That's a good point. But the Florida analysis calculated incidence risk of period A (28 days post-vaccination) compared to period B (25 weeks post-vaccination), not their absolute numbers.

By not excluding deaths from period A in period B, the denominator of period B is overestimated, resulting in underestimated incidence. If period A is compared to the underestimated incidence in period B, period A's incidence would thus be overestimated.