Warfarin is administered as a racemic mixture of S- and R- warfarin. S- warfarin is 3 to 5 times more potent than R- warfarin.
So, what's the logic behind giving a mixture of it? Isn't administration of only one form gives more predictable results?
In general, it's more difficult to produce a specific stereoisomer of a compound than it is to produce the racemate. Producing a pure S- or R- sample of a particular compound requires either enantioselective synthesis, or racemic synthesis followed by a purification step such as chiral chromatography. This tends to be expensive - especially for an older drug like warfarin where it would likely imply either chiral purification (throw away half your product) or redesigning the synthesis to be chiral.
For some drugs it makes sense to do this - some reasons might include adverse effects of the undesired stereoisomer, or taking a drug that's about to go off-patent and re-branding an enantiopure preparation for further patent protection. I think the latter is what happened with Citalopram -> Escitalopram (Lexapro) although this is speculation.
Most likely in the case of warfarin there is no strong argument for producing a pure enantiomer. It has been off-patent for a while, there likely aren't major adverse effects related to the less potent isomer, and most likely economics dictate that producing and selling the racemate is cheaper / more profitable than retooling to produce and sell an enantiopure version. That latter option would only make sense if they could sell the pure S-warfarin for enough to recoup the development / purification costs, and this is unlikely for a drug with such a long history of use in its racemic form.