Initiation of direct oral anticoagulation after reperfusion therapy in ischemic stroke in clinical practice: Results from Sits-International Stroke Registry

Abstract

Introduction:

Data is limited on the safety of early initiation of direct oral anticoagulation (DOAC) treatment after acute ischemic stroke (AIS) receiving reperfusion therapy in patients with atrial fibrillation (AF). We investigated the timing of DOAC initiation and its association with safety and outcomes.

Materials and methods:

We included AIS patients receiving reperfusion therapy with AF diagnosis (prevalent or new) registered in the Safe Implementation of Treatments in Stroke international registry during 2013–2024. Safety outcomes were hemorrhage and death. Secondary outcomes were recurrent AIS, any embolism and functional independence (modified Rankin Scale [mRS] 0–2) at 3 months. We performed descriptive statistics and multivariable analysis for DOAC initiation time as an ordinal variable (0–3, 4–7, and 8–100 days after stroke onset) and its association with outcomes. Explorative analyses were performed to investigate factors associated with DOAC initiation time, as a continuous or ordinal variable.

Results:

In total, 13,389 patients had data on DOAC initiation time, and 7861 patients had new event data by 3-month follow-up. We observed 0.1% intracranial hemorrhage, 0.4% major extracranial hemorrhage, 1.1% recurrent ischemic stroke, and 0.2% systemic embolism. At 3 months, 4.8% patients had died, and functional independence was seen in 60.9%. In multivariable analyses, DOAC initiation after stroke onset was not associated with any outcomes. Higher 24 h NIHSS and lower pre-stroke mRS score were associated with delayed DOAC initiation.

Conclusion:

DOAC initiation time was not associated with any outcomes in AIS patients who received reperfusion therapy. Severe stroke symptoms at 24 h were associated with delayed DOAC initiation. The low incidence of safety outcomes and missing data in this study should lead to cautious interpretations of these results.

Graphical abstract.

Introduction

Atrial fibrillation (AF) is the most common cardiac arrhythmia and patients with AF have a fivefold increased risk of ischemic stroke, often a severe stroke. Major trials evaluating direct oral anticoagulants (DOACs) for stroke prevention in patients with AF excluded patients with stroke within 10–14 days and severe disabling stroke within 3–6 months.^1–4 If untreated, the risk of early recurrence of ischemic stroke in patients with AF can reach up to 7.5% within the first 2 weeks. ⁵ When this study was planned, only few prospective observational and small randomized studies data were available showing that early initiation of DOAC was associated with a low risk of symptomatic intracranial hemorrhage.^6,7 While our study was ongoing, two randomized controlled trials (RCTs) were published, showing no safety issues in early DOAC initiation after AIS.^8,9

In clinical practice, physicians still use a variety of timing regimes commonly stratified according to the severity of the stroke, neuroimaging findings or expert opinion, where OAC treatment is postponed in the more severe cases.^10,11 At this time, physicians must rely on the current guidelines with weak level of evidence, while the guidelines get updated with the results of the finished and ongoing RCTs.^10,11

There are some RCTs still ongoing which are investigating the benefit of early DOAC administration in patients with AF-related ischemic stroke such as (START NCT03021928, USA; and OPTIMAS, EudraCT 018-003859-38, UK). Observational studies may inform practice in patient groups ineligible for trial participation, for whom randomized evidence would continue to be lacking even after the publication of trial results.

We aimed to investigate the timing of initiation of DOACs following AIS treated with acute reperfusion therapy and evaluate the safety of early initiation of DOACs for secondary prevention in patients with AF.

Material and methods

Data were collected from the Safe implementation of treatment in stroke (SITS) international stroke thrombolysis and thrombectomy registry (SITS-ISTR), an academic-driven, international stroke registry. SITS-ISTR has several data entry forms, for example, for intravenous thrombolysis (IVT) treated patients, endovascular thrombectomy (EVT) treated patients, a general stroke registry. Data collection on AF and OAC related variables is an extension of IVT and EVT data entry forms.

Study participants

We included patients with an acute ischemic stroke, who received reperfusion therapy with intravenous thrombolysis (IVT) alone, endovascular thrombectomy (EVT) alone or IVT + EVT combined, between January 1st, 2013 and January 31st, 2024. Patients were included if they had known or newly diagnosed AF at baseline or by discharge. The methodology of the SITS-ISTR has been described previously, including procedures for data collection and management, patient anonymization and verification of source data.^12,13

We collected baseline and demographic characteristics, premorbid modified Rankin Scale (mRS) scores, variables required for calculating stroke risk (CHA₂DS₂-VASc) ¹⁴ and bleeding risk (HAS-BLED) ¹⁵ scores, stroke severity per the NIHSS at baseline and 24 h, medication history, imaging data at admission and follow up, time interval in days between index event and start of DOAC, and physicians’ reasons for delaying DOAC initiation beyond acute hospital discharge. Data from the 3-month follow-up included data on any new clinical events including new ischemia or hemorrhage, a new diagnosis of AF, functional outcome per the mRS, and death. All assessments of imaging studies, neurological, and functional status were done according to clinical routine at centers participating in the SITS-ISTR.

Outcomes

Primary outcomes were safety outcome events: intracerebral hemorrhage, major intracranial hemorrhage and major extracranial hemorrhage defined according to the International Society on Thrombosis and Hemostasis, and death. ¹⁶ Secondary outcomes were recurrent ischemic stroke, pulmonary embolism, systemic embolism, myocardial infarction, and functional outcome per the mRS 0–2. To decrease the likelihood of low numbers of safety outcome events, we also considered composite outcomes: All embolism, which included recurrent ischemic stroke, pulmonary embolism, systemic embolism, and myocardial infarction, and all bleedings, which included intracerebral hemorrhage, major intracranial hemorrhage, and major extracranial hemorrhage.

Additional explorative analyses investigated the number of days from stroke onset to initiation of DOAC treatment as an outcome, both as a continuous variable and as an ordinal variable with time intervals of 0–3, 4–7, and 8–100 days. The timing of DOAC initiation was at the discretion of the treating physicians.

Statistical analysis

Descriptive statistics (means, standard deviations, medians, interquartile ranges, proportions, and percentages, as appropriate) for baseline and demographic characteristics for all included patients were presented. Patients with missing data or unknown values for a variable were excluded from the analysis of that variable.

Analyses of the primary and secondary outcomes were performed by multivariable logistic regression. Days from stroke onset to DOAC start were added to the analyses as an ordinal variable. Other baseline covariates were chosen based on clinical relevance: age, sex, systolic blood pressure (SBP), glucose, cholesterol, onset to door time, NIHSS score at baseline, NIHSS score at 24 h, premorbid mRS score, and medical history of hypertension, hyperlipidemia, diabetes mellitus, congestive heart failure, and previous stroke. All covariates were added into the models and analyzed together.

Separate explorative analyses considered days from stroke onset to DOAC start as a continuous and ordinal outcome separately. We used linear and ordinal regression analysis, respectively, with backwards stepwise selection of covariates into the final models. Other baseline covariates were chosen based on clinical relevance and were all added into the models. The initial variables that were included in the backwards stepwise selection were the same as in the primary analysis.

Additionally, we performed univariate analysis on the median stroke onset to DOAC start in days for combined variable of any hemorrhagic transformation versus no hemorrhage. We also included a summary of the reasons why clinicians chose to delay DOAC initiation at discharge, where this data was available.

Two-sided p values below 0.05 were regarded as statistically significant. Statistical analyses were performed on R version 4.0.4.

Ethical approvals

Data collection in this study was done under the SITS Monitoring study II (SITS-MOST II) which was originally approved by the Stockholm Ethics committee (2015/767-31) and later amendments were approved by the Swedish National Ethics Review Authority (2022-01157-02). Need for ethical approval or patient consent for participation in the SITS-ISTR varied among participating countries. Ethical approval and patient consent were obtained in countries that required this; other countries approved the register for conduct as an anonymized audit.

Results

Figure 1 shows the study flow chart. Our final study population included 13,389 patients receiving DOAC with known time of DOAC initiation (apixaban = 6353, dabigatran = 3530, rivaroxaban = 2388, and edoxaban = 1118). Mean age was 76.7 years, median baseline NIHSS score was 11, 55.5% were female, and 87.3% received IVT ± EVT while 32.0% received EVT ± IVT (Supplemental Table 1).

Figure 1.

Study flow chart.

SITS: safe implementation of treatment in stroke international registry; AF: atrial fibrillation; DOAC: direct oral anticoagulants.

The median time from stroke onset to DOAC initiation was 8 days (IQR 5–14). In the early DOAC initiation group (0–3 days, n = 2374) as compared to the delayed (4–7 days, n = 3801) and late (8–100 days, n = 7214) groups, history of ongoing OAC treatment at index stroke and AF were more common (Table 1). We observed lower NIHSS scores for patients with early initiation of DOAC (median 8, IQR 5–14) compared to patients with delayed (median 9, IQR 6–15) and late (median 13, IQR 8–18) DOAC initiation.

Table 1.

Baseline characteristics of patients for the 3 time-intervals for direct oral anticoagulant start after stroke onset.

Baseline characteristics	DOAC start 0–3 days	DOAC start 4–7 days	DOAC start 8–100 days	p-Value
n	2374	3801	7214
DOAC start from stroke onset (days)	2 (2, 3)	5 (4, 7)	13 (10, 20)	<0.001
Age	76.3 (10.3)	76.5 (9.9)	77 (9.9)	0.006
Systolic blood pressure (mm Hg)	151.6 (23.6)	151.8 (23.4)	152.8 (24)	0.045
Diastolic blood pressure (mm Hg)	83.6 (14.9)	83.8 (14.9)	83.8 (15.1)	0.831
Glucose (mmol/l)	7.2 (2.5)	7.3 (3.5)	7.4 (2.9)	<0.001
Cholesterol (mmol/l)	4.6 (1.2)	4.5 (1.8)	4.5 (1.8)	0.857
Onset to door (min)	88.5 (60, 140)	86 (59, 136)	86 (59, 135)	0.573
Onset to imaging (min)	113 (78, 176)	116 (80, 172)	115 (81, 167)	0.796
Onset to IVT (min)	142 (105, 200)	150 (110, 201)	152 (115, 200)	<0.001
Onset to EVT (min)	216 (151, 325)	218 (159, 296)	230 (170, 330)	<0.001
NIHSS score	8 (5, 14)	9 (6, 15)	13 (8, 18)	<0.001
mRS score	0 (0, 1)	0 (0, 1)	0 (0, 1)	0.001
CHADS-VASc	3 (2, 4)	3 (2, 4)	4 (3, 4)	<0.001
HAS-BLED	2 (1, 2)	2 (2, 2)	2 (2, 2)	0.116
Sex, male	47.6 (1131/2374)	45.1 (1715/3801)	43.1 (3107/7214)	<0.001
Hypertension	72.7 (1722/2369)	77.5 (2923/3772)	78.8 (5645/7167)	<0.001
Hyperlipidemia	30.8 (712/2308)	31.5 (1163/3692)	33.4 (2345/7028)	0.032
Diabetes mellitus	19.0 (451/2371)	17.4 (658/3777)	19.3 (1381/7165)	0.056
Congestive heart failure	12.6 (297/2359)	11.5 (433/3762)	11.7 (831/7122)	0.399
Atrial fibrillation	65.3 (1344/2057)	63.7 (1997/3137)	60.6 (3520/5807)	<0.001
TIA	5.4 (129/2367)	4.8 (179/3763)	4.7 (335/7149)	0.311
Previous stroke	13.2 (312/2370)	12.7 (480/3770)	11.7 (835/7157)	0.085
Smoking, current and previous	17.0 (384/2258)	17.1 (614/3582)	17 (1150/6765)	0.982
Antiplatelet treatment	32.5 (768/2362)	34.4 (1297/3767)	36.7 (2627/7151)	<0.001
Anti-hypertensive treatment	73.4 (1738/2367)	73.4 (2770/3772)	74.3 (5310/7150)	0.552
Statin treatment	30.8 (727/2358)	28.0 (1050/3749)	29.9 (2125/7115)	0.039
Anti-diabetic treatment	15.3 (363/2366)	13.0 (490/3770)	15.0 (1073/7148)	0.008
Anticoagulatant treatment	23.5 (556/2369)	19.1 (722/3775)	16.6 (1186/7151)	<0.001
IVT treatment	84.0 (1994/2374)	88.5 (3362/3801)	87.7 (6326/7214)	<0.001
EVT treatment	32.4 (769/2374)	30.6 (1163/3801)	32.5 (2348/7214)	0.100
IVT and EVT treatment	16.4 (389/2374)	19.0 (724/3801)	20.2 (1460/7214)	<0.001

DOAC: direct oral anticoagulant; n: number of patients; NIHSS: National Institute of Stroke Scale; mRS: modified Rankin scale; CHADS-VASc: ischemic stroke risk scale; HAS-BLED: bleeding risk scale; TIA: transient ischemic attack; IVT: intravenous thrombolysis; EVT: tndovascular thrombectomy.

Mean, median, and percentages presented for continuous, ordinal, and categorical variables, respectively. Within parentheses presented standard deviations, interquartile range, and proportion of patients, for continuous, ordinal, and categorical variables, respectively.

By 3-month follow-up, 0.1% of patients suffered an ICH, 0.4% had suffered a major extracranial hemorrhage, 1.1% had suffered a recurrent stroke, and 1.6% had suffered any other embolism, with no major differences between DOAC initiation groups (Table 2). The DOAC initiation group 8–100 days had a higher proportion of death and a lower proportion of mRS 0–2 by 3-month follow-up as compared to the other two groups. We found that DOAC initiation as an ordinal predictor had no statistically significant effect on any of the outcomes (Table 3).

Table 2.

Outcomes in time interval groups of direct oral anticoagulant start after stroke onset.

Results	DOAC start 0–3 days	DOAC start 4–7 days	DOAC start 8–100 days	p-Value
NIHSS, 24 h	2 (0, 5)	3 (1, 6)	7 (3, 14)	<0.001
New ischemic stroke	0.8 (13/1550)	0.8 (18/2170)	1.3 (56/4292)	0.127
Intracerebral hemorrhage	0.1 (1/1550)	0.0 (1/2170)	0.1 (6/4292)	0.470
Other intracranial hemorrhage	0 (0/1550)	0.1 (2/2170)	0.1 (6/4292)	0.470
Unspecified stroke	0 (0/1550)	0.0 (1/2170)	0.0 (1/4292)	0.677
Major extracranial hemorrhage	0.3 (5/1550)	0.2 (5/2170)	0.4 (19/4292)	0.390
Myocardial infarction	0.1 (2/1550)	0.4 (8/2170)	0.3 (12/4292)	0.386
Systemic embolism	0.1 (2/1550)	0.1 (2/2170)	0.2 (10/4292)	0.393
Pulmonary embolism	0.1 (1/1550)	0 (0/2170)	0.2 (9/4292)	0.059
Any embolism	1.2 (18/1550)	1.3 (28/2170)	2.0 (85/4292)	0.031
Any hemorrhage	0.4 (6/1550)	0.4 (8/2170)	0.7 (30/4292)	0.149
mRS 0–2 at 3-month follow-up	75.5 (1269/1681)	74.0 (1851/2501)	48.9 (2346/4794)	<0.001
Death by 3-month follow-up	4.0 (69/1723)	4.0 (101/2555)	5.5 (268/4910)	0.004

DOAC: direct oral anticoagulant; NIHSS: National Institute of Stroke Scale; mRS: modified Rankin scale.

Percentages presented. Within parentheses, proportion of patients.

Table 3.

DOAC start in days from stroke onset as an ordinal variable (0–3, 4-7, and 8–100 days) in adjusted logistic regression models for the outcomes at 3-month follow-up.

Outcomes	Odds ratio (95% confidence intervals)	p-Value
Recurrent stroke	0.83 (0.27–1.46)	0.749
Any embolism	1.56 (0.59–4.12)	0.373
Any bleed	0.73 (0.17–3.08)	0.671
mRS 0–2 at 3-month follow-up	1.13 (0.84–1.53)	0.420
Death by 3-month follow-up	0.76 (0.43–1.32)	0.328

DOAC: direct oral anticoagulant; mRS: modified Rankin scale; NIHSS: National stroke scale.

Odds ratios presented. Within parentheses, 95% confidence intervals.

Multivariable models were adjusted for: age, sex, systolic blood pressure, glucose, cholesterol, onset to door time, NIHSS score at baseline, NIHSS score at 24 h, premorbid mRS score, and medical history of hypertension, hyperlipidemia, diabetes mellitus, congestive heart failure, and previous stroke.

In the exploratory stepwise multivariable linear regression analysis of DOAC initiation as a continuous outcome (Table 4), we found that higher NIHSS score at 24 h was associated with delayed DOAC initiation. Lower baseline mRS score was associated with delayed DOAC initiation.

Table 4.

Adjusted stepwise backwards linear regression model of days to DOAC start from stroke onset as a continuous outcome, with backwards stepwise selection of covariates.

Variable	Increase in days to DOAC start	p-value
mRS score, baseline	−0.59	0.002
NIHSS, 24 h	0.75	<0.001

DOAC: direct oral anticoagulant; NIHSS: National Institute of Health Stroke Scale; mRS: modified Rankin scale.

Variables included in the initial stepwise backwards model: age, sex, systolic blood pressure, glucose, cholesterol, onset to door time, NIHSS score at baseline, NIHSS score at 24 h, premorbid mRS score, and medical history of hypertension, hyperlipidemia, diabetes mellitus, congestive heart failure, and previous stroke.

In the exploratory stepwise multivariable ordinal regression analysis of DOAC initiation as an ordinal outcome (Table 5), we found that the history of hyperlipidemia and NIHSS score at 24 h were associated with a delayed DOAC initiation. Baseline mRS score and previous stroke were associated with earlier DOAC initiation.

Table 5.

Adjusted ordinal regression model for days to DOAC start from stroke onset as an ordinal outcome (0–3, 4–7, and 8–100 days).

Covariate	Common odds ratios (95% confidence intervals)	p-Value
mRS, baseline	0.93 (0.87–0.99)	0.025
Hyperlipidemia	1.15 (1.01–1.32)	0.038
Previous stroke	0.75 (0.63–0.91)	0.002
NIHSS, 24 h	1.18 (1.16–1.19)	<0.001

DOAC: direct oral anticoagulant; NIHSS: National Institute of Health Stroke Scale; mRS: modified Rankin scale.

Variables included in the initial stepwise backwards model: age, sex, systolic blood pressure, glucose, cholesterol, onset to door time, NIHSS score at baseline, NIHSS score at 24 h, premorbid mRS score, and medical history of hypertension, hyperlipidemia, diabetes mellitus, congestive heart failure, and previous stroke.

We found that there was a statistically significant difference in median stroke onset to DOAC start in days when comparing patients who suffered hemorrhagic infarction (median 13, IQR 8–21), parenchymal hematoma (median 17, IQR 11–28), and no hemorrhagic transformation (median 8, IQR 4–13) (Supplemental Table 2). Our data regarding physician-specified reasons for DOAC initiation delay beyond acute hospital discharge showed that the most common reasons were large infarct and hemorrhagic transformation (Supplemental Table 3).

Discussion

In this observational study based on the international SITS-ISTR database we found that DOAC initiation during 0–3, 4–7, or 8–100 days after stroke onset was not significantly associated with any safety or functional outcomes. Our results at 3-month follow-up indicate that the risk of an intra- or extracranial hemorrhage is low regardless of the timing of initiation of DOAC.

Our results are congruent with the findings of two recently published RCTs investigated early versus late start of DOAC in AIS patients with AF, with a key difference being that all patients in our material had received reperfusion therapy.^8,9 In the TIMING study, AIS patients were randomized to early (⩽4 days) versus delayed (5–10 days) DOAC initiation and in the ELAN study patients were randomized to early (within 48 h for minor/moderate stroke and 6–7 days after a major stroke) versus later (day 3–4 after a minor stroke, day 6–7 after a moderate stroke, or day 12–14 after a major stroke) DOAC initiation.^8,9 Both trials found no difference in composite outcome between the early and delayed groups. One of the main differences between our study and these studies’ populations is the stroke severity (median NIHSS 11 for SITS-ISTR vs 4 for TIMING and 3 for ELAN).

Our study population of moderate to severe stroke patients receiving acute reperfusion therapy showed no statistically significant difference in safety and other outcomes between early and delayed initiation of DOAC, complementing the results of the RCTs.^8,9 There was a statistically significant difference between the DOAC initiation groups and IVT treatment, with or without EVT treatment, yet the difference in absolute numbers was relatively small. Until guidelines are updated with the latest data from the published and ongoing RCTs, physicians have to rely on these trial results together with observational data to inform and guide their decisions.

We found that stroke severity at 24 h and baseline mRS are important variables that seem to guide DOAC initiation after AIS. These findings are in agreement with previous observational studies investigating different DOAC initiation time points and stategies.^17–19 Our data showed that NIHSS score at 24 h was associated with later DOAC initiation time and higher mRS with earlier start of DOAC treatment. Association of stroke severity with delayed initiation was highly statistically significant and most likely reflects clinical practice. This could also be reflected in our results showing hemorrhagic transformations had a delayed start of DOAC after reperfusion therapy.

This study has several limitations. It is observational, which despite attempts using regression models to correct imbalances could entail residual bias and confounding. The low occurrence of safety outcomes may be related to missing data and furthermore, some outcomes were present in only a small number of cases, necessitating caution in interpretation. Of 13,133 patients with data on initiation of DOAC, only 7861 had data on whether there was a new event or not at 3-month follow up. The study was based on patients receiving reperfusion therapy, with relatively severe strokes and results may not apply for patients without reperfusion therapy. In the study population, the decision of when and which DOAC to start with was based on clinician preference based on experience and local guidelines, which may be biased and not easily generalized.

This study also has several strengths. It is to our knowledge the largest multinational cohort of stroke patients treated with DOACs for secondary prevention following acute reperfusion therapy. Our results are also congruent with the findings of two recent published RCTs,^8,9 in that early start of DOAC does seem to be safe.

Conclusions

This large observational study found that the timing of DOAC initiation at 0–3, 4–7, or 8–100 days after stroke was not associated with safety and functional outcomes in AIS patients who had received reperfusion therapy. Stroke severity at 24 h was associated with delayed DOAC initiation, while higher baseline mRS was associated with earlier DOAC initiation. The infrequent occurrence of safety outcomes in this study should be interpreted with caution due to missing data on new hemorrhagic and embolic events.