I would have guessed it would be the other way around. If you know at design time that this value will be a percentage, use the DV_PROPORTION data type with the ‘type’ attribute set to 2 (percent, denominator fixed to 100). On the other hand if you don’t know for sure (such as for some lab results or medication strengths which could be for example mg/ml or % interchangeably), you would use DV_QUANTITY.
As a relative OpenEHR outsider but a data modelling enthusiast, I would agree with Silje that if this is known to be a proportion then using DV_PROPORTION seems more intuitive, as this preserves the semantics of the data. It also would allow confident conversion of that data into other types of proportion (eg per-thousand or PPM)
I think that is the intended use. It is the case that UCUM has a ‘%’ unit as part of DV_QUANTITY which I think I have only ever used in the context of integrating a lab test where the ‘proportionality’ of the value is not really relevant i.e it is in some ways an arbitrary unit.
Also, take into account that if you use DV_PROPORTION to represent this percent, you will always have this double quantity stored in your data, which doesn’t really add nothing of value and just will slow down your queries.