Demographics adjustment for combined counts


My unit of observation was cities within a county. Since asthma counts were too small for each city, I decided to combine counts from various cities by a binary variable (all 'yes' cities combined and 'no' cities combined). How can I then adjust for demographic variables, such as age group, in this situation?

Thanks for your responses, I appreciate it!


Ambassador to the humans
It's not clear to me what is going on. Could you describe your study in more detail? Don't forget to mention what you measured and what your research question actually is.
The objective is to evaluate whether monthly/annual rates of asthma have decreased or not since the implementation of several policies within a county. Policy is a binary variable and varies by year and city. Cities are assigned a ‘1’ after a policy was implemented regardless if additional policies were passed following the initial one. Covariates include: linear time, policy, time*policy. Since age and gender are associated with asthma, it’s necessary to account for them. However, since asthma counts are too sparse for smaller cities, I felt it appropriate to sum counts by policy group (1/0) and then generate rates (which is the outcome). So now I’m unsure of how to account for age/gender since city is no longer the unit of analysis but rather the policy groups. I have demographic data at the individual level (hospitalization records) and census level by city.

Any suggestions on how to account/standardize for age/gender/ethnicity (maybe)?