I somehow suspect she is not even using data conditioned by infection history and looking at a mix including individuals that might be recently infected with covid.