Efficacy and Safety of Direct Oral Anticoagulants for Atrial Fibrillation Across Body Mass Index Categories.

Background Direct-acting oral anticoagulants are now the preferred method of anticoagulation in patients with atrial fibrillation. Limited data on efficacy and safety of these fixed-dose regimens are available in severe obesity where drug pharmacokinetics and pharmacodynamics may be altered. The objectives of this study were to evaluate efficacy and safety in patients with atrial fibrillation taking direct-acting oral anticoagulants across body mass index (BMI) categories in a contemporary, real-world population. Methods and Results We performed a retrospective study of patients with atrial fibrillation at an integrated multisite healthcare system. Patients receiving a direct-acting oral anticoagulant prescription and ≥12 months of follow-up between 2010 and 2017 were included. The primary efficacy and safety outcomes were ischemic stroke or systemic embolism and intracranial hemorrhage. We performed Cox proportional hazards modeling to compute hazard ratios (HRs) adjusted for CHA2DS2-VASc score to examine differences by excess BMI categories relative to normal BMI. Of 7642 patients, mean±SD age was 69±12 years with a median (interquartile range) follow-up of 3.8 (2.2-6.0) years. Approximately 22% had class 1 obesity and 19% had class 2 or 3 obesity. Stroke risks were similar in patients with and without obesity (HR, 1.2; 95% CI, 0.5-2.9; and HR, 0.68; 95% CI, 0.23-2.0 for class 1 and class 2 or 3 obesity compared with normal BMI, respectively). Risk of intracranial hemorrhage was also similar in class 1 and class 2 or 3 obesity compared with normal BMI (HR, 0.64; 95% CI, 0.35-1.2; and HR, 0.66; 95% CI, 0.35-1.2, respectively). Conclusions Direct-acting oral anticoagulants demonstrated similar efficacy and safety across all BMI categories, even at high weight values.

direct‐acting oral anticoagulant

Clinical Perspective

What Is New?

This retrospective study aimed to examine the safety and efficacy of direct‐acting oral anticoagulants in patients with atrial fibrillation.

In 7642 patients, we found similar risks of stroke and intracranial hemorrhage in patients with atrial fibrillation on direct‐acting oral anticoagulants across body mass index categories, even at high weight values.

What Are the Clinical Implications?

Direct‐acting oral anticoagulants may be safe for stroke prevention in patients with atrial fibrillation and obesity.

Stroke prevention is a cornerstone in the management of atrial fibrillation (AF). Direct‐acting oral anticoagulants (DOACs) have transformed the landscape of anticoagulation therapies, as they are given in fixed doses, have minimal food and drug interactions, have more rapid onset/offset, and there is no requirement for invasive monitoring. As such, they have become the preferred method of stroke prevention for most patients, but certain subgroups, such as those with obesity, have been understudied and concerns remain about the efficacy and safety of DOACs in these populations.

Obesity is a well‐established, independent risk factor for AF and stroke, but limited numbers of patients with obesity were included in the phase III clinical trials that demonstrated safety and efficacy outcomes of DOACs compared with warfarin. , , , In particular, patients in the higher weight category (class 2 or 3 obesity) comprised only 13% to 16% of participants in the major clinical trials of DOACs but are estimated to represent nearly a quarter of the general US population. Obesity is likely even more prevalent among those with AF. , , , There are currently no dose adjustment recommendations based on patient weight, but obesity may result in pharmacokinetic and pharmacodynamics changes. While the 2019 Focused Update of the 2014 American College of Cardiology/American Heart Association/Heart Rhythm Society Guideline for the Management of Atrial Fibrillation does not discuss caution of DOAC use in patients with obesity, the International Society of Thrombosis and Haemostasis recommend against their use in patients with severe obesity or advise consideration of the use of drug‐specific serum level monitoring if a DOAC is used in a patient with class 3 obesity or weight >120 kg. ,

As prevalence of both obesity and AF increase, the issue of managing anticoagulation in patients with AF and elevated body weight has become even more essential to address in routine clinical practice. The purpose of this study was to evaluate efficacy and safety outcomes of real‐world usage of DOACs for stroke prevention in AF across body mass index (BMI) categories. The primary efficacy outcome was the occurrence of ischemic stroke or systemic embolism. The primary safety outcome was the occurrence of intracranial hemorrhage.

METHODS

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Study Population

We used data from the Northwestern Medicine Enterprise Data Warehouse, which houses comprehensive demographic, laboratory, and prescription data on patients as well as claims for inpatient or outpatient diagnoses and procedures. , The Northwestern Medicine Enterprise Data Warehouse represents a single integrated healthcare system that includes a large academic medical center and multiple community hospitals in the same region in Illinois.

We performed a retrospective analysis using electronic health record data to identify patients with AF between January 1, 2010, and December 31, 2017 (date of first diagnosis was defined as index date). Other inclusion criteria included being age 18 years or older and evidence of healthcare system use for at least 12 months of follow‐up, defined as an in‐person encounter in internal medicine, cardiology, or neurology after the index date. End of follow‐up was defined by the last in‐person encounter in internal medicine, cardiology, or neurology, event of interest for the primary or secondary analyses, or death. Exclusion criteria included patients with a diagnosis of prevalent stroke and venous thromboembolism events on the index date and patients prescribed warfarin or not prescribed any anticoagulation medication. Patients with missing or extreme values for systolic blood pressure (<80 or >200 mm Hg) and for BMI (<18.5 or >70 kg/m2) were excluded. This study was approved by the Institutional Review Board at Northwestern University who determined that informed consent was not required.

Derivation of Clinical Data

Baseline clinical history and CHA2DS2‐VASc score was determined using multiple areas in the medical record (encounter diagnosis code, problem list, medical history, and billing), and International Classification of Diseases, Ninth Revision or Tenth Revision (ICD‐9 orICD‐10) codes used are outlined in Table S1. Race, ethnicity, and smoking habits were self‐reported. Race was dichotomized as White and non‐White. Ethnicity was also dichotomized as Hispanic/Latinx and non‐Hispanic/Latinx. Smoking status was presented as never‐smoker, former smoker, current every‐day smoker, and never assessed. BMI and systolic blood pressure were abstracted at the index date. Anticoagulation status was determined by the presence of a prescription for a DOAC (apixaban, dabigatran, edoxaban, rivaroxaban) at the index visit. We categorized patients using the definitions from the Centers for Disease Control and Prevention as normal (BMI, 18.5–24.9 kg/m2), overweight (BMI, 25.0–29.9 kg/m2), class 1 obesity (BMI, 30.0–34.9 kg/m2), class 2 obesity (BMI, 35.0–39.9 kg/m2), and class 3 or severe obesity (BMI, ≥40.0 kg/m2).

Primary and Secondary Outcomes

The primary outcome was defined as the occurrence of stroke or systemic embolism based on the following factors:

Patients had to be admitted with ischemic stroke as the primary diagnosis after the index date (ICD‐9/ICD‐10 codes 433.X, 434.X, 436.X, I63.X, I65.X, I66.X).

Additional criteria had to be met during the admission period, including:

Patient underwent mechanical thrombectomy or treatment with tissue plasminogen activator OR

Patient had a computed tomography/magnetic resonance imaging; carotid ultrasound; electroencephalogram; coagulation test; carotid angiography; or administration of heparin, antiplatelet, anticoagulation, or thrombin inhibitor during admission; AND either blood glucose test, chest radiograph, ECG, Holter recording, echocardiogram, blood panel, an injection or infusion, or hospital critical care performed during admission. A list of the Current Procedural Terminology codes used to corroborate the stroke end point are provided below.

The additional criteria were also required to increase the specificity of identification of the outcome to ensure the stroke end point was a new event rather than documentation of a prior event (Table S2).

The occurrence of a systemic embolism was defined as a hospitalization or ambulatory encounter with primary diagnosis involving ICD‐9 or ICD‐10 codes of 444.X or I74.X. Intracranial hemorrhage was defined by a hospitalization encounter with the primary diagnosis involving ICD‐9 or ICD‐10 codes of I61.X. For patients with multiple events, only the first occurrence was included.

Statistical Analysis

For the primary analysis, patients with class 2 or 3 obesity were combined into 1 group given smaller numbers. CHA2DS2‐VASc scores were calculated on the basis of the patient's clinical history before the index date using the diagnosis codes.

Overall rates of incident stroke, systemic embolism, and intracranial hemorrhage events in each BMI category were estimated using the Kaplan–Meier curve, and statistical significance was evaluated by log‐rank test. Incidence rate for each cardiovascular event per 1000 person‐years was also assessed. The proportionality assumption was evaluated for each outcome by 3 different methods: the Schoenfeld residuals, log‐log plot, and Cox model with time‐varying covariates. If covariates had not satisfied the proportionality assumption by all 3 methods they would have been considered as time‐varying covariates, and the Cox model with time‐varying covariates would have been used. Given that all covariates included in the model did satisfy the assumption, the Cox proportional hazard model was used for computing hazard ratio (HR) with 95% CI. HRs for the associations between DOAC use and stroke, systolic embolism, and intracranial hemorrhage were adjusted for CHA2DS2‐VASc score.

A secondary analysis was done based on absolute weight class on the basis of recommendations to avoid DOACs in those with high absolute weight (defined as >120 kg). We also performed a sensitivity analysis using inverse probability weighting to account for potential residual confounding in our primary analysis. The probability of BMI ≥35 kg/m2 (propensity score) was computed using a multivariate logistic regression model including age, sex, race, ethnicity, and the CHA2DS2‐VASc score as independent variables. Inverse probability weighting assigns a weight of 1/propensity score for exposure (BMI, ≥35 kg/m2) and 1/(1−propensity score) compared with nonexposure (BMI <35 kg/m2). A standardized mean difference was used to assess the covariates balance before and after adjusting for inverse probability weighting. Inverse probability ‐weighted Kaplan–Meier curve and log‐rank test were performed to compare the cumulative incidence of stroke/systemic embolism and intracranial hemorrhage. Furthermore, inverse probability‐weighted Cox proportional hazard model was performed to compute inverse probability‐weighted HRs of obesity (BMI ≥35 kg/m2) for incident stroke or systemic embolism and intracranial hemorrhage.

Continuous variables were reported as mean and SD while discrete variables were reported as counts and percentages, as appropriate. Continuous variables were compared using a t test or one‐way analysis of variance for >2 means, and discrete variables were compared using a chi‐square test. Type 1 error rate for all tests was set at 0.05. All statistical analyses were performed using Stata statistical software (StataCorp, College Station, TX).

RESULTS

There were 7642 patients with AF identified who received a prescription for a DOAC and were included in this study (Figure 1). Mean±SD age was 69±12 years. The cohort was predominantly White (87.4%), and more than half of the patients were men (59.5%). Median follow‐up was 3.8 years (interquartile range, 2.2–6.0).

Figure 1

Development of the study cohort.

Overview of the study design with inclusion and exclusion criteria and the number of patients in each category. BMI indicates body mass index; and DOAC, direct acting oral anticoagulants.

Approximately 41% of patients had obesity (BMI ≥30 kg/m2). Baseline characteristics by BMI category are shown in Table 1. Patients with class 2 or 3 obesity (BMI ≥35 kg/m2) were younger than patients with normal BMI (64±11 versus 69±12 years; P<0.001). Patients with obesity also had higher rates of hypertension and diabetes mellitus, but lower rates of prior transient ischemic attack or stroke. Across all BMI categories, the most commonly prescribed DOACs were apixaban or rivaroxaban. Edoxaban was rarely prescribed (<1% of all anticoagulation use) and dabigatran accounted for 8% to 9% of the DOAC use in each BMI category.

Table 1

Baseline Characteristics of the Study Population by BMI Categories

	Overall N=7642	Normal Weight (18.5 ≤ BMI <25.0) N=1720	Overweight (25.0 ≤ BMI <30.0) N=2804	Class 1 Obesity (30.0 ≤ BMI <35.0) N=1697	Class 2+3 Obesity (35.0 ≤ BMI) N=1421
Age, y	69 (12)	72 (13)	69 (12)	67 (12)	64 (11)
Male, n (%)	4546 (59.5)	782 (45.5)	1870 (66.7)	1088 (64.1)	806 (56.7)
White, n (%)*	6678 (87.4)	1494 (86.9)	2464 (87.9)	1483 (87.4)	1237 (87.1)
Hispanic, n (%)	193 (2.5)	23 (1.3)	76 (2.7)	54 (3.2)	40 (2.8)
Non‐Hispanic Black, n (%)	356 (4.7)	58 (3.4)	97 (3.5)	95 (5.6)	106 (7.6)
BMI, kg/m2	28.7 (25.3–33.2)	23.0 (21.5–24.1)	27.5 (26.3–28.7)	32.1 (31.0–33.4)	39.0 (36.6–43.0)
Weight, kg	87.1 (73.5–102.5)	65.8 (58.4–74.8)	83.9 (74.8–91.8)	97.5 (87.3–106.8)	118.7 (105.1–132.4)
Weight >120 kg, n (%)	756 (9.9)	3 (0.2)	6 (0.2)	74 (4.4)	673 (47.4)
Systolic BP	128 (19)	126 (20)	127 (19)	129 (19)	131 (19)
CHA2DS2‐VASc Score	2.7 (2.0)	3.1 (2.0)	2.7 (2.0)	2.6 (2.0)	2.6 (1.9)
Heart failure, n (%)	959 (12.5)	213 (12.4)	304 (10.8)	205 (12.1)	237 (16.7)
Hypertension, n (%)	3826 (50.1)	770 (44.8)	1363 (48.6)	882 (52.0)	811 (57.1)
Diabetes mellitus, n (%)	1128 (14.8)	143 (8.3)	334 (11.9)	290 (17.1)	361 (25.4)
Stroke, n (%)	1289 (16.9)	391 (22.7)	473 (16.9)	253 (14.9)	172 (12.1)
TIA, n (%)	1185 (15.5)	357 (20.8)	436 (15.5)	239 (14.1)	153 (10.8)
VD, n (%)	603 (7.9)	170 (9.9)	228 (8.1)	124 (7.3)	81 (5.7)
DOAC, n (%)
Apixaban, n (%)	3356 (43.9)	810 (47.1)	1271 (45.3)	708 (41.7)	567 (39.9)
Rivaroxaban, n (%)	2785 (36.4)	563 (32.7)	1002 (35.7)	650 (38.3)	570 (40.1)
Edoxaban, n (%)	10 (0.1)	1 (0.1)	5 (0.2)	3 (0.2)	1 (0.1)
Dabigatran, (%)	1491 (19.5)	346 (20.1)	526 (18.8)	336 (19.8)	283 (19.9)

Continuous variables are summarized as mean (standard deviation) or median (interquartile range). Categorical variables are summarized as n (%). Student t test and chi‐square test were used for statistical significance. A P<0.05 was considered statistically significant. BMI indicates body mass index; BP, blood pressure; DOAC, direct oral anticoagulant; TIA, transit ischemic attack; and VD, vascular disease.

There are statistically significant differences among each BMI category in all variables other than White (P value was 0.76).

There were 45 occurrences of the primary efficacy outcome, stroke or systemic embolism. Higher BMI category was not associated with a significantly higher risk of stroke compared with the normal BMI group. HRs compared with the normal BMI group were 1.2 (95% CI, 0.58–2.7) for the overweight group, 1.2 (95% CI, 0.52–2.9) for class 1 obesity, and 0.68 (95% CI, 0.23–2.0) for class 2 or 3 obesity (Table 2). The incident rates per 1000 person‐years were similar in all BMI categories and are shown in Table S3. Stroke rates over the follow‐up period by BMI group are shown in Figure 2.

Table 2

HRs of Stroke/Systemic Embolism Events in Excess BMI Categories Compared With Normal BMI

Stroke/Systemic Embolism Events
BMI Subgroup	HR	95% CI	P Value
18.5 ≤ BMI <25.0	Section References	Section References	Section References
25.0 ≤ BMI <30.0	1.252	0.581–2.701	0.566
30.0 ≤ BMI <35.0	1.217	0.516–2.873	0.654
35.0 ≤ BMI	0.684	0.233–2.008	0.490

HRs with 95% CIs were adjusted for CHA2DS2‐VASc score. BMI indicates body mass index; DOAC, direct oral anticoagulants; and HR, hazard ratio.

Figure 2

Incident stroke/systemic embolism events by body mass index group over the follow‐up period.

Normal body mass index (group A, blue line), overweight (group B, red line), Class 1 obesity (group C, green line), Class 2 or 3 obesity (group D, orange line). BMI indicates body mass index.

For the primary safety outcome of intracranial hemorrhage, there were 92 events with no significant differences by BMI groups. HRs were 0.69 (95% CI, 0.42–1.1) for the overweight group, 0.64 (95% CI, 0.35–1.2) for class 1 obesity, and 0.66 (95% CI, 0.35–1.2) for class 2 or 3 obesity (Table 3) compared with the normal BMI group. The incident rates per 1000 person‐years are shown in Table S4. Incidence of intracranial hemorrhage events over the follow‐up period are shown in Figure 3.

Table 3

HRs of Intracranial Hemorrhage Events by Each Excess BMI Category Compared With Normal BMI

Intracranial Hemorrhage Events
BMI Subgroup	HR	95% CI	P Value
18.5 ≤ BMI <25.0	Section References	Section References	Section References
25.0 ≤ BMI <30.0	0.692	0.418–1.145	0.152
30.0 ≤ BMI <35.0	0.640	0.352–1.161	0.142
35.0 ≤ BMI	0.656	0.347–1.239	0.194

HRs with 95% CIs were adjusted for CHA2DS2‐VASc score. BMI indicates body mass index; DOAC, direct oral anticoagulants; and HR, hazard ratio.

Figure 3

Incident intracranial hemorrhage events by body mass index group over the follow‐up period.

Normal body mass index (group A, blue line), overweight (group B, red line), class 1 obesity (group C, green line), class 2 or 3 obesity (group D, orange line). BMI indicates body mass index; ICH, intracranial hemorrhage; and MB, major bleeding.

In the secondary analysis by absolute weight groups, those with weight >120 kg were younger (60±10 versus 69±12; P<0.001) and had a lower CHA2DS2‐VASc score (2.0±1.7 versus 2.8±2.0; P<0.001). There were fewer women in the higher weight group (20.5% versus 42.7%; P<0.001. All baseline characteristics by absolute weight (> or ≤120 kg) are shown in Table S5. Apixaban and rivaroxaban were the most common DOACs used in both the higher and lower weight groups. Higher weight was not associated with higher rate of stroke (HR, 1.06; 95% CI, 0.38–2.98) or intracranial hemorrhage (HR, 0.65; 95% CI, 0.26–1.62) (Tables S6 and S7). Incident rates of stroke and intracranial hemorrhage by absolute weight are shown in Figures S1 and S2, respectively. In the sensitivity analysis performed with inverse probability weighting, findings were similar (Table S8 and Figures S3, S4).

DISCUSSION

In this large, real‐world study examining patients with AF receiving a DOAC, there was no significant difference in the occurrence of stroke and systemic embolism events across BMI categories. These findings build upon the growing body of evidence supporting the safety and efficacy of DOACs in patients with severe obesity. As the era of generic versions of DOACs begins and the financial barrier is reduced for many patients who may benefit from them, these drugs will be even more commonly used than before. Based on our analysis, weight should not be a barrier to adoption and prescribing of DOACs for patients with AF.

Our findings and our estimated quantitative HRs are consistent with those from post hoc analyses of the randomized trials of the DOACs, which showed similar rates of stroke and systemic embolism in patients with and without obesity. Importantly, we now demonstrate these findings outside of a controlled trial setting. , Our findings using objectively measured height and weight to calculate BMI adds to the available administrative data examining rates of stroke and bleeding in patients with obesity on DOAC therapy. Specifically, one study from a large claims database found similar rates of stroke and systemic embolism (1.5% and 1.7%) for rivaroxaban and warfarin, respectively, in matched pairs of patients with severe obesity, but this analysis was limited by use of ICD‐9 and ICD‐10 coding alone for mobid obesity, which is quite limited and often poorly coded. ,

Monitoring drug levels has been proposed as a strategy to ensure that fixed‐dose anticoagulants reach appropriate therapeutic serum levels in patients with larger volumes of distribution. Such monitoring was performed in the ENGAGE‐TIMI 48 (Edoxaban Versus Warfarin in Patients With Atrial Fibrillation) trial. Trough edoxaban concentrations were checked in patients receiving the higher dose of edoxaban, and there was no significant difference across BMI categories. Similarly, anti‐Xa activity trough levels were monitored as well, and these also did not show a significant difference based on BMI. As serum levels are not routinely checked as part of standard medical practice, they were not available in our study. However, our study findings suggest that there may not be added clinical utility in monitoring trough levels on the basis of BMI, which can increase costs and complexity of care for health systems and patients.

We did note a lower rate of stroke and systemic embolism in patients in higher obesity classes. This seemingly paradoxical finding was noted in several prior studies regardless of specific anticoagulant choice. An analysis of ROCKET‐AF (Rivaroxaban Versus Warfarin in Nonvalvular Atrial Fibrillation) with 5206 patients with a BMI ≥30 kg/m2 showed a lower risk of stroke in these patients with obesity compared with patients with a lower BMI. A meta‐analysis of all of the randomized trials of DOACs in both AF and venous thromboembolism disease demonstrated similar, if not better, outcomes with DOAC usage in those with higher BMI values. These seemingly favorable outcomes have been termed the obesity paradox in the literature and have been reported in multiple chronic conditions, including other cardiovascular diseases (eg, coronary heart disease, heart failure), end‐stage liver and renal disease, and chronic obstructive pulmonary disease. However, in longer‐term analyses examining incident cardiovascular events and adjusting for competing risk of death, the obesity paradox was not observed and may reflect the younger age of onset in those with obesity. The overall low rates of stroke and systemic embolism noted in our study are not unexpected for a cohort of patients already on anticoagulation.

Strengths of our study include the large sample size and contemporary cohort that reflects a real‐world experience pooling data from both an academic medical center as well as several community hospitals, which enhances the generalizability of our findings. BMI was based on directly recorded weight and height in the electronic health record. We believe this approach to be more robust than basing weight classification solely on diagnosis codes.

There are several limitations to our study, notably those inherent to a retrospective analysis and the potential for unmeasured confounders. Electronic health record data are subject to potential coding errors; additionally, a prescription order in the electronic health record may not reflect actual medication dispensing or primary or secondary medication adherence of the patient. We included both patients diagnosed with AF as outpatients and as inpatients for analysis; however, there are unmeasured factors that differ between these populations. In our study, BMI and weight categorizations were based on the patient's recorded weight at the index date. While patients' weights may fluctuate over time, an analysis from ARISTOTLE (Apixaban for Reduction in Stroke and Other Thromboembolic Events in Atrial Fibrillation) of patient weight changes over the 3‐year study period showed minimal change overall. In particular, there was almost no overall change in patient weight among the highest weight group (>120 kg) over a 1 year period. We did not assess differences in dosing on the basis of renal function in this study, but further study is needed given the higher risk of renal dysfunction in patients with obesity. Although this study has a large number of patients, stroke rates are low in patients treated with anticoagulation (as expected) but may have resulted in lack of adequate power to detect some small differences between obesity groups. The population of this study was predominantly White, so generalizability may be limited to other racial groups.

In summary, use of DOACs for anticoagulation was associated with similar rates of stroke and systemic embolism as well as intracranial hemorrhage in this large analysis of patients with obesity and AF. These findings add to the growing body of evidence supporting the safety and efficacy of the use of DOACs in stroke prevention for AF.

Sources of Funding

Research reported in this publication was supported, in part, by the National Institutes of Health's National Center for Advancing Translational Sciences (Grant Numbers KL2TR001424) and the American Heart Association (#19TPA34890060) to Dr Khan. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. The funding sponsor did not contribute to design and conduct of the study, collection, management, analysis, or interpretation of the data or preparation, review, or approval of the manuscript.

Disclosures

None.

Tables S1–S8

Figures S1–S4

Click here for additional data file.