Yes, confounders must be accounted for. But it also works the other way. People who are on death's doorstep typically aren't jabbed. Neither are those who are ill. There are inherent biases both ways.

The interesting thing is that both very conservative and inclusive filtering of the results came up with similar ratios.