Jessica Chasler, PharmD

: Our primary objective was to determine the prevalence of novel anticoagulant use in patients with atrial fibrillation (AF) and venous thromboembolic disease (VTE) in a large multispecialty ambulatory practice.

: Our study included 5632 patients who had at least 1 outpatient provider visit and were actively being treated with an anticoagulant from April 2012 through March 2013. The primary outcome was the difference in proportion of patients receiving warfarin compared with a novel agent as of March 2013.

: Warfarin was an active medication for 83% of patients, while a novel agent was active for 17% of patients. Use of novel anticoagulants was higher among patients with AF compared with those with other indications for anticoagulation (24% of patients with AF were prescribed a novel, 3% of those with VTE, 5% with both conditions, 8% with other indications; 

 <.001). When adjusting for potential confounders, predictors of novel anticoagulant use among those with AF included younger age (odds ratio [OR]: 0.97 per 1-year increase in age; 95% CI, 0.96-0.97), higher risk of stroke (OR: 1.74; 95% CI, 1.27-2.37), and being seen in counties with lower poverty rates (OR: 0.94; 95% CI, 0.92-0.96).

: Despite the availability of novel agents, warfarin remained the dominant anticoagulant prescribed for patients with AF and VTE in a large ambulatory practice. Further work is needed to examine the appropriateness of novel anticoagulant use and methods for optimizing the risks, benefits, and accessibility to therapy with regard to the prevention and treatment of thrombosis.

For more than 50 years, warfarin was the sole oral anticoagulant in the United States. Although it is an effective agent for the prevention and treatment of thromboembolism, it interacts with many drugs and foods and requires close monitoring to prevent complications. Novel anticoagulants, including both direct thrombin inhibitors and Factor Xa inhibitors, entered the prescription drug market in 2010 as alternative agents to warfarin for the treatment and prevention of thrombosis in patients with nonvalvular atrial fibrillation (AF). These agents have a rapid onset of action, are prescribed in a fixed dosing regimen, do not require frequent monitoring, and have fewer drug-drug and drug-food interactions compared with warfarin.

 In clinical trials, they have equal or superior efficacy compared with warfarin while maintaining similar or lower rates of bleeding.

Despite these benefits, these agents are not without risks and come at a considerably greater drug acquisition cost. Their precise role in the pharmacotherapy of thromboembolism has yet to be well defined.

 Recently, several concerns about these agents have emerged, including warnings against use for patients with mechanical valves, and the potential utility of monitoring drug levels to improve the safety of these agents has been discussed.

While much is known about the benefits and risks of these novel anticoagulants, less information is available regarding how frequently these agents are prescribed in clinical practice. One investigation using IMS Health’s National Disease and Therapeutic Index data showed an increase in the use of dabigatran for AF from 4% to 17% from the last quarter of 2010 to the end of 2011 in the United States.

 Another study from a similar time period found that 12% of patients with AF were treated with dabigatran.

We sought to add to the literature by reporting on anticoagulant use for all indications in patients seen within a large ambulatory practice in Maryland. We hypothesized that warfarin would continue to be the most prevalent oral anticoagulant prescribed, as opposed to newer agents, for patients receiving care at this practice site. We performed specific analysis on a subset of patients with nonvalvular AF as a population with a consistent indication for novel agents over the study period.

We included patients from a large, academically affiliated multispecialty ambulatory practice (MAP) in the study. At the time of this study, the MAP cared for 250,000 patients in 35 primary care practices (staffed by approximately 180 primary care providers) and 2 cardiology practices (staffed by 12 cardiologists). We were able to extract data from 24 primary care and both cardiology sites. These sites served diverse patient populations including residents of inner cities, rural counties, and the affluent suburbs of Washington, DC; they are located across the District of Columbia and 13 Maryland counties. Twelve of the primary care practices and both cardiology practices had established anticoagulation clinics, and all practices used the same algorithms and integrated electronic medical record (EMR).

We included patients who had at least 1 office visit, either a provider visit or a provider support visit (ie, nurse-managed anticoagulation clinic visit), to practice sites within the MAP between April 1, 2012, and April 3, 2013, and had warfarin, dabigatran, or rivaroxaban on their active medication list through the outpatient EMR. We excluded patients seen at specialty sites for obstetrics and gynecology and for surgical care, because it is unlikely that these providers were managing patients’ anticoagulation therapy. Data abstracted from the EMR included the last active oral anticoagulant on the medication list during the study period; date of birth; sex; practice site; and 

International Classification of Diseases, Ninth Revision, Clinical Modification 

 codes were grouped into categories to identify major indications (AF, thromboembolic disease, antiphospholipid syndrome, or a valvular heart condition) as well as medical conditions used for the calculation of the risk-scale score known as CHADS2 (an acronym for Congestive heart failure, Hypertension, Age ≥75 years, Diabetes mellitus, Stroke).

From the EMR, we extracted the total numbers of office visits, provider support visits, and whole-blood and point-of-care international normalized ratio (INR) values by the patient during the study time frame, and if the patient was also taking aspirin.

In order to better characterize changes in anticoagulant prescribing over time, we created a comparison group by collecting similar information from the year prior to our study period, including the last active anticoagulant agent on the medication list during that year. A patient was considered to have switched anticoagulant therapy if the active drug from the previous year differed from the active drug during the study period. We grouped practice sites by their county of location, and assigned poverty rates based on US Census Bureau data for Maryland’s counties from 2009-2011.

 We also classified counties as urban or rural based on US Census Bureau data. The study was approved by the Johns Hopkins Institutional Review Board.

We used descriptive statistics to determine baseline patient characteristics and proportions throughout the analysis. We first used 

 tests and Wilcoxon rank sum tests to examine the bivariate association among patient (age, sex, CHADS2 score, heart failure, hypertension, diabetes, stroke history, aspirin use, number of INR values), provider (number of provider office visits), and practice (urban vs rural setting, poverty rate in the county, cardiology vs general medicine site) characteristics, and our outcome of interest. Next, we performed bivariate and multivariate analyses to examine predictors of novel oral anticoagulant use, focusing solely on individuals with AF only, given that this represented 61% of all study subjects and that all of the agents examined were FDA approved for this indication during the study period. Model fit was determined by stepwise model selection. The 2-sided a priori level of significance was 0.05 for all analyses. Analyses were conducted using STATA 13 (StataCorp LP, College Station, TX).

A total of 5632 patients were included in the study. The mean age of eligible patients was 70 years (SD 15), 56% were male, 61% had AF only, 16% had venous thromboembolic disease (VTE) only, 4% had both AF and VTE, and 19% had other indications for anticoagulation (Table 1). Warfarin was an active medication for 4675 (83%) patients, while a novel agent was active for 957 (17%) patients. Of the 3424 patients with only AF, 76% were prescribed warfarin, compared with 24% prescribed a novel anticoagulant. In 881 patients with only VTE, 3% were treated with novel anticoagulants, while 12 (1%) patients with both AF and VTE were treated with these newer agents.

Bivariate Predictors of Novel Oral Anticoagulant Use in Patients with Only Atrial Fibrillation

For those with AF, there were statistically significant differences in all baseline characteristics except for sex, history of stroke, and aspirin use between those receiving warfarin and those receiving novel agents (Table 2). The majority of patients had a CHADS2 score of 2 or 3 (63% for warfarin and 61% for novels). Those with the lowest stroke risk (CHADS2 score of 0 or 1) were more likely to receive a novel agent (22% for novels compared with 18% for warfarin), while those at highest risk (CHADS2 of 4-6) were more likely to receive warfarin (20% for warfarin vs 17% for novels; overall 

 value .02). The 2 cardiology practices accounted for 45% of patients on a novel anticoagulant, while the other patients on novel agents were distributed across the 24 general medicine practices. Those receiving a novel agent tended to reside in a county with a lower poverty rate (mean rate 8.2%; SD 3.9), compared with those receiving warfarin (mean rate 9.8%; SD 6.0).

Multivariate Predictors of Novel Oral Anticoagulant Use in Patients with Atrial Fibrillation

Based on the multivariate model (Table 3), the odds of being prescribed a novel oral anticoagulant decreased by 3% for each additional year in age (odds ratio [OR]: 0.97; 95% CI, 0.96-0.97) when controlling for confounders. Those carrying a diagnosis of heart failure had 46% lower odds of being prescribed a novel agent (OR: 0.54; 95% CI, 0.45-0.64). The odds of being prescribed a novel anticoagulant increased 44% (OR: 1.44; 95% CI, 1.14-1.82) for those with a moderate-risk CHADS2 score compared with a low-risk score, while those with a high-risk score had 74% higher odds (OR: 1.74; 95% CI, 1.27-2.37) of receiving a novel agent. The odds for novel agent use decreased by 6% (OR: 0.94; 95% CI, 0.92-0.96) for every 1% increase in the poverty rate by county of practice site.

Switching Among Novel Oral Anticoagulants and Warfarin in the Overall Population

There were 3906 patients within the ambulatory practice actively receiving an anticoagulant during the April 2012-to-March 2013 time frame who had also received an anticoagulant in the previous year. In this subset of patients, there was no significant change in the overall numbers being treated with warfarin versus a novel anticoagulant, with therapeutic changes occurring in relatively few patients. Between study periods, 11 (0.3%) patients were switched from warfarin to dabigatran, 17 (0.4%) from warfarin to rivaroxaban, 23 (0.6%) from dabigatran to warfarin, and 16 (0.4%) from dabigatran to rivaroxaban. More than 90% of patients who switched from warfarin to dabigatran, dabigatran to warfarin, or dabigatran to rivaroxaban had only AF as an indication for anticoagulation, whereas patients switching from warfarin to rivaroxaban had a higher incidence of VTE (29%). In those with AF who did not switch therapy, 82% were prescribed warfarin while 18% were prescribed a novel anticoagulant.

All Patients Initiating Anticoagulation Therapy in 2013

There were 1726 patients in the overall study population who did not have an anticoagulant on their medication list in the previous year and appeared to be newly started on anticoagulation. Of these patients, 1284 (74%) were prescribed warfarin, 240 (14%) dabigatran, and 202 (12%) rivaroxaban. Compared with the overall study population, novel anticoagulant use increased by 9% in 2013, leading to a reduction in warfarin prescribing (

 <.001). Among these patients newly started on novel agents, 44% were managed by cardiologists and 56% by primary care providers (

When comparing across indications for all new starts, 80% of novel prescribing was for patients with only AF, 4% for patients with VTE, 1% for patients with both conditions, and 15% for patients with indications other than AF or VTE (

 <.001). Dabigatran was initiated more frequently than rivaroxaban in those with only AF (22% vs 16%, respectively) while rivaroxaban was more commonly initiated compared with dabigatran in those with a VTE indication (5% vs 1%, respectively). Warfarin continued to be initiated in 85% of patients with indications for anticoagulation other than AF and VTE; 8% of these patients were prescribed rivaroxaban, while 7% were prescribed dabigatran. Primary care providers initiated novel anticoagulants for 36% of their new patients with only AF, while cardiologists initiated these agents in 41% of similar new patients (

This retrospective cross-sectional study reviewed oral anticoagulant use among 5647 patients who were seen at a large multispecialty practice in Maryland from April 2012 through March 2013, and it helps to shed light on the uptake of these agents in primary and specialty care settings. We found that for patients with AF, 23% were treated with novel anticoagulants, compared with 6% for patients with indications other than AF. These rates may be a result of the timing of use approval for these medications in comparison with our study period. Dabigatran was approved for use in nonvalvular AF in 2010 and in VTE in 2014, while rivaroxaban was approved for use in nonvalvular AF in 2011 and in VTE in 2012. While lower rates of use in VTE are therefore not surprising, limited data exist on the prevalence of novel anticoagulant use for this indication in the ambulatory setting.

After adjustment for potentially confounding covariates, novel anticoagulant use for AF was greater among younger individuals and those seen in counties with lower poverty rates, while lower prescribing of novel agents was seen in patients with heart failure and diabetes. Comparable results were reported by Desai and colleagues in a commercial population, which showed that patients prescribed novel agents tended to be male and younger, to have fewer comorbidities, and to live in areas with higher incomes.

Unlike the Desai study, our report is based on a more diverse population, as our statewide cohort included the uninsured as well as patients with Medicare, Medicaid, and commercial insurance. The Desai study reported an increase in initiation of novels over time in patients with AF. While it was difficult for us to assess trends in use of these agents over time due to the cross-sectional nature of our study, we did see significantly more patients with new anticoagulation initiated on novel agents if they had AF, versus other indications, in 2013.

By utilizing data from the year prior to our study period, we were able to capture limited information on patients who switched therapy. The majority of switching between warfarin and dabigatran, or dabigatran and rivaroxaban, occurred in patients with AF, while we observed more patients with VTE switching from warfarin to rivaroxaban. Overall, few patients switched therapy from one year to the next. This is not necessarily unexpected; studies have shown a mitigated reduction in stroke risk, as well as negligible cost-benefit ratio, when patients with a stable INR are compared with patients treated with dabigatran.

 Additionally, patients on warfarin in our study population are generally followed in clinics staffed by nurses, who are unable to prescribe medications and therefore unable to independently initiate a switch in therapy. Of those patients who did switch therapy, we did not observe a consistent trend toward use of novel agents. Changes in therapy from dabigatran may reflect changing indications during this time, as well as availability of newer agents with daily, rather than twice daily, dosing, which facilitates adherence to therapy.

We noted a trend toward increasing use of novel agents in a subset of 1726 patients who had anticoagulants newly added to their medication lists during the study period. In this subset of patients, warfarin remained the most commonly prescribed anticoagulant agent; however, in those patients with AF, novel anticoagulant use was significantly higher compared with those with other indications for anticoagulant therapy. There was a near doubling of the use of novel agents in newly initiated patients with AF, compared with those patients who continued therapy from the previous year.

This trend indicates an uptake in prescribing of these agents for newly diagnosed AF patients and a continued use of warfarin in those patients who have been on therapy over time, which may be expected. With 2 newly FDA-approved novel agents on the market since our study period, we would expect this trend to continue to increase over time for patients with both AF and VTE, as each of these agents offers recommendations for use in special populations.

Our study has several limitations. First, many baseline characteristics were statistically different between patients on novel anticoagulants versus those on warfarin, and this may reflect a selection bias in our study population. We were limited in the information we were able to abstract from the EMR, and we lacked the cross-section of pharmacy claims available in other studies that used administrative data; however, the EMR also presented an advantage in that clinician-verified data were used in the analysis. Characteristics that were not measured in our study population that may have also influenced choice of therapy include a history of falls, history of major bleeding, renal or hepatic dysfunction, patient preference, or provider comfort with prescribing these different anticoagulant agents.

We also were not able to collect prescription insurance information: Higher medication copays, lack of insurance, or a requirement for prior authorization may have deterred the use of newer agents for some patients, which was not assessed in our study population. Finally, the number of available agents and indications changed during the study period, making the analysis of clinically appropriate therapies more complex. Rivaroxaban gained approval for VTE during our study time period while the other novel agents have been approved for this indication since conclusion of our study. Therefore, use of these agents for VTE was not expected to be high and our data for this indication should be viewed as limited.

A significant proportion of patients seen at a large ambulatory practice in Maryland and the District of Columbia were prescribed warfarin rather than a novel anticoagulant for the prevention and treatment of several thromboembolic conditions. Novel anticoagulants were used more commonly in patients with atrial fibrillation compared with those with other indications for anticoagulation, although less than a quarter of patients in the overall study population with AF received novel therapies. Our study provides further insight into how these agents have been adopted into practice; further studies should be performed to assess the appropriateness of anticoagulant choice, especially in female, urban, and poor populations in light of available data related to the efficacy and safety of newer agents.

 Department of Pharmacy, New York-Presbyterian Hospital (LEE); Department of Pharmacy, Johns Hopkins Hospital (JC), Baltimore, MD; Division of General Internal Medicine, Johns Hopkins University School of Medicine (GCA, MM), Baltimore, MD.

 The authors report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

 Concept and design (LE, JC, MM, CA); acquisition of data (MM, JC); analysis and interpretation of data (JC, MM, CA, LE); drafting of the manuscript (LE, JC, CA, MM); critical revision of the manuscript for important intellectual content (LE, JC, MM, CA); statistical analysis (LE, JC); provision of patients or study materials (JC, MM)

Leigh E. Efird, PharmD, Department of Pharmacy, New York-Presbyterian Hospital, 525 East 68

 St, New York, NY 10065. E-mail: lee9015@nyp.org.

Cove CL, Hylek EM. An updated review of target-specified oral anticoagulants used in stroke prevention in atrial fibrillation, venous thromboembolic disease, and acute coronary syndromes. 

. 2013;2(5);e000136. doi: 10.1161/JAHA.113.000136.

Connolly SJ, Ezekowitz MD, Yusuf S, et al; RE-LY Steering Committee and Investigators. Dabigatran versus warfarin in patients with atrial fibrillation.

 2009;361(12):1139-1151. doi: 10.1056/NEJMoa0905561.

Patel MR, Mahaffey KW, Garg J, et al; ROCKET AF Investigators. Rivaroxaban versus warfarin in nonvalvular atrial fibrillation. 

 2011;365(10):883-891. doi: 10.1056/NEJMoa1009638. 

Granger CB, Alexander JH, McMurray JJ, et al; ARISTOTLE Committees and Investigators. Apixaban versus warfarin in patients with atrial fibrillation. 

 2011;365(11):981-992. doi: 10.1056/NEJMoa1107039.

Zeidan A, Faltas B, Streiff M. Dabigatran etexilate: what do hospitalists need to know?

 2012;7(3):262-269. doi: 10.1002/jhm.996.

Reilly PA, Lehr T, Haertter S, et al; RE-LY Investigators. The effect of dabigatran plasma concentrations and patient characteristics on the frequency of ischemic stroke and major bleeding in atrial fibrillation patients: the RE-LY Trial (Randomized Evaluation of Long-Term Anticoagulation Therapy). 

 2014;63(4):321-328. doi: 10.1016/j.jacc.2013.07.104.

Cohen D. Dabigatran: how the drug company withheld important analyses. 

Van de Werf F, Brueckmann M, Connolly SJ, et al. A comparison of dabigatran etexilate with warfarin in patients with mechanical heart valves: the Randomized, Phase II Study to Evaluate the Safety and Pharmacokinetics of Oral Dabigatran Etexilate in Patients After Heart Valve Replacement (RE-ALIGN). 

 2012;163(6):931-937.e1. doi: 10.1016/j.ahj.2012.03.011.

Kirley D, Qato DM, Kornfield R, Stafford RS, Alexander GC. National trends in oral anticoagulant use in the United States, 2007 to 2011. 

Steinberg BA, Holmes DN, Piccini JP, et al; Outcomes Registry for Better Informed Treatment of Atrial Fibrillation (ORBIT-AF) Investigators and Patients. Early adoption of dabigatran and its dosing in US patients with atrial fibrillation: results of the Outcomes Registry for Better Informed Treatment of Atrial Fibrillation. 

 2013;2(6):e000535. doi: 10.1161/JAHA.113.000535.

Shimabukuro TT, Kramer J, McGuire M. Development and implementation of a nurse-managed anticoagulation program. 

Gage BF, Waterman AD, Shannon W, Boechler M, Rich MW, Radford MJ. Validation of clinical classification schemes for predicting stroke: results from the National Registry of Atrial Fibrillation. 

Maryland’s Money Matters. Maryland Center of Economic Policy. State of Working Maryland: Poverty in Maryland. http://www.mdeconomy.org/state-of-working-maryland-poverty-in-maryland/. Accessibility verified May 13, 2014.

Desai NR, Krumme AA, Schneeweiss S, et al. Patterns of initiation of oral anticoagulants in patients with atrial fibrillation—quality and cost implications. 

2014;127(11):1075-1082.e1. doi: 10.1016/j.amjmed.2014.05.013. 

Coyle D, Coyle K, Cameron C, et al. Cost-effectiveness of new oral anticoagulants compared with warfarin in preventing stroke and other cardiovascular events in patients with atrial fibrillation. 

 2013;16(4):498-506. doi: 10.1016/j.jval.2013.01.009.

Articles

While warfarin remains the most commonly prescribed anticoagulant, novel anticoagulant use appears to be increasing especially among cardiologists. With new indications for these agents, uptake in prescribing by primary care providers is likely to increase.