Figures like this keep coming up. They are drastically skewed by the fact that 100% of cases in children, who very rarely die, are among the unvaxxed, whereas the large majority of cases in the elderly, who often die, are among the vaxxed. When two groups are wildly different in ways unrelated to the intervention you wish to test, you cannot draw any conclusions. (If you have patient-level data, you can correct for age, though you'll never be able to correct for health and comorbidities, or behavior. But here that hasn't been done.)