Addressing Barriers to Optimal Oral Anticoagulation Use and Persistence Among Patients with Atrial Fibrillation: Proceedings, Washington, DC, December 3–4, 2012

2. Lack of awareness of the potential use of novel oral anticoagulants for vitamin K antagonist (VKA)-unsuitable patients

NOACs have several advantages over VKAs, the most salient of which is lower risk of intracranial hemorrhage and hemorrhagic stroke found for all of the NOACs. Others include lack of need for therapeutic monitoring and a modest but worthwhile reduction in mortality found in several clinical trials. Having been in use for more than sixty years, VKA nonetheless remains the dominant treatment for stroke prevention in AF. In the modern therapeutic era, patient selection factors for warfarin therapy compared with NOACs may not be immediately apparent. Moreover, concern over use of VKAs – and often over issues specific to VKA therapy such as the need for close INR monitoring – is a common reason for not using OAC of any type. Historically, a variety of reasons not to use OAC have been put forward.¹⁸ The misperception that aspirin is sufficiently effective for stroke prevention and substantially safer than the novel drugs appears to be a significant contributor to the problem of OAC undertreatment.

3. Lack of recognition of expanded eligibility for oral anticoagulation

The improved side effect profile of NOACs over VKAs may alter OAC eligibility. In fact, guidelines have evolved to recommend OAC for patients with at least 2 CHA₂DS₂-VASc risk factors, and to be preferred or considered for patients with at least one CHA₂DS₂-VASc risk factor.^19,20 Further, patients not commonly thought to be at high risk for stroke may nonetheless derive benefit from the NOACs, which have lower risk of hemorrhagic stroke and intracranial hemorrhage.^21,22

The threshold for OAC initiation is determined by the benefit of treatment balanced against the risk of adverse events, notably serious bleeding. The most recent study comparing warfarin to antiplatelet therapy, ACTIVE-W, found that warfarin had a favorable risk-benefit balance among patients with a CHADS₂ score of 1.²³ Newer data strongly indicate that even relatively low-risk patients with AF benefit substantially from anticoagulant therapy, best shown to date with apixaban.²⁴ Patients at lower risk of stroke with a CHADS₂ score of 1 made up as much as one third of the population in several of the NOAC clinical trials.^10,11 In the AVERROES trial, patients with CHADS₂ score of 0 or 1 had significant relative and absolute reductions in stroke with apixaban versus aspirin (6/1004 (0.54%/year) versus 16/1022 (1.41%/year), hazard ratio (HR) 0.38, 95% confidence interval (CI 0.14–0.93) but a comparable risk of major bleeding (6/1004 (0.54%/year) versus 6/1022 (0.53%/year), HR 1.02, 95% CI −0.32–3.26).²⁵ However, neither professional guidelines nor the Food and Drug Administration (FDA)- approved labeling support initiating OAC among AF patients with a CHADS₂ score of 0, and the most recent European and American College of Cardiology/ American Heart Association guidelines recommend no antithrombotic therapy for patients with CHA₂DS₂-VASc score of 0.^19,20

Improving stroke risk stratification could allow identification of patients for treatment who were not previously thought to benefit from OAC. The discriminatory power of the CHADS₂ score is moderate,²⁶ suggesting that its use may result in patients whose true stroke is low receiving OAC, while those with a relatively high stroke risk may not. Data show that accounting for female sex, age 65 to 74 years, and vascular disease (as in the CHA²DS₂-VASc score) adds discriminatory power,^27,28 particularly among those with a CHADS₂ score of 0 or 1.²⁹ Cardiac biomarkers, including high sensitivity troponin and N-terminal proB-type natriuretic peptide, as well as creatinine clearance may further improve prognostic power.^30–32

4. Lack of reversal agents and lack of ability to monitor effects of novel oral anticoagulants

Clinicians are uncomfortable with the absence of good anticoagulation reversal strategies for the NOACs. Concern exists regarding life-threatening hemorrhage on NOACs, particularly in patients requiring invasive procedures and especially when those are needed emergently.³³ Monitoring of NOAC treatment effect may be desirable in these situations, in the event of an overdose, and in advance of planned procedures such as cardioversion when there are questions about drug compliance and concern with the thrombosis risk associated with inconsistent anticoagulation. Unfortunately, reversal options and monitoring strategies and how they may inform care are not well-defined.^34,35

5. Bleeding risk of oral anticoagulants, particularly in the setting of dual antiplatelet therapy

Dual antiplatelet therapy (including following coronary stent placement) and OAC are indicated for acute coronary syndromes and AF, respectively. The combination of disease processes poses a therapeutic dilemma, as bleeding risk is significantly elevated when antiplatelet and antithrombotic therapies are used simultaneously. Compared with aspirin alone, triple therapy with aspirin, VKA and clopidogrel increases bleeding fourfold.³⁶ When prasugrel rather than clopidogrel is used, bleeding rates may be even higher.³⁷ There is a similar increase in risk when adding a NOAC to aspirin and clopidogrel, with a 3-to 4-fold increased risk in major bleeding events.^38,39 The use of ticagrelor or prasugrel rather than clopidogrel might be expected to further elevate bleeding risk in this setting, although empiric data are lacking. More data on these issues are needed to inform clinical decision making.

6. Suboptimal time in therapeutic range (TTR) for VKA

VKA has a narrow therapeutic window, and the amount of time spent in therapeutic range (TTR) varies. Although cause and effect is unproven in the absence of randomized data, low TTR is generally associated with increased risk of bleeding and of stroke;⁴⁰ conversely, in some datasets, patients with a high TTR are more prone to bleeding, presumably related to higher TTR being mainly due to less time in sub therapeutic range.⁴¹ Achieving a higher TTR, and possibly lower event rates, is possible through regular INR monitoring with timely and appropriate dose adjustment.⁴² However, monitoring can be burdensome for many patients, particularly those who do not have ready access to a medical facility. The inconvenience of monitoring may lead patients to decide against starting VKA, to decreased adherence, or to discontinue treatment altogether.

7. Fear of bleeding events with novel oral anticoagulants

VKA-associated bleeding is common in clinical practice. By contrast, no information was available regarding bleeding rates associated with NOACs in general practice until recently, although bleeding rates appeared generally similar to warfarin in clinical trials of NOACs.⁴³ Following the approval of dabigatran in October 2010, a substantial number of reports of serious and fatal bleeding events were submitted to the US Food and Drug Administration’s Adverse Event Reporting System. The number of reports of bleeding associated with dabigatran was considerably higher than the number of reports with warfarin. These findings contrasted with those of a large controlled trial, Randomized Evaluation of Long-Term Anticoagulation Therapy (RE-LY), which showed that bleeding rates with dabigatran 150 mg twice a day and warfarin were similar, and bleeding was less with dabigatran at the 110 mg twice a day dose. Concern regarding dabigatran-associated bleeding rippled through the medical literature and the lay press alike. Because adverse drug effects that were not detected in clinical trials can appear when a drug is broadly used, regulatory authorities have a responsibility to understand and respond to such concerns. Detailed review of the spontaneous reports did not identify any unknown risk factors for bleeding, and dabigatran was generally used in accordance with its FDA label. FDA responded with careful analyses of insurance-claim and administrative data from the Mini-Sentinel database, which has the advantage, compared to spontaneous reports, of a clearly defined denominator of patients on dabigatran and on warfarin as well as a systematic report of all observed, clinically significant bleeding events associated with each drug. Dabigatran-associated bleeding rates were clearly not increased compared with warfarin, and in fact bleeding on dabigatran appeared similar to if not lower than bleeding on warfarin.⁴⁴

8. Costs and insurance coverage

The total costs of care using NOACs compared with warfarin may not be very different.⁴⁵ While costs of NOACs themselves are substantially higher than VKA, savings associated with NOACs occur in the clinical sphere with fewer intracranial bleeding events and a reduced need for monitoring. As a result, dabigatran has been recommended as a therapeutic option for AF patients in the United Kingdom.⁴⁵ In the United States, VKAs and associated monitoring are affordable for most patients with regard to “out-of-pocket” costs, but NOACs are less so. Whether NOACs are included in various formularies and to what degree their costs are covered is variable. The tiered structure of Medicare Part D, for example, may not fully reflect the clinical benefits (and related health care system savings) of the products. In the instances in which a NOAC is covered, out-of-pocket costs are at times difficult to ascertain prior to filling a prescription. Moreover, with the ‘donut hole,’ patients may be held accountable for a greater amount of the cost of NOACs depending on the costs of their other drugs. Finally, coverage for NOACs (and individual patient’s ability to pay) may vary over time, leading to transitions to VKA and associated adverse outcomes.⁴⁶

1. Define reasons for oral anticoagulant underuse classified in ways that can guide intervention to improve use

Although numerous studies have documented that only about half of AF patients with risk factors for stroke are treated with an OAC in various health care settings, the specific reasons are less well known. There was consensus in the working group that better understanding of why so many patients are not being treated is a high priority. Challenges around safe and effective use of warfarin, such as documented or perceived inability to comply with monitoring, is a commonly given reason. Multiple reasons are often reported in individual patients. Patient and/or physician preference for antiplatelet therapy is a frequently cited reason, but this presumably reflects a lack of understanding of how inferior anti-platelet therapy is compared to OAC. Concern with potential bleeding is an important factor, and the lack of reversal agents for the NOACs is a widely expressed concern among physicians even though this may be less of a problem due to their relatively short half-lives. Reasons to withhold OAC therapy could be categorized in two domains: according to whether it is a patient, a provider, or a system level reason; and according to whether it is appropriate, inappropriate, or of uncertain appropriateness (Figure 2).

Open in a separate window

Figure 2

Reasons for Unsuitability for Oral Anticoagulation Among Patients with AF and CHADS ≥1 and/or CHADS-VASc ≥2

*Applicable to VKA only.

2. Increase the awareness of stroke risk and the value of oral anticoagulant use via impactful educational initiatives

Patient educational efforts should focus on the threat of preventable stroke despite AF’s often asymptomatic nature. Educational initiatives should target a broad array of physician groups involved in the management of AF patients. To maximize the yield of patient-provider interactions, the development of decision aids for shared decision-making,⁴⁷ multifaceted educational materials, and point-of-care decision support is needed. Although it is logical to focus on educating cardiologists, education of primary care physicians, hospitalists, emergency physicians, and advanced practice providers will be essential to guide improved care. Case-based studies, interactive teaching methods, education embedded into patient care environments, and assessment of education effectiveness are important elements of improvement efforts. Identifying barriers to use and opportunities to guide optimal use of OACs at a health system level is an important priority.

3. Collect data and feedback performance regarding oral anticoagulant use among eligible patients to providers

“If you don’t measure it, you can’t improve it” is an appropriate adage for anticoagulation for AF. An important question is defining who should be treated with OAC. While this seems simple on the surface, identification of such patients using electronic health records in the United States has been challenging. Although the CHADS₂ risk factors are easy to measure, they may not be easy to assess in an electronic medical record without specifically collecting information on whether or not they are present. A crucial question, as yet unanswered, is, “How much AF is enough to warrant treatment?” The trials studying anticoagulants have generally included patients with a clinical diagnosis of AF who were either in AF at the time of enrollment or who had two documented episodes two weeks apart. Six minute episodes of silent AF in older patients with cardiac devices are associated with a 2.5 fold risk of subsequent stroke,⁴⁸ but more studies are needed to determine if OAC is beneficial for these patients.

To measure performance in clinical care, patients with AF must be identified, the presence of stroke risk factors must be assessed, and the use of OAC and presence of contraindications should be determined.⁴⁹ Several quality improvement initiatives addressing this are underway, including the American College of Cardiology’s PINNACLE-AF registry,⁵⁰ the Outcomes Registry for Better Informed Treatment of AF (ORBIT-AF),⁵¹ the Global Anticoagulant Registry in the FIELD (GARFIELD) registry,⁵² and the American Heart Association’s Get With The Guidelines program.⁵³ Traditional registries, while helpful in assessing how populations are being treated, are not well suited to providing measurement and feedback in real time to improve care for individual patients. Much more work needs to be done to enhance measurement, feedback, and broad interventions to improve the use of OAC, including the use of appropriate electronic tools and incentives at the health system level. A data warehouse in a health system could be used to identify candidates for OAC treatment and follow-up on a system level rather than solely relying on individual physicians to make appropriate decisions in real-time. The increasing availability of electronic health records represents a rich opportunity for broad, real-time assessment and feedback of therapeutic decisions and clinical outcomes. Improvement in standardized electronic decision support tools may enhance the point of care use of OAC. OAC adherence may be enhanced electronically by providing reminders to patients on their mobile devices that their prescription refill time has lapsed. These strategies, in conjunction with data analytics of the health system data warehouse, may provide a true “safety net” for select patients.

4. Define who should receive vitamin K antagonists rather than novel oral anticoagulants

Patients who have been on warfarin for a significant period of time, are on a stable dose with stable INRs, and can comply with frequent monitoring, may prefer to stay on warfarin despite a higher risk of intracranial hemorrhage. Similarly, a NOAC may not be suitable for patients with advanced renal disease, e.g. creatinine clearance < 25 to 30 ml/min. Rare patients may develop intolerance to NOACs but can tolerate VKA. NOACs are approved for use in “non-valvular” AF. The term “non-valvular atrial fibrillation,” however, needs further definition, since more than one quarter of patients in some of the trials of NOACs had moderate or severe valvular abnormalities, with consistent treatment effects in that subgroup.⁵⁴ The novel agents should not be used in patients with significant mitral stenosis (who were excluded from the trials) or with mechanical prosthetic valves (for which the novel agent tested was neither safe nor effective).⁵⁵ Finally, out-of-pocket expenses may be substantially less with warfarin compared with a NOAC (Table 2), and higher cost will continue to be an important barrier for many patients.

5. Identify NOAC reversal agents and monitoring strategies

It is important to recognize that while vitamin K, fresh frozen plasma, and prothrombin complex concentrates reverse the coagulation test effects of VKAs, their effectiveness on reducing bleeding and its consequences is much less well established. Furthermore, data regarding risk of NOAC-associated periprocedural bleeding are reassuring, including data from the RE-LY trial showing similar or lower serious bleeding with dabigatran than with warfarin, even among patients undergoing emergent procedures.⁵⁶ This may be due, in part, to their shorter half-life in comparison with VKA such that the effect is largely gone 1 to 2 days after the last dose. Nonetheless, research to identify ways to quickly reverse the effect of NOACs and monitor their anticoagulant effect is needed and is underway.^57,58 Andexanet alpha, for example, is a recombinant protein that functions as a factor Xa decoy, and it has shown promise with regard to reversing effects of oral factor Xa inhibitors.⁵⁹ A monoclonal antibody fragment antidote for dabitatran is under development.⁶⁰ As more data become available, there will be a value in making NOAC reversal strategies widely interpretable and accessible, perhaps analogous to the poison control model.⁶¹ A website or hotline to provide guidance on this and other practical issues regarding all NOACs may be useful. It will be important to engage hematology specialists and to demonstrate how institutional protocols can help guide care. Along these lines, reviews providing practical guidance are currently available.^35,62,63 Helping providers deal with every-day practical issues in the use of the novel drugs is important in enhancing their safe and effective use.

6. Minimize the duration of dual antiplatelet therapy and concomitant OAC use

Among AF patients undergoing percutaneous coronary interventions, a bare metal stent is preferable to a drug-eluting stent in the absence of a clear need for the latter. Among patients with a significant bleeding diathesis and acceptably low stroke risk, consideration can be given to temporarily suspending OAC and resuming it when an antiplatelet agent is no longer required. Guidelines have encouraged avoiding aspirin when using VKA, unless there is a clear indication, i.e. within a year of a myocardial infarction. Such advice is likely to be applicable to the novel agents as well. Avoiding aspirin may result in important reductions in bleeding.

7. Improve time in therapeutic range for patients on warfarin

Organized anticoagulation services have been shown to improve care and outcomes for patients on VKA. Scheduling and ensuring appropriate follow-up is part of this systematic approach. INR checks should generally occur at least monthly among those on a stable dose and more often among those requiring dosing adjustments.⁴⁹ Automated telephone or electronic appointment reminders may aid in the process. Alternatively, use of point-of-care INR devices may also improve TTR. In a randomized clinical trial, INR self-testing was comparable to in-office venous blood draws with regard to bleeding and stroke rates.⁶⁴ The safety of self-monitoring has been demonstrated in several other clinical trials.⁶⁵ Accordingly, the Veterans Affairs healthcare system recently revised their anticoagulation policy to allow self-testing in place of venous plasma testing. If outcomes are improved, this strategy may be adopted by other healthcare systems.

8. Leverage observational datasets to refine understanding of oral anticoagulant use and outcomes (effectiveness and safety) in general practice

Using the Mini-Sentinel database, the FDA has demonstrated its ability to assess the rate and impact of spontaneous bleeding reports in a timely way.⁴⁴ As electronic health records become more widespread and interconnected, this process is expected to become more useful and safety signals may even be detected in real-time. The large number of bleeding events attributed to dabigatran versus warfarin in spontaneous reports to the FDA was thought to be driven largely by a heightened awareness of the dabigatran bleeding because of the drug’s novelty, leading to a larger fraction of events that were reported for dabigatran compared with warfarin, a familiar drug known to cause bleeding.⁶⁶ Such a reporting bias is always possible in a spontaneous reporting system. As noted, the mini-Sentinel sites, including all parts of the health care system, did not have such a bias. A similar effect may be observed when other NOACs are used more widely.

9. Better align health system incentives

The full set of patient outcomes over the care cycle should be weighed against total costs for the patient’s condition, rather than the costs borne by a single payor. Educational programs should be designed that include systems improvement to measure, provide feedback on, provide tools for, and establish incentives to guide optimal OAC for AF. An important step is to be able to identify the AF population with electronic health records. A system is needed to categorize patients not being treated in a way that assesses their risks and can lead to specific interventions that will improve their care. The goal is to enhance value for patients, rather than simply focusing on cost containment. Patients with AF should be fully educated regarding risk of stroke and its consequences. Further, out-of-pocket expenses of OACs in various formularies should be made clear. With therapeutic benefit and costs in mind, patients will be better equipped to make appropriate, informed decisions regarding whether to use OAC and what insurance coverage best suits them. An important opportunity will be to advocate for use of OAC performance measures⁴⁹ with feedback to providers and health systems as appropriate.

We give special thanks to Robert Temple, MD, for his important contributions to this manuscript.