Composite Endpoints

A clinical trial needs an endpoint. Something to measure. "Death" is a great endpoint. It is important. It is easy to measure.

But death is rare. To prove a drug prevents death, you need 10,000 patients. That is expensive.

The Trick

So, researchers use a Composite Endpoint. They combine several events into one: "Major Adverse Cardiac Events (MACE): Death + Heart Attack + Stroke + Hospitalization for Chest Pain."

Now, if the drug prevents any of these, the study is a success.

The Problem

These events are not equal. Death is catastrophic. Hospitalization for chest pain is inconvenient.

Often, a drug will reduce "Hospitalization" (the soft endpoint) but have zero effect on "Death" (the hard endpoint).

But the headline says: "Drug Reduces Major Cardiac Events by 20%!"

It sounds like it saves lives. In reality, it just saves bed space.

The Diagnosis

Unpack the composite. Look at the individual components.

If the victory is driven entirely by the least important component, be skeptical.