Timing for the Resumption of Anticoagulants after Intracranial Hemorrhage

Article information

J Korean Neurosurg Soc. 2026;69(2):215-224

Publication date (electronic) : 2025 September 1

doi : https://doi.org/10.3340/jkns.2025.0112

Sejin Choi¹, Myoung-jin Jang², Kangmin Kim¹, Won-Sang Cho¹, Hyun-Seung Kang¹, Jeong Eun Kim¹, Sung Ho Lee^,1,3

¹Department of Neurosurgery, Seoul National University Hospital, Seoul National University College of Medicine, Seoul, Korea

²Medical Research Collaborating Center, Seoul National University Hospital, Seoul, Korea

³Department of Neurosurgery, Sheikh Khalifa Specialty Hospital, Ras Al Khaimah, UAE

Address for correspondence : Sung Ho Lee Department of Neurosurgery, Seoul National University Hospital, 101 Daehak-ro, Jongno-gu, Seoul 03080, Korea Tel : +82-2-2072-1189, Fax : +82-2-744-8459, E-mail : nssungholee@snu.ac.kr

Received 2025 May 21; Revised 2025 August 18; Accepted 2025 August 25.

Abstract

Graphical Abstract

INTRODUCTION

Spontaneous intracranial hemorrhages (sICH) associated with anticoagulants are known to result in larger hematomas, increased growth of the bleed, and higher mortality rates compared to those not related to anticoagulant use [7,23]. When to resume anticoagulants after sICH is highly nuanced as delayed resumption can lead to ischemic complications including the recurrence of underlying disease and precipitous resumption can lead to hemorrhagic complications [15].

Currently, there are no clear recommendations for when patients with intracranial hemorrhage who were taking or receiving anticoagulants should resume therapy. According to the 2022 American Heart Association and American Stroke Association, the optimal time to resume anticoagulant therapy after sICH should be individualized based on factors such as hemorrhage location, size, and patient-specific thromboembolic risks [10]. In Korea, a study using national health insurance data suggested resuming anticoagulant therapy 6–8 weeks after sICH to lower the risk of all-cause mortality, but this study lacked individual patient information and was limited to atrial fibrillation patients [19]. Overall, the level of evidence on this topic is not high, and there is a need for studies in more general situation instead of assuming specific conditions as in previous studies.

This study aims to determine the appropriate timing for resuming anticoagulant therapy that minimizes the risk of rebleeding without increasing the risk of thromboembolic events and to help clinical decisions in more general settings.

MATERIALS AND METHODS

Study population

Anonymized data were collected retrospectively from electronic medical records and from SUPREME^® (SNUH, Seoul, Korea) which is the clinical data warehouse system. The study period was from June 2000 to May 2022. Patients younger than 19 years or older than 90 years at ictus were excluded. Intracranial hemorrhage patients who were taking oral anticoagulants such as vitamin K antagonists (warfarin) or non-vitamin K antagonist oral anticoagulants (NOACs), including direct thrombin inhibitors such as dabigatran and direct Xa inhibitors, were included in the study. Additionally, patients who were taking or receiving injections of indirect thrombin inhibitors (heparin or low-molecular-weight heparin) were included (n=154). We excluded patients who met any of the following criteria : hemorrhages definitively caused by trauma (n=30), death within one week of ictus (n=27), and failure to resume anticoagulant therapy within 6 months of the initial hemorrhage (n=7). The third criterion was applied because our study focused on evaluating outcomes after resumption of anticoagulation; thus, patients who never resumed therapy were outside the scope of analysis. Additionally, patients were followed up till 6 months after anticoagulant resumption. Longer-term follow-up was often unavailable or lacked sufficient clinical detail which may limit the reliability of late event attribution. The final study population was composed of 90 spontaneous intracranial hemorrhage patients who were taking pre-ictal anticoagulants and who resumed anticoagulant treatment within 6 months. Patients were followed up until 6 months after anticoagulant treatment was resumed. Sixty-eight out of 90 patients had no complications before or within 6 months after resumption, nine had hemorrhagic complications, and 13 had ischemic complications (Fig. 1).

Fig. 1.

Flowchart of patient inclusion and exclusion criteria. *Criteria : 1) 19–90 years old; 2) from June 2000 to May 2022; and 3) those who were taking or injecting anticoagulants (vitamin K antagonists, non-vitamin K antagonists including direct thrombin inhibitors and direct Xa inhibitors, indirect thrombin inhibitors) and resumed anticoagulants.

RESULTS

Participant demographics and clinical characteristics

Our study included 90 patients with sICH who resumed anticoagulant therapy within 6 months after the event. The median age was 71.0 years (interquartile range, 62.3–76.8) with 14.4% of the population was 80 years or older, and 56.7% of the patients were male. Most patients presented with IPH (71.1%), with a subset also having IVH (14.4%). A total of 63.3% of patients were taking warfarin and 22.2% of patients were taking NOACs, and 14.4% were taking other anticoagulants (such as low-molecular-weight heparin). Compared to the no complication and ischemic groups, NOAC consumption was significantly more common in the hemorrhagic complication group (77.8%) (p=0.003). The reasons for medication varied, with atrial fibrillation being the most common across all groups, followed by heart valve disease, cerebral infarction and other conditions such as pulmonary thromboembolism or DVT (Table 1). There was no significant difference in laboratory findings and underlying metabolic diseases among groups (Supplementary Table 1).

Table 1.

Participant demographics and clinical characteristics

Anticoagulant resumption

The management strategy was predominantly conservative treatment (57.8%), with fewer patients undergoing surgical interventions such as craniotomy/craniectomy (13.3%) or EVD placement (5.6%). Nine patients experienced hemorrhagic complications after the resumption of anticoagulants, while thirteen experienced ischemic complications. Among the nine hemorrhagic events, three (33.3%) were recurrences at the original site, and six (66.7%) were newly developed intracranial hemorrhages at different locations. No extracranial hemorrhages were observed.

Upon the resumption of anticoagulants, 32.2% (29 out of 90) of patients resumed the same medication at the same dose, and 67.8% (61 out of 90) resumed with less potent medication or a reduced dose. A total of 13.8% (4 out of 29) of those who resumed treatment with the same medication experienced complication. In cases where less potent anticoagulants were resumed or the dose was reduced, complications occurred in 29.5% (18 out of 61) of patients. The type or dose of the resumed medication was not associated with ischemic or hemorrhagic complications (p=0.526). The median time between ictus and the resumption of anticoagulant therapy was 21.0 days, with an interquartile range of 12.0–42.0 days (Table 2).

Table 2.

Post-ictal management

Risk factors for complications and optimal range for anticoagulant resumption

Our multinomial regression analysis identified factors that influence the risk of hemorrhagic and ischemic complications. Age greater than 80 years was not significantly associated with hemorrhagic events (odds ratio [OR], 1.48; 95% CI, 0.22–9.83) but was a significant risk factor for ischemic events (OR, 4.29; 95% CI, 1.01–18.22). Notably, patients on NOACs had a significantly higher risk of hemorrhage (OR, 16.59; 95% CI, 2.61–105.34) than those on other anticoagulants, but this difference was not significant for ischemic events (Table 3).

Table 3.

Multivariable, multinomial logistic regression for risk factors of complications

ROC curve analysis for ischemic events suggested a possible cut-off at day 30 (AUC, 0.62; 95% CI, 0.47–0.77; p=0.16), and for hemorrhagic events at 7.5 days (AUC, 0.72; 95% CI, 0.49–0.95; p=0.15) (Table 4 and Supplementary Fig. 1). While neither AUC reached statistical significance, these cut-offs may indicate potential temporal windows of differing complication risks. We plotted the combined complication risk based on the calculations described in the Methods section. The risk was minimized around days 20–22. This indicates that the risk of hemorrhagic or ischemic complications after anticoagulant resumption is minimized when those were resumed 20–22 days after the initial intracranial hemorrhage (Fig. 2).

Table 4.

Cut-off points of anticoagulants resumption with the best discriminatory power for complications

Fig. 2.

Analysis of competing risks and optimal ranges for anticoagulant resumption.

DISCUSSION

This retrospective cohort study provides valuable insights into the complex decision-making process surrounding the resumption of anticoagulant therapy following sICH. Our findings showed that resuming anticoagulant therapy after approximately 3 weeks post-ictus minimizes the complications. Age above 80 years can be a risk factor for ischemic complications and NOACs increase the risk of hemorrhagic complications.

First, ROC analysis provided insight into the optimal timing. The analysis identified day 30 as the threshold with the best discriminatory power for predicting ischemic complications and day 7.5 for hemorrhagic complications. In other words, restarting anticoagulation within the first week post-ICH may expose patients to a relatively higher bleeding risk, whereas deferring beyond day 30 might increase ischemic vulnerability. In additional LOESS analysis of event rates by anticoagulants-resumption day, we were able to suggest ideal resumption around day 21. While we included patients who resumed anticoagulant treatment within up to 6 months, the graph restricted the X-axis to day 75 (Fig. 2). This truncation was due to the absence of ischemic events in patients who resumed therapy after day 75, which would otherwise artificially flatten the ischemia curve beyond that point. Notably, the hemorrhagic event curve exhibited a rise around day 30. This unexpected fluctuation is likely attributable to sampling bias rather than a true physiological risk increase. A closer inspection of the raw data revealed clustering of restart decisions around days 27 and 29, corresponding to a common clinical practice of restarting anticoagulants approximately 1 month after hemorrhagic events.

Although reduced-dose or less potent anticoagulants are often considered a safer strategy in high-risk patients, our findings suggest that dose reduction alone may not sufficiently mitigate the risk of hemorrhagic complications. Of the nine patients who experienced hemorrhagic complications, seven resumed anticoagulation at a reduced dose—six of them within 7 days of the index hemorrhage. Among all 61 patients who resumed with a reduced dose, 11 experienced ischemic events, and nine out of them had delayed resumption (≥21 days). These observations imply that the timing of resumption—rather than dose intensity alone—may play a more critical role in determining complication risk.

In patients with DVT, especially those at high risk of rebleeding, mechanical alternatives such as inferior vena cava (IVC) filters are often considered. While our cohort included four patients with DVT, documentation regarding the use or consideration of IVC filters was not consistently available. Notably, all four patients underwent craniotomy or craniectomy, likely requiring prolonged care at intensive care unit. In such critically ill neurosurgical patients, logistical limitations might have precluded timely vascular interventions. Although the exact clinical reasoning was not recorded, this highlights the complexity of anticoagulation decisions in real-world neurocritical care settings, where both thromboembolic and hemorrhagic risks must be weighed carefully.

This study is clinically significant considering the burgeoning use of anticoagulants and the aging global population [6]. Globally, the 2019 incidence of sICH was approximately 3.5 million cases (42 cases per 100000 person-years), with the incidence among Asians being roughly double that of Black or White populations [22]. The number of patients in their 80s diagnosed with sICH increased by 26.6% while those in other age groups showed a decreasing trend. Additionally, with aging populations in many countries, including Korea, the number of patients taking anticoagulants is increasing [14].

The prevalence of warfarin use in our study population was high (63.3%), particularly given that the predominant indication was atrial fibrillation, for which NOACs are recommended in current practice. This finding is attributable to the historical context of our dataset. Most atrial fibrillation patients taking warfarin were treated before NOACs were first recommended in the Korean guidelines for non-valvular atrial fibrillation in 2018, and thus the observed prescription pattern reflects the standard of care at that time.

Interestingly, the pre-ictal use of NOACs was significantly associated with hemorrhagic complications. This finding suggests that NOACs, despite being a newer class of anticoagulants, may not confer a lower risk of post-sICH bleeding, which has critical implications for clinical practice. Clinicians should carefully consider alternative anticoagulant strategies for patients with a history of sICH and NOAC therapy. A review concluded that NOACs use should be preferred over warfarin after sICH [24], but specific comparative studies have shown varying results [16]. Also, NOAC is associated with lower risk of sICH but there is no evidence that NOAC reduces the risk of sICH “recurrence” compared to other anticoagulants [13]. In addition, unavailability of reversal agents for NOAC such as idarucizumab, andexanet, and 4-factor prothrombin complex concentrate in Korea might have affected the result [9,10]. Also, it is possible that our study population was well-managed at a tertiary hospital and was strictly titered for warfarin. The average PT INR among warfarin users was 2.19, which falls within the standard therapeutic range for most indications. Notably, 28.1% (16 of 57) had PT INR <1.5, and ischemic events occurred in three of these patients, suggesting that some may have been in a subtherapeutic or hypercoagulable state at the time of presentation. However, the small number of events limits definitive conclusions.

The lack of a significant association between age above 80 years and hemorrhagic complications is intriguing, given previous studies reported higher intracranial bleeding risks in the elderly individuals at age 80 or older, regardless of anticoagulation therapy [5]. However, advanced age has emerged as a significant risk factor for ischemic events, underscoring the necessity of anticoagulants for the prevention of thromboembolic events in this age group. The selection of 80 years as the age cutoff in our analysis aligns with existing literature that identifies this age as a significant threshold for increased risk of intracranial hemorrhage or cerebral infarction [8].

Previous studies have mostly focused on patients with certain conditions, such as atrial fibrillation : Aguilar et al. [1] highlighted the variability in clinical practice, noting a wide range of expert opinions on the acute reversal and resumption of anticoagulant therapy in warfarin-associated sICH. Majeed et al. [18] suggested an optimal timing for resuming warfarin therapy between 10 and 30 weeks post-sICH based on a large retrospective cohort study. This timeframe was associated with a reduced combined risk of recurrent sICH or ischemic stroke [18]. Alkherayf et al. [2] provided a more specific window, suggesting that 4–7 days might be an ideal time for vitamin K antagonist resumption in mechanical heart valve patients post-sICH, balancing the risks of recurrence and thrombosis. Kuramatsu et al. [12] focusing on patients with sICH and mechanical heart valves, reported an increased incidence of hemorrhagic complications with anticoagulant resumption within less than 2 weeks post-sICH. In contrast to these previous studies, our study highlights a more generalized approach to anticoagulant reinitiation following sICH. This general approach allows our findings to have broader applicability in real-world clinical practice, where patients often present with varying hemorrhage types and complexities.

Limitations of our study include its retrospective nature and reliance on a single-center dataset, which may not be generalizable to other populations. Especially, there is the lack of detailed information regarding the clinical rationale for early anticoagulation resumption, especially in those who resumed within 7 days. While early resumption may reflect high thromboembolic risk or clinical stability, this could not be confirmed due to the retrospective nature of the study. Future prospective studies with more comprehensive clinical documentation are needed to clarify these decision-making factors. Also, the relatively small sample size and 6-month follow-up after anticoagulant resumption may have limited our ability to detect all relevant complications. Additionally, the heterogeneity of the anticoagulants used and the various management strategies employed make it challenging to generalize our findings. The CHA₂DS₂-VASc, HAS-BLED, HEMORR2HAGES, and ATRIA scores, which are used to calculate the risk of hemorrhage among atrial fibrillation patients treated with anticoagulants, were not used in the analysis [17,25]. As this study used retrospective data, there were too many missing values to calculate these scores.

This study did not exclude or identify patients with cerebral amyloid angiopathy (CAA) or other contraindications to anticoagulants. Only two patients were later identified as having probable or possible CAA, based on delayed clinical and radiologic findings. In both cases, the diagnosis became evident only well after the index hemorrhagic event and did not influence the decision to resume anticoagulation. One of these patients experienced a hemorrhagic complication, and the other an ischemic event. Given the retrospective nature of the study and the known difficulty of diagnosing CAA we found it infeasible to include CAA status as a meaningful analytical variable.

Another limitation of our study is the lack of detailed stratification by hemorrhage location (e.g., lobar vs. deep IPH). Prior studies, such as Becattini et al. [3], have demonstrated that anatomical location may influence the risk of hemorrhage recurrence, particularly in cases of IPH. We fully agree that further granularity in hemorrhage classification could provide valuable insights. However, given the inclusion of diverse hemorrhage types—not limited to IPH alone—it was not feasible to perform reliable and standardized imaging-based classification across all cases. We have acknowledged this as a critical limitation, and we believe that future prospective studies with systematic radiological review will be essential to refine risk stratification based on hemorrhage subtype.

Although our study provides a generalized guideline for anticoagulant resumption around 21 days post-sICH, this recommendation may not apply to specific subgroups, such as patients undergoing EVD or other surgical interventions. These patients often have distinct clinical trajectories, including a higher likelihood of requiring subsequent procedures (e.g., ventriculoperitoneal shunting), necessitating individualized decision-making. Additionally, subgroup analysis was limited by the small sample size, which constrained our ability to separate these cases for detailed analysis.

Future research should aim to include prospective multicenter trials to validate our findings. Moreover, the integration of comprehensive pharmacogenomic data could further refine risk stratification and anticoagulant management after sICH. Finally, the creation of international registries could help in gathering data to better inform clinical practice guidelines and improve patient outcomes. The pooled mean duration to oral anticoagulant resumption after the index sICH was 31 days, and further research could explore the individual factors affecting the decision to resume anticoagulants in patients with sICH [4].

CONCLUSION

In conclusion, our findings suggest that anticoagulation resumption around 21 days post-sICH may serve as a useful clinical reference, potentially balancing hemorrhagic and ischemic risks. However, given the heterogeneity of patient profiles and the limitations of a retrospective, single-center study, this should not be interpreted as a prescriptive guideline. Individualized assessment remains paramount in clinical practice.

Supplementary materials

Article information Continued