Introduction
Anticoagulants are among the most widely used medications worldwide [
1]. In France, more than 3 million patients (5% of the population) received anticoagulant therapy in 2013 [
2]. These medications are essential for preventing or treating arterial and venous thrombotic events and other widespread cardiovascular conditions associated with significant morbidity, mortality and economic consequences [
3]. Heparin and its derivatives are usually administered during the acute phase of illness, while vitamin K antagonists (VKA), and more recently new molecules labeled “sdirect oral anticoagulants” (DOAC) are available for chronic treatment [
2,
4]. However, anticoagulant medication is associated with a high rate of adverse events. In a large study in the United States, anticoagulants were identified as the primary medication responsible for adverse drug-related events requiring emergency hospitalizations in older adults [
5]. Bleeding is the main life-threatening side effect of anticoagulant therapy, and varies according to the type of molecule [
3]. Patients treated with DOACs experience lower bleeding rates compared to those receiving VKAs. For instance, Steinberg et al. reported a major bleeding incidence of 3.3 per 100 patient-years for patients treated by DOAC, compared to 3.5 per 100 patient-years when patients received VKA [
6]. Similarly, in a multicenter cohort study in France, including patients older than 80 years, Hanon et al. reported significantly lower major bleeding risk in Rivaroxaban-treated patients (7.4/100 patient-years) compared with VKA-treated patients (14.6/100 patient-years) after multivariate adjustment [
7]. International guidelines have been published to help clinicians manage anticoagulant-related bleeding [
8‐
10], but their application remains an ongoing challenge [
11].
Moreover, most of the studies on anticoagulant-related bleeding have been conducted in emergency rooms and medical departments. A retrospective multicenter study encompassing cases of oral warfarin versus DOACs-associated bleeding events involving individuals older than 66 years, and presenting to five tertiary care hospitals, found that in-hospital mortality was lower following DOAC bleeding events (9.8% vs. 15.2%) [
12]. However, there is limited information available regarding the life threatening bleeding events requiring intensive care unit (ICU) admission. This study aimed to describe the epidemiological and clinical characteristics of patients admitted to the ICU for severe extracranial anticoagulant-related bleeding, and to investigate prognosis factors as well as real life therapeutic management.
Methods
We conducted a multicenter retrospective observational study in five ICUs across different locations in France including University Hospitals in Paris, Strasbourg, Nantes, Lille and Poitiers. The inclusion period spanned 12 years (from January 2007 to December 2018) to collect data before and after DOACs commercialization in France. The data were queried using the ICD-10 nomenclature [
13]. We considered all ICD-10 codes indicating potential acute or chronic bleeding diseases and all medical or interventional procedures (Personal list, Additional file
1: Table S1) related to the management of potential bleeding (transfusion, embolization, etc.) for patients receiving therapeutic anticoagulant therapy. Each individual medical record was manually reviewed for patient inclusion. Patients with intracranial bleeding were not included in this study as our focus was to describe severe bleeding responsible for hemorrhagic shock with organ failure. The management and the outcome of intracranial hemorrhage are not dependent on blood volume loss and prognosis factors applicable to intracranial bleeding may not be directly relevant in other cases [
14]. General characteristics of patients at admission were collected, including demographic data, biological data, diagnoses, severity of illness evaluated by the Sequential Organ Failure Assessment (SOFA) score [
15] and Simplified Acute Physiology Score II (SAPS II) [
16]. Additionally, the initial bleeding site and the therapeutic management were recorded. The data are reported in accordance with the STROBE guidelines [
17].
Statistical analysis
Quantitative data are expressed as mean ± standard deviation, qualitative variables as number of recurrences (percentage). Qualitative variables were compared using a Chi2 test. Quantitative variables were compared with a Student's t test or a Mann–Whitney test according to their normal or non-normal distribution (hypothesis of normality systematically tested). The significance threshold was set bilaterally to p < 0.05. Multivariate analysis consisted of a logistic regression including variables available at ICU admission with a p-value < 0.10 in univariate analysis and variables previously identified as predictive for bleeding in patients receiving therapeutic anticoagulant therapy were forced into the model [
18], while collinear variables were excluded. As missing data accounted for less than 5% of the dataset, analyses were performed on the complete cases. All statistical analyses were performed using the R software (v 2.12.0;
http://cran.rproject.org).
This observational retrospective study had the ethical approval of the SRLF commission (CE SRLF 23-032). All patients and families were informed through the admission leaflet that anonymous data could be used for academic research.
Discussion
This multicenter retrospective study showed that over the past decade the number of patients admitted to ICU for anticoagulant-related extracranial bleeding has significantly increased. These patients experienced severe organ failure, requiring frequent organ support therapy and ultimately faced a high mortality rate. One concerning finding of our study is the weak compliance with guidelines for managing anticoagulant-related bleeding. Enhancing adherence to these guidelines could be crucial to improve patient outcomes.
In our cohort, critically ill patients admitted to ICU for anticoagulant-related extracranial bleeding accounted for around 1% of ICU patients, consistent with previous works [
2,
19]. The increase in ICU admission for severe bleeding complications can be imputed to the growing use of anticoagulants in France [
2] driven by the aging population and broader indications for anticoagulants, especially for atrial fibrillation which accounted for 60% of anticoagulant prescriptions in our cohort [
20]. DOAC-related bleedings were identified starting from 2012, 2 years after their introduction to the market. The continuous increase of DOAC prescription suggests that physicians should receive specific training to effectively manage bleeding occurring with dabigatran, rivaroxaban or apixaban-therapy. Previous studies have shown that among DOACs, dabigatran was at higher risk of bleeding events [
6,
21,
22]. In our study dabigatran represented only 1% of anticoagulant related bleeding accidents. We also observed an encouraging decrease in the number of patients admitted with bleeding occurring during VKA-heparin bridge in recent years. This could be explained by the implementation of recent expert international recommendations that propose limiting VKA-heparin bridge for elective surgery [
23].
Consistently with the work of Hauguel et al. [
24] patients were elderly, with a high prevalence of cardiovascular risk factors and comorbidities. In our cohort, hypertension was not found to be independently associated with ICU mortality. This contrasts with the findings of Goodman et al. who observed a significant relationship between hypertension and the outcome in bleeding among VKA- and rivaroxaban-treated patients [
25]. However, Lauzier et al., who focused on heparin-treated patients, did not report such an association [
26]. Here, one third of patients were treated by antiplatelet medication in addition to anticoagulants, but this combination was not associated with in-ICU mortality. Similar results were reported by Beyer-Westendorf et al. [
27] in selected rivaroxaban-treated patients that frequently received antiplatelet agents (16%), with no association with bleeding severity.
Bleeding sites were different according to type of anticoagulant drug. In our cohort and in other studies [
21‐
27], VKA or DOACs-treated patients predominantly experienced bleeding in the digestive tract. The mechanisms underlying such observations is unclear. Some authors have speculated that the preferential localization of bleeding in the digestive tract with DOACs may be related to their activity on the P-glycoprotein transporter [
28]. In contrast heparin-treated patients presented with bleeding in soft tissues such as the psoas muscle or the abdominal wall. In the multicenter study by Litjos et al. [
29] which examined psoas hematomas managed in the ICU, almost half of the patients were treated by heparin. For abdominal anterior wall hematomas, Hauguel et al. described iatrogenic and traumatic mechanisms due to the puncture of hypogastric artery branches during subcutaneous heparin injection [
24].
Upon admission, patients in our study exhibited high severity, frequently presenting with organ failure and requiring both vasopressors and invasive mechanical ventilation. These findings are consistent with the monocentric study by Hauguel et al. [
24]. The overall in-ICU mortality from anticoagulant-related extracranial hemorrhagic accidents admitted to ICU was 27% in our cohort, which aligns with the all-cause mortality in French ICUs (23%) [
30].
Factors found to be associated with mortality in our cohort are very difficult to compare to published works because the definition of severe bleeding varies from one study to another [
31]. Here, these factors reflected the severity of acute circulatory failure (vasopressor need, hyperlactatemia, and impaired consciousness). Furthermore, prolonged aPTT was also independently associated with mortality. Whether this is due to the anticoagulant drug or to hemorrhage-associated liver failure or consumption coagulopathy is uncertain.
Finally, one of the main findings of this work is the lack of compliance with expert recommendations on anticoagulation reversal [
8‐
10]. For instance, only 60% of patients with severe VKA-associated hemorrhagic accidents received PCC, and only 22% of patients admitted for heparin-associated bleeding received protamine. In the emergency ward, Tremey et al. also reported inappropriate use of PCC in these patients, which improved after a teaching program [
11]. Starting from 2012 bleeding accidents associated with DOACs became common in ICU. Before Idarucizumab and Andexanet alfa became available, PCC administration was recommended. In our cohort, PCCs were only used in 29% of cases and fresh frozen plasma was used in 29% of cases. These results underscores the need for developing of teaching programs for anticoagulant-related bleeding management in ICUs.
We did not find any significant relationship between prognosis and time to reversal and/or hemostatic procedures but such an association cannot be excluded. Indeed the retrospective nature of our work is a limitation and despite the exhaustive analysis of each medical record, we cannot rule out the possibility of missing data, particularly regarding reversal therapy before ICU admission. In addition, the ICD-10 code-based approach used in this study to identify the cases is original and not validated by previous studies. Finally, our study included only 5 medical ICUs, thereby limiting the generalizability of its results, particularly for post-operative patients. Therefore, future prospective research should be conducted to validate our observations.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.