The trusted source for
healthcare information and
Source: Phan TG, et al. Arch Neurol 2000;57:1710-1713; Hacke WS. Arch Neurol 2000;57:1682-1684.
Stopping warfarin therapy in patients with prosthetic cardiac valves or atrial fibrillation (AF) may be a source of concern for both physicians and patients who fear that a stroke may occur during this vulnerable time period. These risks are quite exaggerated. Given the overall risk of stroke on a yearly basis in these conditions (4% in patients with a prosthetic valve and approximately 12% in patients with AF), the risk of a short interruption of warfarin is probably minute. The report by Phan and associates from the Mayo Clinic confirms this. As these data suggest, the stroke risk among patients taken off warfarin for about 7-14 days was only 3-5%.
This report retrospectively reviews the records of 141 patients suffering an intracerebral hemorrhage while on warfarin therapy presenting over the past 25 years. Cases were divided into: group 1—patients with prosthetic heart valves (n = 52), group 2—patients with atrial fibrillation and a history of embolic stroke (n = 53), and group 3—patients with a history of recurrent stroke or TIA despite anti-platelet therapy (n = 36). All except nine prosthetic valve patients had mechanical devices. Hemorrhages included: intracerebral hematomas (n = 87), subdural hematoma (n = 43), or subarachnoid hemorrhage (n = 8). Warfarin was withheld from patients for a mean of 10 days, with fresh frozen plasma and vitamin K administered in the acute stages. Three patients suffered ischemic events over 30 days (an occipital embolism in a patient with AF, a lacunar infarct in a patient with a Bjork Shiley aortic valve, and a vertebrobasilar TIA in a patient with a known severe basilar artery stenosis). The risk of stroke in groups 1, 2, and 3 were therefore 2.9%, 2.6%, and 4.8%, respectively. No patient suffered recurrence of ICH with resumption of anticoagulation. Thirty-day mortality rates from the initial hemorrhage were high: 38% in groups 1 and 2 and 50% in group 3. The majority of this mortality took place in the first 14 days.
In an accompanying editorial, Hacke reviews the data from Phan et al. As he observes and as Phan et al acknowledge, the lengthy period during which cases were collected may have introduced bias due to changing clinical practices. In particular, the high mortality rates may reflect less aggressive surgical intervention strategies and a significant proportion of DNR orders. As Hacke comments, recurrent event rates may be underestimated because patients died within the first two weeks or were managed with a primary focus on comfort measures. Also, follow-up imaging is scant. Recurrent bleeding or new embolic events that were not obviously clinical may have been missed.
Hacke’s editorial also draws attention to complementary data from his own institution in Heidelberg, Germany (Bertram M, et al. J Neurol 2000;247:209-214). Among 15 Heidelberg patients studied over three weeks, three patients had recurrent bleeding and three suffered embolic events. Both of these rates, in the 20% range, were significantly higher than in Phan et al’s data. Recurrent bleeding was restricted to patients with only partially corrected prothrombin times and emboli occurred only in patients who were fully reversed and not treated with intravenous heparin.
The need to stop warfarin in the setting of ICH may provide helpful insights into the safety of stopping anticoagulation for short periods of time in other circumstances. In situations much less protean than ICH, it is common clinical practice to briefly withhold warfarin, for instance, before elective surgery or even potentially bloody dental procedures. In rare cases, patients at high risk may be admitted to the hospital to be "covered" with intravenous, unfractionated heparin or treated with subcutaneous low molecular weight heparin injections, but such practices are labor intensive and expensive. If the risk of an intercedent stroke is as low as Phan et al’s data indicate, such practices are probably not justified. As Hacke observes, however, Phan et al’s data may be an underestimation. The true risk, as suggested by the small Heidelberg series, may be many magnitudes higher. Clearly, anticoagulation should be interrupted for as short a period of time as is possible. Frequent monitoring of prothrombin times as warfarin is withdrawn and paying close attention to its resumption when considered safe afford the best protection against rare, but potentially serious, embolic complications. — Alan Z. Segal