Anticoagulation Under-utilization in Atrial Fibrillation Patients is Responsible for a Large Proportion of Strokes Requiring Endovascular Therapy

Abstract

Objectives:

Atrial fibrillation (AF) is responsible for 30-50% of large strokes requiring endovascular thrombectomy (EVT). Anticoagulation (AC) underutilization is a common source of AF-related stroke. We compared antithrombotic medications among stroke patients with AF that did or did not undergo EVT to determine if AC underutilization disproportionately results in strokes requiring EVT, while quantifying the proportion of likely preventable thrombectomies.

Methods:

This retrospective single-center cohort included consecutive patients admitted with acute ischemic stroke between 2016 and 2021. Patients were categorized based on the presence of AF and pre-admission antithrombotic medications were compared between those who underwent EVT and those who didn’t. The reason for not being on AC was abstracted from the medical record, and patients were categorized as either AC eligible or AC contraindicated.

Results:

Of 3092 acute ischemic stroke patients, 644 had a history of AF, 213 of whom underwent EVT. Patients who required EVT were more likely to not be taking any antithrombotics prior to admission (34% vs 24%, p=0.007) or have subtherapeutic INR on admission if taking warfarin (83% vs 63%; p = 0.046). Among the AF-EVT patients, 44% were taking AC, and only 31% were adequately anticoagulated. Only 8% of AF-EVT patients who were not on pre-admission AC had a clear contraindication, and 94% were ultimately discharged on AC.

Conclusions:

Lack of antithrombotic therapy in AF patients disproportionately contributes to strokes requiring EVT. A small minority of AF patients have contraindications to AC, so adequate anticoagulation can prevent a remarkable number of strokes requiring EVT.

1. Introduction

Atrial fibrillation (AF) accounts for an estimated 1 in 5 ischemic strokes (1, 2), but AF is disproportionately responsible for large strokes, accounting for 30-50% of those treated with endovascular thrombectomy (EVT). (3, 4) Anticoagulation (AC) dramatically reduces the risk of acute ischemic stroke in patients with AF, (5–8) and professional society guidelines recommend oral AC in patients with AF and elevated risk of stroke. (9, 10) However, a range of barriers contribute to AC underutilization in patients with AF, including financial barriers, patient decision-making, healthcare provider decision-making, healthcare system issues, and misperception of the risk-benefit profile. (11–13) It has been estimated that more appropriate AC use could prevent 50,000 strokes annually in the United States. (14) Because AF is responsible for such a large proportion of large vessel occlusions, more appropriate AC use may have a particularly potent effect on reducing the demand for EVT, but this has not yet been specifically characterized. In this study, we aimed to leverage a single-center cohort of stroke patients with atrial fibrillation to compare antithrombotic medications among patients that did or did not undergo EVT to determine if AC underutilization disproportionately results in large vessel occlusions requiring EVT, while quantifying the number and proportion of preventable thrombectomies.

2. Material and Methods

2.1. Study design and patients

This is a single-center retrospective cohort study. The study protocol was approved by the University of Pennsylvania institutional review board (IRB) and the need for informed consent was waived. The study conforms to STROBE guidelines for observational studies. Eligible patients were identified from the Hospital of the University of Pennsylvania’s local Get With The Guidelines^® (GWTG). The cohort included patients admitted between January 1, 2016, and December 31, 2021 with a discharge diagnosis of acute ischemic stroke. Patients were categorized based on the presence or absence of AF prior to the stroke admission. Patients with a new diagnosis of AF detected during or after the stroke admission were considered to not have AF prior to admission. Patients were also categorized based on undergoing or not undergoing EVT during the admission.

2.2. Data collection

Patient-level data were retrieved from the GWTG platform, including demographics, presenting NIH stroke scale (NIHSS) score, prior medical history, antithrombotic medication prior to admission, antithrombotic medication at discharge, large vessel occlusion details (presence/absence, which vessel if applicable), acute stroke treatment including intravenous thrombolysis and EVT, modified Thrombolysis in Cerebral Infarction (mTICI) score, discharge disposition, in-hospital mortality, and EVT complications including symptomatic intracranial hemorrhage (sICH) which was defined according to the European Cooperative Acute Stroke Study II (ECASS-2) criteria - any ICH on post-treatment brain imaging associated with at least a 4-point worsening on the NIHSS.(15)

Pre-admission antithrombotics were categorized as antiplatelets, anticoagulants, or none. Specific medication names were also collected. A physician performed a manual chart review to confirm antithrombotic medications and to abstract the reasons for not receiving AC before the admission, when applicable. Patients were subsequently categorized as “AC contraindicated”, “AC eligible”, or “unknown.” “AC contraindicated” patients were not receiving AC because of an established contraindication, including prior major bleeding or thrombocytopenia, or if AC was appropriately interrupted for surgery. “AC eligible” patients had a documented reason for not receiving AC which was not a clear contraindication. Patients were categorized as “unknown” if the reason for not being on AC was not documented in the electronic health record.

Discharge antithrombotics were categorized as antiplatelets, anticoagulants, or none. Patients were considered to be discharged on AC if AC was initiated prior to discharge, at the time of discharge or if there was a plan to initiate AC typically within 28 days after discharge (i.e. if there was a brief delay due to infarct size).

2.2.1. Data availability

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

2.3. Statistical analysis

Continuous variables were reported as mean and standard deviation (SD), ordinal variables were reported as median and interquartile range (IQR), and categorical variables were reported as relative frequencies. AF patients who underwent EVT were compared with those who did not undergo EVT. Among AF patients who underwent EVT, those with and without pre-admission AC were compared. Between-group differences were evaluated by t-tests for continuous variables, Wilcoxon-Mann-Whitney tests for ordinal or nonparametric variables, and chi-squared or Fisher’s exact tests for categorical variables. All tests were evaluated at a significance level of 0.05. Statistical analyses were performed using STATA 15.1 (StataCorp LLC., College Station, TX).

3. Results

3092 patients had a discharge diagnosis of acute ischemic stroke during the study period, 2427 did not undergo EVT and 665 underwent EVT. AF was more common in patients who underwent EVT (32% versus 18%, p<0.001).

In total, 644 stroke patients (20.8%) had a history of AF prior to the stroke. Table 1 summarizes the demographics and baseline antithrombotic use, comparing AF patients who underwent EVT and those who did not undergo EVT. EVT patients were more likely to be on no antithrombotic therapy (34% vs 24%; p = 0.007) or have a subtherapeutic INR if taking warfarin (83% vs 63%; p = 0.046). Of the 95 AF-EVT patients taking AC at the time of the stroke, 35 (37%) were taking warfarin and 55 (58%) were taking a DOAC. Similarly, of the 205 non-EVT patients taking AC at the time of the stroke, 76 (37%) were taking warfarin and 121 (59%) were taking a DOAC. Overall, only 31% of AF-EVT patients were adequately anticoagulated (receiving DOAC or Warfarin with therapeutic INR). Table S1 (supplemental table) reports a comprehensive list of medication names and combinations of therapy.

Table 1.

Baseline characteristics of stroke patients with AF compared based on EVT

	EVT (n=213)	No EVT (n=431)	p-value
Age, years	77 (12)	73 (12)	0.0004
Sex, % female	116 (55%)	208 (48%)	0.23
Race:			0.055
African American	73 (34%)	114 (27%)
White	113 (53%)	263 (61%)
Asian	3 (1%)	15 (4%)
Undetermined	24 (11%)	38 (9%)
Pre-admission Antithrombotic Therapy:
Anticoagulation (± antiplatelet therapy)	95 (44%)	205 (48%)	0.478
Antiplatelet therapy only	45 (21%)	122 (28%)	0.050
No antithrombotic therapy	73 (34%)	104 (24%)	0.007

Continuous variables are reported as mean (standard deviation). Categorical variables are reported as percentages. P-values were calculated by t-test for continuous variables and by chi-squared or Fisher’s exact test for categorical variables. EVT indicates endovascular thrombectomy. VKA indicates vitamin K antagonist. DOAC indicates direct oral anticoagulant. LMWH indicates low molecular weight heparin.

The 213 AF-EVT patients are described in Table 2, contrasting those receiving or not receiving anticoagulation. Patients not taking AC were slightly older, but baseline characteristics were otherwise largely similar. Procedural outcomes were also grossly similar. AC use did not impact the rate of sICH after EVT (Table 2). Among 118 EVT patients not on AC at the time of the stroke (Table 3), the reason for not being on AC was unknown in 25%. Only 8% were categorized as AC contraindicate. Overall, 94% of patients who were not on AC prior to admission were discharged on AC, excluding those who died during the hospitalization or were discharged with hospice care. The proportion of patients discharged on AC was not dependent on the pre-admission reason for not being on AC: 96% of “AC eligible,” 80% of “AC contraindicated,” and 91% of “unknown” patients were discharged on AC (p= 0.119). While warfarin accounted for 37% of AC before admission, only 14% of those on AC at discharge were prescribed warfarin (p < 0.001).

Table 2.

Demographics, stroke characteristics, and outcomes following endovascular therapy: Comparing patients receiving or not receiving anticoagulation prior to the stroke

	Prior Anticoagulation (n=95)	Not on anticoagulation (n=118)	p-value
Age, years	75 (13)	78 (11)	0.033
Sex, % female	54 (57%)	62 (53%)	0.53
Race, %			0.088
African American	41 (43%)	32 (27%)
White	44 (46%)	69 (58%)
Asian	1 (1%)	2 (2%)
Undetermined	9 (9%)	15 (13%)
Previous stroke or TIA	27 (29%)	26 (22%)	0.26
Presenting NIHSS	17 (10, 22)	18 (12, 24)	0.04
Vessel occlusion, %
MCA, M1 segment	43 (46%)	68 (58%)	0.10
MCA, M2 segment	25 (27%)	29 (25%)	0.70
Internal carotid artery	13 (14%)	16 (14%)	0.93
Clinical and Radiographic Outcome
Successful recanalization, % mTICI 2b or 3	68 (72%)	91 (77%)	0.42
sICH	4 (4.2%)	9 (7.6%)	0.39
sICH (Excluding patients who received IV tPA)	4 (4.4%)	8 (7.1%)	0.55
In-hospital mortality	19 (20%)	21 (18%)	0.68
Length of hospitalization, days	11 (9)	10 (8)	0.59
Discharge disposition:			0.43
Home	21 (22%)	19 (16%)
Other facilities	50 (53%)	62 (53%)
Death or hospice	24 (25%)	37 (31%)
Discharged on anticoagulation (Excluding hospice/death):	69 (99%)	76 (94%)	0.22
Anticoagulant type:			0.077
Warfarin	14 (20%)	7 (9%)
DOAC	55 (79%)	68 (91%)
Enoxaparin	1 (1.4%)	0 (0.0%)

Continuous variables are reported as mean (standard deviation). Ordinal variables are reported as median (interquartile range). Categorical variables are reported as percentages. P-values were calculated by t-test or Wilcoxon-Mann-Whitney test for continuous or ordinal variables. P-values were calculated by chi-squared or Fisher’s exact test for categorical variables, respectively. MCA indicates middle cerebral artery. M1 indicates the first segment of the MCA. M2 indicates the second segment of the MCA. mTICI indicates the modified thrombolysis in cerebral ischemia score. sICH indicates symptomatic intracranial hemorrhage. DOAC indicates direct oral anticoagulant.

Table 3.

AF patients that underwent EVT: reasons for not receiving AC prior to admission

Reasons for not receiving AC at baseline	Frequency (N=118)
AC eligible candidates	79 (67%)
Fall risk	8 (6.8%)
Elderly/dementia	3 (2.5%)
Patient preference	12 (10.2%)
Noncompliance	12 (10.2%)
AF was thought to have resolved	7 (5.9%)
AC discontinued after prior ablation/ awaiting ablation	5 (4.2%)
Recent diagnosis (>1 week)	3 (2.5%)
INR lability	2 (1.7%)
Previous gastrointestinal bleeding	14 (11.9%)
Previous hematuria	6 (5.1%)
Other non-life-threatening prior bleedings	6 (5.1%)
Low risk for stroke (CHA2DS2-VASc = 1)	1 (0.8%)
AC contraindicated	10 (8.4%)
Prior ICH (including SAH/SDH/IPH)	3 (2.5%)
Other life-threatening bleeding/severe anemia	2 (1.7%)
Thrombocytopenia	2 (1.7%)
Recent surgery	3(2.5%)
Unknown
No contraindication to AC documented	29 (25%)

Frequencies are reported as n (%). AF indicates atrial fibrillation. AC indicates anticoagulation. INR indicates the international normalized ratio. ICH indicates intracranial hemorrhage. SAH indicates subarachnoid hemorrhage. SDH indicates subdural hemorrhage. IPH indicates intraparenchymal hemorrhage.

4. Discussion

AC underutilization is a known contributor to stroke in patients with AF, and this cohort demonstrates that AC underutilization disproportionately contributes to strokes with LVO that require EVT. Prior studies indicate AF patients who suffer a stroke while on AC experience less severe symptoms as compared to those not on AC. (16, 17) Similarly, here we found that a lack of antithrombotic therapy disproportionately results in strokes that require EVT. Only 31% of AF patients that underwent EVT were adequately anticoagulated at the time of the stroke (DOAC or warfarin with INR in the therapeutic range). The majority of AF-EVT patients not on AC prior to the stroke lacked a clear contraindication to AC, and 94% were treated with AC after the stroke. Improved AC utilization not only stands to prevent a large proportion of strokes in patients with AF, but also reduces the demand for EVT, associated healthcare costs, and long-term disability.

At this center, 21% of all ischemic stroke patients had a history of AF (known prior to the stroke), which is consistent with national estimates. (1, 2) 47% of AF patients were using AC at the time of the stroke. In a nationwide study, Xian et al. reported that 70% of stroke patients with AF between October 2012 and March 2015 were not on AC at the time of the stroke admission, and an additional 13.5% were treated with warfarin but had a subtherapeutic INR. (18) Meinel et al. similarly reported that more than 60% of AF patients between 2014 and 2017 were not on AC at the time of stroke. (19) Our reported rate of AC utilization may reflect a trend toward better AC utilization over time, or may be a result of institutional prescribing habits or local patient preferences. Another recent study on stroke patients who underwent EVT showed that only 28% of patients with pre-stroke indications for AC were adequately anticoagulated (20) which is in line with our results. AC utilization varies regionally, (21, 22) so there may be a greater opportunity in regions with particularly poor AC utilization.

Similar to other studies, (23, 24) administration of DOACs was more common than warfarin both prior to admission, and even more common after discharge. Further, DOAC use increased over the course of the study. This shift is consistent with a general trend toward more DOAC use. (19, 25) Among EVT patients on warfarin, subtherapeutic INR was common (68%), similar to a large national cohort of stroke patients with AF. (18) This reinforces the need to re-consider DOAC eligibility when encountering a patient on Warfarin, particularly if there is a history of INR lability. (26)

Among the AF patients who underwent EVT, those who were on pre-admission AC were significantly younger (75 versus 79 years old). This is consistent with what others have reported in stroke patients with AF, (18) and likely reflects the provider and patient hesitancy with AC due to a perceived higher risk of bleeding complications in older age groups. (27) However, this hesitancy is often driven by an overvaluing of the bleeding risk and a failure to recognize that stroke risk increases with age and net the benefit of AC is greatest in the elderly. (28) In this cohort, only 8% of AF-EVT patients had an AC contraindication but 56% were not receiving AC prior to the stroke (more if including those with incomplete AC or subtherapeutic INR). This discrepancy is largely the result of provider and patient misperceptions regarding suitability for AC despite the absence of a true contraindication. In the AVERROES trial, AF patients perceived to be bad warfarin candidates (but without an AC contraindication) were treated with either apixaban or aspirin, in which apixaban reduced the risk of stroke without increasing the risk of major bleeding. (29) Still, AC underutilization remains a significant problem. In this cohort, despite a high proportion of patients not treated with AC prior to the stroke, 94% were discharged on AC after the stroke. The decision to discharge on AC may also be influenced by an increase in CHA₂DS₂VASc score, but that is unlikely seeing as only one patient was not on AC because of a CHA₂DS₂VASc score of 1. It is possible that the stroke impacts the provider’s, or more likely the patient’s perception of the AC risk-benefit profile.

Adequate AC in AF patients can be expected to achieve an estimated 70-75% relative risk reduction of acute ischemic stroke.(30) Based on the rate of AC utilization and the proportion of EVT patients with AF, we extrapolate that an estimated 15% of thrombectomies could be prevented (Figure 1). The potential impact is substantial when considering more than 20,000 patients nationwide are estimated to be eligible for EVT annually with an average inpatient healthcare cost of $45,761 per patient. (31–33) Further, an estimated 10-fold present with large vessel occlusion, (32, 34) so as EVT eligibility likely expands to include patients in later time windows, medium vessel occlusion, large infarct cores, or low presenting NIHSS, the number of thrombectomies performed on AF patients can be expected to increase. This analysis focuses on the impact of AC underutilization on EVT, but as noted, a small minority of stroke patients with large vessel occlusion undergo EVT, and thus appropriate AC utilization stands to have an even greater impact beyond EVT. The estimated impact of AC is based on a relative risk reduction of 70% derived from recently published data in a real-world setting. (30) This discussion is not intended to provide a precise number of preventable thrombectomies, but rather to highlight the scope of the issue.

Figure 1.

Flowchart of antithrombotic therapy in patients with AF who underwent EVT: Of 213 AF patients who underwent EVT, 147 were not adequately anticoagulated. Only 10 had a contraindication to AC. Appropriate AC could have prevented 45% of AF-associated EVTs. AF indicates atrial fibrillation. AC indicates anticoagulation. INR indicates the international normalized ratio.

This study has additional limitations. This is a single-center study which limits generalizability, but the prevalence of AF, AC utilization rates, and proportion with subtherapeutic INR are representative of previously reported national cohorts. The GWTG database does not capture the reason patients are not on AC prior to admission (only the reason patients aren’t discharged on AC), so this was abstracted from the electronic health record, but the documentation was insufficient in 25%. Thus, the proportion of AC-contraindicated patients may be higher than reported. Further, determining who is an eligible candidate for AC is not always clear, further introducing bias. Fortunately, the rate of AC use at discharge supports the idea that the vast majority of patients did not have a true contraindication to AC, regardless of how AC eligibility is categorized prior to the stroke. In addition, this cohort does not differentiate between persistent and paroxysmal AF, which may impact AC prescribing habits. In a national registry of AF patients, providers were less likely to prescribe AC if AF was paroxysmal rather than persistent (35), but it is important to recognize that the benefit of AC is similar in both types of AF. (36, 37)

5. Conclusions

In conclusion, AC underutilization disproportionately contributes to strokes that require EVT. A minority of AF patients were adequately anticoagulated when they presented with an acute stroke, but the vast majority are eligible AC candidates. Adequate AC utilization stands to prevent a remarkable number of strokes, subsequent EVTs, along with the associated cost and disability.