Periprocedural antithrombotic and bridging therapy: recommendations for standardized reporting in patients with arterial indications for chronic oral anticoagulant therapy


Alex C. Spyropoulos, Department of Medicine, Division of Hematology/Oncology, University Of Rochester Medical Center, James P. Wilmot Cancer Center, 601 Elmwood Avenue, Box 704, Rochester, NY 14642, USA.
Tel.: +1 585 276 6103; fax: +1 585 276 0350.


The periprocedural management of patients on chronic oral anticoagulant (OAC) therapy, especially vitamin K antagonist (VKA) therapy such as warfarin, is a common clinical problem that is estimated to affect approximately 250 000 patients annually in North America alone [1]. ‘Bridging therapy’, usually with parenteral unfractionated heparin (UFH) or low-molecular-weight heparin (LMWH), is commonly used to allow for continued anticoagulation in patients needing temporary interruption of VKA for an elective procedure or surgery [2]. A systematic review of clinical studies assessing VKA-treated patients who required an elective procedure or surgery with or without interruption of VKA conducted from January 1966 to June 2001 did not provide firm conclusions on the safety and efficacy of different management strategies due to the variations in the definitions of patient populations and procedures, anticoagulant regimens, events and durations of follow-up [3]. A more recent systematic review of the periprocedural management of such patients needing temporary interruption of VKA with and without the use of bridging therapy found similar results: most studies were observational with methodological concerns and bias, there were no uniform definitions of patient and procedure-related thromboembolic (TE) and bleed risks, no standardized definitions of outcomes, and heterogeneity of time frames for capturing adverse events [4]. Both reviews discussed the need for standardization of patient/procedural TE and bleed risks, outcomes and durations of follow-up, and the need for large randomized controlled trials (RCTs) or well-designed cohort studies to establish the need for an aggressive periprocedural antithrombotic approach (such as the use of bridging therapy) vs. a minimalist approach of simply withholding and resuming OAC in the periprocedural period.

The objective of this ISTH communication is to recommend a standardized approach for reporting patient and procedural TE and bleed risk, anticoagulant regimens, key outcome definitions including arterial TE and bleeding, and standardized reporting of the duration of follow-up when conducting future studies in periprocedural antithrombotic and bridging therapy. This recommendation would pertain to patients with arterial indications for OAC and would encompass both currently used agents such as VKA and use of novel OACs (NOACs) such as oral direct thrombin inhibitors (dabigatran) and oral direct factor Xa inhibitors (apixaban, rivaroxaban and edoxaban).

Patient thromboembolic risk stratification and periprocedural bleeding risk

A recent systematic review of moderate-to-large (> 100) antithrombotic management studies (including bridging therapy) since 2001 assessed patient TE and procedural bleed risk stratification of the patient groups on chronic VKA most commonly studied in the periprocedural period [4]. A total of 11 studies with 6173 patients were reviewed, including patients with mechanical heart valve (MHV) and chronic atrial fibrillation (CAF), indications for VKA. For TE risk stratification, patients were classified according to a two-tier or a three-tier TE risk stratification scheme of low-to-intermediate and high TE risk, or low, moderate and high TE risk, respectively. Many of the studies adopted the suggested periprocedural TE risk stratification scheme of the 2008 American College of Chest Physicians (ACCP) Antithrombotic Guidelines [1]. The TE risk stratification included the prosthetic valve type (caged-ball, tilting disk and bileaflet) and valve position (aortic or mitral) as well as other major stroke risk factors such as prior stroke or transient ischemic attack (TIA) in patients with an MHV, and the CHADS2 score in patients with CAF. Although a two-tiered risk scheme may be simpler, we suggest that future studies adopt the ACCP-endorsed TE risk scheme due to its widespread use, its transparency, and its possibility of distinguishing, pending formal validation, high-risk groups (who may warrant an aggressive periprocedural antithrombotic approach such as bridging) and low-risk groups (who warrant a minimalist approach), from moderate-risk groups, in which there is clinical equipoise and the greatest uncertainty as to the optimal management strategy. Preliminary data suggest that the three-tiered ACCP periprocedural TE risk scheme predicts a clinically meaningful increase in the 30-day postoperative risk of stroke among patients with CAF for each of the three risk tiers [5].

With respect to periprocedural bleed risk, most studies explicitly divided procedures/surgeries into a two-tier risk scheme of high post-procedural bleed risk (2-day risk of major bleed of 2–4% without bridging therapy) including most major surgeries lasting > 45 min, cardiothoracic and orthopedic surgeries, and those of low or non-high post-procedural bleed risk (2-day risk of major bleed 0–2%), including invasive procedures and same-day office procedures. Many of the studies also suggested that special emphasis be placed on certain ‘low-bleed risk’ procedures (such as pacemaker/defibrillator insertion or minor neurospinal procedures) due to the potential serious consequences of a bleed. We suggest that studies state explicitly the general type of procedure or surgery performed (i.e. cardiac catheterization or joint replacement) and avoid broad categorizations such as ‘orthopedic’ or ‘cardiovascular’ procedures, to more effectively assess the effect of the surgery or procedure type on the risk of bleeding.

Periprocedural anticoagulant regimens

The majority of studies in the systematic review explicitly described the type of VKA (warfarin or phenprocoumon), the type and dose of heparin used as bridging therapy (LMWH or UFH), the timing of the last pre-procedural and first post-procedural heparin dose, and the post-procedural VKA dose [4]. Approximately one-third of studies made no specific mention of dose or timing of periprocedural anticoagulant regimens. We propose that studies make explicit mention of the type, dose and timing of pre- and post-procedural antithrombotic agents used (including VKAs, NOACs and bridging therapy).

Arterial thromboembolic and major bleeding outcome definitions

For arterial TE, most studies (64%) used stroke, TIA and systemic embolism as the primary outcome because OAC therapy is used to prevent cardioembolic events [4]. A minority of studies included myocardial infarction (MI) and acute coronary syndrome (ACS) as part of the primary outcome definitions. We propose that studies do not include MI and ACS as part of the primary outcome due to the difficulty in attributing the occurrence of such events to cardioembolism in the periprocedural setting. However, MI and ACS may be considered secondary outcomes in anticoagulant-based or primary outcomes of antiplatelet-based periprocedural studies.

Most studies defined major bleeding as overt bleeding with a fall in hemoglobin > 2 g dL−1 or unexpected transfusion of ≥ 2 units of packed red blood cells, bleeding into a critical organ, fatal bleeding, and bleeding requiring re-hospitalization, prolonged hospitalization or need for repeat surgery. We propose that future studies adopt the ISTH surgical definition of major bleeding [6] with one caveat: the time-frame to capture bleed events needs to be relevant to the patient population being studied, as described below.

Time-frame to capture events

The majority of studies (73%) in the systematic review used a 30-day post-procedural time-frame to capture events such as arterial TE and major bleeding, with the view that this time-frame better represents adverse outcomes attributable to periprocedural antithrombotic management [4]. Very few studies used a 15-day time-frame and one study included five half-lives after study drug discontinuation for major bleed definitions [6,7]. A minority of studies used a 90-day time period with the view that this prolonged time-frame can better capture delayed TE events related to periprocedural bleeding. One such study found only 2/15 major bleeds (0.13%) and 1/8 arterial TE (0.13%) events after 4 weeks [8]. We propose that all studies adopt a 30-day post-procedural time-frame to report both arterial TE and major bleeds due to the potential to better discriminate periprocedural-related complications (i.e. better ‘signal-to-noise’ ratio). A follow-up period from 30 to 90 days may be considered for secondary reporting of outcomes.

Recommendations for standardized reporting of periprocedural antithrombotic therapy

We propose the following recommendations for future RCTs and cohort studies on periprocedural antithrombotic and bridging therapy, with emphasis on patients with arterial indications for OAC:

  • 1 There should be a standardized description of a patient’s arterial TE risk based on the ACCP suggested periprocedural TE risk stratification, including: (i) the type and position of valve in patients with MHV indications, as well as inclusion of major stroke risk factors (as included in the CHADS2 score); and (ii) the use of the CHADS2 score in patients with CAF indications for OAC.
  • 2 There should be a general description of each patient’s elective invasive procedure or surgery (with time of surgery based on anesthesia records if available). Broad and non-specific surgical descriptions such as ‘orthopedic’ and ‘cardiovascular’ surgery should be avoided.
  • 3 There should be a complete description of the type and dose of antithrombotic therapy, as well as timing of pre- and post-procedural discontinuation and re-initiation of periprocedural antithrombotics used. This should encompass both VKAs and NOACs as well as bridging therapy.
  • 4 The primary outcome of arterial TE should be defined as stroke, TIA and systemic embolism. MI and ACS may be considered secondary outcomes.
  • 5 The ISTH surgical definitions of major bleeding should be used.
  • 6 A 30-day post-procedural follow-up period should be used to report events, including major bleeding, which should be linked to a specific patient group as per TE risk and procedure as per bleed risk.


Harmonized reporting would facilitate across-study comparisons of outcomes and would enable study-level meta-analysis, thereby allowing more robust assessments of benefits and risks of different periprocedural antithrombotic management strategies. We suggest that the recommendations from this ISTH statement should be applied to both RCT and cohort studies, although we acknowledge that the very low ATE event rate found in the systematic review (1.02%, 95% CI, 0.50–1.55) necessitating large sample sizes [4], the heterogeneity of types of procedures, and the concern regarding generalizability of findings from RCTs, are likely to favor observational cohort studies to assess risk/benefit of a particular periprocedural antithrombotic strategy in the future. Future research in the field should validate an aggressive or minimalist periprocedural antithrombotic approach for subgroups of patients with low, moderate and high TE risk (as per the ACCP risk scheme) undergoing specific procedures. In addition, proof-of-concept that bridging therapy as an aggressive approach to prevent cardioembolic events is needed, and towards this end two large placebo-controlled trials of bridging therapy are underway [9,10]. Lastly, an acceptable upper boundary for the absolute ATE event rate, and its relation to major bleeding, possibly as bivariate endpoints, is required to design and power future studies in the field of periprocedural antithrombotic therapy.

Disclosure of Conflict of Interests

The authors state that they have no conflict of interest.