How I manage anticoagulant therapy in older individuals with atrial fibrillation or venous thromboembolism

Ischemic stroke is a major cause of morbidity and mortality in patients with atrial fibrillation (AF), and pulmonary embolism (PE) is a major cause of death in hospitalized patients.^1,2  Both cardioembolic stroke and recurrent venous thromboembolism (VTE) can be prevented with oral anticoagulation. The most common side effect of anticoagulation is bleeding, and bleeding risk rises sharply with increasing age.³  Concern about the risk of bleeding with anticoagulation in older patients with AF or at risk of recurrent PE presents an important challenge for clinicians, because both these conditions are common in older individuals.^4,5 

The non–vitamin K antagonist oral anticoagulants (NOACs) represent a major advance in anticoagulant therapy (Table 1), particularly for older patients, because they are safer than vitamin K antagonists (VKAs), such as warfarin, and at least as effective.⁶  Nevertheless, many older patients with AF or at risk of recurrent PE do not receive anticoagulant therapy, and when treated with a NOAC, they frequently receive lower doses than those recommended by the guidelines, because clinicians fear bleeding.^3,7-11  Undertreatment of older patients is a significant cause of preventable morbidity and mortality.^12,13 

In this article, we use clinical cases to illustrate challenges associated with anticoagulant use in older patients and describe how we approach these challenges.

An 81-year-old white male is diagnosed with AF. Additional risk factors for stroke include type 2 diabetes, hypertension, and prior stroke. He has stable angina, and chronic kidney disease (CKD). His medications include aspirin, metformin, hydrochlorothiazide, and ramipril. He weighs 75 kg, blood pressure is 146/88 mm Hg, and estimated creatinine clearance (CrCl) using the Cockcroft-Gault formula is 37 mL/min.

The goal of anticoagulant therapy in patients with AF is to prevent stroke. Randomized trials have demonstrated that compared with placebo or untreated control, warfarin adjusted to a target international normalized ratio (INR) of 2 to 3 reduces the risk of stroke or systemic embolism by 64%.¹⁴  Although warfarin increases major bleeding, including intracranial bleeding, it reduces all-cause mortality by 26%, indicating a clear net clinical benefit over no treatment.¹⁴  There were few patients aged ≥75 years in these trials, but the results indicate that there is a consistent benefit of warfarin in older patients with AF.^15-17 

Prior to the introduction of NOACs, only ∼50% of eligible AF patients received treatment with a VKA, and only one-half of those who received treatment had a time in therapeutic range >60%.¹⁸  There is some evidence that patients who fail to achieve a time in therapeutic range >60% to 65% do not derive a benefit from warfarin compared with dual antiplatelet therapy.¹⁹ 

NOACs have been tested as alternatives to warfarin in four randomized trials involving 71 683 patients.^20-23  The pooled data indicate that NOACs compared with warfarin significantly reduce stroke or systemic embolism by 19%, major bleeding by 14%, fatal bleeding by 51%, and mortality by 10%.^24,25  The reduction in stroke was driven primarily by a 51% reduction in hemorrhagic stroke.²⁴  Importantly, the treatment effects of the NOAC compared with warfarin for stroke prevention were consistent irrespective of age (Table 2).^26-29  Because the risk of intracranial and fatal bleeding increases with age, the absolute reduction in bleeding with NOACs was greatest in older patients.

Clinicians frequently recommend aspirin in older patients because of a perceived safety advantage over anticoagulants. Although meta-analysis data indicate that aspirin compared with placebo or untreated control reduces stroke or systemic embolism by 21% in patients with AF, it is much less effective than warfarin, which produces a 38% stroke reduction vs aspirin, with no evidence of an increase in extracranial or intracranial bleeding.¹⁵  Furthermore, in the only randomized trial directly comparing a NOAC with aspirin in AF, apixaban halved the risk of stroke or systemic embolism without significantly increasing major bleeding or intracranial hemorrhage (ICH).³⁰  Thus, aspirin is not a suitable alternative to anticoagulation for stroke prevention in AF.

Guidelines recommend that clinicians assess stroke risk in patients with AF using a validated tool (Table 3) and that they prescribe anticoagulation in those with an annual stroke risk ≥1% to 2%.^31-33  When using CHA₂DS₂-VASc to evaluate stroke risk, this corresponds to a score of ≥1 for men and ≥2 for women. Patients with AF ≥75 years of age score 2 points for age and thus qualify for anticoagulant therapy, irrespective of additional stroke risk factors. Bleeding risk can also be assessed using validated tools, but most older patients at high bleeding risk should not be denied anticoagulation, because stroke risk and bleeding risk often track closely, and those at highest risk of bleeding derive the greatest overall benefit of anticoagulation.³²  Clinicians should instead focus efforts on addressing modifiable bleeding risk factors (eg, uncontrolled hypertension) and avoiding concomitant therapies that increase bleeding (eg, antiplatelet agents and nonsteroidal antiinflammatory drugs) (Table 3). Rare patients in whom anticoagulation should be avoided because of an unacceptable risk of bleeding might include those with recurrent major bleeding, or those with a first episode of major bleeding who are deemed to have an unacceptable risk of recurrent bleeding if treatment is restarted (eg, ICH in patients with amyloid angiopathy).^32,34 

Our patient has a CHA₂DS₂-VASc score of 7, which corresponds to an estimated ischemic stroke risk of 11.2% per year (Table 3).³⁵  Risk factors for bleeding include age, moderate CKD, prior stroke, uncontrolled blood pressure, and aspirin use. To minimize bleeding risk, he requires improved control of blood pressure (BP). We do not delay starting anticoagulation unless BP is >160/90 mm Hg, because the benefits of starting anticoagulation below this threshold are likely to substantially outweigh the risk of bleeding. We recommend stopping antiplatelet drugs at the time of starting anticoagulation, because there is no evidence that the addition of an antiplatelet drug provides additional benefit for prevention of cardioembolic stroke or arterial vascular events in patients receiving full dose anticoagulation, but there is evidence of increased bleeding. Exceptions may include patients within 1 year of an acute coronary syndrome or coronary revascularization (stenting or bypass surgery), or within 3 months of carotid or peripheral artery stenting.

Consistent with guideline recommendations, we prefer a NOAC over warfarin in older AF patients who do not have a contraindication (eg, major drug–drug interactions, end-stage kidney disease).^31-33  In choosing among the NOACs, we consider their potential for drug–drug interactions, convenience of dosing, and efficacy and safety as well as patient characteristics and preferences (Table 4). All 4 available NOACs provide important convenience and safety advantages over warfarin.

In our patient, we prefer apixaban 2.5 mg twice daily, dabigatran 110 mg twice daily, or edoxaban 30 mg daily, because they are at least as effective as warfarin for stroke prevention and significantly reduce major and intracranial bleeding (Table 4). We use the lower dose of these agents because our patient meets the criteria for dose reduction (Table 1). These criteria identify patients at risk of anticoagulant overexposure if the standard NOAC dose is used. Dabigatran 110 mg twice daily is not licensed in the United States. Dabigatran 150 mg twice daily (in the United States) and rivaroxaban 15 mg daily are also good choices, because they are at least as effective as warfarin for stroke prevention and reduce the risk of ICH.

We generally see patients 3 months after anticoagulant initiation to evaluate adherence and possible adverse events, check renal function, and plan interruption for any scheduled procedures (Table 5).³⁶ 

A 78-year-old African American widow is hospitalized following acute right middle cerebral artery territory stroke causing left-sided hemiparesis and dysphasia. She has hypertension and was diagnosed with AF 1 year earlier but was not anticoagulated and instead was prescribed aspirin 81 mg daily. She lives alone and has had 3 falls in the last 6 months. Her calculated CrCl using the Cockcroft-Gault formula is 57 mL/min.

Observational studies conducted prior to the introduction of NOACs have shown that <50% of older patients with AF who fulfilled guideline criteria for anticoagulant therapy were treated with a VKA. Furthermore, compared with white patients, African Americans are less likely to be treated with an anticoagulant (odds ratio, 0.28; 95% confidence interval [CI], 0.13-0.60), have poorer INR control when taking a VKA, and are at higher stroke risk.^37-39  Although the introduction of NOACs has led to an increased uptake of oral anticoagulants, underuse remains a major concern, especially in the older population, and cost remains a barrier.^10,40  Comorbidities (eg, CKD and cognitive impairment), frailty, polypharmacy, and a history of falls further contribute to the underuse of anticoagulants in older patients.

The risk of falls increases sharply with age, and older patients with falls have a two- to fourfold increase in the risk of traumatic ICH compared with those without falls.⁴¹  This risk is even higher in patients with prior stroke or major bleeding and in those with neuropsychiatric impairment. Fear of ICH during warfarin therapy is the major reason why physicians prescribe aspirin for older patients with a history of falls, despite the lack of evidence that aspirin is associated with a lower risk of major bleeding.^12,41  In a subanalysis of the PREvention oF thromboembolic events–European Registry in Atrial Fibrillation (PREFER in AF) involving 6412 patients, of whom 505 were aged ≥ 85 years, Patti et al reported that major bleeding was not significantly different in older patients receiving anticoagulants compared with those receiving antiplatelet agents (4.1% vs 3.9%).⁴²  In a study of 1245 older patients with AF at risk of falls, Gage et al reported that the rate of ICH in patients treated with antiplatelet agent was not significantly different to that of warfarin-treated patients.⁴¹  Furthermore, their analyses suggested that warfarin provided a net clinical benefit if the estimated annual risk of stroke was >5% (CHADS₂ score ≥2). Thus, compared with no anticoagulation, patients treated with warfarin had a lower risk of death or hospitalization for stroke, myocardial infarction, or hemorrhage (hazard ratio [HR], 0.75; 95% CI, 0.61-0.91) if the CHADS₂ score is ≥2.

A separate modeling study confirmed the findings of Gage et al and suggested that patients with a CHADS₂ score ≥2 would need to fall more than 295 times per year for the increase in trauma-induced subdural hemorrhages to offset the reduction in stroke with warfarin therapy.⁴³  Although there are no such analyses for other anticoagulants, these results are likely to also apply to NOACs, because they have a better risk–benefit profile than either warfarin or aspirin in older patients.

Our patient has a CHA2DS2-VASc score of 6, which corresponds to an estimated ischemic stroke risk of 9.7% per year (Table 3).³⁵  She has several risk factors for bleeding, including age, recent stroke, and a heightened risk of falls in the setting of the new neurological deficits, but despite this, the benefits of anticoagulation clearly outweigh the risks.

We delay the start of anticoagulation in our patient until at least 6 days after the acute stroke because of the risk of hemorrhagic transformation. Our decision on when to start is guided by consensus recommendations (eg, the “1-3-6-12 day rule”), which suggest that anticoagulation can be initiated on day 1 after transient ischemic attack and after 3, 6 and 12 days in AF patients with mild, moderate, and severe stroke, respectively.⁴⁴  Before starting anticoagulation, we repeat brain imaging to ensure that there is no hemorrhagic transformation. We assess falls risk in all patients and recommend multidisciplinary interventions to prevent falls in those at risk.

Upon initiation of anticoagulation, we do not lower the dose of the NOAC in patients at increased risk of bleeding unless established criteria for dose reduction are met (Table 1). We evaluate the potential for drug–drug interactions and assess renal function for NOAC dosing. With a CrCl >50 mL/min, we could use apixaban 5 mg twice daily, dabigatran 110 mg twice daily, or edoxaban 60 mg daily, because they are at least as effective as warfarin and significantly reduce major and intracranial bleeding (Table 4). Dabigatran 110 mg twice daily is not licensed in the United States. Dabigatran 150 mg twice daily and rivaroxaban 20 mg daily also reduce the risk of ICH compared with warfarin and are also acceptable choices.

To maximize adherence, we engage the patient and her carers in the decision to initiate anticoagulation, provide education about the benefit and risk of anticoagulant therapy, and review adherence at every follow-up visit (Table 5).

We believe that 2 important barriers to the use of anticoagulant in older patients with AF are lack of awareness of stroke risk and lack of understanding of the potential benefit of anticoagulant therapy. These issues need to be addressed at the level of patients, health care providers, and the health care system.⁴⁰  By providing specific advice on the benefits of anticoagulation in older patients with AF, guidelines could promote the appropriate use of anticoagulant therapy.

An 87-year-old male presents with sudden onset of pleuritic chest pain and breathlessness. One week earlier, he had been diagnosed with a right leg deep vein thrombosis and was prescribed rivaroxaban 15 mg twice daily. He lives alone. Comorbidities include mild cognitive impairment, macular degeneration, coronary artery disease with previous bypass surgery, hypertension, benign prostatic hypertrophy, and right total knee replacement for osteoarthritis. Computer tomography pulmonary angiography reveals bilateral segmental and lobar PE. He has a poor understanding of the deep vein thrombosis diagnosis and the rationale for treatment, and does not know whether he is taking a “blood thinner”. Although he was dispensed 3 weeks of rivaroxaban, the pill bottle still contained 19 of the 21 pills issued by the pharmacy. Consequently, the therapeutic failure was attributed to lack of adherence to anticoagulant therapy.

More than one half of patients with acute VTE are aged >70 years.⁵  In the RIETE registry, older patients with acute VTE treated with VKA had higher 3-month rates of fatal PE (3.7% vs 1.1%) and major bleeding (3.4% vs 2.1%) than younger patients.⁴⁵ 

NOACs are now replacing warfarin for acute VTE treatment because of the convenience of fixed oral dosing and superior safety, particularly in older patients. Pooled analyses of randomized trials in acute VTE treatment indicate that in patients aged ≥75 years, NOACs, when compared with warfarin, are associated with a large reduction in major bleeding and fewer VTE recurrences (Table 6).⁴⁶ 

Adherence is the extent to which patients take their medications as prescribed and consists of 3 components: initiation, implementation, and discontinuation.⁴⁷  Thus, some patients do not initiate treatment, others start treatment but do not take their medications according to the prescribed dosing regimen, and others prematurely discontinue their medications. Adherence barriers can be patient related (eg, poor understanding of the diagnosis and rationale for therapy, misunderstanding instructions, or cognitive impairment), medication related (eg, complex dosing regimens or fear of side effects such as bleeding), disease related (eg, severity of conditions or level of disability), social/economic (eg, medication cost), and system related (eg, lack of time or resources to provide education about diagnosis and treatment and lack of follow-up).^48,49  Factors predisposing to nonadherence are more common in older patients.⁵⁰  Our patient has several factors that may have contributed to nonadherence, including cognitive impairment, multiple comorbidities, and social isolation.

Adherence is an important determinant of the effectiveness of anticoagulant therapy and failure to adhere to the prescribed regime increases the likelihood of therapeutic failure. Nonadherence to warfarin is associated with poor INR control and an increased risk of recurrent VTE. In a study of 136 patients with VTE treated with warfarin, 36% of patients missed >20% of bottle openings, and these patients were twice as likely to have a subtherapeutic INR as other patients.⁵¹  In a medical claims database study of 8040 patients with an unprovoked VTE, approximately one-third of patients were nonadherent, defined as a warfarin availability <80% of the time based on prescription refill data.⁵²  These patients had a 1.6-fold increase in the risk of recurrent VTE. Although adherence to NOACs is generally better than adherence to VKA therapy, not all reports are consistent, and superior adherence with VKA has been reported in several studies. Furthermore, when there is concern about adherence, some clinicians prefer using a VKA, because it provides a means to monitor the quality of anticoagulation by measuring the INR.⁵³ 

Adherence in older patients may be enhanced by ensuring that they understand the risks and potential consequences of recurrent thrombosis and the benefits and risks of anticoagulant therapy.⁵⁴  Adherence is likely to be better with the provision of clear dosing instruction, use of a pill organizer, and electronic reminders. Pharmacists play a critically important role by providing education, reviewing dosing instructions, and monitoring prescription refills.⁵⁴  After addressing the aforementioned issues, we restarted rivaroxaban with close family and pharmacy oversight. If such oversight is not possible, then some clinicians may choose to use a VKA instead of a NOAC, because this allows for laboratory monitoring of the quality of anticoagulation. In this patient, we organized an earlier clinic review at 6 weeks to reinforce adherence and check for complications and then at the completion of treatment (3 months) to assess the need for secondary VTE prevention.

An 84-year-old male requires anticoagulation for stroke prevention in the setting AF. Comorbidities include 2 previous ischemic strokes, coronary artery disease and New York Heart Association class II heart failure, CKD, hypertension, and diabetes. Two years ago, he was switched from warfarin to dabigatran 150 mg twice daily because of poor INR control. One week ago, he presented to the emergency department with urinary retention. His estimated CrCl was 29 mL/min. Anticoagulation was stopped for 3 days prior to insertion of a suprapubic catheter.

Renal function declines with age, and one-third of older patients have CKD (estimated glomerular filtration rate <60 mL/min per 1.73 m²).⁵⁵  Over 50% of AF patients aged ≥75 have CKD and, compared with those with normal renal function, have a 1.5- to 1.8-fold higher risk of stroke and a 1.3- to 1.6-fold higher risk of major bleeding when treated with warfarin.⁵⁶ 

The evidence for anticoagulation in patients with AF who have severe CKD (estimated glomerular filtration rate <30 mL/min per 1.73 m²) comes from observational studies.⁵⁷  In a meta-analysis of 11 cohort studies, which included >48 500 AF patients with severe CKD, warfarin compared with no anticoagulation was associated with a lower risk of ischemic stroke/thromboembolism (HR, 0.70; 95% CI, 0.54-0.89) and mortality (HR, 0.65; 95% CI, 0.59- 0.72), with no increase in major bleeding (HR, 1.15; 95% CI: 0.88−1.44) in the subset of patients not requiring dialysis, but increased the risk of major bleeding (HR, 1.30; 95% CI, 1.08-1.56) and did not reduce stroke/systemic embolic events (HR, 1.12; 95% CI, 0.69-1.82) or mortality (HR, 0.96; 95% CI, 0.81-1.13) in those requiring dialysis.

Randomized trials of NOACs in patients with AF excluded those with estimated CrCl <30 mL/min (dabigatran, edoxaban, and rivaroxaban) or with creatinine level >221 µmol/L (apixaban). Despite the almost complete lack of evidence from clinical outcome studies, all 4 NOACs are licensed in United States and Europe in patients with a CrCl of 15 to 30 mL/min based on data from pharmacokinetic studies. Furthermore, the United States Food and Drug Administration approved apixaban for use in patients receiving hemodialysis. By contrast, most treatment guidelines recommend caution when using NOACs in patients with a CrCl between 15 and 30 mL/min and suggest that their use should be avoided in patients with a CrCl <15 mL/min.⁵⁷ 

Our patient was receiving dabigatran at a dose of 150 mg twice daily despite having a CrCl of 29 mL/min. In patients with this degree of renal impairment, we prefer using apixaban, edoxaban, or rivaroxaban, because they are less dependent on renal function for clearance. In the United States, dabigatran 75 mg twice daily is approved for patients with a CrCl of 15 to 30 mL/min (Table 1). Many clinicians still prefer using warfarin if the CrCl is <30 mL/min, because it is nonrenally cleared.⁵⁸ 

Appropriate periprocedural management of anticoagulation is important to minimize the risk of bleeding and thrombotic complications. The timing and duration of treatment interruption is influenced by the half-life of the anticoagulant and the bleeding risk of the procedure.⁵⁹  We recommend interrupting NOACs for 2 to 3 half-lives prior to procedures associated with a low or moderate bleeding risk and for 4 to 5 half-lives prior to procedures associated with a high bleeding risk. To account for the prolonged half-life of dabigatran (∼27 hours) in the presence of CrCl <30 mL/min, we recommend stopping treatment of 3 days prior to the elective suprapubic catheter insertion. We resume anticoagulant therapy the day after the procedure. Our approach is supported by data from the PAUSE study, which showed low 30-day postoperative major bleeding and thrombotic complication rates of 0.90% (95% CI, 0% to 1.73%) and 0.6% (0 to 1.33%), respectively, if anticoagulant therapy is resumed 1 to 2 days after low-bleeding-risk procedures.⁶⁰ 

A 75-year-old female is admitted with hematemesis. She has AF and is taking apixaban 5 mg twice daily for stroke prevention. Her last dose was 6 hours earlier. Her other medications include aspirin, atorvastatin, and hydrochlorothiazide. On presentation she is hemodynamically unstable, with a pulse rate of 110 and blood pressure of 90/40 mm Hg and receives fluid resuscitation. Her hemoglobin is 76 g/L, and she has acute kidney injury with a creatinine of 169 μmol/L. Her medications were stopped. The patient was given 2000 U of a 4-factor prothrombin complex concentrate and underwent endoscopy of the upper GI tract.

Whereas NOACs have a lower risk of ICH and fatal bleeding than VKAs, 3 NOAC regimens (dabigatran 150 mg twice daily, edoxaban, and rivaroxaban) have been associated with an increased risk of GI bleeding compared with warfarin.⁶¹  The increase in GI bleeding is greatest in older patients and in those with GI tract abnormalities or other risk factors for NOAC overexposure (eg, severe CKD). Older patients have higher risk of morbidity and mortality following a major GI bleed.

Interruption of antithrombotic therapy in patients with major bleeding is associated with a fourfold increase in ischemic events.⁶²  Low-quality evidence from observational studies suggests that early resumption of anticoagulation may protect against the risk of thrombotic events after major GI bleeding.^63-65  In our patient, the timing of anticoagulant resumption depends on the cause of GI bleeding and the risk of rebleeding. If endoscopic features suggest a low risk of rebleeding (eg, small gastric ulcer with no active bleeding), then anticoagulation could be restarted within 24 to 48 hours. For lesions at higher risk of bleeding (eg, variceal bleed and complex or large ulcers), a longer interruption (1-2 weeks) may be necessary, and a second look endoscopy may assist in the decision to restart.

In our patient, endoscopy found a large bleeding ulcer, which was injected. With a high rebleeding risk, we delayed resumption of full-dose anticoagulation for 1 week and after a second-look endoscopy. We prescribed long-term therapy with a proton pump inhibitor and discontinued aspirin.

Although age alone is not a contraindication for anticoagulant therapy, underuse and underdosing of anticoagulants are common in older patients. We believe that the benefits of treatment outweigh the risks in most older patients with AF or at risk of VTE unless there is a history of recurrent major bleeding or an unacceptable risk of ICH (eg, amyloid angiopathy). An exception may be older patients with end-stage CKD in whom evidence of benefit is lacking. We encourage close clinical follow-up of all patients receiving anticoagulant therapy to maximize adherence and the benefit of anticoagulant therapy.

N.C.C. holds a McMaster University, Department of Medicine Internal Research Career Award.

Contribution: J.W.E. and N.C.C. developed the first draft and wrote the manuscript.

Conflict-of-interest disclosure: J.W.E. reports honoraria and grant support from Astra-Zeneca, Bayer Boehringer-Ingelheim, Bristol-Myer-Squibb, Daiichi-Sankyo, Eli-Lilly, Glaxo-Smith-Kline, Pfizer, Janssen, and Sanofi-Aventis outside of the submitted work. N.C.C. reports honorarium for a speaker fee from Bayer outside of the submitted work.

Correspondence: Noel C. Chan, Thrombosis & Atherosclerosis Research Institute, 237 Barton St E, Hamilton, ON L8L 2X2, Canada; e-mail: noel.chan@taari.ca.