How to read a weight-loss trial headline without being misled

Weight-loss trials are reported with a single eye-catching percentage. Before comparing two of them, ask these five questions.

1. Which estimand?

• Treatment regimen (sometimes "efficacy" or "if adherent"): what happens if everyone keeps taking the drug. Usually the bigger number.
• Treatment policy: everyone randomised is counted, including people who stopped. Usually smaller and closer to real life.

Companies often lead with the first. Check which one a headline quotes.

2. Raw or placebo-adjusted?

A 20% loss on the drug with 3% on placebo is a 17-point difference. Some headlines report the drug arm only, others the difference.

3. How long?

A 48-week result and a 72-week result are not comparable. Weight often keeps falling for a year or more.

4. Who was enrolled?

People with type 2 diabetes usually lose less weight on the same drug than people without it. A trial in a specific group, such as people with knee osteoarthritis, may not generalise.

5. What dose, and was it forced?

Some trials push everyone to the top dose; others let people stay lower if side effects bite. Flexible designs can lower the average.

A made-up example

Trial C may describe the strongest drug of the three. Neither headline number is wrong; they just answer different questions.