https://www.medrxiv.org/content/10.1101/2020.03.11.20031096v...
Type A is: 32% in the population 38% of those infected 41% of those that die.
Type O is: 34% in the population 26% of those infected 25% of those that die.
Also given that there seems to be an important coagulation component in the disease, this can make sense.
But interesting enough, this is not too obvious. Because O type blood has both Anti-A and Anti-B antibodies, so if there was something in the virus that looks like the A or B antigen this would make sense, but it's the opposite.