And the number of people dying per (large) unit of time is almost certainly well approximated by a normal distribution, modulo seasonal variation and events such as this (which break the "independence" assumption of the CLT).