Tweetorial: Intro to Case-Control Studies

This is my first attempt at a #tweetorial, so taking up something relatively simple for it… Case-Control Studies. Let me begin with a reminder of one of the most classic #CCStudies of all times, where study 7 patients led to a conclusive finding! 1/14

Case-Control Studies investigate the relationship between a health/disease outcome and one or more putative factors (risk/protective) associated with the disease. They start from the point where the outcome of interest has already happened, and look back in time. 2/14

This is different from cohort or clinical trials, where the exposure (risk/protective/interventions) factors are known, and then the individuals are followed up over time to see if the outcome of interest (health/disease condition) take place. 3/14

Case-control studies can be done quickly, using very little resources and can help in hypothesis generation. This is the preferred choice when the disease is rare overall (<10% prevalence). It is a good starting point for epidemiological investigations into emerging threats. 4/14

The case-control design is susceptible to several limitations. One of them is selection bias. Selected cases should be representative of all cases, but sometimes, they may not be so! Selecting controls maybe an even more difficult task! 5/14

Controls should come from the same population/risk group as cases. They differ from cases in that they did not develop the disease – and case-control designs try to unmask why this difference happened. Which factors protected them, while cases got sick? 6/14

Selection of cases could be straightforward, esp for a rare condition. But selection of appropriate controls could be difficult. Different groups may be selected, but each has limitations. No perfect controls, then, eh? 7/14

Another problem in case-control studies is Recall Bias. Case groups tend to recall more details; they lived through the health/disease condition in question. Controls may not be able to recall as much information. 8/14

Since the whole case-control comparison hinges on retrospective data extraction, this recall bias can be a big problem. Bypass by having a short recall period, using recorded data and blinding. 9/14

The great Dave Sackett has elucidated beautifully the biases associated with case-control designs. Check it out for more details: 10/14

So, to summarize, pros of case-control studies: good for rare conditions; quick to do; cheap; gives odds ratio (which can be similar to relative risk for rare outcomes); hypothesis generation. 11/14

And cons of case-control studies: bias and confounding from sampled populations; recall bias; limited to one outcome; unclear sequence of events; no causality; no prevalence, incidence, excess risk measure. 12/14

Good to remember the golden rule of epidemiology: Association/Correlation does not imply causation. 13/14

"Case–control studies are the ‘‘house red’’ on the research design wine list: more modest and a little riskier than the other selections but much less expensive and sometimes surprisingly good." 14/14

That's it folks!

Originally tweeted by Pranab Chatterjee (@scepticemia) on May 30, 2020.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.