Estimands are a way to describe the treatment effect that a study sets out to quantify. Our work focuses on how to choose, report, and estimate different estimands.

What are estimands and why are they relevant?

Research studies can answer different questions, but it is often unclear which question a study has set out to address. Estimands help to resolve the issue by providing researchers with a structured way of describing the exact treatment effect they hope to quantify in their study. This can:

  • Make study aims clearer to patients, healthcare professionals, and healthcare payers
  • Help investigators choose which study design or statistical analysis is most appropriate.

What have we done?

Our work focuses on identifying how to best incorporate estimands into research studies to ensure they are appropriately designed and analysed, and to ensure study interpretation is clear. We have:

  • Provided overviews and tutorials for implementing the estimands framework
  • Reviewed trial protocols and published reports to examine the clarity of the research question, which identified an urgent need for estimands
  • Described additional considerations for defining estimands for trial designs such as cluster randomised trials, factorial trials, non-inferiority, and re-randomisation trials
  • Described the types of estimands that should be used for clinical areas such as tuberculosis and COVID-19
  • Co-developed tools to help involve public partners in discussions around estimands

How will this make a difference?

Our work provides practical guidance on how estimands can be used in practice. Greater uptake of estimands can lead to improved design and analysis of studies, and better interpretation of results from stakeholders.