Unicode characters are not stored as ususal after change to 9.15.2

Hi,

I switched from 9.14.2 to 9.15.2. Usually Chinese characters and other latin ones were stored as they are. Now the utf8-codes are stored instead:

For example: 单能 will become "& #x5355;& #x80FD;" This happens in the test data base and the online database.

What can I do to avoid this.

This happend in a live study with 22 languages. The manual correction is horrible. I need a script to convert the wrong data directly in the database, too.