Presumably there are also people who die abnormally early. Unfortunately I have not seen what we really need to see: a breakdown of dead/recovered by day since infection. Without it there is no way to accurately assess the deathrate in an environment where new people are quickly falling ill.