I have a nationally representative data set of ~30,000 emergency department visits in a year; using the provided weights, these represent ~200 million visits. It is straightforward to use SAS or R to perform linear regression to calculate the proportion of ED visits from individuals with private vs public insurance, controlling for confounders like race, gender, and age, and accounting for complex survey design using the provided cluster and strata variables. In this manner, we would be answering the question "Are emergency department visits more likely to be from patients with private insurance than public insurance, controlling for race, gender, age, etc."

What I want to do, though, is to compare RATES of emergency department visits differ between privately insured and publicly insured individuals, using population denominators for people with private and public insurance from the Census Bureau. I believe I could use PROC SURVEYMEANS in SAS with the RATIO statement (or potentially svyratio in R) to calculate rates for people with public and private insurance, but I am unsure how to implement a regression that compares these two rates after controlling for important covariates.

Any advice would be greatly appreciated - happy to use either SAS or R. Thanks.