Only BF is required but CF makes sense commercially.
Type CF limits are derived from the assumption of small contact area with the heart, where very small currents (<100µA region) can cause fibrillation. The assumption in the standard is 1mm² (from memory, I need to check this). For a general purpose infusion pump, the point of access is well away from the heart and the current density at the heart will be very low. For reference you need about 40mA applied to a limb to get a risk of fibrillation around 1%. This is why 5mA is allowed for BF in the "mains on applied part" test (although in practice no equipment ever goes near this, it would pretty hard to design anything approaching 5mA @ 250V, that can also handle the 1500V test, i.e. 30mA!).
There is a related issue which is that the test of "mains on applied part" can only pass CF limits if the infusion set (tubing) is kept away from any earthed parts and connects only to the single infusion pump. In reality there's no guarantee of this, it could be touching bed frames, stands etc, and there can be other stuff connected to the circuit such as other pumps, IBP sensors, CO sensors and so on. So in reality, no infusion pump should ever pass CF. However, most test labs seem to be happy to overlook this, and test with the tube suspended in air and just the one pump. Obviously, noting the explanation above, it's not really a safety issue. But this could explain the odd BF device: it could be rare test labs that insist on the "mains on applied part" test being performed in a more realistic environment, and thus exceeding the CF limit and forcing a BF classification.
Commercially though, it makes sense to use CF if you can get away with it. Obviously CF == cardiac , and most people associate the blood circuit as more or less a direct path to the heart. If you don't know the details above, it kind of makes sense. Thus, if you don't want to have to repeatedly explain why BF is OK every time you want to make a sale, go with CF.
