Comparison of a direct vs consultative advanced heart failure role in the outcomes of extracorporeal membrane oxygenation patients

Abstract

BACKGROUND

Advanced heart failure and transplant (AHFTC) teams are crucial in the management of patients in cardiogenic shock. We sought to explore the impact of AHFTC physicians on outcomes in patients receiving extracorporeal membrane oxygenation (ECMO) support.

AIM

To determine whether outcomes differ in the care of ECMO patients when AHFTC physicians serve in a primary vs consultative role.

METHODS

We conducted a retrospective cohort study of 51 patients placed on veno-venous (VV) and veno-arterial (VA) ECMO between January 2015 and February 2023 at our institution. We compared ECMO outcomes between teams managed primarily by intensivists vs teams where AHFTC physicians played a direct role in ECMO management, including patient selection. Our primary outcome measure was survival to 30 days post hospital discharge.

RESULTS

For combined VA and VV ECMO patients, survival to 30 days post discharge in the AHFTC cohort was significantly higher (67% vs 30%, P = 0.01), largely driven by a significantly increased 30-day post discharge survival in VA ECMO patients in the AHFTC group (64% vs 20%, P = 0.05).

CONCLUSION

This study suggests that patients in shock requiring VA ECMO support may have improved survival 30 days after hospital discharge when an AHFTC team serves in a direct role in the selection and management of patients. Further studies are needed to validate this impact.

Key Words: Heart failure; Extracorporeal membrane oxygenation; Critical care; Advanced heart failure and transplant cardiology; Mortality

Core Tip: Patients receiving extracorporeal membrane oxygenation (ECMO) support are critically ill, and suffer from high morbidity and mortality. Advanced heart failure and transplant cardiology physicians have been shown to improve outcomes in patients suffering from severe cardiovascular disease. We found that at our center, advanced heart failure and transplant cardiology physicians improved post hospital discharge survival in patients receiving ECMO support when serving as the primary ECMO attending instead of in a consultative role. This was largely due to improved outcomes in patients receiving veno-arterial ECMO support.

INTRODUCTION

Extracorporeal membrane oxygenation (ECMO) support is an option for patients in cardiac and pulmonary failure refractory to conventional management[1]. Indications for respiratory support include acute respiratory distress syndrome (ARDS), status asthmaticus, diffuse alveolar hemorrhage, and/or respiratory viral illness such as coronavirus disease 2019. Indications for cardiac support include but are not limited to cardiogenic shock, post-cardiotomy syndrome and/or myocarditis. ECMO support can also be used as a bridge to advanced cardiac or pulmonary therapies.

Recent survival outcomes from the ELSO national registry are 44%, 50%, and 57% for veno-arterial (VA), veno-venous (VV), and combined VA and VV ECMO support respectively[2], despite advances in ECMO management, and an increased number of dedicated ECMO centers. While the specific components of the ECMO team have not been completely standardized, guidelines have emphasized quality assurance and clinical case reviews with a multi-disciplinary approach[2].

Studies have demonstrated that the implementation of a formal multi-disciplinary ECMO team has positive benefits on ECMO outcomes[3-5]. Dalia et al[4] showed that a multidisciplinary ECMO team led to a significant improvement in survival to hospital discharge (37.7% vs 52.3%) at their institution. One prior study examined the impact of a heart failure team providing input on decision making on VA ECMO outcomes, finding no difference in outcomes with that team’s involvement[6].

Advanced heart failure and transplant (AHFTC) teams play an integral role in management of patients in cardiogenic shock, due to their content expertise in the management of such shock as well as indications for advanced cardiac therapies[7]. To date, there have been no prior studies to assess the potential benefit of an AHFTC team serving in a direct role for patients receiving ECMO. This study assessed the impact of direct rather than consultative involvement by an AHFTC team in the outcomes of ECMO patients at our institution. We hypothesized that patients receiving ECMO would have better outcomes when AHFTC physicians served in a more direct role in their care.

MATERIALS AND METHODS

Study design

We conducted a retrospective cohort study of 51 patients on VV and VA ECMO between January 2015 and February 2023 at the Queen’s Medical Center (a 575-bed hospital and the main tertiary care center for the state) in Honolulu, Hawaii. We identified 53 patients in total who underwent VA and/or VV ECMO dating back to 2015. If a patient received VA ECMO support at any time during their hospitalization, we placed them in the VA ECMO cohort (e.g., patients initiated on VV ECMO, but subsequently transitioned to VA ECMO, were placed in the VA ECMO cohort). Two patients were excluded from our analysis as they were transferred to a local institution < 24 hours after cannulation to receive their care on ECMO support (one of which did not meet ECMO inclusion criteria for our program). We included patients who were placed on ECMO at other institutions after consultation with and at the request of our ECMO team, and transferred to ours < 24 hours after cannulation. For patients transferred to mainland institutions for advanced therapies including transplantation and left ventricular assist device, chart review of their course at the outside institution was performed. Out of the 51 patients included in our analysis, patients were classified into two cohorts based on the role of the AHFTC team (consultative vs direct).

This study was approved by the institution board of the Queen’s Medical Center.

ECMO team composition

The structure of the ECMO team at our institution has varied. During certain eras, ECMO patients were managed by intensivist-led teams with the AHFTC team serving as consultants for VA ECMO only. During another time period, ECMO patients were managed in a model where AHFTC physicians served as the attending of record. In this model, AHFTC physicians played a direct role in ECMO patient selection, ECMO weaning, and the decision to pursue advanced therapies. All other components of the ECMO team including nursing staff, ECMO coordinators, and the cannulation team (cardiac surgery and interventional cardiology) remained unchanged.

The AHFTC physicians caring for ECMO patients had all trained at institutions (Stanford, UCLA, and Yale) where the cardiac critical care unit at the time of training was staffed by AHFTC physicians as the primary attending. The intensivists caring for ECMO patients included critical care, emergency medicine/critical care, and pulmonary critical care trained physicians, with varying degrees of experience in caring for ECMO patients. Some of the intensivists caring for ECMO patients were not board certified, and others had no exposure to patients receiving ECMO support during their fellowship training.

Statistical analysis

Data were analyzed by STATA software. Baseline characteristics and clinical parameters listed in Table 1 were compared between the two cohorts. Survival rates after de-cannulation (defined as > 24-hour survival after de-cannulation of ECMO support) and to 30 days post -discharge were compared. Survival parameters were also compared in subgroup analyses of patients on VA and VV ECMO. Continuous data was presented as averages with standard deviation, and categorical data was presented as percentages.

Table 1 Baseline characteristics in a cohort of extracorporeal membrane oxygenation patients with direct vs consultative primary advanced heart failure and transplant involvement, n (%).

	Total (n = 51)	With direct AHFTC role (n = 21)	With consultative AHFTC role (n = 30)	P value
Age (year)	47.2 (SD 14.5)	47.0 (SD 13.3)	47.4 (SD 15.5)	0.92
RESP score	3.4 (SD 3.3)	3.2 (SD 1.9)	3.5 (SD 4.1)	0.81
SAVE score	-5.7 (SD 3.9)	-4.6 (SD 4.3)	-6.5 (SD 3.5)	0.22
Time on ECMO (days)	11.8 (SD 17.3)	15.6 (SD 21.8)	9.3 (SD 13.3)	0.21
Total days of MV prior to cannulation (days)	3.6 (SD 5.2)	3.0 (SD 3.6)	4.0 (SD 6.0)	0.51
Renal function	2.1 (SD 1.7)	1.9 (SD 1.8)	2.1 (SD 1.7)	0.68
On pressors	44 (86)	15 (71)	29 (97)	0.01
Requiring RRT	23 (45)	5 (24)	18 (60)	0.01
Requiring mechanical ventilation	49 (96)	19 (90)	30(100)	0.09
Requiring tMCS	15 (29)	5 (24)	10 (33)	0.46
With Swan-ganz placement	20 (39)	9 (43)	11 (37)	0.66

AHFTC: Advanced heart failure and transplant; ECMO: Extracorporeal membrane oxygenation; MV: Mechanical ventilation; RRT: Renal replacement therapy; tMCS: Temporary mechanical circulatory support.

Baseline characteristics were compared with Student’s t-tests and χ² tests for continuous and categorical data respectively. Logistic regression modeling was used to calculate odds ratios and 95% confidence intervals. Multivariable logistic regression modeling was used to adjust for potential confounding factors. All statistical tests were performed with two-sided tests at 0.05 level of significance.

RESULTS

From January 2015 to February 2023, 51 patients were placed on VV (25) and VA (26) ECMO at our institution. Twenty-one patients were managed with AHFTC playing a direct role (AHFTC direct cohort); 30 patients were managed with AHFTC playing a consultative role (AHFTC consultative cohort). Patients were predominantly placed on VV ECMO for ARDS due to bacterial or viral pneumonia in both groups (19/25 total, 8/9 in AHFTC direct group, 11/16 in consultative group). Patients were placed on VA ECMO for post cardiotomy shock (10/26 total, 2/11 in AHFTC direct group, 8/15 in the consultative group), myocarditis (7/26 total, 4/11 in AHFTC direct and 3/15 in consultative group), or other cardiogenic shock (9/26 total, 5/11 in AHFTC direct group, 4/15 in consultative group). The cannulation strategy was peripheral in the majority of cases (43/51, 19/21 in AHFTC direct group, and 24/30 in the consultative group). The use of left ventricular venting for VA ECMO was not significantly different between groups (12/26 total, 6/11 in the AHFTC direct group, 6/15 in the consultative group). Venting was provided by an axial flow pump in most cases, with a surgical vent placed in 1 patient in each group.

Selection for ECMO in the intensivist era largely occurred during informal conversations between the intensivist and the cannulator. In the AHF era, we used a more formal approach to ECMO selection, including incorporating the RESP and SAVE scores into the process and convening a tripartite committee including the AHF specialist, intensivist, and cannulating surgeon to make the ultimate decision.

The baseline characteristics of the two cohorts were compared and are listed in Table 1. Age (47.0 vs 47.4 for the direct and consultative cohorts respectively, P = 0.92), RESP risk score (3.2 vs 3.5, P = 0.81), and SAVE risk score (-4.6 vs -6.5, P = 0.22) were not significantly different. Duration on ECMO support, total days of mechanical ventilation prior to cannulation, and renal function were also not significantly different (15.6 vs 9.3, P = 0.21; 3.0 vs 4.0, P = 0.51; 1.9 vs 2.1, P = 0.68 respectively). The percentage of patients in each cohort requiring Swan-Ganz catheter placement, mechanical ventilation, and temporary mechanical circulatory support were similar (43% vs 37%, P = 0.66; 90% vs 100%, P = 0.09; 24% vs 33%, P = 0.46 respectively). A significantly lower percentage of patients in the AHFTC direct cohort were placed on pressor support and renal replacement therapy (71% vs 97%, P = 0.01; 24% vs 60%, P = 0.01).

For combined VA and VV ECMO patients, survival after de-cannulation for the AHFTC direct cohort was 71% vs 57% for the AHFTC consultative cohort (P = 0.49, Table 2). For VV ECMO patients, survival after de-cannulation was 70% vs 53% (P = 0.45) for the AHFTC direct and consultative cohorts respectively. For VA ECMO patients, survival after de-cannulation was 73% vs 60% respectively (P = 0.43). The cause of death at decannulation in the direct cohort was predominantly progressive shock (4/6), with one patient suffering a fatal stroke. The cause of death at decannulation in the consultative cohort was predominantly fatal stroke (7/13), followed by progressive shock (6/13).

Table 2 Comparison of survival outcomes in a cohort of extracorporeal membrane oxygenation patients with direct vs consultative advanced heart failure and transplant involvement.

	Total (n = 51)	With direct AHFTC role (n = 21)	With consultative AHFTC role (n = 30)	Unadjusted odds ratio	P value	Adjusted odds ratio1	P value
Total survival after decannulation (%)	32/51 (63)	15/21 (71)	17/30 (57)	1.6 (0.5-5.6)	0.46	1.6 (0.4-5.6)	0.49
Total survival to 30 days post discharge (%)	23/51 (45)	14/21 (67)	9/30 (30)	4.7 (1.4-15)	0.01	5.3 (1.5-19)	0.01
VA survival after decannulation (%)	17/26 (65)	8/11 (73)	9/15 (60)	2.25 (0.4-14.6)	0.40	2.20 (0.3-16)	0.43
VA survival to 30 days post discharge (%)	10/26 (38)	7/11 (64)	3/15 (20)	7.0 (1.2-40.8)	0.03	6.4 (1.0-41.7)	0.05
VV survival after decannulation. (%)	15/25 (60)	7/10 (70)	8/15 (53)	1.7 (0.6-5.9)	0.43	1.8 (0.7-6.1)	0.45
VV survival to 30 days post discharge (%)	13/25 (52)	7/10 (70)	6/15 (40)	3.5 (0.6-19.2)	0.15	4.8 (0.71-32.2)	0.11

¹Adjusted for age and gender.

AHFTC: Advanced heart failure and transplant; VV: Veno-venous; VA: Veno-arterial.

One patient in the AHFTC direct cohort died 2 months after decannulation due to gastrointestinal bleeding, prior to hospital discharge. Eight patients in the AHFTC consultative cohort died > 24 hours after decannulation and prior to 30 days post discharge, with time to death from decannulation ranging from 4 days to 4 months. The predominant cause of death after decannulation in the consultative cohort was progressive shock (6/8), though 1 patient died from fatal stroke.

For combined VV and VA ECMO patients, survival to 30 days post hospital discharge in the AHFTC direct cohort (67%) was significantly higher compared to the AHFTC consultative cohort (30%) with a P value of 0.01. Similarly, for VA ECMO patients only, survival to 30 days post discharge in the AHFTC direct cohort (64%) was significantly higher compared to the AHFTC consultative cohort (20%) with a P value of 0.05 and adjusted odds ratio of 6.4. Survival to 30 days post hospital discharge for VV ECMO patients was higher in the AHFTC direct cohort (70% vs 40%), but this result did not reach statistical significance (P = 0.1).

Four patients were transferred on VA ECMO support to mainland institutions from the AHFTC direct cohort vs 3 in the consultative cohort, on average 8 days after initiation of ECMO. Three out of 4 of the patients in the AHFTC direct cohort survived to mainland hospital discharge, one dying due to progressive cardiogenic shock. None of the patients in the consultative cohort who were transferred to the mainland survived to hospital discharge (on average > 10 days after initiation of ECMO), with 2 deaths occurring due to progressive shock, and 1 due to fatal stroke.

Overall, there were 9 strokes (30%) identified by radiographic studies in the consultative cohort, 8 of which were fatal (27%), and 1 in the AHFTC direct cohort (5%), which was also fatal.

Within the intensivist group, patients treated by intensivists with critical care training alone (vs emergency medicine/critical care or pulmonary critical care) had worse outcomes, but given the sample size this difference was not statistically significant.

DISCUSSION

AHFTC teams play a vital role in the management of patients with refractory cardiogenic shock[7]. With expertise in the hemodynamic management of critically ill patients and indications for advanced cardiac therapies, we find that AHFTC teams are also suited to lead and play a directive role in the management of VA ECMO patients.

In our study, there was higher numerical survival in the group with direct involvement of AHFTC in all groups. The most significant difference was in survival outcomes to 30 days post hospital discharge among combined VV and VA ECMO patients, driven largely by a significant survival advantage in VA ECMO patients when AHFTC teams were directly involved in patient selection and management (P = 0.01 and 0.05 respectively).

There are a few possible explanations for these findings. First, the most common indications for VA ECMO patients were myocarditis, cardiogenic shock and post-cardiotomy shock. Cardiogenic shock and myocarditis are disease states in which AHFTC physicians are content experts, and having AHFTC specialists direct clinical care has led to better outcomes in other similar clinical cohorts, such as patients in the Cardiac Intensive Care unit[8]. For example, for patients on VA ECMO due to myocarditis, our AHFTC team introduced a successful treatment protocol utilizing plasmapheresis and IVIG therapy[9].

Another possible explanation is better patient selection in the AHFTC cohort. Though baseline characteristics among the two groups were similar with respect to RESP and SAVE scores (P value 0.81 and 0.22 respectively), in the AHFTC cohort more patients were successfully bridged to advanced cardiac therapies (3 in the direct cohort vs 0 in the consultative cohort). AHFTC specialists receive formal training in assessing the candidacy of patients for advanced heart failure therapies such as left ventricular assist devices and cardiac transplantation, which may have positively informed patient selection and candidacy of patients for such therapies after hospital transfer. In addition, patients placed on ECMO support for post-cardiotomy shock in general suffer higher mortality. There were relatively more patients placed on VA ECMO for post cardiotomy shock in the consultative cohort (53%) than in the direct AHFTC cohort (18%).

Better selection in the AHFTC cohort may have occurred to substantial ECMO experience and formalized training in ECMO during fellowship, allowing the AHFTC physicians to determine which patients could bridge to end stage therapies for lung or heart failure. In the intensivist cohort, a number of the physicians had no exposure to the care of patients receiving ECMO during the fellowship training, and/or were not board certified in their specialty. That lack of knowledge could have contributed to worse outcomes, as none of the patients transferred to the mainland in the intensivist cohort survived to hospital discharge.

More patients died post ECMO decannulation in the consultative cohort than in the AHFTC direct cohort, usually due to progressive shock or stroke. There were also a higher proportion of patients with stroke, mostly fatal, in the consultative cohort (26%) vs the direct AHFTC cohort (5%). Appropriate timing of decannulation of ECMO, including management of cardiogenic shock after ECMO decannulation, and stroke prevention strategies are areas where AHFTC physicians may offer particular value.

As compared to the previous study which showed no improvement in ECMO outcomes with HF team involvement[5], in our paradigm AHFTC physicians were the attending of record, and directing care from ECMO initiation to hospital discharge; that in depth involvement may have positively impacted outcomes. For example, the direct involvement of AHFTC teams may lead to more expedited evaluation for advanced cardiac therapies. This is particularly relevant at our institution, as the transportation of patients from Hawaii to mainland centers, the nearest of which are 4000 km away, requires a highly coordinated effort.

Our study suggests the training of physicians caring for patients receiving ECMO may affect outcomes, and could impact the composition of ECMO teams (e.g., ensuring the presence of AHFTC physicians in the care of patients receiving VA ECMO). It may also be that specific aspects of AHFTC training better equip those physicians in caring patients receiving VA ECMO, and that those learned competencies could improve the care of non AHFTC physicians caring these patients.

In our physician cohort, the AHFTC physicians were trained in centers where AHFTC physicians serve as the attending of records in the cardiac critical care unit. Thus, the above results may not hold in institutions where AHFTC physicians do not receive such training. Similarly, the non AHFTC critical care physicians who cared for ECMO patients during this time period had variable ECMO training, including some with no exposure to ECMO patients during their fellowship training and others who lacked board certification; critical care physicians with more formal ECMO training and/or experience may have better outcomes.

Future directions for research in this area may include multi-center studies examining the role of AHFTC physicians in caring for patients receiving ECMO, as well as standardization of the training across disciplines for physicians caring for patients receiving ECMO. In particular, health care systems may consider integrating AHFTC physicians directly into the selection of patient being considered for ECMO cannulation, and involvement of these physicians earlier on in a patient’s hospital course.

Study limitations include the single-center retrospective design and small sample size. The small sample size limits generalizability of our findings as it reduces study power; also, our study cohort may not be identical to patients receiving ECMO at larger programs in terms of demographics and severity of illness. As noted above, AHFTC specialists were trained at institutions where AHFTC physicians served as attendings of record in the cardiac intensive care unit, so our results may not be applicable if AHFTC physicians are without such training. Lastly, our study was a ‘natural experiment,’ leaving it open to the possibility of unmeasured confounders that could have impacted outcomes teams in the management of patients on ECMO.

CONCLUSION

In summary, our study suggests that there is an association between AHFTC team serving in a direct role in the selection and management of patients requiring VA ECMO support and improved patient outcomes, including survival 30 days post hospital discharge. Further studies are needed to validate this impact, although our findings may assist other institutions in the composition of new and existing ECMO programs.

ACKNOWLEDGEMENTS

We would like to acknowledge Erick Itoman, Chris Fiack, Rick Bruno, and Whitney Limm, without whom this manuscript would not have been possible.