I have calculated the crude incidence rates per 1000 population per year of strokes in patients with and without kidney disease e.g.18.37 (16.94-19.9) vs 6.37 (5.78-7.0). How do I calculate the p value for difference?
You don't need a pvalue. Just calculated the rate difference and present with SE and 95% CI, that will convey more information. Methodologically, most will say you cannot use risk estimates if your sample or study is retrospective and recommend the use of odds ratios - if this is applicable to your context.