Warfarin can safely be held for the short term

Clinical Question

Can warfarin therapy be withheld safely for a short time in patients undergoing elective procedures?

Bottom Line

Current guidelines from several European and US cardiology groups allow for the cessation of warfarin for up to 1 week in patients at risk for bleeding because of an invasive procedure such as dental surgery or colonoscopy. Although we still need more research to provide a definitive answer, this study does not refute these recommendations, finding that 0.59% of patients who stop treatment develop a thromboembolism within the subsequent 30 days. (LOE = 2b)


Garcia DA, Regan S, Henault L, et al. Risk of thromboembolism with short-term interruption of warfarin therapy. Arch Intern Med 2008;168(1):63-69.  [PMID:18195197]

Study Design

Cohort (prospective)




Outpatient (any)


Warfarin treatment sometimes must be interrupted in patients undergoing surgery or dental procedures. The researchers conducted a large observational study to determine the impact of short-term warfarin discontinuation on the likelihood of developing a thromboembolic event over the subsequent 30 days. To assemble the data, the researchers contacted community-based physicians in 101 offices who were using a computer-based warfarin medical record (CoumaCare) and asked them to participate. These physicians treated patients as they normally would, stopping warfarin treatment (91.6%) or stopping but using a bridging therapy (8.4%) for elective procedures such as colonoscopy, dental work, or ophthalmic surgery. They excluded patients who were hospitalized for a procedure, since they couldn't track whether a bridging treatment was used. More than half the patients were receiving warfarin because of atrial fibrillation and 15.5% were receiving warfarin following a prosthetic valve replacement (29% of the latter group received heparin as bridging therapy). In 1293 interruptions of warfarin treatment in 1024 patients, 7 patients -- all in the no treatment cohort -- developed a clinically apparent thromboembolic event. This prevalence translates into a rate of 0.54%. Most of the patients in the withheld warfarin group had treatment stopped for 5 days or fewer. Two of the 7 patients were at high risk of thrombosis because of a recent history of thromboembolism or active malignancy. The low number of thromboembolic events, though good for patients in the study, limits our confidence in these data, and a larger study is needed. That being said, it is reassuring that most patients who stop taking warfarin for 5 or fewer days are at little risk of an thromboembolic event.