Binary Endpoints: Simple Definition, Examples

Randomized Clinical Trials > Binary Endpoints

What are Binary Endpoints?

binary endpoints
Binary endpoints have two choices.
A binary endpoint is a conclusion with two clear choices. For example, if a clinical trial is trying to find out if a drug cures a disease, the endpoints might be:

  • “Cure” and “no cure”,
  • “Remission” or “No remission”,
  • “Symptom relief” or “no symptom relief”.

Endpoints can also be phrased as a simple yes/no option:

  • Major cardiac event: “Yes” or “No”,
  • Disease recurrence: “Yes” or “No.”

They can also be created from continuous variables by splitting the variables into two. For example, a trial investigating a weight loss drug might split weight loss into “Under 20 lbs” and “20 lbs or more”.

Binary choices are used because they are simple to understand, but they can result in larger sample sizes. They also tend be be more clinically relevant; patients and physicians want to know if a drug cures, or doesn’t cure. They are usually coded as 0 for No and 1 for Yes.

Multiple Binary Endpoints

Sometimes a trial will have multiple binary endpoints (sometimes called binary-event composite endpoints). Instead of one possible Yes/No event, there are several. For example, a researcher might be investigating preventive care choices in the inner city. Endpoints might include:

  • Mammogram: Yes / No,
  • Pap Smear: Yes / No,
  • Routine Bloodwork: Yes / No.

Studies often have multiple endpoints to better capture the process being studied, compared to a single outcome.

Composite outcomes typically increase statistical power, which basically means the study’s result will hold more weight. The higher the power, the more a study’s results can be trusted. However, this doesn’t mean that a study’s power can be increased simply by adding more binary outcomes; outcomes with small numbers of data points or a very small relative risk will reduce power.

References

:
Evans & Ting. Fundamental Concepts for New Clinical Trialists.
Lee et. al. (2009). Analysis of Group Randomized Trials with Multiple BE and Small Number of Groups. PLOS One.


Comments? Need to post a correction? Please Contact Us.

Join us on our YouTube Channel