I am one month in to my new rotation at the Cole Eye Institute and have already begun working on several studies! One project allowed me to build a research proposal on Retinal Vein Occlusions (RVO). To say I was flattered to be entrusted to build an experiment was an understatement!

But the fact is whenever you a designing a new population health study, it is challenging to figure out the best research design for your study. But why is it important? The type of design you choose will have its own sources of error and bias limiting certain types of calculations. If you mistakenly choose a different type of design, you may bias your own results or present them as stronger or weaker associations than they really are. Choosing the correct design is vital so we can do the correct analysis the first time. So let’s dive in and look at two of the most commonly confused study designs.

We all know that carrots are good for your eyes. So lets say I am want to study if eating carrots makes you less likely to need glasses when you are older. The exposure I am studying is eating carrots and my outcome is if an individual needs glasses or not.

A cohort study is when you start by evaluating your exposure, gather participants, and then stratify into your two outcome groups, in this case not needing glasses or having glasses. Cohort studies can be both prospective or retrospective, but even if you are doing a retrospective chart review, your inclusion criteria should be based on your exposure, NOT your outcome! Stratifying always comes after data collection and in the analysis phase.

Now, lets look at the Case-Control study.

In this study design you START with the outcome and then define exposure groups. In contrary to Cohort studies, this is a great design choice for rare disease. For instance, Idiopathic Intracranial Hypertension is a rare vision threatening disorder and we have no idea what causes it. This type of design allows us to gather these cases, the outcome, first and then study many different exposures that may be associated with the disease. Remember, the only study design that can truly tell us causality is a Randomized Control Trial. But for a rare disease like this, it can be extraordinarily challenging and expensive, so alternative designs are used. 

Although study design can be challenging, the most important thing to remember is to spend time designing your study and building an analysis plan in advance.