StATS: What is a retrospective study?

A retrospective study is a study that looks backwards in time. For example, we find people that are already dead and try to figure out why they died. A retrospective study is fast. Since the subjects are already dead; we just have to tabulate all the results. The one problem is that it's hard to interview a dead person.

In contrast, a prospective study looks forward in time. For example, we select a group of subjects and sit around and watch them for a decade. A prospective study is slow. Unless you are studying a rapidly fatal disease, you have to wait years or even decades to accumulate sufficient data to draw any strong conclusions. On the other hand, live subjects make for a more informative interview.

That's a flip answer, but it summarizes the situation well. When we are studying a disease that takes a long time to appear, we usually need to use a retrospective study. The classic example is a study of smoking and lung cancer. We could find people as they start smoking and watch them from that point onward. This has been done, but it takes a couple of decades for the smoke to produce a tumor large enough to be noticed. Most studies look backwards in time, by selecting patients who have lung cancer and asking how much they smoked.

The biggest problem in a retrospective study is that some of the information that we need may be hard to get. When we use death certificates, for example, we usually need to interview family members in order to get some of our information. Even if the patient is alive, we have to rely on them to recall things that may have happened many years ago. Memory is a selective thing, and it can introduce all sorts of biases into our study.

For example, asking patients if they smoked is a tricky issue. Because the risk of cigarette smoking is well publicized, lung cancer patients are more likely to remember (and regret) their previous smoking history than the controls would. One safeguard is to make sure that the interviewer is blinded to the lung cancer status so that he or she doesn't probe harder for smoking history in the lung cancer cases. We can also look at those cases where patients were misdiagnosed with lung cancer. If those patients have the same smoking prevalence as the controls, then we have some assurance that the effect is real and not a product of faulty memories.

Bad joke

How many epidemiologists does it take to change a light bulb? None if it's a retrospective study, because the light bulb has already changed itself.

This page was written by Steve Simon while working at Children's Mercy Hospital. Although I do not hold the copyright for this material, I am reproducing it here as a service, as it is no longer available on the Children's Mercy Hospital website. Need more information? I have a page with general help resources. You can also browse for pages similar to this one at Category: Definitions, Category: Research designs.