SOCR logoBPAD

Competing Risks: KM vs. CIF Explorer

Simulate cancer and cardiovascular deaths together and see why treating a competing event as censoring inflates a naive Kaplan-Meier curve.

  • competing risks
  • cumulative incidence function
  • Aalen-Johansen
  • survival analysis

Cancer mortality: naive KM vs. true CIF

051015202530354000.20.40.60.81YearsCumulative incidence
Naive KM (censors CV deaths)Cumulative incidence (Aalen-Johansen)

Cumulative incidence by cause (stacked)

051015202530354000.20.40.60.81YearsCumulative incidence
CancerCardiovascular

Simulation summary

Total patients
1000.000
Cancer deaths
524.000
CV deaths
384.000
Censored
92.000
High competing-risk ratio
With CV/cancer hazard ratio 0.80, censoring competing deaths in a naive KM curve substantially overestimates cancer-specific mortality. The CIF is the correct estimand here.