From my posts in the previous thread on this topic:
Variations from ISA temperature start becomming really applicable at or below ISA-15°. If it is not represented graphically in your Ops manual, you can use the 'rule of thumb':
+4 feet for each -1°C temperature variation from ISA, per 1000' altitude above the airport (QNH datum)
The airport QNH is adjusted so that it is accurate at the field, it is only the airspace above the field elevation that needs to be corrected for density variation.
e.g. For a 200' Cat I minima at, say ISA-20°, this produces a correction of (4 x 20 x 0.2) 16ft. Round it up to 20ft. and, remembering the rhyme: "ISA low? Watch out below!", set the published Decision Altitude PLUS 20ft. on the minima bug.
At an OM crossing height of say 1300ft AGL the variation is much larger (4 x 20 x 1.3 =)104ft. so your crossing height, on glide slope, would be more like 1404ft. indicated above the field elevation.
This assumes that the ISA-20° is constant throughout the atmosphere, and not affected by an inversion, which would mess up the calculations.
Question: So, you are landing at Madrid, 2000' AMSL, DH 200', is the correction based on 200' of cold air, or 2200' of cold air?
Ans: If you are using the Madrid QNH (as I expect you would) then that QNH is not the "actual" pressure at sea level.
If it was you would require a 2200' correction, and indeed the altimeter would not read 2000' when sitting on the taxiway. The Airport QNH is adjusted, so that an altimeter reads the airport elevation when you are at the airport.
The correction from sea level up to the airport elevation has already been done for you, so you only need to apply a density correction to the height above the airport elevation. (ie 200' in the question).