Zenodo
One trial, several numbers: how the estimand, follow-up and enrolled population change cross-trial comparisons of incretin obesity agents. A meta-research study of twenty trials
By Tommy Carstensen.
Abstract
Objectives
To measure how far estimand and follow-up move headline weight-loss results compared across trials of incretin-based obesity agents, and to illustrate the effect of comparator and population.
Design
Meta-research study of published aggregate results: a closed convenience sample of agents, one index trial per agent and regimen, chosen by a rule written after assembly.
Data sources
Primary publications of the index trials: endpoints, estimands, baseline populations, placebo arms and published weekly weight curves. Eligibility criteria: Randomized trials of incretin-based or amylin-based agents in adults with overweight or obesity, not restricted to type 2 diabetes.
Main outcome measures
The gap between efficacy-family and policy-family estimates for the same arm; the order of agents under each; weight loss accrued after week 24; placebo-arm weight change; naive against placebo-anchored differences.