How to do a reverse modeling for cases and controls to identify increased % of risk factor in cases

[sms1891]

Dear SAS experts,

I have a set of cancer cases and age and sex matched (1:5) controls in the dataset. I am looking back in time to see if the cancer cases have a significantly increased % of diabetes leading up to cancer compared to the controls. For cancer cases, I used cancer date as the index date and for controls, I have selected an arbitrary date when they came for follow-up in the system. For both cancer cases and controls, I looked back in time to see if they had a diabetes diagnosis. Among cancer cases and controls who had a diabetes diagnosis, I calculated the time from index date to diabetes date. When I plot the results, I definitely see increased % of diabetes in cancer cases compared to controls.

                                               How do I test the significance of this increased % in cancers compared to controls? I need to adjust for race and other cofactors as well. I want to show that each year prior to index date, the % of diabetes in cancer cases are significantly higher compared to controls and fit a overall model and also maybe a graph (like reverse KM curve or cumulative incidence curve?). I have attached an excel sheet with the % of diabetes identified in cancer cases and controls. 

Any help with this is greatly appreciated!

Thank you so much,

Sat  

[sbxkoenk]

Hello,

For modelling incidence rate and comparing between groups, have a look at the below !

Usage Note 24188: Modeling rates and estimating rates and rate ratios (with confidence intervals)
https://support.sas.com/kb/24/188.html

Usage Note 37344: Estimating rate differences (with confidence interval) using a Poisson model
https://support.sas.com/kb/37/344.html

You can also screen this webpage :
Usage Note 30333: FASTats: Frequently Asked-For Statistics
for the key-word "rate" or whatever other key-word you would like
https://support.sas.com/kb/30/333.html

Koen