Case-control study

From RationalWiki
Jump to: navigation, search
The poetry of reality
Science
Icon science.svg
We must know.
We will know.
A view from the
shoulders of giants.

A case-control study, or retrospective study, is a study design that looks for relationships between types of data.[1] It is used especially in epidemiology. It differs from randomized controlled trials (RCTs) in several ways, and is usually considered a somewhat weaker type of evidence.

Rather than randomly assigning a group of subjects to different exposures (treatment vs. placebo), it takes groups of people who already have the condition in question, matches them with a similar group that does not, and uses statistics to look for commonalities and differences between the groups.

For example, one particular study[2] was designed to look at the relationship between head injuries and wearing bicycle helmets. By looking a groups of people who were injured in bicycle accidents, and comparing the rates of head injuries in groups that did or did not wear helmets, the authors were able to find a strong association between not wearing a helmet and sustaining a serious head injury.

There are limits to this type of study, especially in establishing causation (vs. correlation). Perhaps the group of people who wore helmets had more careful habits in general and were more careful riders.

This type of study is invaluable, however, as designing an RCT to look at this association would be unethical. (Imagine randomly assigning 300 people to ride with helmets, and 300 without for a year, and seeing which group had more head injuries — yuck).

References[edit]

  1. Case-control and cross sectional studies, Epidemiology for the uninitiated, 4th ed, Coggon, Rose, and Barker, Ch 8
  2. A case-control study of the effectiveness of bicycle safety helmets by R. S. Thompson et al. N. Engl. J. Med. 1989 May 25;320(21):1361-7.