Direct standardization - age

I have data with 250 events during an 10-year period. I have calculated incidence rates based on the general population. I now want to standardize the annual IRs for age using a standard-population. My age groups are split in 10-yearinterval groups=9 groups. now when calculating the strata-specific IRs, I have a few strata with 0 events. What should I do about this? Is it a must that all strata need to have events to do this age-standardization? What are the "rules"?