The CML method to be used for ADPE contains, in its original form as reproduced in Table C2 of EN15804+A1, contains characterisation factors (CF) for elements only. But databases (certainly ecoinvent and I'm pretty sure GaBIi too) contain - in the resources in ground category of elementary flows - compounds of these elements, and indeed many of the elements themselves barely occur in nature in an uncombined state. Boron is a good example. This occurs in nature almost entirely as borax (a hydrated sodium borate), and borax is the boron-containing elementary flow in the "resource-in-ground" category that you'll find if you create an LCI for products such as detergents, some fire retardants. Without a CF for borax any calculation of ADPE for a product containing a reasonable amount of borax-based fire retardant will be seriously flawed. We therefore include CF for compounds of the elements covered by Table C2 of EN15804+A1. I think GaBi does too, because I see consistency of ADPE results for products with known borax content calculated using GaBi and our database. The old (now obsolete) implementation of the CML2001 method in ecoinvent also included CF for compounds.
I hope that helps. As Andreas notes, we will release a new version of the database, also covering EN15804+A2 very soon.