Can I do regression analysis or statistical tests with estimated data?

#1
I have two sets of estimates from two different studies. Suppose my data look like the following (it's a made-up data as an example). Is there a way I can test if results from study 1 is statistically different from study 2.
Please note that all I have is the following table, and I've simplified it to just having three rows so it's not too messy to show. I don't have the original data that produced these results. I know how to test the difference within each row (smoking status), but I'd like an overall test to see if proportions in study one are identical to/correlated with that in study two. Is there such a test? And a more general question, can I somehow do a linear regression using the estimates from such a table, taking into account the standard errors of the two sets of estimates? Your inputs will be greatly appreciated.
1517870703645.png