I have this article that I was reminded of today which I think is well worth the read (I've distributed it a few times). However I'd like to put this down again because its just such a good summary to remind people of the important things (emphasis on the last point mine). It is the most succinct summary I've encountered.
• Variability in the reported INR is normal. This may be the result of limitations around the measurement reliability of the test, or of subtle changes in diet or exercise and requires monitoring but no action unless the patient is at risk of either bleeding or thromboembolic events; i.e. substantially above or below their target INR.
• Clinically important changes in the INR may reflect changes in drugs, adherence, diet or co- morbidities.
• Drug–drug interactions with warfarin are widely reported, are manifest through a variety of mechanisms and are often unpredictable in both their occurrence and the magnitude of the effect.
• Where possible drugs with a well-established record of interaction should be avoided. Where alternatives to interacting drugs are not available, or are inferior, the interacting drug may be prescribed and the INR more closely monitored.
• Where there is intercurrent illness or a worsening of conditions with the potential to impact on liver function, such as congestive heart failure, the INR should be more closely monitored until the patient is stabilised.
• Patients should be counselled clearly on the potential for interaction between warfarin and other medicines, whether prescribed, over-the-counter, traditional, herbal or complementary. Patients should be encouraged to discuss their use of other medicines rather than being told to avoid them.