Abstract
Aims
Shock severity predicts mortality in patients with cardiogenic shock (CS). We evaluated the association between pre-cannulation Society for Cardiovascular Angiography and Intervention (SCAI) shock classification and mortality among patients receiving venoarterial (VA) extracorporeal membrane oxygenation (ECMO) support for CS.
Methods and results
We included Extracorporeal Life Support Organization (ELSO) Registry patients from 2010 to 2020 who received VA ECMO for CS. SCAI shock stage was assigned based on hemodynamic support requirements prior to ECMO initiation. In-hospital mortality was analyzed using multivariable logistic regression. We included 12 106 unique VA ECMO patient runs with a median age of 57.9 (interquartile range: 46.8, 66.1) years and 31.8% were females; 3472 (28.7%) were post-cardiotomy. The distribution of SCAI shock stages at ECMO initiation was: B, 821 (6.8%); C, 7518 (62.1%); D, 2973 (24.6%); and E, 794 (6.6%). During the index hospitalization, 6681 (55.2%) patients died. In-hospital mortality increased incrementally with SCAI shock stage (adjusted OR: 1.24 per SCAI shock stage, 95% CI: 1.17–1.32, P < 0.001): B, 47.5%; C, 52.8%; D, 60.8%; E, 65.1%. A higher SCAI shock stage was associated with increased in-hospital mortality in key subgroups, although the SCAI shock classification was only predictive of mortality in non-surgical (medical) CS and not in post-cardiotomy CS.
Conclusion
The severity of shock prior to cannulation is a strong predictor of in-hospital mortality in patients receiving VA ECMO for CS. Using the pre-cannulation SCAI shock classification as a risk stratification tool can help clinicians refine prognostication for ECMO recipients and guide future investigations to improve outcomes.
Keywords: Cardiogenic shock, Shock, Vasopressors, Mechanical circulatory support, Extracorporeal membrane oxygenator , Extracorporeal life support
Graphical Abstract
Graphical Abstract.
In line with the Journal's conflict of interest policy, this paper was handled by Borja Ibanez. See the editorial comment for this article ‘First steps taken, but many more ahead’, by B. N. Beer and B. Schrage, https://doi.org/10.1093/ehjacc/zuac140.
Introduction
Cardiogenic shock (CS) is a haemodynamic state resulting in end-organ hypoperfusion that remains associated with high short-term mortality despite contemporary therapy including a growing array of temporary mechanical circulatory support (MCS) devices.1–4 Venoarterial (VA) extracorporeal membrane oxygenation (ECMO) can provide robust biventricular haemodynamic and respiratory support, making ECMO the ideal temporary MCS modality for patients with severe CS, refractory CS, or biventricular CS.1,4
Shock severity is among the primary prognostic determinants in patients with CS, and clinicians aim to match the degree of haemodynamic support to the degree of haemodynamic compromise.5,6 The Society for Cardiovascular Angiography and Intervention (SCAI) shock classification classifies CS into five stages of escalating severity from A to E to enable clinical decision-making.6 The SCAI shock classification predicts short-term mortality in patients with CS, and risk modifiers can portend worse outcomes independent from shock severity.7–16 The scalability of the SCAI shock classification to large muti-centre registry data sets remains uncertain given the need to incorporate multiple clinical variables that are not always available.7
The severity of shock at the time of ECMO initiation is a primary determinant of mortality.17–19 Vasopressor load has been used to quantify shock severity in patients receiving ECMO but no published analyses have used the more comprehensive SCAI shock classification to assess shock severity in ECMO recipients.5,17,18 The objective of this study was to evaluate the association between the SCAI shock classification prior to VA ECMO initiation with in-hospital mortality in a large multinational registry, and to further explore reported risk modifiers.
Methods
Data sources
The Extracorporeal Life Support Organization (ELSO) Registry is a voluntary international database of patients who received ECMO support since 1989 and includes >120 000 patients from 430 centres making it the largest collection of ECMO data in the world.20 The ELSO Registry is utilized for quality improvement by member centres, and extensively for research.21 Data accuracy and quality within the ELSO Registry are ensured through real time data validation.22
De-identified patient ECMO run data were extracted electronically from the ELSO Registry. We included all adult VA ECMO patient runs from 2010 to 2020 for cardiac support with a diagnosis of CS based on International Classification of Diseases (ICD)-9/10 codes (785.51 and R57.1, respectively); we included both non-surgical (medical) ECMO and post-cardiotomy ECMO. Post-cardiotomy CS was defined as the presence of any Clinical Procedural Terminology (CPT) code for cardiac surgery or use of cardiopulmonary bypass (CPB) prior to ECMO, and patients not meeting these criteria were considered to have non-surgical (medical) CS. We excluded extracorporeal cardiopulmonary resuscitation (ECPR) patient runs and patient runs without data on the use of vasoactive drugs and MCS devices. We collected data regarding details of the ECMO run, clinical outcomes and patient condition prior to ECMO initiation, including vital signs, laboratory studies, ventilator support and the use of vasoactive drugs and MCS devices.
SCAI shock classification
The SCAI shock stage used in this analysis was assigned retrospectively using a modification of the classification scheme proposed by the Cardiogenic Shock Working Group (CSWG) based on the number of vasopressors and the number of MCS devices used (including all forms of MCS) prior to VA ECMO initiation (Figure 1).12 The CSWG application of the SCAI shock classification has been validated in multiple studies for mortality risk stratification in patients with CS.12,13,16 Vasopressors were defined as drugs with vasoconstrictive properties (with or without inotropic properties), while inodilators were defined as inotropic drugs with vasodilator properties and considered separately. All patients were considered to have at least SCAI shock stage B based on a diagnosis of CS with the need for VA ECMO; we performed a sensitivity analysis where all patients were classified as SCAI shock stage C or higher.9 Vasopressor doses were not available, precluding a more detailed SCAI shock classification.7,9,12 For the minority of patients with an available lactate level prior to ECMO initiation, we recreated the modified CSWG SCAI shock classification using lactate levels (see Supplementary material online, Table S1).12
Figure 1.
Society for Cardiovascular Angiography and Intervention shock classification used for this analysis (top), with observed in-hospital outcomes in each Society for Cardiovascular Angiography and Intervention shock stage in the final study population (bottom).
Outcomes
The primary outcome of interest was all-cause in-hospital mortality; patients who remained alive on ECMO were not included in this primary outcome. The key secondary outcome of native heart survival was defined as patients who survived to hospital discharge without inpatient transplantation or ongoing ECMO support. Discharge disposition was recorded among hospital survivors, and reason for ECMO discontinuation was recorded for all patients. Hospital length of stay was recorded and used to determine the duration of in-hospital survival; post-hospital outcomes were not available.
Statistical analysis
Categorical variables were summarized as number (percent), and the Pearson chi-square test was used to compare groups. Continuous variables were summarized as median (interquartile range, IQR), and the Kruskal–Wallis test was used to compare groups. Kaplan–Meier analysis was used to estimate in-hospital survival (censored at discharge), with groups compared using the log-rank test. Odds ratio (OR) and 95% confidence interval (CI) estimates for in-hospital mortality were generated using logistic regression, before and after adjustment. Covariates included in all multivariable models selected a priori based on clinical experience, published literature and availability included: age, sex, race, first vs. later ECMO run, year, weight, post-cardiotomy, primary diagnosis of CS, receipt of heart or lung transplant, receipt of renal replacement therapy (RRT), use of inodilators and cardiac arrest before ECMO. Due to high missingness of physiological variables such as laboratory data and haemodynamics, these variables were not included in the multivariable models; imputation was not performed due to excessive missingness. Subgroups of interest included those separated by age (<50 vs. 50–64 vs. ≥65 years), sex, year (2010–2015 vs. 2016–2018 vs. 2019–2020), post-cardiotomy vs. medical, cardiac arrest before ECMO, use of RRT, and pre-specified cut-off values of lactate (5 mmol/L) and blood pH (7.2). A sensitivity analysis was performed using only data from the first ECMO run. All P-values were two-tailed. Analyses were performed using BlueSky version 7.40 (BlueSky LLC, Chicago, IL).
Results
Study population
Of 14 037 eligible adult patient runs who received VA ECMO for cardiac support with a diagnosis of CS between 2010 to 2020 (not including ECPR), we excluded 635 without VA ECMO as the primary ECMO mode and 1296 others without data available on vasoactive agents (see Supplementary material online, Figure S1). The final study population of 12 106 patient runs, including 11 843 (97.8%) first runs, had a median (IQR) age of 57.9 (46.8, 66.1) years; 31.8% were females and 36.1% were non-White race (Table 1). Only 20.0% of patients were cannulated prior to 2016. A primary diagnosis of CS was present in 62.8%. A total of 3472 (28.7%) patients were post-cardiotomy, and the remaining 9861 (81.5%) were non-surgical (medical) ECMO. The median duration of ECMO support was 120 (64, 206) h, and 5.0% were converted from VA ECMO to another ECMO mode.
Table 1.
Baseline characteristics of the final study population, and comparison of patients in each society for cardiovascular angiography and intervention shock stage including data obtained prior to extracorporeal membrane oxygenator initiation
| SCAI Stage B (N = 821) | SCAI Stage C (N = 7518) | SCAI Stage D (N = 2973) | SCAI Stage E (N = 794) | Total (N = 12 106) | N missing | P-value | |
|---|---|---|---|---|---|---|---|
| Demographics | |||||||
| Age | 55.1 (41.2, 64.6) | 57.2 (45.8, 65.7) | 59.3 (49.0, 66.8) | 60.8 (51.7, 68.6) | 57.9 (46.8, 66.1) | 0 | < 0.001 |
| Age group | |||||||
| <50 years | 330 (40.2%) | 2427 (32.3%) | 789 (26.5%) | 165 (20.8%) | 3711 (30.7%) | 0 | < 0.001 |
| 50–65 years | 294 (35.8%) | 3055 (40.6%) | 1258 (42.3%) | 347 (43.7%) | 4954 (40.9%) | ||
| ≥65 years | 197 (24.0%) | 2036 (27.1%) | 926 (31.1%) | 282 (35.5%) | 3441 (28.4%) | ||
| Female sex | 291 (35.4%) | 2358 (31.5%) | 941 (31.7%) | 245 (30.9%) | 3835 (31.8%) | 29 | 0.192 |
| Race | 556 | 0.001 | |||||
| White | 499 (64.6%) | 4508 (62.7%) | 1830 (64.9%) | 538 (70.0%) | 7375 (63.9%) | ||
| Black | 106 (13.7%) | 842 (11.7%) | 354 (12.6%) | 79 (10.3%) | 1381 (12.0%) | ||
| Hispanic | 48 (6.2%) | 501 (7.0%) | 192 (6.8%) | 46 (6.0%) | 787 (6.8%) | ||
| Asian | 48 (6.2%) | 695 (9.7%) | 212 (7.5%) | 39 (5.1%) | 994 (8.6%) | ||
| Multiple/other | 71 (9.2%) | 643 (8.9%) | 232 (8.2%) | 67 (8.7%) | 1013 (8.8%) | ||
| Year | 0 | < 0.001 | |||||
| 2010–2015 | 290 (35.3%) | 1722 (22.9%) | 309 (10.4%) | 96 (12.1%) | 2417 (20.0%) | ||
| 2016–2018 | 288 (35.1%) | 2900 (38.6%) | 1117 (37.6%) | 358 (45.1%) | 4663 (38.5%) | ||
| 2019–2020 | 243 (29.6%) | 2896 (38.5%) | 1547 (52.0%) | 340 (42.8%) | 5026 (41.5%) | ||
| Weight (kg) | 82 (69, 100) | 84 (70, 100) | 88 (73, 104) | 87 (74, 104) | 85 (71, 101) | 229 | < 0.001 |
| ECMO details | |||||||
| First episode | 802 (97.7%) | 7363 (97.9%) | 2905 (97.7%) | 773 (97.4%) | 11 843 (97.8%) | 0 | 0.380 |
| Hours on ECMO | 120 (60, 209) | 121 (66, 209) | 117 (59, 198) | 130 (66, 213) | 120 (64, 206) | 3 | 0.030 |
| Hours on ECMO—survivors | 120 (70, 199) | 124 (74, 198) | 120 (70, 187) | 122 (78, 191) | 122 (73, 195) | 0 | 0.267 |
| Conversion to another ECMO mode | 36 (4.4%) | 326 (4.3%) | 197 (6.6%) | 50 (6.3%) | 609 (5.0%) | 0 | < 0.001 |
| Cardiac arrest before ECMO | 315 (38.8%) | 2627 (35.5%) | 1115 (38.0%) | 242 (31.0%) | 4299 (36.1%) | 185 | < 0.001 |
| Primary diagnosis of CS | 490 (59.7%) | 4713 (62.7%) | 1912 (64.3%) | 492 (62.0%) | 7607 (62.8%) | 0 | 0.129 |
| Post-cardiotomy | 93 (11.3%) | 1742 (23.2%) | 1022 (34.4%) | 615 (77.5%) | 3472 (28.7%) | 0 | < 0.001 |
| Outcomes | |||||||
| In-hospital mortality | 390 (47.5%) | 3966 (52.8%) | 1808 (60.8%) | 517 (65.1%) | 6681 (55.2%) | 0 | < 0.001 |
| Native heart survival | 363 (44.2%) | 3066 (40.8%) | 1036 (34.9%) | 241 (30.4%) | 4706 (38.9%) | 1 | < 0.001 |
| Discharged alive | 412 (50.2%) | 3342 (44.5%) | 1087 (36.6%) | 259 (32.6%) | 5100 (42.1%) | 1 | < 0.001 |
| Discharge home—survivors | 179 (42.7%) | 1358 (39.4%) | 360 (31.8%) | 73 (26.8%) | 1970 (37.4%) | 6838 | < 0.001 |
| Heart or lung transplant | 75 (9.1%) | 490 (6.5%) | 116 (3.9%) | 36 (4.5%) | 717 (5.9%) | 0 | < 0.001 |
| Length of stay | 20 (9, 40) | 18 (8, 34) | 16 (6, 33) | 18 (8, 35) | 18 (7, 34) | 124 | 0.038 |
| Length of stay—survivors | 29 (16, 49) | 28 (17, 48) | 30 (18, 50) | 34 (21, 54) | 29 (18, 49) | 37 | < 0.001 |
| Renal replacement therapy | 48 (5.8%) | 631 (8.4%) | 412 (13.9%) | 103 (13.0%) | 1194 (9.9%) | 0 | < 0.001 |
| Reason for discontinuation | 279 | < 0.001 | |||||
| Died or poor prognosis | 276 (33.9%) | 2840 (38.3%) | 1385 (47.3%) | 374 (47.5%) | 4875 (40.8%) | ||
| Expected recovery | 469 (57.5%) | 3995 (53.9%) | 1321 (45.1%) | 356 (45.2%) | 6141 (51.4%) | ||
| ECMO complication | 11 (1.3%) | 81 (1.1%) | 23 (0.8%) | 5 (0.6%) | 120 (1.0%) | ||
| Heart/lung transplant | 19 (2.3%) | 117 (1.6%) | 30 (1.0%) | 8 (1.0%) | 174 (1.5%) | ||
| Ventricular assist device | 34 (4.2%) | 289 (3.9%) | 126 (4.3%) | 35 (4.4%) | 484 (4.1%) | ||
| Resource limitation | 1 (0.1%) | 18 (0.2%) | 12 (0.4%) | 2 (0.3%) | 33 (0.3%) | ||
| Respiratory support prior to ECMO | |||||||
| Ventilator | 476 (87.3%) | 5368 (92.3%) | 2372 (95.0%) | 627 (98.0%) | 8843 (93.1%) | 2606 | < 0.001 |
| Ventilator rate | 17 (14, 22) | 17 (14, 22) | 18 (14, 24) | 16 (12, 20) | 18 (14, 22) | 4434 | 0.178 |
| FiO2 | 100 (60, 100) | 100 (60, 100) | 100 (70, 100) | 100 (70, 100) | 100 (60, 100) | 3869 | < 0.001 |
| PaO2/FiO2 ratio | 118 (67, 252) | 138 (74, 278) | 136 (73, 283) | 197 (87, 336) | 139 (74, 283) | 4224 | < 0.001 |
| Peak inspiratory pressure | 26 (20, 32) | 25 (20, 30) | 26 (21, 31) | 25 (22, 30) | 25 (21, 30) | 6252 | 0.021 |
| Positive end-expiratory pressure | 7 (5, 10) | 8 (5, 10) | 8 (5, 10) | 7 (5, 10) | 8 (5, 10) | 4534 | 0.052 |
| Mean airway pressure | 13 (10, 18) | 13 (10, 17) | 14 (11, 18) | 13 (10, 17) | 13 (10, 18) | 7580 | 0.553 |
| Intubation to ECMO (hours) | 5 (1, 17) | 7 (2, 20) | 10 (4, 26) | 11 (6, 27) | 8 (2, 22) | 2066 | < 0.001 |
| Prolonged intubation (≥30 hours) | 100 (14.7%) | 1077 (17.3%) | 542 (22.2%) | 158 (23.2%) | 1877 (18.7%) | 2066 | < 0.001 |
| Vital signs and haemodynamics prior to ECMO | |||||||
| Systolic blood pressure | 91 (77, 107) | 86 (72, 101) | 85 (72, 101) | 80 (68, 97) | 86 (72, 101) | 3023 | < 0.001 |
| Mean blood pressure | 66 (56, 78) | 65 (55, 75) | 64 (54, 74) | 61 (51, 71) | 65 (55, 75) | 4188 | < 0.001 |
| Diastolic blood pressure | 56 (45, 68) | 54 (44, 65) | 53 (43, 63) | 50 (40, 61) | 54 (43, 64) | 3076 | < 0.001 |
| Pulse pressure | 33 (23, 43) | 30 (20, 42) | 31 (20, 44) | 30 (18, 43) | 30 (20, 43) | 3078 | 0.663 |
| Systolic PA pressure | 45 (31, 59) | 40 (31, 50) | 40 (31, 50) | 37 (29, 45) | 40 (31, 50) | 8990 | < 0.001 |
| Diastolic PA pressure | 27 (20, 32) | 24 (18, 30) | 24 (18, 29) | 22 (17, 28) | 24 (18, 30) | 9011 | < 0.001 |
| Mean PA pressure | 34 (24, 41) | 30 (24, 37) | 30 (24, 37) | 29 (22, 34) | 30 (24, 37) | 9352 | < 0.001 |
| Cardiac index | 1.83 (1.50, 2.32) | 1.80 (1.50, 2.30) | 1.80 (1.40, 2.39) | 1.80 (1.40, 2.20) | 1.80 (1.46, 2.30) | 10 409 | 0.774 |
| Cardiac power index | 0.29 (0.21, 0.37) | 0.27 (0.21, 0.36) | 0.26 (0.20, 0.35) | 0.25 (0.19, 0.33) | 0.27 (0.20, 0.35) | 10 594 | 0.048 |
| Laboratory data within 6 h prior to ECMO | |||||||
| PO2 (mmHg) | 90 (64, 162) | 98 (68, 186) | 102 (67, 212) | 150 (77, 297) | 100 (68, 199) | 2766 | < 0.001 |
| SaO2 | 96 (88, 99) | 96 (90, 99) | 96 (90, 99) | 97 (93, 99) | 96 (90, 99) | 3970 | < 0.001 |
| Blood pH | 7.28 (7.17, 7.38) | 7.29 (7.18, 7.38) | 7.27 (7.17, 7.36) | 7.30 (7.21, 7.37) | 7.29 (7.18, 7.37) | 2518 | 0.923 |
| Blood pH <7.2 | 181 (31.2%) | 1635 (27.8%) | 739 (30.0%) | 129 (19.7%) | 2684 (28.0%) | 2518 | < 0.001 |
| PCO2 (mmHg) | 42 (33, 50) | 40 (32, 49) | 40 (34, 49) | 41 (36, 49) | 40 (33, 49) | 2726 | 0.112 |
| Bicarbonate (mEq/L) | 19 (15, 24) | 19 (15, 23) | 19 (15, 22) | 21 (17, 24) | 19 (15, 23) | 2853 | 0.052 |
| Lactate (mmol/L) | 5.2 (2.3, 9.8) | 5.8 (2.9, 10.2) | 7.6 (3.9, 12.0) | 6.5 (3.7, 11.4) | 6.2 (3.1, 10.8) | 6939 | < 0.001 |
| Lactate ≥5 mmol/L | 129 (53.3%) | 1723 (56.1%) | 992 (67.2%) | 245 (65.0%) | 3089 (59.8%) | 6939 | < 0.001 |
| Lactate and blood pH group | 7057 | < 0.001 | |||||
| Lactate <5 and pH ≥7.2 | 89 (38.2%) | 1184 (39.6%) | 421 (29.0%) | 118 (31.9%) | 1812 (35.9%) | ||
| Lactate <5 and pH <7.2 | 18 (7.7%) | 126 (4.2%) | 58 (4.0%) | 12 (3.2%) | 214 (4.2%) | ||
| Lactate ≥5 and pH ≥7.2 | 64 (27.5%) | 1000 (33.4%) | 600 (41.3%) | 179 (48.4%) | 1843 (36.5%) | ||
| Lactate ≥5 and pH <7.2 | 62 (26.6%) | 682 (22.8%) | 375 (25.8%) | 61 (16.5%) | 1180 (23.4%) | ||
Data are displayed as median (interquartile range) for continuous variables and number (percent) for categorical variables. P-values represent the across-groups comparison for the Society for Cardiovascular Angiography and Intervention shock stages using the Kruskal–Wallis test for continuous variables and Pearson chi-square test for categorical variables.
Cardiac arrest preceding ECMO occurred in 36.1% of patients. Ventilator support was required in 93.1% for a median of 8 (2, 22) hours when ECMO was initiated, including 18.7% who had intubation duration ≥30 h. ECMO was discontinued due to complications in 1.0%, and 6.5% received a heart or lung transplant. In the 6 h prior to ECMO initiation, lactate [n = 5167 (42.7%)] was ≥5 mmol/L in 59.8% and pH [n = 9588 (79.2%)] was <7.2 in 28.0%. Prior to ECMO initiation, 83.2% of patients were on vasoactive drugs (Table 2), including 78.5% on vasopressors and 41.1% on inodilators. MCS devices were used prior to ECMO initiation in 59.4% of patients, including intra-aortic balloon pump in 31.7% and percutaneous ventricular assist device (VAD) in 13.0%.
Table 2.
Pre-ECMO haemodynamic support strategies used in the final study population, and comparison of patients in each Society for Cardiovascular Angiography and Intervention shock stage
| SCAI Stage B (N = 821) | SCAI Stage C (N = 7518) | SCAI Stage D (N = 2973) | SCAI Stage E (N = 794) | Total (N = 12 106) | P-value | |
|---|---|---|---|---|---|---|
| Vasoactives prior to ECMO | 257 (31.3%) | 6348 (84.4%) | 2678 (90.1%) | 794 (100.0%) | 10 077 (83.2%) | < 0.001 |
| # of vasoactives | 0 (0, 1) | 2 (1, 2) | 3 (3, 4) | 3 (3, 4) | 2 (1, 3) | < 0.001 |
| Vasoactives | < 0.001 | |||||
| 0 | 564 (68.7%) | 1170 (15.6%) | 295 (9.9%) | 0 (0.0%) | 2029 (16.8%) | |
| 1 | 214 (26.1%) | 1700 (22.6%) | 145 (4.9%) | 0 (0.0%) | 2059 (17.0%) | |
| 2 | 43 (5.2%) | 2951 (39.3%) | 112 (3.8%) | 185 (23.3%) | 3291 (27.2%) | |
| 3 | 0 (0.0%) | 1506 (20.0%) | 1045 (35.1%) | 309 (38.9%) | 2860 (23.6%) | |
| 4 | 0 (0.0%) | 187 (2.5%) | 993 (33.4%) | 211 (26.6%) | 1391 (11.5%) | |
| 5+ | 0 (0.0%) | 4 (0.1%) | 383 (12.9%) | 89 (11.2%) | 476 (3.9%) | |
| Vasopressors prior to ECMO | 0 (0.0%) | 6058 (80.6%) | 2653 (89.2%) | 794 (100.0%) | 9505 (78.5%) | < 0.001 |
| # of vasopressors | 0 (0, 0) | 1 (1, 2) | 3 (3, 3) | 2 (2, 3) | 2 (1, 2) | < 0.001 |
| Vasopressors | < 0.001 | |||||
| 0 | 821 (100.0%) | 1460 (19.4%) | 320 (10.8%) | 0 (0.0%) | 2601 (21.5%) | |
| 1 | 0 (0.0%) | 2742 (36.5%) | 251 (8.4%) | 0 (0.0%) | 2993 (24.7%) | |
| 2 | 0 (0.0%) | 3316 (44.1%) | 0 (0.0%) | 410 (51.6%) | 3726 (30.8%) | |
| 3 | 0 (0.0%) | 0 (0.0%) | 1815 (61.0%) | 287 (36.1%) | 2102 (17.4%) | |
| 4+ | 0 (0.0%) | 0 (0.0%) | 587 (19.7%) | 97 (12.2%) | 684 (5.7%) | |
| Epinephrine | 0 (0.0%) | 3385 (45.0%) | 2324 (78.2%) | 695 (87.5%) | 6404 (52.9%) | < 0.001 |
| Dopamine | 0 (0.0%) | 809 (10.8%) | 890 (29.9%) | 174 (21.9%) | 1873 (15.5%) | < 0.001 |
| Norepinephrine | 0 (0.0%) | 4257 (56.6%) | 2385 (80.2%) | 684 (86.1%) | 7326 (60.5%) | < 0.001 |
| Phenylephrine | 0 (0.0%) | 207 (2.8%) | 686 (23.1%) | 153 (19.3%) | 1046 (8.6%) | < 0.001 |
| Vasopressin | 0 (0.0%) | 706 (9.4%) | 1809 (60.8%) | 378 (47.6%) | 2893 (23.9%) | < 0.001 |
| Inodilators prior to ECMO | 257 (31.3%) | 3097 (41.2%) | 1173 (39.5%) | 448 (56.4%) | 4975 (41.1%) | < 0.001 |
| # of inodilators | 0 (0, 1) | 0 (0, 1) | 0 (0, 1) | 1 (0, 1) | 0 (0, 1) | < 0.001 |
| Inodilators | < 0.001 | |||||
| 0 | 564 (68.7%) | 4421 (58.8%) | 1800 (60.5%) | 346 (43.6%) | 7131 (58.9%) | |
| 1 | 214 (26.1%) | 2690 (35.8%) | 1006 (33.8%) | 378 (47.6%) | 4288 (35.4%) | |
| 2+ | 43 (5.2%) | 407 (5.4%) | 167 (5.6%) | 70 (8.8%) | 687 (5.7%) | |
| Dobutamine | 158 (19.2%) | 1951 (26.0%) | 737 (24.8%) | 229 (28.8%) | 3075 (25.4%) | < 0.001 |
| Milrinone | 137 (16.7%) | 1445 (19.2%) | 570 (19.2%) | 281 (35.4%) | 2433 (20.1%) | < 0.001 |
| MCS prior to ECMO | 0 (0.0%) | 4449 (59.2%) | 1942 (65.3%) | 794 (100.0%) | 7185 (59.4%) | < 0.001 |
| # of MCS devices | 0 (0, 0) | 1 (0, 1) | 1 (0, 1) | 2 (2, 2) | 1 (0, 1) | < 0.001 |
| MCS | < 0.001 | |||||
| 0 | 821 (100.0%) | 3069 (40.8%) | 1031 (34.7%) | 0 (0.0%) | 4921 (40.6%) | |
| 1 | 0 (0.0%) | 4449 (59.2%) | 1371 (46.1%) | 0 (0.0%) | 5820 (48.1%) | |
| 2+ | 0 (0.0%) | 0 (0.0%) | 571 (19.2%) | 794 (100.0%) | 1365 (11.3%) | |
| IABP | 0 (0.0%) | 2140 (28.5%) | 1045 (35.1%) | 649 (81.7%) | 3834 (31.7%) | < 0.001 |
| Percutaneous VAD | 0 (0.0%) | 890 (11.8%) | 478 (16.1%) | 209 (26.3%) | 1577 (13.0%) | < 0.001 |
| Surgical VAD | 0 (0.0%) | 446 (5.9%) | 237 (8.0%) | 197 (24.8%) | 880 (7.3%) | < 0.001 |
| Cardiopulmonary bypass | 0 (0.0%) | 948 (12.6%) | 763 (25.7%) | 581 (73.2%) | 2292 (18.9%) | < 0.001 |
Data are displayed as median (interquartile range) for continuous variables and number (percent) for categorical variables. P-values represent the across-groups comparison for the Society for Cardiovascular Angiography and Intervention shock stages using the Kruskal–Wallis test for continuous variables and Pearson chi-square test for categorical variables.
SCAI shock stages
Using our SCAI shock classification schema, the distribution of SCAI shock stages at the time of ECMO initiation was: B, 821 (6.8%); C, 7518 (62.1%); D, 2973 (24.6%); and E, 794 (6.6%). The distribution of SCAI shock stages differed for post-cardiotomy vs. medical patients, with a shift toward higher SCAI shock stages in the post-cardiotomy group (see Supplementary material online, Figure S2). As the SCAI shock stage increased, patients were older, were more likely to require ventilator or RRT and had lower blood pressure; preceding cardiac arrest was less common in SCAI shock stage E. Metabolic parameters (pH, lactate, bicarbonate) were worst in SCAI shock Stage D. In the 1697 (14.0%) patients with available data, the cardiac index and cardiac power index did not vary across SCAI shock stages.
In-hospital mortality and other outcomes
During the index hospitalization, a total of 6681 (55.2%) patients died and 4706 (38.9%) had native heart survival. Among the 5268 (97.1%) hospital survivors with available data, 1970 (37.4%) were discharged home. Predictors of unadjusted in-hospital mortality are displayed in Supplementary material online, Table S2. The risk of in-hospital mortality increased incrementally with each vasopressor (adjusted OR: 1.14, 95% CI: 1.10–1.18, P < 0.001) or MCS device (adjusted OR: 1.09; 95% CI: 1.02–1.16; P = 0.01) prior to ECMO initiation (see Supplementary material online, Figure S3). Conversely, native heart survival decreased incrementally with each vasopressor (adjusted OR: 0.92, 95% CI: 0.89–0.95, P < 0.001) or MCS device (adjusted OR: 0.88; 95% CI: 0.82–0.94; P < 0.001). Use of inodilators prior to ECMO was associated with slightly lower in-hospital mortality (adjusted OR: 0.92, 95% CI: 0.85–1.00, P = 0.05) and was not associated with native heart survival (adjusted OR: 0.96, 95% CI: 0.90–1.02, P = 0.20).
The risk of in-hospital mortality increased progressively with higher SCAI shock stage (adjusted OR: 1.24 per SCAI shock stage, 95% CI: 1.17–1.32, P < 0.001; Figure 1). Compared with patients in SCAI shock stage B, patients in each higher SCAI shock stage were at incrementally greater risk of adjusted in-hospital mortality (Table 3). A sensitivity analysis including only data from the first ECMO run demonstrated similar findings (adjusted OR: 1.25 per SCAI shock stage, 95% CI: 1.17–1.33, P < 0.001). A further sensitivity analysis where all patients with SCAI shock stage B were reclassified as SCAI shock Stage C was consistent (adjusted OR: 1.27 per SCAI shock stage, 95% CI: 1.18–1.36, P <0.001).
Table 3.
Predictors of in-hospital mortality on multivariable logistic regression
| Variable | Adjusted OR | Upper 95% CI | Upper 95% CI | P-value |
|---|---|---|---|---|
| Age (per year) | 1.028 | 1.025 | 1.031 | <0.001 |
| Female sex | 1.062 | 0.973 | 1.159 | 0.175 |
| White race | 0.871 | 0.801 | 0.946 | 0.001 |
| Year (per year) | 0.930 | 0.915 | 0.946 | < 0.001 |
| Weight (per kg) | 1.005 | 1.003 | 1.006 | < 0.001 |
| First run | 0.706 | 0.537 | 0.927 | 0.012 |
| Primary diagnosis of cardiogenic shock | 0.977 | 0.901 | 1.058 | 0.561 |
| Post-cardiotomy | 1.177 | 1.073 | 1.291 | 0.001 |
| Cardiac arrest | 1.141 | 1.050 | 1.239 | 0.002 |
| Heart/lung transplant | 0.642 | 0.544 | 0.759 | < 0.001 |
| Renal replacement therapy | 1.833 | 1.597 | 2.102 | < 0.001 |
| Inodilators | 0.925 | 0.854 | 1.001 | 0.054 |
| SCAI shock stage (per each) | 1.242 | 1.169 | 1.320 | < 0.001 |
| Versus SCAI shock Stage B: | ||||
| SCAI shock Stage C | 1.162 | 0.990 | 1.364 | 0.066 |
| SCAI shock Stage D | 1.574 | 1.322 | 1.875 | < 0.001 |
| SCAI shock Stage E | 1.637 | 1.305 | 2.054 | < 0.001 |
Data displayed as adjusted odds ratio (OR) and 95% confidence interval (CI). This analysis included 11 134 patients, as 972 had missing data for one or more variables. Final model C-statistic 0.65 for discrimination.
Native heart survival decreased progressively with higher SCAI shock stage (adjusted OR: 0.85 per SCAI shock stage, 95% CI: 0.80–0.90, P < 0.001; see Supplementary material online, Table S3). Discharge disposition among hospital survivors differed across the SCAI Shock stages (see Supplementary material online, Figure S4), as did reasons for ECMO discontinuation (Table 1). Among hospital survivors, the likelihood of being discharged home decreased progressively with higher SCAI shock stage (adjusted OR: 0.85 per SCAI shock stage, 95% CI: 0.77–0.94, P = 0.002). In-hospital survival decreased incrementally with higher SCAI shock stages (see Supplementary material online, Figure S5). The median in-hospital survival by SCAI shock stage was: B, 44 days; C, 30 days; D, 22 days; E, 21 days (P <0.001 by log-rank).
Subgroup analyses
The SCAI shock classification produced a graded relationship with in-hospital mortality in medical CS (adjusted OR: 1.34, 95% CI: 1.24–1.46, P < 0.001) but not in post-cardiotomy CS (adjusted OR: 1.09, 95% CI: 0.99–1.20, P = 0.06; P < 0.001 for interaction). Post-cardiotomy CS was associated with higher in-hospital mortality overall (adjusted OR: 1.18, 95% CI: 1.07–1.29, P = 0.001) and in SCAI shock Stage B through D but not SCAI shock stage E (Figure 2). The SCAI shock classification displayed a similar inverse relationship with native heart survival in medical CS (adjusted OR: 0.83, 95% CI: 0.77–0.90, P < 0.001) and post-cardiotomy CS (adjusted OR: 0.87, 95% CI: 0.79–0.96, P = 0.006); post-cardiotomy CS was not associated with native heart survival (adjusted OR: 0.94, 95% CI: 0.86–1.03, P = 0.19).
Figure 2.
Observed in-hospital mortality as a function of Society for Cardiovascular Angiography and Intervention shock stage in patients with non-surgical/medical versus post-cardiotomy cardiogenic shock. * denotes P < 0.05 between groups.
Patients with pre-ECMO cardiac arrest were at higher risk of in-hospital death overall (adjusted OR: 1.14, 95% CI: 1.05–1.24, P = 0.002; Figure 4) and in each SCAI shock stage (P < 0.05 in SCAI shock stage B/C and P >0.1 in SCAI shock stage D/E). After adjustment, the SCAI shock classification remained associated with higher in-hospital mortality in patients with (adjusted OR: 1.23, 95% CI: 1.12–1.36, P < 0.001) and without (adjusted OR: 1.25, 95% CI: 1.16–1.34, P < 0.001) pre-ECMO cardiac arrest. Similarly, pre-ECMO cardiac arrest was associated with lower native heart survival (adjusted OR: 0.90, 95% CI: 0.93–0.98, P = 0.01), and SCAI shock classification displayed a similar inverse relationship with native heart survival in patients with (adjusted OR: 0.85, 95% CI: 0.76–0.94, P = 0.001) and without (adjusted OR: 0.84, 95% CI: 0.78–0.91, P < 0.001) pre-ECMO cardiac arrest.
Figure 4.
Observed in-hospital mortality as a function of Society for Cardiovascular Angiography and Intervention shock stage in patients with and without cardiac arrest prior to extracorporeal membrane oxygenator initiation. * denotes P < 0.05 between patients with and without cardiac arrest.
The unadjusted association between higher SCAI shock stage with increased in-hospital mortality was present in all other tested subgroups including groups stratified by age, sex, and year of cannulation (Table 4). Although in-hospital mortality decreased in the population over time, the SCAI shock classification was incrementally associated with in-hospital mortality in each era (Figure 3), and this association was stronger in the most recent years (P = 0.001 for interaction). The SCAI shock classification remained associated with in-hospital mortality across age groups (see Supplementary material online, Figure S6) and in both males and females, with no difference in mortality between the sexes in any SCAI shock stage (data not shown). Patients who received RRT during hospitalization were at elevated risk of in-hospital mortality overall and in each SCAI shock stage (see Supplementary material online, Figure S7; all P < 0.05); patients in SCAI shock Stage E who received RRT had nearly 80% in-hospital mortality.
Table 4.
Unadjusted odds ratio and 95% confidence interval values for prediction of in-hospital mortality by logistic regression using the Society for Cardiovascular Angiography and Intervention shock stages, stratified by subgroup with unadjusted interaction P-values
| Subgroup | Unadjusted OR per SCAI shock stage | Lower 95% CI | Upper 95% CI | P-value | Number missing | Interaction P-value |
|---|---|---|---|---|---|---|
| Overall | 1.32 | 1.25 | 1.39 | < 0.001 | 0 | |
| Age | 0 | 0.99 | ||||
| <50 years | 1.24 | 1.12 | 1.37 | < 0.001 | ||
| 50–65 years | 1.28 | 1.18 | 1.40 | < 0.001 | ||
| >=65 years | 1.25 | 1.13 | 1.38 | < 0.001 | ||
| Sex | 0.52 | |||||
| Male | 1.30 | 1.22 | 1.39 | < 0.001 | 29 | |
| Female | 1.35 | 1.23 | 1.48 | < 0.001 | ||
| Race | 556 | 0.56 | ||||
| Non-White | 1.29 | 1.17 | 1.41 | < 0.001 | ||
| White | 1.33 | 1.25 | 1.43 | < 0.001 | ||
| Year | 0 | 0.001 | ||||
| 2010–2015 | 1.18 | 1.04 | 1.35 | 0.0125 | ||
| 2016–2018 | 1.31 | 1.21 | 1.43 | < 0.001 | ||
| 2019–2020 | 1.50 | 1.38 | 1.63 | < 0.001 | ||
| Primary diagnosis of cardiogenic shock | 0 | 0.35 | ||||
| No | 1.36 | 1.25 | 1.48 | < 0.001 | ||
| Yes | 1.29 | 1.21 | 1.38 | < 0.001 | ||
| Post-cardiotomy | 0 | < 0.001 | ||||
| No | 1.38 | 1.28 | 1.48 | < 0.001 | ||
| Yes | 1.07 | 0.98 | 1.17 | 0.11 | ||
| Ventilator at ECMO initiation | 2606 | 0.02 | ||||
| No | 1.82 | 1.39 | 2.39 | < 0.001 | ||
| Yes | 1.32 | 1.24 | 1.41 | < 0.001 | ||
| Heart or lung transplant | 1104 | 0.49 | ||||
| No | 1.30 | 1.22 | 1.37 | < 0.001 | ||
| Yes | 1.41 | 1.12 | 1.77 | 0.0032 | ||
| Cardiac arrest | 185 | 0.37 | ||||
| No | 1.34 | 1.26 | 1.43 | < 0.001 | ||
| Yes | 1.28 | 1.17 | 1.40 | < 0.001 | ||
| Renal replacement therapy | 0 | 0.11 | ||||
| No | 1.28 | 1.21 | 1.35 | < 0.001 | ||
| Yes | 1.49 | 1.24 | 1.78 | < 0.001 | ||
| Lactate >=5 mmol/L | 6939 | 0.81 | ||||
| No | 1.41 | 1.24 | 1.61 | < 0.001 | ||
| Yes | 1.38 | 1.24 | 1.54 | < 0.001 | ||
| Blood pH <7.2 | 2518 | 0.70 | ||||
| No | 1.35 | 1.26 | 1.44 | < 0.001 | ||
| Yes | 1.38 | 1.22 | 1.56 | < 0.001 |
Figure 3.
Observed in-hospital mortality as a function of year of extracorporeal membrane oxygenator initiation and Society for Cardiovascular Angiography and Intervention shock stage.
Lactic acidosis prior to ECMO initiation
In-hospital mortality was higher in those with lactate ≥5 mmol/l overall and in each SCAI shock stage (see Supplementary material online, Figure S8A; P = 0.08 for SCAI shock Stage B, all others P < 0.05). In-hospital mortality was higher in those with pH <7.2 overall and in each SCAI shock stage (see Supplementary material online, Figure S8B; P < 0.001 in SCAI shock Stage C/D and P > 0.1 in SCAI shock Stage B/E). Among patients with both pH and lactate data (n = 5049), in-hospital mortality was higher in those with pH <7.2 and/or lactate ≥5 mmol/L overall and in each SCAI shock stage (Figure 5; P < 0.001 in SCAI shock stage C/D and P > 0.1 in SCAI shock stage B/E). Further adjusting for pre-ECMO lactate did not diminish the association between SCAI shock stage and either in-hospital mortality (adjusted OR: 1.26, 95% CI: 1.14–1.38, P < 0.001) or native heart survival (adjusted OR: 0.80, 95% CI: 0.73–0.89, P < 0.001).
Figure 5.
Observed in-hospital mortality as a function of Society for Cardiovascular Angiography and Intervention shock stage and group based on lactate level (</≥5 mmol/L) and blood pH (</≥7.2).
Among patients with available lactate data (n = 5167), the distribution of modified SCAI shock stages was: B, 40 (0.8%); C, 523 (10.1%); D, 1515 (29.3%); E, 3089 (59.7%). The modified SCAI shock classification incorporating lactate (see Supplementary material online, Figure S9) was strongly associated with in-hospital mortality (adjusted OR: 1.49, 95% CI: 1.37–1.63, P < 0.001) and native heart survival (adjusted OR: 0.71, 95% CI: 0.65–0.78, P < 0.001). Among patients with available lactate data, the modified SCAI Shock Classification had a slightly higher C-statistic than the original SCAI shock classification for discrimination of in-hospital mortality (0.58 vs. 0.56) or native heart survival (0.56 vs. 0.55).
Discussion
Summary of findings
In this analysis of a multinational registry of VA ECMO recipients with CS encompassing more than 12 000 ECMO patient runs, we found that the SCAI shock classification prior to ECMO initiation (based on the modified CSWG definition) provided robust mortality risk stratification. Most included patients were classified as SCAI shock stage C and D, suggesting that most providers initiate ECMO prior to development of refractory CS. In-hospital mortality was incrementally higher, and native heart survival incrementally lower, as SCAI shock stage increased, with patients classified as SCAI shock Stage E having the worst adjusted outcomes. The SCAI shock classification produced mortality risk stratification across all age groups, in both sexes, after cardiac arrest, and among patients separated by pH and lactate levels. However, the SCAI shock stage was only associated with mortality in patients receiving ECMO for non-surgical (medical) CS and was not predictive in post-cardiotomy patients. While a higher number of vasopressors and MCS devices were independently associated with increasing in-hospital mortality, the simplified SCAI shock classification provides better ease of use in clinical practice and incorporation of lactate levels when available appears to improve prognostication. This is among the largest outcome studies in CS patients receiving VA ECMO, underscoring the potential usefulness of the SCAI shock classification for predicting subsequent in-hospital mortality for CS patients who receive ECMO.
Application of the SCAI shock classification in ECMO patients
Since its publication in 2019, the SCAI shock classification has been validated as a useful mortality risk stratification tool in acutely ill populations, including patients with CS or cardiac arrest.7,8,11–13,15 Our results extend these findings to CS patients receiving VA ECMO, mirroring a recent study that demonstrated that the SCAI shock classification effectively stratified the risk of early and late mortality, the likelihood of myocardial recovery, and the occurrence of adverse events in 245 patients who received VA ECMO for CS.19 Importantly, we demonstrated the scalability of a simple SCAI shock classification based on the previously validated CSWG approach to a large multicentre registry that represents one of the largest analyses of the SCAI shock classification. By incorporating fewer variables, this SCAI shock classification can be easily applied to large datasets, allowing mortality risk stratification even if detailed information regarding other relevant variables is not available.
Our SCAI shock classification was only effective for mortality risk stratification among non-surgical (medical) CS patients, and not among those with post-cardiotomy CS. The aetiology of CS is a known predictor of outcomes in patients receiving VA ECMO, and post-cardiotomy CS patients typically have worse outcomes as we observed.23–25 The pathophysiology of post-cardiotomy CS is unique including differing underlying disease mechanisms and often manifesting a vasodilatory or mixed shock component.1 Strategies for vasoactive drug and temporary MCS use may differ from medical CS and based on our data it appears that shock severity (as quantified based on the number of vasopressors and MCS devices) is not the primary determinant of outcome in post-cardiotomy CS. As such, the SCAI shock classification should primarily be applied to non-surgical (medical) CS rather than post-cardiotomy CS when considering outcomes after ECMO support. Our proposed application of the SCAI shock classification may be used to quantify shock severity at the time of ECMO initiation for outcomes research.
This analysis and prior studies have not demonstrated a robust association between SCAI shock stage and standard measures of systemic hemodynamics in patients with CS, and the SCAI shock classification carries a stronger association with outcomes than traditional haemodynamic measurements.12,14 Patients with similar haemodynamics may reside in any of the SCAI shock stages, and each patient’s physiological compensation determines their ability to maintain perfusion in the face of an acute insult causing haemodynamic compromise. This analysis suggests that the degree to which patients need haemodynamic support (as reflected by the SCAI shock stage) is the more essential determinant of shock severity than measurements such as cardiac index, and the SCAI shock classification may be more helpful for determining the need for ECMO than haemodynamics alone.
Risk modifiers beyond the SCAI shock classification
Prognostic variables that are not directly related to shock severity yet provide additional risk stratification when combined with the SCAI shock classification are labelled risk modifiers.6,7 Age is the quintessential non-modifiable risk modifier, and has been consistently identified as a predictor of higher mortality even when patients are stratified by SCAI shock stage.7,9,14,26 We observed a strong gradient of mortality when patients were stratified by both age and SCAI shock stage, consistent with prior analyses highlighting the importance of age as a risk marker in patients receiving ECMO.23,27 Cardiac arrest, with the potential for anoxic brain injury as a determinant of outcomes, is another strong predictor of adverse outcomes across the SCAI shock classification.6–8,11,15 Cardiac arrest is a risk factor for adverse outcomes in patients receiving ECMO, particularly non-shockable rhythms.23,27 We observed a modest association between preceding cardiac arrest and higher in-hospital mortality that was primarily restricted to patients with lower shock severity (SCAI shock stage B/C); this may reflect selection of cardiac arrest patients with more favourable characteristics for ECMO.
The severity of lactic acidosis is a marker of hypoperfusion associated with higher mortality in patients with CS, including patients receiving ECMO.10,27–30 Patients with CS and severe lactic acidosis (defined by a high lactate and/or low blood pH) had increased mortality across the SCAI shock stages in this ECMO cohort, likely reflecting haemometabolic CS which is a known predictor of adverse outcomes.10,13 We observed particularly poor outcomes for VA ECMO patients who received RRT, consistent with the known association between the need for RRT and poor outcomes in CS.31,32 Consistent with prior studies, mortality for patients receiving VA ECMO improved over time, but the association between SCAI shock stage and mortality was stronger in recent years with a significant interaction.33 The severity of shock at the time of VA ECMO initiation remains an important predictor of prognosis in the contemporary era.
Vasoactive drug load, temporary MCS, and ECMO outcomes
Prior studies of shock severity in patients receiving ECMO have predominantly focused on vasopressor load, quantified using the Vasoactive-Inotropic Score (VIS).17,18 Patients who require greater vasopressor doses (defined by a higher VIS) prior to ECMO have higher mortality.18 Vasopressor load (including the VIS) is a risk factor for mortality in critically ill patients with cardiac disease, and can be integrated into the SCAI shock classification.7,8,34 We observed an incremental association between a higher number of vasopressors and in-hospital mortality in this cohort, although inodilator use was not associated with higher mortality. While it remains possible that use of a greater number of vasopressors is directly harmful in patients receiving ECMO, instead we believe that the need for greater degrees of vasopressor support defines more severe shock that itself predicts worse outcomes.12 The SCAI shock classification has important advantages over the simple quantification of vasopressor load, which can be influenced by provider treatment choices and blood pressure goals.6,7 By including the use of other MCS devices before ECMO initiation, our SCAI shock classification (modified from that used by the CSWG) is potentially more informative than vasopressor dosing alone.12,13,16 The current retrospectively derived SCAI shock classification would ideally include objective markers of tissue perfusion (such as lactate levels) to define SCAI shock stages C and above, and when we integrated lactate into the SCAI shock classification among the minority of patients with available data, the performance for risk stratification appears to have improved.6,8
Temporary MCS devices, particularly percutaneous VADs, have been associated with a substantial risk of complications in patients with CS.35,36 The higher mortality in patients who received a greater number of temporary MCS devices prior to initiation of ECMO could be interpreted as evidence that these devices are potentially harmful. However, several studies have suggested a benefit from combining other temporary MCS devices with ECMO to unload the left ventricle.37,38 Therefore, we believe that the need for multiple temporary MCS devices identifies patients with truly refractory shock who would be expected to have poor outcomes, and we propose that earlier initiation of VA ECMO for patients with CS who are failing an initial temporary MCS device may be preferred over waiting until a patient has failed more than one temporary MCS device before initiating VA ECMO. We speculate that initiation of VA ECMO at an earlier SCAI shock stage might be associated with improved outcomes, and we eagerly await the results of currently enrolling randomized trials examining the use of ECMO in CS. Because we only included patients who were selected to receive ECMO, our results should not be used to justify providing or refusing ECMO for individual patients. Future studies examining CS patients who receive VA ECMO should integrate the SCAI shock classification, potentially to stratify randomization in clinical trials.
Strengths and limitations
The multicentre, multinational design of the ELSO registry is an important strength of this analysis, which greatly improves the generalizability of our findings although these data may not reflect outcomes in centres that do not participate in the ELSO registry.21,22 This study represents one of the largest reported cohorts of patients receiving VA ECMO for CS, allowing us to examine several relevant subgroups of interest; to our knowledge, this is largest study of shock severity in ECMO recipients, and one of the largest studies examining the SCAI shock classification. We examined a heterogeneous CS cohort including patients receiving VA ECMO for diverse indications, including both non-surgical (medical) CS and post-cardiotomy CS, and had limited data regarding the cause of CS or details such as revascularization status and type of surgery. We excluded patients with missing data to determine the SCAI shock stage, which could have led to selection bias. Due to the high prevalence of missing data for physiological and other variables, we could not perform an extensive multivariable adjustment or calculate the SAVE score (which was derived to predict mortality risk for patients who receive VA ECMO).24 The CSWG SCAI classification that we emulated carries its own limitations, namely that it does not assess vasopressor doses (which can be influenced by provider preferences and local practice patterns) and does not directly incorporate standard measures of hypoperfusion; nonetheless, this approach has been validated in multiple prior studies and was incrementally associated with mortality in our analysis.12,13,16 Data regarding the sequence or duration of temporary MCS device use and doses of vasoactive drugs were not available. The C-statistics for discrimination of mortality by our regression models were modest; while we believe that this reflects the challenges in mortality risk stratification among ECMO recipients with a high baseline risk of death, it may represent poor model fit due to unmeasured confounders or other determinants of prognosis that we could not include. We did not have data on cause of death, neurological complications, or other relevant outcome measures. Finally, we lacked information regarding the specific indication for ECMO, pre-ECMO comorbidities, cardiac arrest circumstances, markers of frailty, cannula configuration, post-cannulation complications, complete laboratory values or standard measures of severity of illness, all of which are prognostically important and influence ECMO candidacy.
Conclusions
In the ELSO registry, VA ECMO is utilized across the spectrum of CS severity and not just limited to patients with refractory shock. Shock severity prior to ECMO initiation, as quantified using the SCAI shock classification, was a strong independent predictor of in-hospital mortality in non-surgical (medical) CS patients who received VA ECMO for cardiac support. The pre-cannulation SCAI shock classification, particularly when coupled with established risk modifiers, can help physicians refine prognostication for ECMO recipients and allow for mortality benchmarking. Further studies are needed to determine whether using the SCAI shock stage to tailor the approach to temporary MCS selection will result in an improvement in clinical outcomes.
Supplementary Material
Contributor Information
Jacob C Jentzer, Department of Cardiovascular Medicine, Mayo Clinic, 200 First Street SW, Rochester, MN 55905, USA.
David A Baran, Heart and Vascular Institute, Cleveland Clinic Florida, 2950 Cleveland Clinic Blvd, Weston, FL 33331, USA.
J Kyle Bohman, Department of Anesthesiology and Perioperative Medicine, Mayo Clinic, 200 First Street SW, Rochester, MN 55905, USA.
Sean van Diepen, Extracorporeal Life Support Organization (ELSO), ELSO Office, 3001 Miller Road, Ann Arbor, MI 48103, USA.
Misty Radosevich, Department of Anesthesiology and Perioperative Medicine, Mayo Clinic, 200 First Street SW, Rochester, MN 55905, USA.
Suraj Yalamuri, Department of Anesthesiology and Perioperative Medicine, Mayo Clinic, 200 First Street SW, Rochester, MN 55905, USA.
Peter Rycus, Department of Critical Care Medicine and Division of Cardiology, Department of Medicine, University of Alberta, 8440 112 St NW, Edmonton, AB T6G 2B7, Canada.
Stavros G Drakos, Divisions of Cardiothoracic Surgery and Emergency Medicine, University of Utah Hospital, 50 Medical Dr N, Salt Lake City, UT 84132, USA.
Joseph E Tonna, Divisions of Cardiothoracic Surgery and Emergency Medicine, University of Utah Hospital, 50 Medical Dr N, Salt Lake City, UT 84132, USA.
Supplementary material
Supplementary material is available at European Heart Journal: Acute Cardiovascular Care online.
Disclosures
Dr. Drakos has been a Consultant for Abbott and has research support from Novartis, Merck, NIH, AHA, Department of Veterans Affairs and Nora Eccles Treadwell Foundation. Dr. Baran reports consulting for Getinge, Livanova, Abbott, Abiomed, Impulse Dynamics; serving on Steering committee for Procyrion and CareDx; and has been a speaker for Pfizer. Dr. Tonna received research support from the NIH. The other authors have no relevant financial disclosures.
Data availability
All data are incorporated into the article and its online Supplementary material.
References
- 1. van Diepen S, Katz JN, Albert NM, Henry TD, Jacobs AK, Kapur NK, Kilic A, Menon V, Ohman EM, Sweitzer NK, Thiele H, Washam JB, Cohen MG, American Heart Association Council on Clinical Cardiology, Council on Cardiovascular and Stroke Nursing, Council on Quality of Care and Outcomes Research; and Mission: Lifeline . Contemporary management of cardiogenic shock: a scientific statement from the American heart association. Circulation 2017;136:e232–e268. [DOI] [PubMed] [Google Scholar]
- 2. Jentzer JC, Ahmed AM, Vallabhajosyula S, Burstein B, Tabi M, Barsness GW, Murphy JG, Best PJ, Bell MR. Shock in the cardiac intensive care unit: changes in epidemiology and prognosis over time. Am Heart J 2021;232:94–104. [DOI] [PubMed] [Google Scholar]
- 3. Tyler JM, Brown C, Jentzer JC, Baran DA, van Diepen S, Kapur NK, Garberich RF, Garcia S, Sharkey SW, Henry TD. Variability in reporting of key outcome predictors in acute myocardial infarction cardiogenic shock trials. Catheter Cardiovasc Interv 2022;99(1):19–26. [DOI] [PubMed] [Google Scholar]
- 4. Rihal CS, Naidu SS, Givertz MM, Szeto WY, Burke JA, Kapur NK, Kern M, Garratt KN, Goldstein JA, Dimas V, Tu T, Society for Cardiovascular Angiography and Interventions (SCAI), Heart Failure Society of America (HFSA), Society of Thoracic Surgeons (STS), American Heart Association (AHA), American College of Cardiology (ACC) . 2015. SCAI/ACC/HFSA/STS clinical expert consensus statement on the use of percutaneous mechanical circulatory support devices in cardiovascular care: endorsed by the American heart association, the cardiological society of India, and sociedad latino Americana de cardiologia intervencionista; affirmation of value by the Canadian association of interventional cardiology-association canadienne de cardiologie d'intervention. J Am Coll Cardiol 2015;65:2140–2141. [DOI] [PubMed] [Google Scholar]
- 5. Wiley BM, Eckman PM, Jentzer JC. Vasopressor load: sounding the alarm in management of cardiogenic shock associated with acute myocardial infarction. Crit Care Med 2021;49:865–869. [DOI] [PubMed] [Google Scholar]
- 6. Baran DA, Grines CL, Bailey S, Burkhoff D, Hall SA, Henry TD, Hollenberg SM, Kapur NK, O'Neill W, Ornato JP, Stelling K, Thiele H, van Diepen S, Naidu SS. SCAI Clinical expert consensus statement on the classification of cardiogenic shock: this document was endorsed by the American college of cardiology (ACC), the American heart association (AHA), the society of critical care medicine (SCCM), and the society of thoracic surgeons (STS) in April 2019. Catheter Cardiovasc Interv 2019;94:29–37. [DOI] [PubMed] [Google Scholar]
- 7. Naidu SS, Baran DA, Jentzer JC, Hollenberg SM, van Diepen S, Basir MB, Grines CL, Diercks DB, Hall S, Kapur NK, Kent W, Rao SV, Samsky MD, Thiele H, Truesdell AG, Henry TD. SCAI SHOCK stage classification expert consensus update: a review and incorporation of validation studies: this statement was endorsed by the American college of cardiology (ACC), American college of emergency physicians (ACEP), American heart association (AHA), European society of cardiology (ESC) association for acute cardiovascular care (ACVC), international society for heart and lung transplantation (ISHLT). society of critical care medicine (SCCM), and society of thoracic surgeons (STS) in December 2021. J Am Coll Cardiol. 2022. [DOI] [PubMed] [Google Scholar]
- 8. Jentzer JC, van Diepen S, Barsness GW, Henry TD, Menon V, Rihal CS, Naidu SS, Baran DA. Cardiogenic shock classification to predict mortality in the cardiac intensive care unit. J Am Coll Cardiol 2019;74:2117–2128. [DOI] [PubMed] [Google Scholar]
- 9. Jentzer JC, Schrage B, Holmes DR, Dabboura S, Anavekar NS, Kirchhof P, Barsness GW, Blankenberg S, Bell MR, Westermann D. Influence of age and shock severity on short-term survival in patients with cardiogenic shock. Eur Heart J Acute Cardiovasc Care 2021;10(6):604–612. [DOI] [PubMed] [Google Scholar]
- 10. Jentzer JC, Kashani KB, Wiley BM, Patel PC, Baran DA, Barsness GW, Henry TD, Van Diepen S. Laboratory markers of acidosis and mortality in cardiogenic shock: developing a definition of hemometabolic shock. Shock 2022;57(1):31–40. [DOI] [PubMed] [Google Scholar]
- 11. Jentzer JC, Henry TD, Barsness GW, Menon V, Baran DA, Van Diepen S. Influence of cardiac arrest and SCAI shock stage on cardiac intensive care unit mortality. Catheter Cardiovasc Interv 2020;96:1350–1359. [DOI] [PubMed] [Google Scholar]
- 12. Thayer KL, Zweck E, Ayouty M, Garan AR, Hernandez-Montfort J, Mahr C, Morine KJ, Newman S, Jorde L, Haywood JL, Harwani NM, Esposito ML, Davila CD, Wencker D, Sinha SS, Vorovich E, Abraham J, O'Neill W, Udelson J, Burkhoff D, Kapur NK. Invasive hemodynamic assessment and classification of in-hospital mortality risk among patients with cardiogenic shock. Circ Heart Fail 2020;13:e007099. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Zweck E, Thayer KL, Helgestad OKL, Kanwar M, Ayouty M, Garan AR, Hernandez-Montfort J, Mahr C, Wencker D, Sinha SS, Vorovich E, Abraham J, O'Neill W, Li S, Hickey GW, Josiassen J, Hassager C, Jensen LO, Holmvang L, Schmidt H, Ravn HB, Moller JE, Burkhoff D, Kapur NK. Phenotyping cardiogenic shock. J Am Heart Assoc 2021;10:e020085. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Baran DA, Long A, Badiye AP, Stelling K. Prospective validation of the SCAI shock classification: single center analysis. Catheter Cardiovasc Interv 2020;96(7):1339–1347. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Sarma D, Tabi M, Jentzer JC. Society for cardiovascular angiography and intervention shock classification predicts mortality after out-of-hospital cardiac arrest. Resuscitation 2022;172:101–105. [DOI] [PubMed] [Google Scholar]
- 16. Garan AR, Kanwar M, Thayer KL, Whitehead E, Zweck E, Hernandez-Montfort J, Mahr C, Haywood JL, Harwani NM, Wencker D, Sinha SS, Vorovich E, Abraham J, O’Neill W, Burkhoff D, Kapur NK. Complete hemodynamic profiling with pulmonary artery catheters in cardiogenic shock is associated with lower in-hospital mortality. JACC Heart Fail 2020;8:903–913. [DOI] [PubMed] [Google Scholar]
- 17. Choi KH, Yang JH, Park TK, Lee JM, Song YB, Hahn JY, Choi SH, Ko YG, Yu CW, Chun WJ, Jang WJ, Kim HJ, Bae JW, Kwon SU, Lee HJ, Lee WS, Jeong JO, Park SD, Cho S, Gwon HC. Differential prognostic implications of vasoactive inotropic score for patients with acute myocardial infarction complicated by cardiogenic shock according to use of mechanical circulatory support. Crit Care Med 2021;49(5):770–780. [DOI] [PubMed] [Google Scholar]
- 18. Hyun J, Kim AR, Lee SE, Hong JA, Kang PJ, Jung SH, Kim MS. Vasoactive-inotropic score as a determinant of timely initiation of venoarterial extracorporeal membrane oxygenation in patients with cardiogenic shock. Circ J 2022;86(4):687–694. [DOI] [PubMed] [Google Scholar]
- 19. Mehta S, Nemeth S, Kurlansky P, Fried J, Melehy A, Masoumi A, Sayer G, Uriel N, Naka Y, Char S, O'Connell G, Takeda K. SCAI Shock classification to predict outcomes of venoarterial extracorporeal membrane oxygenation. J Heart Lung Transplant 2021;40(4S):S184. [Google Scholar]
- 20. Thiagarajan RR, Barbaro RP, Rycus PT, McMullan DM, Conrad SA, Fortenberry JD, Paden ML and centers Em. Extracorporeal life support organization registry international report 2016. ASAIO J. 2017; 63:60–67. [DOI] [PubMed] [Google Scholar]
- 21. Tonna JE, Barbaro RP, Rycus PT, Wall N, Raman L, Nasr VG, Paden ML, Thiagarajan RR, Dalton H, Conrad SA, Bartlett RH, Toomasian JM, Alexander PMA. On the academic value of 30 years of the extracorporeal life support organization registry. ASAIO J 2021;67:1–3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Lorusso R, Alexander P, Rycus P, Barbaro R. The extracorporeal life support organization registry: update and perspectives. Ann Cardiothorac Surg 2019;8:93–98. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Alba AC, Foroutan F, Buchan TA, Alvarez J, Kinsella A, Clark K, Zhu A, Lau K, McGuinty C, Aleksova N, Francis T, Stanimirovic A, Vishram-Nielsen J, Malik A, Ross HJ, Fan E, Rac VE, Rao V, Billia F. Mortality in patients with cardiogenic shock supported with VA ECMO: a systematic review and meta-analysis evaluating the impact of etiology on 29,289 patients. J Heart Lung Transplant 2021;40:260–268. [DOI] [PubMed] [Google Scholar]
- 24. Schmidt M, Burrell A, Roberts L, Bailey M, Sheldrake J, Rycus PT, Hodgson C, Scheinkestel C, Cooper DJ, Thiagarajan RR, Brodie D, Pellegrino V, Pilcher D. Predicting survival after ECMO for refractory cardiogenic shock: the survival after veno-arterial-ECMO (SAVE)-score. Eur Heart J 2015;36:2246–2256. [DOI] [PubMed] [Google Scholar]
- 25. Kowalewski M, Zielinski K, Brodie D, MacLaren G, Whitman G, Raffa GM, Boeken U, Shekar K, Chen YS, Bermudez C, D’Alessandro D, Hou X, Haft J, Belohlavek J, Dziembowska I, Suwalski P, Alexander P, Barbaro RP, Gaudino M, Di Mauro M, Maessen J, Lorusso R. Venoarterial extracorporeal membrane oxygenation for postcardiotomy shock-analysis of the extracorporeal life support organization registry. Crit Care Med 2021;49:1107–1117. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Padkins M, Breen T, Anavekar N, van Diepen S, Henry TD, Baran DA, Barsness GW, Kashani K, Holmes DR Jr, Jentzer JC. Age and shock severity predict mortality in cardiac intensive care unit patients with and without heart failure. ESC Heart Fail 2020;7(6):3971–3982. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Sohail S, Fan E, Foroutan F, Ross HJ, Billia F, Alba AC. Predictors of mortality in patients treated with veno-arterial ECMO for cardiogenic shock complicating acute myocardial infarction: a systematic review and meta-analysis. J Cardiovasc Transl Res 2022;15(2):227–238. [DOI] [PubMed] [Google Scholar]
- 28. Burstein B, Vallabhajosyula S, Ternus B, Barsness GW, Kashani K, Jentzer JC. The prognostic value of lactate in cardiac intensive care unit patients with cardiac arrest and shock. Shock 2021;55:613–619. [DOI] [PubMed] [Google Scholar]
- 29. Omar HR, Handshoe JW, Tribble T, Guglin M. Survival on venoarterial extracorporeal membrane oxygenation in cardiogenic shock: which lactate is most useful? ASAIO J 2022;68(1):41–45. [DOI] [PubMed] [Google Scholar]
- 30. Beer BN, Jentzer JC, Weimann J, Dabboura S, Yan I, Sundermeyer J, Kirchhof P, Blankenberg S, Schrage B, Westermann D. Early risk stratification in patients with cardiogenic shock irrespective of the underlying cause - the cardiogenic shock score. Eur J Heart Fail 2022;24:657–667. [DOI] [PubMed] [Google Scholar]
- 31. Keleshian V, Kashani KB, Kompotiatis P, Barsness GW, Jentzer JC. Short, and long-term mortality among cardiac intensive care unit patients started on continuous renal replacement therapy. J Crit Care 2020;55:64–72. [DOI] [PubMed] [Google Scholar]
- 32. Lauridsen MD, Gammelager H, Schmidt M, Rasmussen TB, Shaw RE, Botker HE, Sorensen HT, Christiansen CF. Acute kidney injury treated with renal replacement therapy and 5-year mortality after myocardial infarction-related cardiogenic shock: a nationwide population-based cohort study. Crit Care 2015;19:452. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Jang EJ, Jung SY, Lee HJ, Kim MS, Na KJ, Park S, Park IK, Kang CH, Kim YT. Trends in extracorporeal membrane oxygenation application and outcomes in Korea. ASAIO J 2021;67:177–184. [DOI] [PubMed] [Google Scholar]
- 34. Jentzer JC, Wiley B, Bennett C, Murphree DH, Keegan MT, Kashani KB, Bell MR, Barsness GW. Temporal trends and clinical outcomes associated with vasopressor and inotrope use in the cardiac intensive care unit. Shock 2020;53:452–459. [DOI] [PubMed] [Google Scholar]
- 35. Vallabhajosyula S, Subramaniam AV, Murphree DH, Jr., Patlolla SH, Ya'Qoub L, Kumar V, Verghese D, Cheungpasitporn W, Jentzer JC, Sandhu GS, Gulati R, Shah ND, Gersh BJ, Holmes DR, Jr., Bell MR and Barsness GW. Complications from percutaneous-left ventricular assist devices versus intra-aortic balloon pump in acute myocardial infarction-cardiogenic shock. PLoS One. 2020; 15:e0238046. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Ternus BW, Jentzer JC, El Sabbagh A, Eleid MF, Bell MR, Murphy JG, Rihal CS, Barsness GW. Percutaneous mechanical circulatory support for cardiac disease: temporal trends in use and complications between 2009 and 2015. J Invasive Cardiol 2017;29:309–313. [PubMed] [Google Scholar]
- 37. Russo JJ, Aleksova N, Pitcher I, Couture E, Parlow S, Faraz M, Visintini S, Simard T, Di Santo P, Mathew R, So DY, Takeda K, Garan AR, Karmpaliotis D, Takayama H, Kirtane AJ, Hibbert B. Left ventricular unloading during extracorporeal membrane oxygenation in patients with cardiogenic shock. J Am Coll Cardiol 2019;73:654–662. [DOI] [PubMed] [Google Scholar]
- 38. Grandin EW, Nunez JI, Willar B, Kennedy K, Rycus P, Tonna JE, Kapur NK, Shaefi S, Garan AR. Mechanical left ventricular unloading in patients undergoing venoarterial extracorporeal membrane oxygenation. J Am Coll Cardiol 2022;79:1239–1250. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
All data are incorporated into the article and its online Supplementary material.