The Gompertz–Makeham law the author uses, actually allows also to modelize the accidental death rate. I don't know how hard it is to fit this additionnal parameter with real-world data though.