<b> Exchangeability assumption needed for death (competing events)? </b> |
Not needed |
At each k + 1, conditional on the measured past, death is independent of future counterfactual outcomes had everyone followed A = a and death was eliminated. |
<b> Positivity assumption needed for death (competing events)? </b> |
Not needed |
For any possibly observed level A = a and covariate history amongst those remaining uncensored (alive) and free of dementia diagnosis through k, some individuals continue to remain alive through _k + 1_. |
<b> Consistency assumption needed for death (competing events) </b> |
Not needed |
An intervention that “eliminates death (competing events)” is well-defined. |