Finally fixed the crazy bug in HubMed where international characters weren't displaying properly. All it took was changing the charset in the HTML header from iso-8859-1 to utf-8. The problem with Endnote/Procite remains though, as while I'm reluctant to strip the extra information and replace the international characters with plain Latin characters, the bibliographic managers are still unable to handle Unicode (which is just laziness, really). The only solution offered so far is to write to them and get them to sort it out.