To Bridge or Not To Bridge? Is it the Wrong Question?

December 7, 2015

Results of two recent studies suggest that perioperative "bridge" anticoagulation may do more harm than good in some patients.

One of the most common dilemmas primary care physicians face is how best to manage interruption of oral anticoagulation therapy with warfarin in patients with atrial fibrillation (AF) during the perioperative period. Frequently, we “bridge” our patients temporarily using low molecular weight heparin (LMWH), which has a shorter half-life than warfarin, in order to minimize time the patient is without anticoagulation and presumably, minimize the risk of stroke. However, this practice is somewhat empiric with a dearth of data demonstrating the impact of bridging on perioperative stroke or bleeding. Two recent studies, the BRIDGE1 randomized controlled trial and the ORBIT-AF2 registry study, both attempted to answer the question of whether bridging is necessary or not.  

The BRIDGE trial was a noninferiority study with 1884 patients randomized to either LMWH or placebo, with cessation of warfarin 5 days before surgery and restarted within 24 hours after surgery; LMWH was given 3 days before procedure until 24h before procedure and resumed either 24h (for low bleeding risk surgeries) or 48-72h (for high bleeding risk surgeries) after and continued for 5-10 days after the procedure. Patients were excluded if they had one or more of the following: a mechanical heart valve; stroke, systemic embolism, or transient ischemic attack within the previous 12 weeks; major bleeding within the previous 6 weeks; creatinine clearance <30 mL/min; platelet count <100 x 109/L; or planned cardiac, intracranial, or intraspinal surgery.

The incidence of the primary outcomes was: arterial thromboembolism (0.3% LMWH vs 0.4% placebo, p=0.01 for non-inferiority) and major bleeding (3.2% LMWH vs 1.2% placebo, p=0.005 for superiority).

There are some important caveats and take-aways from this study:

(1) Despite patients having an average CHADS2 score of 2.3, the overall risk of perioperative events was quite low. This suggests that either the CHADS2 score may not be a good predictor of perioperative thromboembolism or that we may have previously overestimated the hypercoagulable state that exists in the perioperative period. Note that the CHADS2 score was used in this study, not the CHA2DS2-Vasc score, which we know to be a better predictor of thromboembolic risk, especially in those with low CHADS2 scores; (2) there were few patients with CHADS2 scores of 5-6 and those having cardiac, intracranial, or intraspinal surgery were excluded so these patients still pose a clinical dilemma; (3) the timing of the initiation of LMWH (24-72h after surgery) was left to the discretion of the treating physician and this critical decision, which was not standardized in this trial, is likely to have played a significant role in the bleeding rates; and finally (4) because event rates were low, the sample size and power calculation for noninferiority were adjusted during the study and this may have  influenced the noninferiority conclusion.

The ORBIT-AF prospective registry (US-based outpatient registry of AF) offered a nonrandomized perspective on the same question: is perioperative bridging necessary or beneficial?  The authors concluded that about 24% of all patients were bridged perioperatively, mostly using LMWH (73%) but unfractionated heparin (15%) and other agents were also used. Patients who were bridged were more likely to be higher risk overall, with more prior CVA and mechanical valves but there actually was no difference between bridged vs nonbridged groups with respect to the proportion with high CHA2DS2-VASc score (≥2). Similar to the findings of BRIDGE, there was more bleeding (5% vs. 1.3%, adj. OR 3.84, p<.001) and more MI/stroke/systemic embolism/major bleeding/hospitalization/death within 30 days (13% vs 6.3%, adj. OR 1.94, p=0.001) when bridging anticoagulation was used.

There are caveats to keep in mind regarding the ORBIT-AF registry study resutls as well: (1) it is a nonrandomized observational study, which is subject to residual confounding; (2) there is a high degree of variability in timing, dose, and choice of drug for periprocedural bridging therapy, which is a critical component of bleeding outcomes. The advantage of the registry is its inclusion of patients with mechanical valves and prior CVA.

Despite their limitations, these two trials do provide compelling evidence that bridging therapy (which seems to be used in one out of four patients currently) may be over utilized and may be leading to an increase in periprocedural bleeding without providing additional efficacy. Based on these results, I know I will feel more comfortable bridging fewer of my low-risk patients.

Many electrophysiologists now perform invasive intravascular venous procedures (eg, pacemaker implantation) without interruption of warfarin. Perhaps, then, the question we really should be asking is not “to bridge or not to bridge” but “to interrupt or not to interrupt?” As we learn more about perioperative risk (and collect more data, including data on the novel oral anticoagulants), the field of perioperative medicine will continue to evolve and perhaps our surgical/gastrointestinal/dental colleagues will become more comfortable doing procedures without stopping anticoagulation altogether.



Douketis JD, Spyropoulos AC, Kaatz S; BRIDGE Investigators. Perioperative bridging anticoagulation in patients with atrial fibrillation.N Engl J Med. 2015; 373:823-833. DOI: 10.1056/NEJMoa1501035

Steinberg BA, Peterson ED, Kim, S. Use and outcomes associated with bridging during anticoagulation interruptions in patients wtih atrial fibrillation. Findings from the Outcomes Registry for Better Informed Treatment of Atrial Fibrillation (ORBIT-AF). Circulation. 2015; 3:131(5):488-94. doi: 10.1161/CIRCULATIONAHA.114.011777. Epub 2014 Dec 12.