Sensitivity Analyses

TipExecutive summary

This page highlights how I use sensitivity analyses to evaluate whether study conclusions remain robust under reasonable alternative assumptions or analysis choices.

Why it matters

Primary analyses depend on assumptions. Sensitivity analyses help determine whether conclusions are stable or whether they rely on assumptions that materially affect interpretation.

What I focus on

  • Identifying key assumptions in the primary analysis
  • Designing targeted sensitivity analyses (not checkbox extras)
  • Distinguishing prespecified sensitivity analyses from exploratory analyses
  • Reporting results in a review-friendly and interpretable way

Typical assumption areas assessed

  • Missing data assumptions
  • Model specification choices
  • Population definitions
  • Endpoint derivation rules
  • Intercurrent event handling
  • Time-to-event/censoring rules (when applicable)

Practical contribution areas

  • Sensitivity analysis planning and rationale support
  • SAP alignment and traceable implementation
  • Reviewer-friendly summary of consistency vs divergence
  • Interpretation support for robustness conclusions

Common pitfalls I help prevent

  • Too many low-value sensitivity analyses
  • Poor linkage to primary assumptions
  • Ad hoc interpretation after results are produced
  • Under-reporting meaningful divergence from primary conclusions