exposure and the outcome at a specific point in time. In other words, a cross-sectional study, allows the researcher to take a “snapshot” of the study population.

When is it useful to conduct a cross-sectional study?

Cross-sectional studies allow us to assess the prevalence of diseases in a given population. These studies are not ideal for diseases or illnesses that have a short duration. Imagine attempting to determine the prevalence of the flu, but conducting your study after flu season. On the other hand, cross-sectional studies skew us towards observing diseases or illness that have a long duration.


What is a cohort study?

Cohort studies are observational studies that compare outcomes (disease/illness) between groups (exposed and unexposed) over time. These types of studies are efficient for rare exposures and can either be retrospective (looking backward in time) or prospective (looking forward in time). In retrospective cohort studies, some participants may already have the outcome of interest, whereas, in prospective cohort studies, the researcher will follow the group until the outcome(s) of interest develop.

What’s the difference between a retrospective cohort study and a case-control study?

In cohort studies, you already know what the exposure(s) are. Your objective is to determine if those exposures led to the outcome of interest (disease, death or some other outcome).

In a retrospective cohort study, the exposed group may already have the outcome of interest, but you would examine them via historical medical records to observe them at a time before the outcome occurred. The goal here is to determine the risk of developing the outcome, given that they were exposed to the variable of interest.

In a case-control study, on the other hand, our primary concern is to pinpoint the exposure that caused the outcome of interest. In case-control studies, you already know the outcome, but you do not know the exposure. A case-control study is the most efficient study design when the outcome is rare.

When is it useful to conduct a cohort study?

Cohort studies are useful for studying rare exposures. However, conducting a cohort study might not be the best approach when there is a long time to the development of a disease. Cohort studies give us the real incidence rates and relative risks.

Example of a cohort study

The Framingham Heart Study (FHS) is an example of a cohort study. The goal of FHS is to examine the factors that lead to cardiovascular disease (CVD). The study enrolled people who had not developed CVD and followed them and their children over time.

In cohort studies, information can be collected on multiple exposures, and various outcomes of interest can be analyzed with minimal increase in costs.


The most significant threat to cohort studies is the loss to follow-up.

Cohort studies can also be very time consuming, and expensive, especially if the researcher has to wait several years before diseases start to manifest in enough individuals to obtain accurate results.

I hope this post is helpful to readers. I started MyStudentHq as a way to learn and write about what I learn. My foray into epidemiologic study designs on this site is related to my commitment to learning and writing boldly.

You may be interested in these posts…

You may support me with a generous cup of coffee.

What is an ecological study?

Ecological studies, sometimes called correlational studies, are observational studies that rely on population/group-level data rather than individual-level data. The units of comparison in these types of studies are often countries, family units, or neighborhoods.

When is it useful to conduct an ecological study?

Ecological studies are useful for determining correlations. This type of study only allows you to suggest that an association may be present. However, you would need to conduct more rigorous studies to be sure. Ecological studies are also good for generating hypotheses.

Ecological study example: Average wine sales data in high-income countries and the rate of throat cancer in each country.

Ecological studies are inexpensive and easy to conduct because the population-level data most often already exists. However, you may commit an ecological fallacy if you ascribe observed group-level associations to the individual. In our example, if you find that high-income countries have increased sales of wine and high rates of throat cancer, you commit an ecological fallacy if you conclude that an individual who frequently buys wine will also have an increased chance of getting throat cancer. This is an issue because ecological studies are not based on individual-level data.


It is important to note that ecological studies do not adjust for individual-level confounders.

Ecological studies have several limitations, including, misclassification of data, missing data, and errors in the data collection or measurement method. Ecological studies only suggest associations, and as a result, you are not able to determine relative risks or odds.

I hope this post is helpful to readers. I started MyStudentHq as a way to learn and write about what I learn. My foray into epidemiologic study designs on this site is related to my commitment to learning and writing boldly.

You may be interested in these posts…

You may support me with a generous cup of coffee.