This isn't correct in general; 1) both the rates are random (from last year and this year) so you need a CI on the difference of the two rates. The CI on a difference will be approximately 40% larger than the individual CIs (if they're about the same width), because the difference is more variable than either quantity. 2) Also, 1.96 is based on an assumption of normality. Poissons aren't normal (they're asymmetric and have heavier tails). Under some conditions the difference will be approximately normal, but you still need to deal with (1).