Perioperative Management of Patients who are Receiving a Novel Oral Anticoagulant (Dabigatran, Rivaroxaban or Apixaban) and Require an Elective Surgery or Medical Procedure: A Prospective Cohort Study (PAUSE)*

2.1 Overview

The NOACs, which include dabigatran, rivaroxaban and apixaban, are used for
stroke prevention in atrial fibrillation (SPAF) and to prevent and treat venous
thromboembolism (VTE). These drugs are appealing alternatives to warfarin
because they are given in fixed-doses that do not require routine laboratory
monitoring and have fewer drug interactions (1). The NOACs are contraindicated
in patients with severe renal insufficiency: dabigatran and rivaroxaban are
avoided if creatinine clearance (CrCl) <30 mL/min; apixaban is avoided if CrCl
<25 mL/min.
Dabigatran, an oral factor IIa inhibitor given 110 mg or 150 mg BID, is
dependent on the kidney for 80% of its clearance and has an elimination
half-life of ~14 hours (2, 3). Dabigatran was studied for SPAF in the RE-LY
trial, which involved 18,113 patients: the 110 mg dose was as effective as
warfarin (international normalized ratio [INR]: 2.0-3.0) for stroke prevention
and caused less bleeding than warfarin; the 150 mg dose was more effective than
warfarin for stroke prevention and had comparable bleeding (4). Rivaroxaban, an
oral factor Xa inhibitor given 20 mg OD (15 mg OD if CrCl 30-50 mL/min), is
dependent on the kidney for 33% of its clearance and has a half-life of ~8
hours (5, 6). Rivaroxaban was studied for SPAF in the ROCKET-AF trial, which
involved 14,264 patients; it was as effective as warfarin (INR: 2.0-3.0) for
stroke prevention and with comparable bleeding (7). Apixaban, an oral factor Xa
inhibitor given 5 mg BID (2.5 mg BID if 2 of 3 of: creatinine >133 *mol/L; age
*80 years; weight *60 kg), is dependent on the kidney for 25% of clearance and
has a half-life of ~8 hours (8-10). Apixaban was studied for SPAF in the
ARISTOTLE trial, which involved 18,281 patients; it was more effective than
warfarin (INR: 2.0-3.0) for stroke prevention and with less bleeding (11).
The *general enthusiasm* for the NOACs and their endorsement by practice
guidelines as a first-line treatment option for SPAF (12-15) should be tempered
by several considerations (16, 17), including the potential for misuse in the
perioperative setting (18). NOACs have two properties which can adversely
affect patient safety perioperatively if these drugs are not administered in a
careful, patient-focused manner (1): First, all NOACs depend to some, but
variable, extent on renal clearance which means the timing of pre-operative
NOAC interruption should be adjusted based on patient renal function to ensure
a minimal anticoagulant effect at surgery and, in turn, to minimize bleeding.
Second, all NOACs have a rapid, 1-3 hour, onset of peak anticoagulant effect
which means that their post-operative resumption should be adjusted and, if
necessary, delayed based on surgery/procedure-related bleed risk to minimize
bleeding.

2.2 Definition of the Clinical Problem
* The primary clinical problem is how to safely manage patients who are
receiving a NOAC for SPAF and require an elective surgery/procedure. Data are
lacking to inform best clinical practices, which are currently driven by expert
opinion (19-22).
* Related, secondary clinical problems are how to interpret, in the
perioperative setting after NOAC interruption, *everyday* coagulation tests
(comprising the INR, aPTT, TT, and prothrombin time [PT]), and *specialized*
coagulation tests (comprising the dilute TT for dabigatran and anti-factor Xa
levels for rivaroxaban/apixaban).

2.3 Scope of the Clinical Problem
The perioperative management of NOAC-treated patients with AF who require an
elective surgery/procedure is a common problem that is likely to increase for
two reasons: First, NOAC use for SPAF is increasing. As seen in Appendix 1,
this claim is supported by an audit of types of oral anticoagulants used by
patients who were assessed in the Hamilton Bridging Anticoagulation Clinics and
by data on NOAC prescriptions filled in Ontario. Second, an ageing population
will increase the prevalence of AF and such elderly patients are most likely to
require a surgery/procedure (23-25). Considering there are 2.5 million people
in North America who are taking anticoagulants for SPAF, of whom 25-50%
(estimated) will be taking a NOAC in the next 5 years and 20% will require a
procedure yearly (1, 26), this translates to 100,000-250,000 NOAC-treated
patients/year (divide by 10 for Canada) in whom the study results can be
applied.

2.4 Importance of this Study
Without this study, patient care may be adversely affected based on two
considerations:
i) Lack of standardized management may adversely affect patient safety. When
first considering warfarin-treated patients, physician surveys showed wide
variability in perioperative management before evidence-based recommendations
were developed (27, 28). Non-standard management resulted in perioperative MB
rates of 4-8% (29, 30), whereas in similar patients standardized management
conferred 1-3% rates of MB (31, 32). The importance of minimizing MB is shown
by studies in the perioperative (33) and non-perioperative (34) settings in
which prolonged anticoagulant interruption due to bleeding increases the risk
for thromboembolism. We acknowledge that physician surveys of real world
perioperative NOAC practices are lacking although contemporary response rates
are typically <50%, which limits the interpretation of such surveys (35, 36).
However, based on our interactions with health care professionals during
multiple educational talks across Canada, we have observed uncertainty and
variability in practices. For example, **we stop NOACs 5 days pre-op in all
patients** may reflect what is done for warfarin-treated patients (26), or **we
restart NOACs the evening post-op** may reflect what is done when NOACs are
used for VTE prophylaxis (37). Such practices, although well-intentioned, are
likely to adversely affect patient safety, as seen with warfarin patients.
ii) Uncertainty about coagulation tests interpretation may adversely affect
health care delivery. In warfarin patients, normal hemostasis (e.g., INR <1.4)
is expected before many surgeries and a similar expectation applies to NOAC
patients. The lack of data to inform the interpretation of coagulation tests
will create uncertainty about proceeding with surgery and may lead to
postponements. For example, what aPTT level is a patient *good to go* for
surgery? In dabigatran patients, does an increased TT indicate a significant
anticoagulant effect? Probably not but supportive clinical data are needed
(38). Uncertainty about abnormal tests also may increase the use of blood
products such as plasma or prothrombin complex concentrate, which is costly (up
to $2,480 @$0.62/U, source: Hamilton Health Sciences Blood Bank). This may be
done to reverse a presumed residual NOAC effect although the efficacy of such
treatments is questionable (39, 40), and are associated with a 1-2% risk for
thrombotic complications (41).

2.5 Recognition of the Importance of the Problem by Clinicians and Experts
The applicants have observed, subjectively through educational activities, that
clinicians consider the lack of evidence-based guidelines for perioperative
NOAC management an important knowledge gap. Interest in this area is evident by
visits to the pertinent sections of the Thrombosis Canada
(www.thrombosiscanada.ca) and UpToDateTM websites (see: Appendix 2). The
importance of this problem is also recognized by experts in this field (38,
42-44).

1.1 Overall Aim: To establish a safe, standardized protocol for the
perioperative management of patients with atrial fibrillation (AF) who are
receiving a novel oral anticoagulant (NOAC) drug, either dabigatran,
rivaroxaban or apixaban, and require an elective surgery/procedure.

1.2 Primary Aim: To demonstrate that a standardized but patient-focused
protocol for the perioperative management of each NOAC is safe, with acceptably
low rates of perioperative major bleeding (MB) and arterial thromboembolism
(ATE). The perioperative protocol is adjusted based on patient renal function
and surgery/procedure-related bleed risk, to optimize patient safety, and does
not involve heparin bridging anticoagulation.
Primary hypothesis: We hypothesize that the proposed perioperative management
protocol is *safe for patient care* in patients with AF, defined by a low risk
for MB (1%; 95% confidence to exclude 2%) and a low risk for ATE (0.5%; 95%
confidence to exclude 1.5%).

1.3 Secondary Aim: To determine the effect of the pre-operative NOAC
interruption protocol on the level of residual anticoagulation, when measured
by *everyday* coagulation tests that are not NOAC-specific (e.g., activated
partial thromboplastin time [aPTT]) and *specialized* coagulation tests that
are NOAC-specific (dilute thrombin time [TT] - HemoclotTM, and anti-factor Xa
assays).
Secondary (exploratory) postulate: We postulate that the standardized protocol
will be associated with a high incidence (*90%) of low pre-surgery/procedure
residual NOAC levels, defined as <50 ng/mL, when measured by NOAC-specific
coagulation tests. We acknowledge the exploratory nature of this postulate,
which is based on empiric observations, but deem a laboratory component to this
study important to address an unmet clinical need.

7.1 Study Design
We propose a prospective cohort study, with 3 parallel cohorts for each NOAC,
to assess the safety (MB and ATE risk) of a standardized perioperative
management protocol in patients with AF who are receiving a NOAC and require an
elective surgery/procedure.

7.2 Justification of Study Design
The proposed study design was chosen based on the following considerations:
i) A cohort design is appropriate to assess the safety of a clinical management
that is expected to be associated with low rate of adverse outcomes (with
sufficient statistical power to exclude a clinically important higher event
rate). Although cohort studies have the potential for patient selection bias
(86), for example if *high-risk* patients are excluded, such bias is unlikely
in our proposal for two reasons: First, in the Perioperative Dabigatran Study,
which has a similar cohort design, >90% of eligible patients were recruited.
Second, as there is no *standard-of-care* for the perioperative management of
NOAC patients, we believe clinicians and patients will welcome the opportunity
to participate in a study that aims to optimize patient safety. Overall, a
cohort design is efficient (high screen-to-recruitment ratio), reliable
(prospective documentation of patient data, perioperative management and
outcomes), and optimizes generalizability to *real world* practice (includes a
representative population, not just *healthier* patients).
ii) We considered a cohort study with a control group of warfarin-treated
patients who have an elective surgery/procedure but rejected this design
because some of these patients will receive LWMH bridging whereas our protocol
does not include bridging. Such a comparator would be acceptable only if all
warfarin patients are not bridged. It is not appropriate to mandate this
because guidelines support LWMH bridging in selected warfarin patients with AF
(26, 87).
iii) We considered a randomized trial but rejected this design as any compactor
group, for example stopping NOACs 5 days pre-procedure in all patients (as with
EU study), would constitute an experimental arm. Alternatively, a cluster
randomized trial, where clinical sites are randomly allocated to manage
patients according to the proposed protocol or *usual care*, also is infeasible
because our protocol, that is largely outlined in clinical reviews (1, 19),
would be the de facto *usual care* at many clinical sites. Moreover, a
randomized trial is likely to pose patient recruitment challenges as seen in
the BRIDGE (NCT00786474) and PERIOP-2 (NCT00786474) bridging trials, which have
been prolonged by 3-5 years because of poor patient recruitment.
iv) We considered studying a subgroup of patients who continue NOACs but
decided against this because the rationale for continuing warfarin
perioperatively (e.g., as in pacemaker insertion) is predicated on the unproven
premise that the perioperative risk for ATE is clinically significant (61) and
increases (59) during a short period of anticoagulant interruption. This point
is less relevant for NOACs, which have a much shorter sub-therapeutic
anticoagulation interval than warfarin. Moreover, if patients* risk for
perioperative ATE is low (<0.5%) (62), there is no compelling reason for
anticoagulation continuation, which may increase patient harm (bleeding). We
acknowledge preliminary data in 25 patients suggesting that continuing NOACs
around pacemaker implantation may be safe (88), but this issue requires
additional and careful study.

- Risks of the extra blood drawl are minimal; hematoma or thromboflebitis of
the affected vein. These risks/complications are easily to correct en will not
result in a prolonged admission in hospital.
- Patients will be assessed on the procedure day and contacted 4 times (weekly)
by telephone for 1 month to assess for outcomes. Hospitalized patients will be
seen daily.