Selection bias.
The only cleanish data we have on mortality rates is the ruby princess. 3700 passengers ( admittedly an elderly population) , 710 known to be infected but there were still a few hundred not tested.
11 dead, about 80 currently waiting to discover what the real mortality rate for covid is and 10 of those are critical.
The rest of the worlds data is junk. Take Australia - in the early days you could only be tested if you had come from wuhan, then overseas, then " close contact" with a known case, or perhaps work in a " high risk" industry ( unless you work in a supermarket...lets not scare those workers with the truth )
Anyone watching death rates is looking at 3/4 week old irrelevant and dirty data . Do you wait until the cars in front of you have dead passengers, or hit the brakes when the lights turn red? Oh, and before hitting the accelerator...look for traffic who ignored those other lights....