Example citation

Landmark analysis is when you select a time point during follow-up and analyze only the subjects who have survived until the landmark time.

It is used to avoid guarantee-time bias where responders are guaranteed to survive until the time of response.

Key takeaways

  • Landmark analysis is a useful sensitivity analysis in survival analysis studies
  • Pick the landmark carefully or use multiple landmarks
  • There is an extension of the Cox proportional hazard models called a time-varying Cox model that can be used as an alternative, but results are hard to interpret.