That's why you need a control. How many people in an unrelated similar sample also died during the period. Though I have no readily available evidence for this, I suspect that people vaccinated vs unvaccinated would be more similar than people diagnosed with COVID vs people undiagnosed.