At the very least, you need deaths/(deaths+recoveries) to calculate the mortality rate. That is, deaths / (total cases); as other comments note, the problem is

At the very least, you need deaths/(deaths+recoveries) to calculate the mortality rate. That is, deaths / (total cases); as other comments note, the problem is that computing "total cases" as just deaths+recovered might not be right if one outpaces the other, and if there are a bunch of cases we don't know about because we're not testing. (E.g., are we only seeing severe cases as they send people to the hospital?)