I would not say it is a general preference, but in this specific case I would prefer a simple regular expression over QDomDocument, because it is just a waste of resources. QXmlStreamReader might be a viable option, but it still requires me to write a multitude of code just to remove some character padding around a string and an integer value (and I can hardly think of anything more extendable then a nongreedy regular expression match combined with <code>*.**</code>).
Yes, regular expressions should not be used as a replacement for XML parsing in general, when it comes to different encodings, a varying document structure, schema validation, optional elements and attributes and all the other gruesomeness of XML, but in this specific case ("... if it is just the two fields ..."), where two basic datatypes are paddad by some well-defined and static characters, a simple regular expression is a readable (subjective, I agree on that to a certain degree), extension-resistant single-line construct.
This conclusion will change if the conditions change.