Abstract
Background
Haemorrhage and blood product usage are common in venoarterial extracorporeal membrane oxygenation (VA ECMO) and associated with increased mortality.
Methods
A prospective, investigator-initiated, longitudinal observational cohort study on major haemorrhagic events in 12 ECMO centres from 3 continents for three predefined subgroups (VA ECMO initiated during cardio-pulmonary resuscitation (ECPR), after cardiothoracic surgery (CTS), for cardiogenic shock (CS)). The aim was to describe haemorrhagic complications as well as transfusion practice and anticoagulation for the whole population as well as the subgroups. In addition, independent baseline predictors for red blood cell (RBC) transfusions were evaluated.
Results
545 prospective patients were included between 2019 and 2022 (ECPR 149, CTS 169, CS 227). Hospital mortality was 46%. Over 2796 days 406 major haemorrhage events in 286 (52%) patients were recorded. CTS and ECPR patients had more frequent events occurring earlier in their course. 88% received RBC transfusions (1.27 (95%CI 1.22 – 1.31) units/day) with significantly more transfusions for CTS and ECPR patients. Platelet transfusion rates were highest in the CTS group (0.58 (95%CI 0.53–0.64) units/day). Haemoglobin and platelet count prior to transfusion were independent of subgroups and averaged (78 g/L (IQR 73, 84), 58 × 10^9/L (IQR 37, 85), respectively). However, platelet count prior transfusion was only marginally higher on days with major haemorrhage (74 × 10^9/L (IQR 50, 104). Systemic anticoagulation was started within the first 24 h in 83% (95% CI 80–87%) of patients, most frequently in CS patients (90%, CI 85–95%). Independent baseline predictors for RBC transfusion were ECPR (IRR 1.50, 95%CI 1.19–1.89) and prior use of antiplatelets (IRR: 1.43, 95%CI 1.13–1.80). Myocarditis and pulmonary embolism were associated with a lower rate of transfusion when compared to myocardial infarct (IRR: 0.57 (0.37–0.89), IRR: 0.67 (0.45–1.00, respectively).
Conclusion
Haemorrhagic complications differ in clinical subgroups and RBC transfusion exposure is by far higher than in other critically unwell populations. We identified ECPR and antiplatelet therapy as additional predictors. Transfusion practice for RBC and platelets is variable and does not always follow international guidelines.
Introduction
Venoarterial extracorporeal membrane oxygenation (VA ECMO) is the most frequently used mechanical circulatory support (MCS) for severe forms of cardiogenic shock when refractory to conventional measures such as inotropes [1]. Despite improvements in technology and growing clinical experience, mortality among patients supported with ECMO remains high at 30–70% and major haemorrhagic events are among the most frequent and life-threatening complications [2–4].
Significant perturbations in haemostatic function are related to underlying pathology, blood-surface contact, pump mechanics and the requirement for anticoagulation [5]. Cardiogenic shock with or without cardiac surgery contributes to coagulopathy and results in platelet dysfunction and consumption, hyperfibrinolysis and disseminated intravascular coagulation (DIC). Transfusion of blood products is frequent but lacks precise guidance [6].
Both haemorrhage and transfusion are associated with poor outcome in the general ICU population [7, 8]. Complications such as transfusion-related acute lung injury, transfusion-related circulatory overload and transfusion-related immunomodulation are well described [9]. The transfusion of coagulation factors also increases the risk of ECMO-circuit malfunction and thromboembolism. A systematic literature review of 159 mostly retrospective and single centred observational studies with a more than moderate risk of bias suggest that major haemorrhage is present in almost half of all patients treated with VA ECMO [2]. Raasveld et al. have documented a high requirement for red blood cell (RBC) transfusion in VA ECMO patients and could associate these with transfusion practice in 16 international ECMO centres [10]. Patients received a median total of eight RBC units (IQR 3–17) over a median of only 5 days.
VA ECMO is used in various very distinct clinical settings involving severe cardiogenic shock. Initiating VA ECMO after cardiothoracic surgery (CTS) is a common scenario which has intrinsic haemorrhagic complication risks but also specific demography, complications and outcome [11]. Similarly, VA ECMO initiation during cardiopulmonary resuscitation (ECPR), initiated as a rescue therapy during refractory cardiac arrest, is associated with high mortality and specific challenges during the establishment of VA ECMO [12]. In contrast, VA ECMO may also be indicated in cardiogenic shock outside these two categories and hence may present with a different haemorrhagic risk profile. No prospective detailed multicentre data exists to compare these subgroups and investigate the impact of other baseline characteristics as well as its implication for transfusion practice.
We designed the International Observational Study on Blood Management for Mechanical Circulatory Support using Extracorporeal Membrane Oxygenation (OBLEX, ClinicalTrials.gov: NCT03714048) to prospectively collect granular data on major haemorrhagic events and blood product usage in VA ECMO patients from high-volume international centres across the world. We aimed to identify baseline risk factors for red cell transfusion and to describe haemorrhagic pattern, transfusion practice and anticoagulation in subpopulations of cardiogenic shock defined by ECPR, CTS or neither.
Method
Study design and participants
OBLEX is a prospective, investigator initiated, longitudinal observational cohort study collecting detailed information on blood management and haemorrhagic events in VA ECMO patients. In Australia the study was embedded in the national ECMO registry (EXCEL) and internationally endorsed by the International ECMO Network (ECMONet) [13, 14].
A convenient sample size of greater than 500 was defined in the absence of solid baseline data at the time of study design. Participating centres were asked to screen all consecutive patients who received VA ECMO for mechanical circulatory support using a temporary device containing an oxygenator and a blood pump (study definition for VA ECMO) were included. The first patient was enrolled in November 2019 but centres were allowed to initiate data sampling at individual time points based on local regulations and circumstances. Study centres were asked to enrol to a maximum of 75 consecutive patients to allow a larger number of centres to participate. The duration of the enrolment period per centre was determined by the time taken to enrol 75 consecutive patients or the study target sample size was reached (whichever came first). Patients were excluded if VA ECMO was initiated for respiratory support only, were < 18 years old, received VA ECMO for > 24 h in a non-participating centre, were previously enrolled in the study during the same hospital admission or VA ECMO was managed outside the intensive care unit only. Co-enrolment in other studies was allowed except if a randomized intervention involved anticoagulation or transfusion strategies.
Three pre-specified subgroups were analysed: Patients who received ECPR (ECPR group), patients who underwent CTS in the 24 h prior to VA ECMO but had no ECPR (CTS group), and patients who had cardiogenic shock in the absence of ECPR and cardiothoracic surgery (CS group).
Ethics approval was obtained in Australia (St Vincent’s Health, Sydney Human Research Ethics Committee, HREC/18/SVH/202), which granted a waiver of consent. Non-Australian centres acquired local approval based on their legal requirements and waiver of consent was granted. Human Ethics and Consent to Participate declarations: not applicable.
This study is reported according to the Strengthening the Reporting of Observational Studies in Epidemiology statement [15].
Procedures
Baseline characteristics, daily data on transfusions anticoagulation and haemorrhagic events were collected for the first seven days of VA ECMO or until cessation of VA ECMO (whichever came first).
Baseline data included demographics, diagnostic classification and pre-specified potential predictors for RBC transfusion during VA ECMO in the 24 h prior initiation of VA ECMO. These factors were pre-specified based on the literature, expert options, and their prevalence in the study. Diagnostic classification was based on EXCEL data definitions which allows to identify reason best describing the need for VA ECMO from a list of predefined diagnosis including perioperative support [14]. However, another diagnose (like acute myocardial infarct) could be used even during the first 24 h after surgery if this was more indicative of the reason to initiate VA ECMO. If no specific reason for acute cardiomyopathy could be identified, acute (non-ischaemic) heart failure was offered.
For each day on VA ECMO the presents of a major haemorrhage was identified. Haemorrhagic events were categorised according to the Bleeding Academic Research Consortium (BARC) criteria and a BARC class > = 3 was classified as major haemorrhage (Table 1) [16]. Haemorrhage sites were categorised by adapting extracorporeal life support organisation (ELSO) definitions [17]. Additionally, any interventions needed to manage these events were recorded. Systemic anticoagulation was defined as any anticoagulation given to achieve a target coagulation marker range.
Table 1.
Bleeding definition
| Type | Definition |
|---|---|
| 0 | No bleeding |
| 1 | Non-actionable bleeding |
| 2 | Requiring nonsurgical, medical intervention or prompting evaluation only |
| Major haemorrhage | |
| 3a | Hb drop of 30 to 50 g/l related to bleed, or any transfusion with overt bleeding |
| 3b | Hb drop > 50 g/l related to bleed, cardiac tamponade, bleeding requiring surgical intervention OR bleeding requiring intravenous vasoactive agents |
| 3c | Intracranial haemorrhage, intraspinal haemorrhage (confirmed by autopsy or imaging or lumbar puncture) |
| 4 | Procedure related bleeding (within 48 h of procedure): perioperative intracranial bleeding, reoperation for controlling of bleeding, transfusion of > = 5 Units of RBC |
| 5a | Probably fatal bleeding based on clinical suspicion |
| 5b | Definite fatal bleeding (autopsy, or imaging confirmation) |
Bleeding definitions categorised according to the Bleeding Academic Research Consortium (BARC) criteria used for each individual day observed on VA ECMO (Hb = haemoglobin, RBC = red blood cell transfusion).
Survival status was recorded at discharge from the study hospital. Data entry was facilitated through a secure online portal (REDCAP) [18].
Statistical analysis
For descriptive purpose, medians and interquartile ranges (IQR) were reported for continuous variables and counts and percentages for categorical variables. To describe major haemorrhage, transfusion and anticoagulation events, we reported incidence rates (number of events per VA ECMO day). A VA ECMO day was defined as a single day a participant was treated with ECMO during the observation period.
To identify predictors for RBC transfusion, we modelled the total number of transfusions per patients using a multivariable negative binomial regression. This model estimated incidence rate ratios (IRRs), which represent the change in the rate of transfusion per one-unit increase in continuous predictors, or per one-category change in categorical predictors, while holding all other variables constant. To account for different time spent on VA ECMO, we included the (logarithm) length of follow-up per patient as the offset term. Predictors were age, gender, body mass index, continent, surgery prior to VA ECMO, ECPR, renal replacement therapy, any open cannulation, any heparin bolus for cannulation, massive transfusion prior to VA ECMO, prior use of low-molecular-weight heparin and antiplatelet agents, prior platelet transfusion, transfusion of coagulation factors, and the diagnostic groups of myocardial infarct, pulmonary embolism and myocarditis. These factors were pre-specified based on the literature, expert options, and their prevalence in the study.
Due to the low level of missingness of these selected potential predictors (< 1% for each potential predictor), we removed patients with missing data from the analysis. All analyses were performed using statistical software R version 4.4.0 [19].
Role of funding source
This is an investigator-initiated study, not supported by external funding.
Results
Five hundred and sixty subsequent patients from 12 ECMO centres were enrolled from November 2019 to December 2022, when the target sample size was reached. The median number of cases per centre was 46 (IQR 16, 75). The median annual VA ECMO caseload reported by the participating centres was 32 (IQR 18, 125). Fifteen patients were excluded for missing mortality or daily data. As a result, 545 patients from three continents were included (Australia 267, Europe 219, North America 59). Forty-two percent were in the CS group, 31% in the CTS group and 27% had ECPR. Hospital mortality was 46%. The number of patients who remained on VA ECMO decreased from 545 on day 1 to 218 on day 7 and a total of 2796 days on VA ECMO were studied.
Baseline characteristics
Baseline characteristics before initiation of VA ECMO are summarised in Table 2. Median age was 57 (IQR 45, 65), 33% were female. Ten percent had decompensated chronic cardiomyopathy noted as the reason to initiate VA ECMO. For the remaining 90% the most common diagnose was acute myocardial infarction (AMI, 28%).
Table 2.
Baseline characteristics and hospital outcome
| Characteristic | Overall (n = 545) |
CS group (n = 227) |
CTS group (n = 169) |
ECPR group (n = 149) |
|---|---|---|---|---|
| Demographics (n (%) or median (IQR) | ||||
| Age (years) | 57 (45, 65) | 54 (43, 62) | 62 (52, 71) | 56 (45, 65) |
| Female | 182 (33%) | 69 (30%) | 68 (40%) | 45 (30%) |
| BMI (kg/m2) |
27.2 (24.2, 30.9) |
27.1 (24.2, 30.6) |
27.3 (24.6, 31.2) |
27.1 (24.2, 31.2) |
| Diagnostic group (n (%)) | ||||
| Acute myocardial infarct | 152 (28%) | 73 (33%) | 23 (14%) | 56 (39%) |
| Perioperative support | 76 (14%) | 0 (0%) | 67 (40%) | 9 (6.0%) |
| Pulmonary embolism | 43 (8.1%) | 20 (8.9%) | 6 (3.6%) | 17 (12%) |
| Myocarditis | 41 (7.7%) | 31 (14%) | 1 (0.6%) | 9 (6.2%) |
| Septic cardiomyopathy | 14 (2.6%) | 9 (4.0%) | 4 (2.4%) | 1 (0.7%) |
| Arrhythmia | 13 (2.4%) | 4 (1.8%) | 1 (0.6%) | 8 (5.4%) |
| Toxic cardiomyopathy | 13 (2.4%) | 7 (3.1%) | 0 (0%) | 6 (4.0%) |
| Acute (non-ischaemic) heart failure | 128 (23%) | 44 (19%) | 53 (31%) | 31 (21%) |
| Decompensated chronic cardiomyopathy | 54 (10%) | 35 (16%) | 10 (6.1%) | 9 (6.2%) |
| Missing diagnosis | 11 (2.0%) | 4 (1.8%) | 4 (2.4%) | 3 (2.0%) |
| Cannulation characteristics (n (%)) | ||||
| Open cannulation | 86 (16%) | 21 (9.3%) | 53 (31%) | 12 (8.1%) |
| Heparin bolus | 397 (74%) | 155 (70%) | 135 (80%) | 107 (73%) |
| Conditions present 24 h prior to cannulation (n (%)) | ||||
| Cardiac arrest not followed by ECPR | 93 (17%) | 73 (32%) | 20 (12%) | 0 (0%) |
| Other MCS | 88 (16%) | 44 (19%) | 35 (21%) | 9 (6.0%) |
| Renal replacement therapy | 51 (9.4%) | 27 (12%) | 15 (8.9%) | 9 (6.0%) |
| Massive transfusion | 34 (6.3%) | 1 (0.4%) | 29 (17%) | 4 (2.7%) |
| Platelet transfusion | 56 (10%) | 2 (0.9%) | 47 (28%) | 7 (4.7%) |
| Coagulation factor transfusion | 52 (9.6%) | 4 (1.8%) | 44 (26%) | 4 (2.7%) |
| Low molecular weight heparin | 131 (24%) | 45 (20%) | 58 (35%) | 28 (19%) |
| Antiplatelet medications | 144 (27%) | 71 (32%) | 40 (24%) | 33 (22%) |
| Hospital Outcome | ||||
| Hospital mortality | 250 (46%) | 73 (32%) | 80 (47%) | 97 (65%) |
Demographics, diagnosis leading to cardiogenic shock and VA ECMO initiation, VA ECMO cannulation characteristics, conditions possibly associated with increased haemorrhagic risk and hospital mortality are illustrated for the whole cohort and the pre-defined subgroups
ECPR group = Patients who received ECPR (VA ECMO initiation during cardiopulmonary resuscitation), CTS group = patients who underwent cardiothoracic surgery in the 24 h prior to VA ECMO but had no ECPR, CS group = patients who had cardiogenic shock in the absence of ECPR and major surgery, BMI = body mass index, MCS = mechanical circulatory support. Percentages calculated as fraction of available data
The subgroup of CTS was older and more frequently female, and the majority (76%) had no other specific diagnosis leading to the initiation of VA ECMO. Transfusions prior to VA ECMO and an open cannulation were also more common. ECPR patients were less likely to receive other forms of MCS or renal replacement therapy prior to initiation of VA ECMO. Hospital mortality was highest in the ECPR group followed by CTS.
Major haemorrhage and transfusion practice
A total of 406 major haemorrhage events in 286 (52%) patients were recorded during the observation period. By day 5 (95% CI 4–7), half of the cohort experienced at least 1 major haemorrhage. Patients in the ECPR and in the CTS group reached this time point earlier (median time: 3 days) compared to those in the CS group (median 7 days). Patients in the CTS group had a higher haemorrhage rate, followed by ECPR.
The most common bleeding sites were thoracic cavity and cannula site. Thoracic haemorrhage rate per day was significantly more common not only in the CTS group (0.19, 95% CI 0.16–0.22) but also in the ECPR group (0.14, 95% CI 0.11–0.17) when compared to the CS group (0.07, 95% CI 0.05–0.08). The incidence of thoracic haemorrhage decreased from 26 to 7% over the week while haemorrhage from other sites remained stable.
Surgical intervention and drain insertion were required to manage major haemorrhage on 142 and 83 occasions respectively. However, on 58% of major haemorrhage days (236 days) no procedures were recorded (Table 3).
Table 3.
Patients characteristics during the observation period by subgroup
| Overall (n = 545) | CS group (n = 227) | CTS group (n = 169) | ECPR group (n = 149) | |
|---|---|---|---|---|
| Major haemorrhage | ||||
| % of patients with at least one major haemorrhage | 286 (52%) | 106 (47%) | 100 (59%) | 80 (54%) |
| Median survival time (days) to first major haemorrhage (95% CI) | 5.0 (4.0–7.0) | 7.0 (6.0, —) | 3.0 (2.0–5.0) | 3.0 (2.0, —) |
| Incidence rate (number of events per VA ECMO day, 95% CI) |
0.15 (0.13–0.16) |
0.12 (0.10–0.14) |
0.18 (0.15–0.21) |
0.15 (0.13–0.19) |
| Number of patients (%) with major haemorrhage by site of haemorrhage and incidence rate by site (number of events per VA ECMO day, 95% CI) | ||||
| Thoracic | 190 (35%) | 65 (29%) | 72 (43%) | 53 (36%) |
|
0.12 (0.11–0.14) |
0.07 (0.05–0.08) |
0.19 (0.16–0.22) |
0.14 (0.11–0.17) |
|
| Cannula | 135 (25%) | 53 (23%) | 50 (30%) | 32 (22%) |
|
0.13 (0.11–0.14) |
0.12 (0.10–0.14) |
0.14 (0.12–0.17) |
0.12 (0.10–0.15) |
|
| GI | 41 (7.5%) | 12 (5.3%) | 16 (9.5%) | 13 (8.7%) |
|
0.03 (0.02–0.04) |
0.02 (0.01–0.03) |
0.03 (0.02–0.05) |
0.04 (0.03–0.06) |
|
| Airway | 30 (9.5%) | 8 (3.5%) | 13 (7.7%) | 9 (6.0%) |
|
0.04 (0.03–0.05) |
0.03 (0.02–0.05) |
0.04 (0.03–0.06) |
0.05 (0.03–0.06) |
|
| Lung | 20 (3.7%) | 2 (0.9%) | 10 (5.9%) | 8 (5.4%) |
|
0.02 (0.01–0.02) |
0.01 (0.00–0.02) |
0.03 (0.02–0.04) |
0.02 (0.01–0.03) |
|
| Muscle/limb | 24 (4.4%) | 8 (3.5%) | 10 (5.9%) | 6 (4.0%) |
|
0.02 (0.01–0.02) |
0.01 (0.01–0.02) |
0.02 (0.01–0.04) |
0.01 (0.01–0.03) |
|
| CNS | 6 (1.1%) | 3 (1.3%) | 1 (0.6%) | 2 (1.3%) |
|
0.00 (0.00–0.01) |
0.00 (0.00–0.01) |
0.00 (0.00–0.01) |
0.01 (0.00–0.01) |
|
| Procedures to manage major haemorrhage | ||||
| Surgery | 142 (35%) | 45 (31%) | 61 (40%) | 36 (33%) |
| Drain insertion | 83 (20%) | 18 (12%) | 39 (25%) | 26 (24%) |
| Endoscopy | 11 (2.7%) | 5 (3.4%) | 3 (2.0%) | 3 (2.8%) |
| Other interventional procedure | 5 (1.2%) | 2 (1.4%) | 2 (1.3%) | 1 (0.9%) |
| Major haemorrhage days with none of the above interventions n, (%) | 236 (58%) | 89 (61%) | 83 (54%) | 64 (59%) |
| Transfusions received over the observational period (n (%)) and transfusion incidence rate (number of events per VA ECMO day, 95% CI) | ||||
| RBC | 481 (88%) | 191 (84%) | 159 (94%) | 131 (88%) |
|
1.27 (1.22–1.31) |
1.00 (0.95–1.06) |
1.53 (1.45–1.62) |
1.40 (1.32–1.49) |
|
| Platelets | 309 (57%) | 125 (55%) | 106 (63%) | 78 (52%) |
|
0.47 (0.45–0.50) |
0.38 (0.34–0.41) |
0.58 (0.53–0.64) |
0.49 (0.44–0.55) |
|
| Transfusion threshold | ||||
| Hb (g/L, median (IQR)) | 78 (73, 84) | 78 (73, 83) | 78 (73, 83) | 79 (75, 86) |
| Platelet count (× 10^9/L, median (IQR)) | 58 (37, 85) | 53 (39, 83) | 63 (34, 86) | 57 (35, 85) |
| Platelet transfusion threshold by days with major haemorrhage present (× 10^9/L, median (IQR)) | ||||
| No major haemorrhage | 49 (32, 71) | 47 (37,68) | 58 (30, 80) | 43 (30, 58) |
| Major haemorrhage | 74 (50, 104) | 67 (50, 93) | 75 (46, 102) | 83 (58, 102) |
| Systemic anticoagulation (AC) | ||||
| Median time (days) to first systemic anticoagulation (95% CI) | 1.0 (1.0–1.0) | 1.0 (1.0–1.0) | 1.0 (1.0–1.0) | 1.0 (1.0–1.0) |
| Incidence rate of AC (number of days under AC per VA ECMO day, 95% CI) |
0.83 (0.80–0.87) |
0.90 (0.85–0.95) |
0.79 (0.74–0.86) |
0.76 (0.70–0.83) |
Descriptive characteristics of major haemorrhage, transfusion practice and anticoagulation are illustrated for the entire cohort and pre-defined subgroups. Incidence, time to event and haemorrhage rate are shown as well as bleeding sites and management. Transfusion rates in units per patient day and the known Hb and platelet count prior to prescription of a transfusion (threshold) demonstrate transfusion practice and systemic anticoagulation targeting a specific coagulation parameter are described as time to first initiation and rate per VA ECMO day
ECPR group = Patients who received ECPR (VA ECMO initiation during cardiopulmonary resuscitation), CTS group = patients who underwent cardiothoracic surgery in the 24 hours prior to VA ECMO but had no ECPR, CS group = patients who had cardiogenic shock in the absence of ECPR and major surgery. RBC= red blood cells, Hb=haemoglobin
Four-hundred-and-seventy-seven patients (88%) received RBC transfusions during the observation period. A total of 3,539 units of RBC transfusion were recorded on 1,160 days of the 2,796 VA ECMO days (41%). More than 47% of all RBC transfusions (1669 units) were administered over the first 24 h after initiation of VA ECMO. During the observation period, the transfusion rate was 1.27 (95% CI 1.22 – 1.31) RBC units per person-day. The CS group had a significantly lower rate of RBC transfusion compared to the other groups (1.00 (95% CI 0.95–1.06)).
The average haemoglobin (Hb) prior to RBC transfusion was consistent at 78 g/L with no difference between subgroups (Table 3) or by bleeding status although marked variations in practice was noted (Fig. 1).
Fig. 1.
Histogram of haemoglobin measurement prior to red blood cell (RBC) transfusion. A total of 2069 episodes were recorded. The majority of haemoglobin values triggering an RBC transfusion were above the recommended restrictive transfusion practice (median 78 g/L, IQR 73, 84). Overflow bin for haemoglobin < 50 g/L and > 120g/L used
A total of 1315 platelet transfusions were given during the study period to 312 patients (57%). Platelet transfusion rates were about 1/3 of RBC rates, with the highest rate observed in the CTS group (0.58 (95% CI 0.53–0.64) units per person-day), followed by ECPR (0.49 (95%CI 0.44–0.55). Overall, the median number of platelet units transfused per patient was 1 (IQR 0, 3). The median transfusion trigger did not differ across the subgroups, but the IQR was relative wide compared to the median suggesting substantial variability in transfusion triggers across clinical practice (Table 3, Fig. 2).
Fig. 2.
Histogram of platelet count in × 10^9/L prior to platelet transfusion. A total of 729 transfusion episodes were recorded. Overflow bin used for platelet count > 200 × 10^9/L. See Table 3 for differences in Haemorrhage status
The platelet count recorded before a transfusion occurred was 74 × 10^9/L (IQR 50, 104) on days with major haemorrhage and 49 (IQR 32, 71) when this was not the case. This difference was noted in all groups (Table 3).
Systemic anticoagulation was used at a rate of 83% (95% CI 80–87%) but more frequently in CS patients (90%, CI 85–95%) (Table 3).
Systemic anticoagulation was started during the first 24 h after initiation of VA ECMO in 392 of 545 patients (median days to initiate 1, 95% CI 1–1) and continuous unfractionated heparin was used most frequently during 2151 ECMO days (77%).
Predictors of RBC transfusion
Risk factors associated with a higher rate of RBC transfusion were ECPR (IRR 1.50, 95% CI 1.19–1.89) and prior use of antiplatelets (IRR: 1.43, 95% CI 1.13–1.80). Compared to a diagnosis of AMI, a diagnosis of myocarditis and pulmonary embolism were associated with a lower rate of transfusion (IRR: 0.57 (95% CI 0.37–0.89 and IRR: 0.67 (95% CI 0.45–1.00, respectively) (Fig. 3).
Fig. 3.
Prognostic factors for the total number of RBC transfusions during the first 7 days of VA ECMO. Results of the negative binomial models, looking at risk factors for RBC transfusions. Prior use refers to the 24 h preceding VA ECMO. Risk factors associated with a higher rate of RBC transfusion were ECPR (IRR 1.50, 95%CI 1.19–1.89) and prior use of antiplatelets (IRR: 1.43, 95%CI 1.13–1.80) but not surgery (IRR: 0.98, 95%CI 0.98–1.26). Compared to a diagnosis of AMI, a diagnosis of myocarditis and pulmonary embolism were associated with a lower rate of transfusion (IRR: 0.57 (0.37–0.89 and IRR: 0.67 (0.45–1.00, respectively). RBC = red blood cell, ECPR = VA ECMO initiation during cardiopulmonary resuscitation, LMWH = low molecular weight heparin. Patient with incomplete baseline data were removed
Discussion
To our knowledge, this is the largest prospective cohort study focussing on the management of haemorrhage and transfusion practice in patients receiving VA ECMO. The main findings were: First, more than half of the cohort experienced at least one major haemorrhage. Second, major haemorrhage was associated with high blood cell transfusions usage which was observed not only in patients post cardiotomy but in all predefined subgroups. Third, in addition to the predefined subgroups, the pre-VA ECMO use of antiplatelets and the diagnostic groups influenced transfusion risks. However, systemic anticoagulation is universally initiated early and rarely withheld for prolonged periods, while transfusion practice for RBC and platelets shows large variability with only small differences between subgroups.
We found major haemorrhage present in 52% of patients and 15% of observed days on VA ECMO. The PROTECMO study on blood management in 604 patients who received venovenous (VV) ECMO reported any haemorrhage in 56% of patients and a rate of 16.5% of all ECMO days [20]. However, 60% of these were classified as minor haemorrhage. This difference may be partially due to a higher exposure to major surgery prior to ECMO (31% versus 8%) in our cohort. Nevertheless, we found higher rates and incidence of haemorrhage also in the subgroups not exposed to surgery, suggesting that shock-induced coagulopathy as well as higher rates of anticoagulation and antiplatelet medications might have played a role. This is further supported by a study on 959 patients on VA ECMO were acute haemorrhagic anaemia was recorded in 68% of cases and a systematic review of 159 studies which found a pooled estimate of 40% of major bleeding events (95% CI 36–44) [2, 21].
However, haemorrhage definitions remain inconsistent in the literature. We have used the BARC criteria as it aims to include risk scenarios after cardiac procedures as well as cardiac surgery and provides an ordinal scale [16]. For this study, we have classified a BARC category of > = 3 as major haemorrhage. As this definition is strict, we assume that our incidence of major haemorrhage might have even underestimated the incidence compared with other definitions that could have been used. To compare across studies, RBC transfusion rates may be used as a surrogate.
The most common sites of haemorrhage were the thoracic cavity and cannula insertion sites. Surgery or other procedures to control haemorrhage were reported in 45% of cases which may indicate that medical management was commonly sufficient to control haemorrhage. This is also supported by the fact that platelet transfusions were frequently but not universally used during major haemorrhage events. Decision making processes to manage major haemorrhage during ECMO is not well understood.
Compared with patients randomized in RBC transfusion threshold trials on non-ECMO patients, our cohort had an exposure to RBC exceeding patients in the liberal transfusion arm. In the MINT trial on patients with anaemia and myocardial infarct the mean transfused RBC over the entire hospitalisation was 2.5 ± 2.3 (liberal transfusion arm) versus 0.7 ± 1.6 (restrictive arm) units while we measured a rate of 1.27 units over a single ECMO day [22]. In patients after cardiac surgery the median was 3 units (IQR, 2, 5) in the liberal group and 2 units (IQR, 1, 4) in the restrictive group while we found 5 units (IQR: 2, 9) [23]. In comparison to VV ECMO, the proportion of patients who received RBC transfusion overall and per day was higher (88% over 7 days vs 83% over 28 days and 41% vs 31%, respectively) [24]. Our reported RBC transfusion rate (mean transfusion rate 1.27 units/day) was consistent with findings by Raasveld et al. (1.6 Units per day on ECMO), but lower compared to a systematic literature review involving patients treated with VA ECMO prior to 2016 (mean transfusion rate 3.86 units/day) and supports a registry study suggesting that haemorrhage rates have dropped over the last decade [2, 10].
We found that restrictive transfusion guidelines for RBC were not generally followed. Many clinical trials have addressed RBC transfusion targets and a recent Cochrane review included 45 randomized controlled trials of patients with varying underlying clinical conditions and concluded that a restrictive transfusion protocol is safe [25]. However, the MINT trial found a possible benefit for patients with myocardial infarct to target an Hb as high as 100 g/l [22]. Of note, no ECMO patients were included in any of these trials. Hb values prior to RBC transfusion in our study on VA ECMO were consistent with large cohort studies in VV ECMO,[24] while a mixed cohort of 763 patients found an even higher median threshold of 91 g/L.[26] In 16 VA ECMO centres, only four followed a restrictive transfusion strategy (Hb < 75 g/L) but transfusion rates were very similar to our cohort [10]. A European Society of Intensive Care Medicine guideline concluded that there is not enough evidence to recommend a specific Hb transfusion target in patients during ECMO [27]. It is biologically plausible that VA ECMO patients may respond differently to higher Hb levels as their oxygen delivery remains impaired even when recovering from cardiogenic shock. We demonstrated that they received much higher levels of transfusion of blood products compared to other patient cohorts and hence the risk of transfusion-related complications compared with potential benefits to oxygen delivery might be different. When using our dataset to emulate a restrictive (70 g/l) versus liberal (90 g/l) transfusion trial we found a signal towards an improved survival during the first three days of VA ECMO [28]. A plausible benefit of a higher Hb in ECMO will be the subject of two randomised controlled trials (NCT06526533 and NCT06560164).
In line with existing guidelines, a higher platelet count was targeted on days with major haemorrhage but that difference was relatively small because even on days without major haemorrhage platelet counts were defended above recommended levels [29]. This could have been with the intention to treat non-major haemorrhage or as a primary preventive measure. This practice has been reported by other centres and might be based on the higher overall bleeding rate as well as the knowledge that platelet dysfunction is common in ECMO [30]. However, platelet transfusion in settings where antiplatelet medication is often recommended may also be harmful and no high-level evidence exists to support this practice.
Although some data exists that no or low anticoagulation strategies may be used over short periods of time in VA ECMO, our data suggest that systemic anticoagulation was initiated early and during the time of high incidence for haemorrhagic complications while we could not find evidence that this was tailored towards specific risk profiles.
ECPR and major surgery had very similar haemorrhage and transfusion patterns and could be distinguished from patients who were not exposed to either. Direct surgical trauma as well as trauma secondary to prolonged chest compressions and in association with coagulopathy secondary to the cardiopulmonary bypass and severe cardiogenic shock play an important role in VA ECMO-related haemorrhage [8]. This was reflected by more frequent use of blood products and a lower rate of anticoagulation. However, time to first use of anticoagulation and platelet transfusion thresholds were not different. As ECPR and CTS patients also experienced major haemorrhage earlier compared to CS patients, tailoring anticoagulation and transfusion practice according to these risk factors should be investigated in prospective trials.
While surgery just failed to be identified as an independent predictor of RBC transfusions, ECPR was identified to increase that risk by 50%. Myocardial infarction demonstrated a higher risk when compared to myocarditis and pulmonary emboli. Since antiplatelet use prior to VA ECMO was also found to independently increase RBC transfusions during VA ECMO, these classes of drugs may play an important role for the need of RBC transfusion and possible VA ECMO associated haemorrhage. Dosing anticoagulation accordingly may be important in this scenario but further studies are needed to address the impact of antiplatelet medications during VA ECMO and its interplay with ECMO related platelet dysfunction.
Our study has several limitations. First, because of the observational nature of this research, direct conclusions regarding causation cannot be made. However, most research in this field is observational and this study, to our knowledge, represents the largest and most detailed prospective dataset on VA ECMO patients to date. Second, the participating centres were all experienced, and there was an overrepresentation of Australian patients. Third, the study received no funding, and hence only well-resourced centres were able to participate. This may limit the external validity of the data. Last, because of limited resources, we were unable to collect detailed data beyond day seven of VA ECMO. However, VA ECMO duration was shorter than seven days for the majority of patients.
In conclusion, haemorrhagic risk in VA ECMO is complex and detailed knowledge on current practice is crucially important to improve the management of VA ECMO patients. RBC transfusion exposure is by far higher than in other critically unwell populations and we identified ECPR and antiplatelet therapy as additional risk factors. Transfusion practice for RBC and platelets is variable and does not always follow international guidelines. This study contributes to our knowledge on how to individualise blood management during VA ECMO and may inform future interventional trials by adding additional detailed data on haemorrhage rates, transfusion practice and potential predictors of RBC transfusions.
Acknowledgements
This work was completed with thanks to the EXCEL Management Committee, EXCEL Investigators, International ECMO Network, Australian and New Zealand Intensive Care Society Clinical Trials Group, Alfred Hospital, Melbourne, Vic: Jasmin Board, Aidan Burrell, Annalie Jones, Emma Martin, Phoebe McCracken, Vincent Pellegrino, Jayne Sheldrake, Shirley Vallance and Meredith Young; Gold Coast, Qld: Dee Figures, Maimoonbe Gough, Maree Houbert, James McCullough, Julie Pitman, Mandy Tallott and James Winearls; Liverpool Hospital, Sydney, NSW: Anders Aneman, Danielle Austin, Peter McCanny and Jennene Miller; Princess Alexandra Hospital, Brisbane, Qld: Meg Harward, Chris Joyce, Josie McKay, Jason Meyer, and James Walsham; Prince Charles Hospital, Brisbane, Qld: Rachel Bushell, John Fraser, Jayshree Lavana, Dawn Lockwood and Raymond Marteene; Royal Prince Alfred Hospital, Sydney, NSW: Heidi Buhr, Ruaidhri Carey, Jennifer Coles, David Gattas and Richard Totaro; St Vincent’s Hospital Sydney, Sydney, NSW: Hergen Buscher, Priya Nair, Sally Newman and Claire Reynolds
Author contributions
HB, CH, VP, TM and DB conceptualised the study; LTPT, GM, MG, AS, KT, JSP, KR, RV, ML, MS, JW, JW, BF, SN and CR formally analysed the data; HB, LTPT and ZM wrote the original draft of the manuscript; MG, JG, GT, MS, AR, DG, AA, AA, PN, KS, ZM, CH, VP, TM and DB reviewed and edited the manuscript; HB, CH, ZM, BF and CR were involved in project administration. All authors had full access to all the data in the study and accept final responsibility to submit for publication.
Funding
This is an investigator-initiated study, not supported by external funding.
Data availability
Individual participant data that underlie the results reported in this article, after de-identification (text, tables, figures, and appendices) are available from the corresponding author on reasonable request.
Declarations
Competing interests
GT is a Consultant for Resuscitec GmbH; JG has received honoraria, research funding, and travel reimbursement from Alexion, AstraZeneca, Boehringer Ingelheim, CSL Behring, Instrumentation Laboratory, Johnson & Johnson, Mitsubishi Tanabe Pharma, Octapharma, Portola and Takeda.s; MS received lecture fees from GETINGE, SOLVENTUM, BAXTER, and FRESENIUS outside the submitted work; CH is supported by a NHMRC Investigator Grant, Chair of the EXCEL Registry and Co-President of the International ECMO Network (ECMONet); DB previously consulted for LivaNova. He has been on the medical advisory boards for Medtronic, Inspira, Cellenkos, HBOX Therapies and Vantive. He is the President of the Extracorporeal Life Support Organization (ELSO) and the Chair of the Board of the International ECMO Network (ECMONet), and he writes for UpToDate; AS received research grant funds from the National Institutes of Health (UH3HL166785, R01HL168102), Canadian Institutes for Health Research, and Quantum Leap Healthcare Collaborative; and research funding for work as a participating site in clinical trials from Sedana Medical and CalciMedica; all paid to Columbia University and New York University. All other authors declare no competing interests.
Footnotes
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Carol Hodgson, Vincent Pellegrino, Thomas Mueller and Daniel Brodie were involved in planning, implementation, analysing and publication and contributed evenly as senior authors.
References
- 1.Schrage B, Becher PM, Gossling A, et al. Temporal trends in incidence, causes, use of mechanical circulatory support and mortality in cardiogenic shock. ESC Heart Fail. 2021;8(2):1295–303. 10.1002/ehf2.13202. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Vlok R, Buscher H, Delaney A, et al. Anticoagulation and associated complications in veno-arterial extracorporeal membrane oxygenation in adult patients: a systematic review and meta-analysis. Crit Care Resusc. 2024. 10.1016/j.ccrj.2024.10.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Tonna JE, Boonstra PS, MacLaren G, et al. Extracorporeal life support organization registry international report 2022: 100,000 survivors. ASAIO J. 2024;70(2):131–43. 10.1097/MAT.0000000000002128. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Bartlett R, Arachichilage DJ, Chitlur M, et al. The history of extracorporeal membrane oxygenation and the development of extracorporeal membrane oxygenation anticoagulation. Semin Thromb Hemost. 2024. 10.1055/s-0043-1761488. [DOI] [PubMed] [Google Scholar]
- 5.Kalbhenn J, Wittau N, Schmutz A, Zieger B, Schmidt R. Identification of acquired coagulation disorders and effects of target-controlled coagulation factor substitution on the incidence and severity of spontaneous intracranial bleeding during veno-venous ECMO therapy. Perfusion. 2015;30(8):675. 10.1177/0267659115579714. [DOI] [PubMed] [Google Scholar]
- 6.Scott EJ, Rotar E, Dahl JJ, et al. Discordance among assays for monitoring? Anticoagulation during extracorporeal life support. Perfusion. 2023;38(8):1714–21. 10.1177/02676591221129741. [DOI] [PubMed] [Google Scholar]
- 7.Hughes T, Zhang D, Nair P, Buscher H. A systematic literature review of packed red cell transfusion usage in adult extracorporeal membrane oxygenation. Membranes (Basel). 2021. 10.3390/membranes11040251. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Smith A, Hardison D, Bridges B, Pietsch J. Red blood cell transfusion volume and mortality among patients receiving extracorporeal membrane oxygenation. Perfusion. 2013;28(1):54–60. 10.1177/0267659112457969. [DOI] [PubMed] [Google Scholar]
- 9.Ramanathan K, Peek G, Martucci G, et al. Blood transfusion during extracorporeal membrane oxygenation: an ELSO position statement. ASAIO J. 2024;70(9):719–20. 10.1097/MAT.0000000000002275. [DOI] [PubMed] [Google Scholar]
- 10.Raasveld SJ, Karami M, Schenk J, et al. Transfusion of red blood cells in venoarterial extracorporeal membrane oxygenation: a multicenter retrospective observational cohort study. Transfusion. 2023;63(10):1809–20. 10.1111/trf.17505. [DOI] [PubMed] [Google Scholar]
- 11.Mariani S, Heuts S, van Bussel BCT, et al. Patient and management variables associated with survival after postcardiotomy extracorporeal membrane oxygenation in adults: the PELS-1 multicenter cohort study. J Am Heart Assoc. 2023;12(14):e029609. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Belohlavek J, Smalcova J, Rob D, et al. Effect of intra-arrest transport, extracorporeal cardiopulmonary resuscitation, and immediate invasive assessment and treatment on functional neurologic outcome in refractory out-of-hospital cardiac arrest. JAMA. 2022;327(8):737. 10.1001/jama.2022.1025. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Brodie D, Vincent JL, Brochard LJ, et al. Research in extracorporeal life support: a call to action. Chest. 2018;153(4):788–91. 10.1016/j.chest.2017.12.024. [DOI] [PubMed] [Google Scholar]
- 14.Hodgson C, Fulcher B, Mariajoseph F, et al. A core outcome set for research in patients on extracorporeal membrane oxygenation. Crit Care Med. 2021;49(12):e1252–4. 10.1097/CCM.0000000000005110. [DOI] [PubMed] [Google Scholar]
- 15.von Elm E, Altman DG, Egger M, et al. The strengthening the reporting of observational studies in epidemiology (STROBE) statement: guidelines for reporting observational studies. Lancet. 2007;370(9596):1453–7. 10.1016/S0140-6736(07)61602-X. [DOI] [PubMed] [Google Scholar]
- 16.Mehran R, Rao SV, Bhatt DL, et al. Standardized bleeding definitions for cardiovascular clinical trials: a consensus report from the Bleeding Academic Research Consortium. Circulation. 2011;123(23):2736–47. 10.1161/CIRCULATIONAHA.110.009449. [DOI] [PubMed] [Google Scholar]
- 17.Nunez JI, Gosling AF, O’Gara B, et al. Bleeding and thrombotic events in adults supported with venovenous extracorporeal membrane oxygenation: an ELSO registry analysis. Intensive Care Med. 2022;48(2):213–24. 10.1007/s00134-021-06593-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Harris PA, Taylor R, Thielke R, Payne J, Gonzalez N, Conde JG. Research electronic data capture (REDCap)--a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform. 2009;42(2):377–81. 10.1016/j.jbi.2008.08.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.R: A language and environment for statistical computing. . R foundation for statistical computing. 2024;
- 20.Martucci G, Giani M, Schmidt M, et al. Anticoagulation and bleeding during veno-venous extracorporeal membrane oxygenation: insights from the PROTECMO study. Am J Respir Crit Care Med. 2024;209(4):417–26. [DOI] [PubMed] [Google Scholar]
- 21.Bogerd M, Ten Berg S, Peters EJ, et al. Impella and venoarterial extracorporeal membrane oxygenation in cardiogenic shock complicating acute myocardial infarction. Eur J Heart Fail. 2023;25(11):2021–31. 10.1002/ejhf.3025. [DOI] [PubMed] [Google Scholar]
- 22.Carson JL, Brooks MM, Hebert PC, et al. Restrictive or liberal transfusion strategy in myocardial infarction and anemia. N Engl J Med. 2023;389(26):2446–56. 10.1056/NEJMoa2307983. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Mazer CD, Whitlock RP, Fergusson DA, et al. Restrictive or liberal red-cell transfusion for cardiac surgery. N Engl J Med. 2017;377(22):2133–44. 10.1056/NEJMoa1711818. [DOI] [PubMed] [Google Scholar]
- 24.Martucci G, Schmidt M, Agerstrand C, et al. Transfusion practice in patients receiving VV ECMO (PROTECMO): a prospective, multicentre, observational study. Lancet Respir Med. 2023;11(3):245–55. [DOI] [PubMed] [Google Scholar]
- 25.Carson JL, Stanworth SJ, Dennis JA, et al. Transfusion thresholds for guiding red blood cell transfusion. Cochrane Database Syst Rev. 2021;12(12):CD002042. 10.1002/14651858.CD002042.pub5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Ng PY, Chan HCV, Ip A, et al. Restrictive and liberal transfusion strategies in extracorporeal membrane oxygenation: a retrospective observational study. Transfusion. 2023;63(2):294–304. 10.1111/trf.17221. [DOI] [PubMed] [Google Scholar]
- 27.Vlaar APJ, Dionne JC, de Bruin S, et al. Transfusion strategies in bleeding critically ill adults: a clinical practice guideline from the European Society of Intensive Care Medicine. Intensive Care Med. 2021;47(12):1368–92. 10.1007/s00134-021-06531-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Le TTP, Buscher H, Nguyen T-L, et al. Liberal or restrictive transfusion for veno-arterial Extracorporeal Membrane Oxygenation patients: a target trial emulation using the OBLEX study data. 2025; [DOI] [PMC free article] [PubMed]
- 29.Estcourt LJ, Birchall J, Allard S, et al. Guidelines for the use of platelet transfusions. Br J Haematol. 2017;176(3):365–94. 10.1111/bjh.14423. [DOI] [PubMed] [Google Scholar]
- 30.Mazzeffi M, Bathula A, Tabatabai A, et al. Von willebrand factor concentrate administration for acquired Von Willebrand syndrome- related bleeding during adult extracorporeal membrane oxygenation. J Cardiothorac Vasc Anesth. 2021;35(3):882–7. 10.1053/j.jvca.2020.06.083. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
Individual participant data that underlie the results reported in this article, after de-identification (text, tables, figures, and appendices) are available from the corresponding author on reasonable request.