There are many links in this chain, and if things don't work as a whole you have to go over them link by link to see where the problem(s) happen.

View Unicode in hex offers a nice way of seeing what your actual data is. Adapt the code there to print what you get from LWP. Then make sure what gets fetched from the database is still UTF-8. Finally don't trust your browser, download the page that the web server handed you and see what information is actually there. (It might be your server is sending *two* Content-type headers, in which case only one of them (the wrong one, by Murphy's law) is honored by your browser.

Oh, you also need to tell DBD::Pg that your data needs to be treated as UTF-8. Check out the pg_enable_utf8 attribute. (If you move to mysql one day, contact me for a patch giving similar functionality.)

I suggested many things above but I recommend you tackle them one at a time, not all at once. That way if the first link in the chain was the only bad one you don't waste your time with the others.