You cannot conclude anything about the 16x difference (TFA says 13x, by the way) from the number of dead patients.

As a made up example to illustrate this point, assume that people are either "lucky" or "unlucky". Unlucky people die when they get the virus. Lucky people never die. Assume that one person per million is unlucky, and assume that the vaccine does absolutely nothing. Then this experiment on one million people would find one death in the vaccinated and zero deaths in the control group, inferring that natural immunity is infinitely better.

It doesn't have to do with the percentage who die, but rather the variation in susceptibility.

If 2% of people are 80% likely to die from COVID, and the rest have a baseline 0.1% risk. Assume prior infection provides no protection:

* 1000 (of whom 20 are particularly vulnerable) people are naturally infected; .001 * 990 =~ 1 people of average susceptibility die; 16 maximally susceptible people die. Total of 17 deaths.

* Then, you are left with 983 people, (of whom 4 are particularly vulnerable). Upon reinfection, .001 * 983 =~ 1 person of average susceptibility dies; 3.2 people of high susceptibility die. There's a total of ~4 deaths.

This is a 4x reduction in deaths even if there's no protection from prior infection.

Why do you believe "a small percentage of COVID-19 patients die" (which is only true compared to, say, Ebola) has any relevance to the results of this study?

The argument I was replying to is the numbers are skewed by survivorship bias, but given >98% of people survive that's only going to skew the numbers 2% or so not 600% or whatever.