How to read weight-loss trial numbers without being misled

Headline numbers are usually a mean percentage change in body weight across everyone in a treatment group. Some people lose much more, some lose little, and some gain.

The placebo-adjusted difference subtracts what happened in the placebo group, who typically also received lifestyle counseling. It isolates the drug's added effect.

Trials often report two estimands: a 'treatment-regimen' estimate that counts everyone regardless of whether they stopped the drug, and an 'efficacy' estimate assuming people stayed on treatment. The second is usually larger. We use treatment-regimen values where available.

Cross-trial comparisons are imperfect: populations, durations, and dose schedules differ. Only head-to-head randomized trials directly compare two drugs.