Intracerebral hemorrhage during anticoagulation with vitamin K antagonists: a consecutive observational study
Intracerebral hemorrhage (ICH) is the most devastating complication of oral anticoagulation (OAC). As the number of patients on long-term OAC is expected to rise, the proportion of intracerebral hemorrhage related to OAC (OAC-ICH) in relation to spontaneous ICH (spont-ICH) is expected to increase as well. We determined the proportion of OAC-ICH in consecutive stroke patients and explored differences between OAC-ICH and spont-ICH regarding initial volume, hematoma expansion and outcome. Our prospective study consecutively enrolled patients with supra- and infratentorial ICH. The National Institute of Health Stroke Scale Score and the modified Rankin Scale (mRS) score at baseline and after 3 months, medical history and demographic variables were recorded. All admission and follow-up CTs/MRIs were analysed regarding ICH volume using the ABC/2-method. Intraventricular hemorrhage (IVH) was quantified using the Graeb score. Within 19 months, 2,282 patients were admitted to our ER. 206 ICH patients were included. Overall, 24.8 % of all ICH were related to OAC. Compared to patients with spont-ICH, OAC-ICH patients were older (p = 0.001), more frequently had initial extension of ICH into the ventricles (p = 0.05) or isolated primary IVH (p = 0.03) and a higher Graeb score upon admission (p = 0.01). In contrast, initial ICH volume (p = 0.16) and ICH expansion (p = 0.9) in those receiving follow-up imaging (n = 152) did not differ between the two groups. After correction for age, there was a trend towards poorer outcome in OAC-ICH (p = 0.08). One-fourth of all ICH are related to OAC. Initial extension of ICH into the ventricles and primary IVH are more frequent in OAC-ICH. The rate of hematoma expansion in OAC-ICH patients is similar to non-anticoagulated ICH patients.
KeywordsIntracerebral hemorrhage Oral anticoagulation Outcome Mortality Warfarin
Oral anticoagulation (OAC) with vitamin K antagonists (VKA) is an effective preventive measure for ischemic stroke in atrial fibrillation (AF) . As the prevalence of AF is expected to rise, the number of patients on long-term anticoagulation is expected to double over the next decades . However, new oral anticoagulants had a considerably lower rate of ICH compared to VKA in three recent randomized trials [3, 4, 5].
Secondary hematoma expansion and initial and/or subsequent extension of bleeding into the intraventricular compartment have been speculated to be responsible for the higher mortality and worse outcome in OAC-ICHs [6, 7]. The primary goal in the acute treatment of OAC-ICH is to restore anticoagulation by replacing coagulation factors or by hemostatic agents . In current guidelines, OAC reversal is recommended by using either prothrombin complex concentrate (PCC) or fresh frozen plasma (FFP) in combination with vitamin K [9, 10].
The primary purpose of our study was to determine the proportion of OAC-ICH among all ICH patients in a representative, consecutively enrolled cohort with ICH. Moreover, we examined whether size of the initial hematoma and secondary hematoma expansion differed between OAC-ICH and other ICH, and which factors determined functional outcome after spont-ICH and OAC-ICH.
Materials and methods
We prospectively included consecutive patients admitted to the neurological ER of the Department of Neurology, University of Heidelberg, with supratentorial or infratentorial ICH in our exploratory analysis as part of a stroke registry . Patients with subarachnoid, subdural and traumatic and symptomatic hemorrhages were excluded.
The diagnostic work-up encompassed vital signs, a neurological examination, laboratory work including bedside point-of-care coagulometry of the international normalized ratio (INR). All patients underwent neuroimaging at baseline (CT and/or MRI). The medical history and basic demographic variables were recorded. The baseline National Institute of Health Stroke Scale Score (NIHSSS) and the premorbid modified Rankin Scale (mRS) score were assessed at baseline and after 3 months. To achieve better reliability, the mRS was dichotomized  into 0–3 (no or mild to moderate deficit) and 4–6 (severe deficit or death). Follow-up imaging was performed in 152/206 patients. Data from 54 patients were not included in the analysis because of intermittent surgical hematoma evacuation (n = 6), participation in an interventional trial (n = 9), no performance of follow-up scanning because of an early decision to limit therapy to palliative treatment (n = 31), early referral to another hospital (n = 3), or very small initial ICH volume without subsequent clinical deterioration (n = 4). Insufficient quality of follow-up imaging precluded volumetric image analysis in one patient.
Reversal of anticoagulation
The diagnosis of OAC-ICH was made if the patient was treated with a VKA, i.e., phenprocoumon (Marcumar®, Roche) at the time of the ICH and the INR was ≥1.4. Rapid stepwise reversal of anticoagulation was performed according to the decision of the treating physician following a standardized in-house protocol as previously published [13, 14] using a 4-factor prothrombin concentrate (PCC, Octaplex® 500–1,000 I.E., Octapharma, Wien, Austria) or FFP in combination with vitamin K. During this procedure, the INR is measured using a point of care coagulometer (Coaguchek XS®, Roche, Penzberg, Germany) upon initial presentation in the ER and thereafter during stepwise reversal with PCC or FFP in combination with vitamin K until the INR of the prothrombin time was <1.4.
Assessment of intracerebral hemorrhage
Intracerebral hemorrhage located in the subcortical white matter of the frontal, parietal, occipital, or temporal lobes was defined as “lobar” hemorrhage; hemorrhages located in the basal ganglia, thalamus, brainstem, or cerebellum were classified as “deep” hemorrhages. IVH was diagnosed if the hematoma was only intraventricular (i.e., primary IVH) or if parenchymal ICH extended into the ventricles.
Baseline and follow-up scans were performed using either a 16-section CT scanner (Somatom Sensation 16, Siemens, Erlangen, Germany) or a 3 Tesla MRI (Magnetom Verio or Magnetom Trio, Siemens, Erlangen, Germany).
Follow-up imaging was performed within 24–48 h after initial imaging based on the treating physician’s judgement. Hematoma volume was measured applying the ABC/2 formula . Relevant ICH expansion was defined as an increase in ICH volume either ≥33 % or >6 mL on follow-up CT as suggested by previous studies . Extension of IVH was semi quantitatively assessed using the Graeb score .
All study procedures were approved by the local ethics committee of Heidelberg. Informed consent was obtained from the patient or the patient’s legal representative.
Data were analyzed using the statistical package for the social sciences (SPSS 18.0). Descriptive data are reported in absolute and relative frequencies; ordinal and continuous data as medians and interquartile ranges. We used the χ2 test and the Mann–Whitney U-test to compare binary and ordinal variables. Baseline clinical variables were included in univariate logistic regression analysis. A p value <0.05 was considered significant.
Additionally, a multivariable logistic regression model was used to evaluate factors independently associated with the dichotomized mRS after 3 months and 3 months mortality after ICH in the entire cohort. The mRS was dichotomized into 0–3 (no or mild to moderate deficit) and 4–6 (severe deficit or death). All factors with a p value of <0.05 in univariate regression analysis entered the analysis.
Patients characteristics, differences OAC-ICH and spont-ICH
All patients (n = 206)
OAC-ICH (n = 51)
Spont-ICH (n = 155)
95 % CI
108 (52.4 %)
32 (62.7 %)
76 (49.00 %)
Age: median (IQR)
170 (82.5 %)
46 (90.2 %)
124 (80.0 %)
42 (20.4 %)
17 (33.3 %)
25 (16.1 %)
History of ischemic stroke
27 (13.1 %)
5 (9.8 %)
22 (14.8 %)
64 (31.1 %)
41 (80.4 %)
22 (14.2 %)
67 (32.5 %)
5 (9.8 %)
62 (40.0 %)
51 (24.8 %)
51 (100 %)
INR baseline: median (IQR)
Time to admission CT: median (IQR)
Time to follow-up CT (h): median (IQR)
6 (2.9 %)
4 (7.8 %)
2 (1.3 %)
Initial ICH volume (mL): mean ± SD
27.6 ± 31.3
31.5 ± 30.2
26.4 ± 31.7
Initial ICH volume (mL): median (IQR)
Extension of ICH into the ventricles
96 (46.6 %)
30 (58.8 %)
66 (42.6 %)
Initial Graeb Score: median (IQR)
Graeb score, primary IVH: median (IQR)
NIHSS baseline: median (IQR)
Premorbid mRS: median (IQR)
mRS 3 months: median (IQR)
mRS 3 months, dichotomized (0-3)
78 (37.9 %)
11 (21.6 %)
67 (43.2 %)
3 months mortality
73 (35.4 %)
23 (45.1 %)
50 (32.4 %)
Intracerebral hemorrhage associated with oral anticoagulants
Multivariate logistic regression analysis in the entire cohort to assess independent risk factors associated with poor outcome revealed independent predictive factors after correction for confounding factors: age (p = 0.001), initial ICH volume (p = 0.001) and ICH expansion (p = 0.004).
Hematoma size and expansion
Comparing initial ICH volume, IVH volume and ICH expansion in OAC-ICH and spont-ICH led to the following results: Baseline volume of ICH did not differ significantly between OAC-ICH and spont-ICH (p = 0.12). In contrast, OAC-ICH patients more frequently had intraventricular extension (p = 0.05) and the intraventricular hematoma was more extensive according to the Graeb score (p = 0.01). Similarly, OAC-ICH patients more often suffered from isolated primary IVH (p = 0.03, Table 1).
Determination of the effect of OAC on secondary hematoma expansion and quantitative intraventricular involvement
All patients (n = 152)
OAC-ICH (n = 32)
Spont-ICH (n = 120)
Initial ICH volume (mL): mean ± SD
21.9 ± 23.7
24.4 ± 25.5
21.2 ± 23.2
ICH volume follow-up (mL): mean ± SD
23.4 ± 25.7
27.5 ± 27.7
22.3 ± 25.2
18 (11.8 %)
4 (12.5 %)
14 (11.7 %)
Patients without follow-up CT were significantly older (77.9 vs. 69.4 years, p = 0.001), ICH volume was significantly larger (50.7 vs. 22.5 mL, p = 0.007), and Graeb score was significantly higher (3.0 vs. 1.6, p = 0.02). Moreover, NIHSS at baseline (23 vs. 14, p = 0.002) and premorbid mRS (2 vs. 1, p = 0.007) were significantly higher.
The major findings of our study are that (1) one-fourth of ICHs is related to OAC treatment. (2) Initial ICH volume does not differ between OAC-ICH and spont-ICH. (3) The rate of ICH expansion is similar in both groups (4) Extension of ICH into the ventricles on initial imaging and primary IVH were more frequent in OAC-ICH. (5) After correction for confounding factors there was a trend towards a worse outcome in OAC-ICH.
The high proportion of OAC-ICHs in our study is comparable with the limited number of previous, equally designed cohort studies [6, 7]. Performance at a university hospital may be suspected to bias patient selection to more severe strokes (e.g., OAC-ICH). However, our center is the primary provider for all acute strokes in the area supporting the representativeness of our data. Moreover, the proportion of ICH in the entire stroke cohort is compatible with population-based reports . Comparing our data with the prevalence rates of OAC-ICH reported in earlier studies (range from 12 to 18 %) [2, 18] rather suggests that the proportion of OAC-ICH has considerably increased over the last years . The most likely reason for this development is the increased use of OAC for stroke prevention in AF  for which the prevalence has increased in the last two decades . In particular, OAC are more frequently prescribed in elderly patients [2, 21] who carry a higher annual risk of both spont-ICH and OAC-ICH .
Similar to a previous study , initial ICH volume did not differ significantly between OAC-ICH and spont-ICH in the present study and we could rule out that this resulted from different time windows between symptom-onset and admission CT in both groups. However, differences of baseline clinical deficit and more frequent hemorrhage extension into the ventricles—an important prognostic factor for ICH in general—suggest, that the spontaneous course of OAC-ICH is more severe than in spont-ICH. A recent study on OAC-ICH that also included IVH, reported that OAC-ICH patients had a worse outcome that was attributed to baseline ICH, IVH volumes and ICH/IVH expansion .
Hematoma expansion in OAC-ICH and spont-ICH was substantially less frequent (11.8 %) in our cohort compared to previous case series and clinical trials (range 20 to 32 %) [6, 13, 23, 24]. One potential reason for this discrepancy is the considerably later median time from symptom onset to admission CT and the lack of standardized follow-up imaging in our study. Furthermore, ICH expansion is more likely in large hematomas  and a majority of ICH volume in our study was substantially smaller compared to previous reports [23, 24] (median, Table 1). Additionally, surgical interventions are more frequent in large hematomas leading to an underrepresentation in our follow-up cohort. Importantly, hematoma expansion was as frequent in the OAC-ICH group as in the spont-ICH group. A potential explanation for this could be the rapid and effective reversal of anticoagulation using repetitive bedside point of care monitoring of INR during repeated administration of PCC as described [13, 14]. Using this standardized in-house procedure, INR can be almost normalized (≤1.4) within about 30 min . Furthermore, octaplex—a four-factor PCC—was used for reversal therapy, which might have had an impact on INR normalization  We did not analyse intensity of OAC with respect to ICH expansion, however, a subgroup analysis comparing patients with OAC-ICH (INR ≥2) and spont-ICH did not affect our results significantly, although the proportion of ICH expansion in OAC-ICH compared to spont ICH numerically was larger (data not shown). This potentially indicates a higher risk of ICH expansion in patients with higher INR values. Nevertheless, our finding should be interpreted with caution because follow-up CT was performed only in 73.8 % of all ICH and only in 62.7 % of the OAC-ICH patients, and patients with and without follow-up CT differed which may have introduced a bias. Also, this analysis was not prespecified but exploratory.
Hematoma expansion has been observed over a time period of up to 7 days in previous studies, whereas the median time to follow-up CT of 18 h in our extension cohort was considerably shorter (Table 1). However, regarding hematoma expansion rates in patients with more than one follow-up CT yielded comparable results.
In previous reports, OAC treatment contributes to an unfavourable outcome in ICH patients, mediated in part by intensity of anticoagulation . We did not analyse intensity of OAC with respect to mortality, however, more than two-third of OAC-ICH occurred within or even below the therapeutic range, which parallels findings from previous reports . Similarly, in our cohort initial ICH volume, extension of ICH into the ventricles, secondary hematoma expansion and older age were predictive factors for worse outcome and mortality in the entire cohort, as previously described [6, 7, 28, 29]. Extension of ICH into the ventricles and intraventricular hematoma according to the Graeb score in our cohort was significantly more frequent/more important in the OAC-ICH patients, presuming a worse outcome in OAC-ICH patients. However, outcome according to the dichotomized modified Rankin scale was not statistically significant after correction of age, but there was a trend towards worse outcome in OAC-ICH.
Our study has some limitations. First, our observational study did not use a prospective follow-up imaging protocol. Secondly, therefore we could not include approximately one-fourth of patients for an analysis of potential hematoma growth because of surgical and medical interventions. Nevertheless, availability of follow-up CT scans compared to previous studies was high [9, 10]. Data analysis was exploratory and therefore not specifically powered to answer the questions concerning mortality and outcome parameters.
In conclusion, our data underlines the increasing challenges imposed by OAC-ICH, which presently already contributes one-fourth to all ICHs. Although the introduction of new OAC is expected to reduce the relative risk of ICH compared to VKA, interventional clinical trials are needed to establish appropriate management of OAC-ICH.
Conflicts of interest
On behalf of all authors, the corresponding author states that there is no conflict of interest.
- 9.Morgenstern LB, Hemphill JC 3rd, Anderson C, Becker K, Broderick JP, Connolly ES Jr et al (2010) Guidelines for the management of spontaneous intracerebral hemorrhage: a guideline for healthcare professionals from the American Heart Association/American Stroke Association. Stroke 41:2108–2129PubMedCrossRefGoogle Scholar
- 10.Steiner T, Kaste M, Forsting M, Mendelow D, Kwiecinski H, Szikora I et al (2006) Recommendations for the management of intracranial haemorrhage—part I: spontaneous intracerebral haemorrhage. The European Stroke Initiative Writing Committee and the Writing Committee for the EUSI Executive Committee. Cerebrovasc Dis 22:294–316PubMedCrossRefGoogle Scholar