Need to Know - anticoagulation
Professor David Fitzmaurice answers GP Dr Kathryn Griffith’s questions on warfarin initiation, near-patient testing and which antibiotics to avoid
Professor David Fitzmaurice answers GP Dr Kathryn Griffith's questions on warfarin initiation, near-patient testing and which antibiotics to avoid
1 Most patients taking oral anticoagulants will be taking them for atrial fibrillation (AF). How should we best assess AF patients for those who will benefit from warfarin?
• chronic heart failure (one point)
• hypertension (one point)
• age greater than 70 (one point)
• diabetes (one point)
• history of stroke (two points)
• transient ischaemic attack (two points).
The more points you get, the more your chance of having a stroke associated with AF. Most people would start treatment for patients with AF and two or more CHADS2 points.
But the problem is that increasing CHADS2 scores also predict risk of bleeding. Formal bleeding risk scores include the same factors as those used to predict stroke risk. Thus in the NICE guidelines, the very risk factors that are used to predict stroke are given as precautions for bleeding.
The NICE risk assessment tool, in my opinion, currently allows clinicians to opt too easily for aspirin for moderate stroke risk patients, which is probably ineffective, and would not be the case using CHADS2.
It is important to assess patients on an individual basis and to remember to reassess decisions, at the very least on an annual basis.
2 What do you consider to be the absolute contraindications to warfarin?
Absolute contraindications are few but would include a recent history of major bleeding, particularly intracerebral or GI bleeding, and malignant hypertension. Pregnancy used to be considered an absolute contraindication but the incidence of warfarin-associated embryopathy is very low with doses of warfarin under 5mg, and the efficacy of low-molecular weight heparin for the treatment of mechanical heart valves is less than that of warfarin.
Relative contraindications are much more difficult to manage as many are also indications for warfarin therapy, such as heart failure. This is much more difficult to assess and needs regular review.
One common source of concern is cognitive impairment. As oral anticoagulation is a potentially dangerous medication it is important to be confident the patient is taking the agreed dose.
If there is doubt about this it may be necessary to involve family or carers to ensure medication is being taken appropriately.
3 What are the additional risks when people are taking an antiplatelet drug and warfarin, and what should we do when patients have had a recent acute cardiac event and are taking both aspirin and clopidogrel along with warfarin?
I generally assume that co-prescribing of anti-platelet therapy with warfarin is a mistake. Antiplatelets double – at least – the risk of bleeding with warfarin so the risk of the event you are trying to prevent needs to be sufficiently high to outweigh it.
Warfarin is as effective as aspirin at secondary prevention of coronary heart disease so there is rarely a need for both.
However, patients with drug-eluting coronary artery stents will be discharged on warfarin, aspirin, and clopidogrel, for six to 12 months. As long as a responsible clinician has taken this decision it is necessary to keep the warfarin as well controlled as possible.
Contrary to what is sometimes recommended, reducing the therapeutic target does not reduce bleeding risk, it simply increases thrombotic risk.
4 Is age alone ever a bar to anticoagulant therapy?
Essentially the answer to this is no, as elderly patients actually have more to gain through the use of oral anticoagulation. But we do carry out annual reviews of all patients taking warfarin and the principal question addressed at this review is whether there is still a need for warfarin. If the patient is poorly controlled or having serious side-effects or expresses a wish to discontinue warfarin, it may be stopped.
This can be particularly difficult for patients with mechanical heart valves or a strong history of recurrent thrombotic episodes. There is no evidence for the effectiveness of aspirin in either of these scenarios and the only real alternative currently is low molecular weight heparin, which needs to be injected daily.
This has become the treatment of choice for patients with thrombosis associated with intravenous drug abuse but has not been used widely in patients who are poorly adherent in other situations.
I would suggest that in such cases it is perfectly acceptable to ask for an opinion from a specialist, usually a haematologist.
5 Do all antibiotics have the same effect on INR? How quickly will they have an effect and therefore how frequently should we monitor INR after starting an antibiotic?
The data for antibiotics and warfarin interactions are actually quite scarce, although those of us running clinics are aware that it is a real phenomenon, with most interactions causing a raised INR.
Broad-spectrum antibiotics are thought to increase INR values through eradication of gut flora which produces around 80% of the body's daily vitamin K requirements. The effect is generally small and short-lived.
Rifampicin is the only antibiotic I am aware of that reduces INR – through enzyme induction – and these patients may need specialist supervision.
There have been several case reports of warfarin and macrolide interactions – erythomycin, clindamycin – resulting in major bleeding and occasionally death, so these are best avoided with warfarin.
I do not increase frequency of INR testing unless the antibiotic is continued for more than a week.
6 Are there OTC products we should be aware of that may affect anticoagulant control?
Again information on this is scarce and generally conflicting. I would assume that everything interacts with warfarin unless proven otherwise. The watchword here is consistency. The danger times for anything – be it prescribed drugs, food, alcohol, or OTC products – are starting, stopping and changing dose. So we need to ask our patients if they have changed anything each time we see them.
Some agents, such as St John's wort and co-enzyme Q, have a demonstrable effect on INR and need closer than normal monitoring. Both of these agents are taken relatively commonly and may not be mentioned by the patient unless directly asked.
There are many herbal remedies available that are difficult to characterise as their constituents may vary between batches.
There are common herbs such as ginseng and feverfew that can have an unpredictable effect on anticoagulant control.
As long as the patient is consistent in whatever it is they are taking, virtually nothing is forbidden and can be managed. Patients do get concerned over things like cranberry juice and grapefruit juice, generally unnecessarily so, and often their concerns are made worse by health professionals. There have been case reports of interactions with these items but there is no consistency of interaction and again the main principle should be consistency of intake.
Probably the most important and common interaction is with alcohol, which tends to raise the INR in the short term and reduce it with chronic alcohol misuse. Patients should be advised to be moderate in their drinking and to try to be consistent across time. Binge-drinking tends to cause most acute problems.
There is no problem in administering the flu vaccine to patients taking warfarin, and no close monitoring is required.
7 How accurate are the near-patient testing systems?
Most commercially available near-patient INR systems have been evaluated by the regulatory agencies and found to be accurate if used properly. This includes using effective quality control procedures as well as accurate performance of the test itself.
Newer devices may not have undergone this kind of evaluation as the Department of Health no longer views this as necessary. If a device has not had a formal evaluation it is important to ensure validation is undertaken locally before procurement.
Training is therefore vital to the clinical use of the near-patient system. If testing is performed well, there should be reasonable agreement for INR values within the therapeutic range – up to about five – and agreement between near-patient systems and reference laboratories is similar to that between any two laboratories.
One area where discrepancies do occur is when the INR is high – say, above five. In this situation it is common for clinicians to send a venous sample to confirm the result. There is invariably a large difference in the INR obtained, making dosing decisions difficult. The same difference would be found between any two laboratory systems and merely reflects the instability of the INR measurement at a higher value. The difficulty is in knowing which is the correct result.
The correct course of action, therefore, should be to ensure the INR has been measured accurately with the near-patient device, ensuring quality control procedures have been adhered to correctly, and manage the patient according to this result.
8 What are your opinions on primary care-based anticoagulation services?
Primary care oral anticoagulation management can be effective as long as adequate and ongoing training has been undertaken and quality assurance procedures are in place.
The only model with a robust evidence base is the so-called ‘Birmingham Model', that is, nurse-led clinics using near-patient INR testing and computerised decision support software for dosing support and recall advice. This is level four in the anticoagulation monitoring national enhanced service. This can only provide a safe and clinically effective service if high-quality, regular training is available, and is not a source of easy revenue.
An important issue here is whether primary care will truly provide an oral anticoagulation service with a holistic approach to thrombosis management and prevention or simply provide an INR monitoring service. There is a difference.
The cheapest way to provide an INR monitoring service is through a centralised service – that is, a hospital-based clinic. The huge advantages available through providing a first-class oral anticoagulation service will be lost if we simply transfer the hospital clinic into the community.
9 There are new oral agents being developed. Are there any near to launch and how are they likely to be used?
Two products are now licensed – dabigatran and rivaroxaban – for thromboprophylaxis in orthopaedic surgery and may be available for wider use towards the end of 2009. These agents do not require therapeutic monitoring. But warfarin is likely to be with us for some years to come.
10 Do all patients newly diagnosed with AF need urgent anticoagulation? If not which ones do?
Nobody requires urgent anticoagulation and it may be appropriate to start aspirin while deciding whether or not to commence warfarin. I would reiterate that the decision to anticoagulate or not should be reviewed regularly, at least annually.
11 Do you have a protocol that we could use in primary care to initiate warfarin safely, and any sources of patient information leaflets?
Professor David Fitzmaurice is professor of primary care research at the department of primary care and general practice at the University of Birmingham
Competing interests Professor Fitzmaurice works for the University of Birmingham, which produces commercially available software for oral anticoagulation management.Warfarin What I Will Do Now What I Will Do Now
Dr Griffith considers the answers to her questions
Clearly warfarin is underused for stroke prevention in AF and it is important to reassess each patient's risk of stroke, probably on an annual basis. The CHADS2 method appears a straightforward way of doing this – patients scoring two or more should be considered for warfarin and absolute contraindications are few.
But there is potential for confusion as some of the relative contraindications to warfarin are also listed as indications or stroke risk factors. Elderly patients have more to gain from warfarin and we may need to enlist the support of carers where there are concerns about cognition.
I'll feel reassured that the majority of antibiotics will have only a small effect on INR but aware that particular care should be taken with rifampicin and macrolides. It's useful to know alcohol is fine on a regular and moderate basis. There are some effective, validated devices which can be used for near-patient testing for INR.
Dr Kathryn Griffith is a GP in York and board member of the Primary Care Cardiovascular Society