I would like to construct a survival curve from retrospectively gathered data that represents the time to onset ($t$) of disease ($d$) after some specific event ($x$).
We know that $x$ predisposes to $d$ so we want to explore how $t$ and $d$ are related after $x$ occurs.
The problem is that $x$ can occur at any age and the risk of $d$ (even without $x$ occurring) increases with age. Furthermore, some of the study population already had $d$ when $x$ occurred.
Below is a plot I constructed that represents the diagnosis of $d$ in all patients that were not already diagnosed with $d$ and suffered $x$ at $t=0$.
However, some of these patients likely would have developed $d$ anyway just given the age of that patient at $t=0$. I am unsure of the most responsible way to control for this confounding variable in my survival analysis.
I'm sure this is actually a more complicated question than what I am making it out to be. Additional resources would great.
