Skip to main content
Skip main navigationClose Drawer MenuOpen Drawer Menu

Cardiogenic Shock Classification to Predict Mortality in the Cardiac Intensive Care UnitFree Access

Original Investigation

J Am Coll Cardiol, 74 (17) 2117–2128
Sections

Central Illustration

Abstract

Background:

A new 5-stage cardiogenic shock (CS) classification scheme was recently proposed by the Society for Cardiovascular Angiography and Intervention (SCAI) for the purpose of risk stratification.

Objectives:

This study sought to apply the SCAI shock classification in a cardiac intensive care unit (CICU) population.

Methods:

The study retrospectively analyzed Mayo Clinic CICU patients admitted between 2007 and 2015. SCAI CS stages A through E were classified retrospectively using CICU admission data based on the presence of hypotension or tachycardia, hypoperfusion, deterioration, and refractory shock. Hospital mortality in each SCAI shock stage was stratified by cardiac arrest (CA).

Results:

Among the 10,004 unique patients, 43.1% had acute coronary syndrome, 46.1% had heart failure, and 12.1% had CA. The proportion of patients in SCAI CS stages A through E was 46.0%, 30.0%, 15.7%, 7.3%, and 1.0% and unadjusted hospital mortality in these stages was 3.0%, 7.1%, 12.4%, 40.4%, and 67.0% (p < 0.001), respectively. After multivariable adjustment, each higher SCAI shock stage was associated with increased hospital mortality (adjusted odds ratio: 1.53 to 6.80; all p < 0.001) compared with SCAI shock stage A, as was CA (adjusted odds ratio: 3.99; 95% confidence interval: 3.27 to 4.86; p < 0.001). Results were consistent in the subset of patients with acute coronary syndrome or heart failure.

Conclusions:

When assessed at the time of CICU admission, the SCAI CS classification, including presence or absence of CA, provided robust hospital mortality risk stratification. This classification system could be implemented as a clinical and research tool to identify, communicate, and predict the risk of death in patients with, and at risk for, CS.

Introduction

Cardiogenic shock (CS) continues to be associated with high rates of morbidity and mortality, posing a therapeutic challenge for clinicians (1–4). Although the mortality among patients with CS may be decreasing over time, short-term mortality rates remain 35% to 40% in recent studies (1–9). Other than culprit vessel revascularization for patients with acute myocardial infarction (MI), no other intervention has demonstrated an improvement in short-term survival among patients with CS, and no established beneficial therapies exist for patients with non-MI etiologies of CS (1,5–9). A recent scientific statement from the American Heart Association highlighted several potential priorities for the future of CS research, to address the limited therapeutic options and uncertainty about the efficacy of standard treatment modalities in CS (1).

Although diagnostic criteria for CS have been clearly defined in the literature, a major gap in the field of CS research has been the lack of a standard schema to uniformly characterize CS severity across research protocols and individual centers (1). CS populations encompass a broad spectrum of hemodynamic derangement ranging from isolated hypoperfusion that is easily reversed with initial therapies to refractory shock with multiorgan failure and hemodynamic collapse (1). Patients with differing degrees of shock severity may have varying responsiveness to therapeutic interventions and different clinical outcomes, yet they are considered the same for the purposes of clinical trial and registry enrollment, leading to substantial heterogeneity in CS study populations. Multiple risk scores exist to predict mortality in patients with CS, but these apply primarily to CS complicating acute MI and the need for multiple input variables reduces their clinical applicability (10,11). Although these scores can provide mortality risk stratification, they fail to provide meaningful characterization of the severity of CS in a way that can be easily communicated between providers and inform treatment and transfer decisions. Prior studies have failed to determine the effects that overall illness severity may have on the risk-benefit profile of available therapeutic interventions (7–11).

To overcome these limitations, the Society for Cardiovascular Angiography and Intervention (SCAI) developed an expert consensus statement, endorsed by multiple relevant societies, proposing a novel CS classification scheme, which categorizes patients with or at risk of CS into worsening stages of hemodynamic compromise for the purposes of facilitating patient care and research (12). The SCAI CS classification consensus statement describes 5 stages of CS, each of which may have an “A” modifier signifying the occurrence of cardiac arrest (CA) (12). This classification schema was developed based on expert consensus opinion and its ability to discriminate among levels of mortality risk in critically ill patients remains to be established. The goal of this study was to examine the construct validity of the SCAI CS staging schema by demonstrating the ability of a simple functional classification of SCAI shock stages at the time of cardiac intensive care unit (CICU) admission to predict mortality in unselected CICU patients.

Methods

Study population

The Institutional Review Board of the Mayo Clinic (IRB # 16-000722) approved the study as posing minimal risk to patients, and it was performed under a waiver of informed consent. We analyzed a database of consecutive unique adult patients ≥18 years of age admitted to the CICU at Mayo Clinic Hospital St. Mary’s Campus between January 1, 2007, and December 31, 2015 (13–15). The Mayo Clinic CICU is a closed, 16-bed unit serving critically ill cardiac medical patients. Post-operative cardiac surgery patients and patients receiving extracorporeal membrane oxygenation (ECMO) support are cared for in a separate cardiovascular surgical intensive care unit. To minimize the risk of survival and treatment biases associated with CICU readmission, only data from each patient’s first CICU admission were analyzed. According to Minnesota state law statute 144.295, patients may decline authorization for inclusion in observational research studies; patients who declined Minnesota Research Authorization were excluded from the study.

Data sources

We recorded demographic, vital sign, laboratory, clinical, and outcome data, as well as procedures and therapies performed during the CICU and hospital stay, as previously described; radiographic, invasive hemodynamic, and physical examination data were not available (13–15). All relevant data were extracted electronically from the medical record using the Multidisciplinary Epidemiology and Translational Research in Intensive Care Data Mart, a repository storing clinical data from all intensive care unit admissions at the Mayo Clinic Rochester (16). The admission value of all vital signs, clinical measurements, and laboratory values was defined as either the first value recorded after CICU admission or the value recorded closest to CICU admission. In addition, vital signs were recorded every 15 min during the first hour after CICU admission. Peak vasoactive medication (vasopressor and inotrope) doses were used to calculate the vasoactive-inotropic score and norepinephrine-equivalent vasopressor dose (17–19). Admission diagnoses included all International Classification of Diseases-9th Revision diagnostic codes recorded on the day of CICU admission and the day before or after CICU admission; these admission diagnoses were not mutually exclusive, and the primary admission diagnosis could not be determined. Admission diagnoses of interest included acute coronary syndrome (ACS), heart failure (HF), supraventricular tachycardia, atrial fibrillation, ventricular fibrillation, ventricular tachycardia, shock, CA, respiratory failure, and sepsis.

The APACHE (Acute Physiology and Chronic Health Evaluation)-III score, APACHE-IV predicted hospital mortality, and Sequential Organ Failure Assessment score were automatically calculated for all patients using data from the first 24 h of CICU admission using previously validated electronic algorithms, with missing variables imputed as normal as the default (13–15,20–22). The Charlson Comorbidity Index (CCI) and individual comorbidities were determined from the medical record using a previously validated electronic algorithm (23). Severe acute kidney injury (AKI) during the CICU stay was defined as doubling of serum creatinine from baseline, increase in serum creatinine to ≥4.0 mg/dl, or new dialysis initiation; patients with a prior history of dialysis were excluded from this analysis (14).

Definition of shock stages

We defined hypotension or tachycardia, hypoperfusion, deterioration, and refractory shock using data from CICU admission through the first 24 h in the CICU (Table 1). Hypotension and tachycardia were defined within the first 1 h of CICU admission. The definition of hypoperfusion included an elevated lactate level on admission or AKI developing within 24 h after admission. Deterioration was defined as increasing vasoactive drug requirements after the first hour or a rising lactate level after admission. We used pragmatic and simplified definitions to divide patients into the 5 SCAI shock stages with increasing severity (A through E) using combinations of these variables (Central Illustration). Importantly, the SCAI shock classification system (Table 2) involves details such as rapid escalation of inotropes or addition of temporary mechanical circulatory support (MCS) devices, which were not available in the database (12) Late deterioration was defined as increasing vasopressor requirements after 24 h.

Table 1. Study Definitions of Hypotension, Tachycardia, Hypoperfusion, Deterioration, and Refractory Shock

TermDefinition
Hypotension/tachycardiaPresence of any of the following criteria:

Admission systolic BP <90 mm Hg

Minimum systolic BP <90 mm Hg during first 1 h

Admission MAP <60 mm Hg

Minimum MAP <60 mm Hg during first 1 h

Admission HR >100 beats/min

Maximum HR >100 beats/min during first 1 h

Admission HR > admission systolic BP

Mean HR > mean systolic BP during first 1 h

HypoperfusionPresence of any of the following criteria:

Admission lactate >2 mmol/l

Urine output <720 ml during first 24 h

Creatinine increased by ≥0.3 mg/dl during first 24 h

DeteriorationPresence of any of the following criteria:

Maximum lactate > admission lactate

Number of vasoactives during first 24 h > number of vasoactives during first 1 h

Maximum VIS during first 24 h > VIS during first 1 h

Maximum NEE during first 24 h > NEE during first 1 h

Refractory shockPresence of any of the following criteria:

Mean systolic BP during first 1 h <80 and on vasoactives

Mean systolic MAP during first 1 h <50 and on vasoactives

Number of vasoactives during first 1 h >2

Number of vasoactives during first 1 h >1 and IABP during first 24 h

Admission lactate ≥10 mmol/l

VIS is calculated as using vasoactive drug doses (in μg/kg/min), as follows: VIS = dobutamine + dopamine + (10 * phenylephrine + milrinone) + (100 * [epinephrine + norepinephrine]) + (10,000 * units/kg/min vasopressin). NEE is calculated using the dose equivalency as follows: 0.1 μg/kg/min norepinephrine = 0.1 μg/kg/min epinephrine = 15 μg/kg/min dopamine = 1 μg/kg/min phenylephrine = 0.04 U/min vasopressin.

BP = blood pressure; HR = heart rate; IABP = intra-aortic balloon pump; MAP = mean arterial pressure; NEE = norepinephrine-equivalent vasopressor dose; VIS = vasoactive-inotropic score.

Central Illustration.
Central Illustration.

Definitions of SCAI Shock Stages A Through E, With Associated Cardiac Intensive Care Unit and Hospital Mortality in Each SCAI Shock Stage

Cardiac intensive care unit and hospital mortality increased as a function of higher Society for Cardiovascular Angiography and Intervention shock stage.

Table 2. Definition of CS Stages Used in this Study, Based on the SCAI Consensus Statement Classification

CS StageSCAI Definition
Stage A (“at risk”)Patients without CS who are hemodynamically stable but have acute cardiovascular disease putting them at risk of developing CS
Stage B (“beginning”)Patients without CS who display hemodynamic instability, including hypotension and/or tachycardia, but with normal perfusion
Stage C (“classic”)Patients with CS, manifested by hypoperfusion (lactic acidosis, oliguria, cool/clammy periphery, or altered mentation) requiring intervention
Stage D (“deteriorating)”Patients with CS whose hemodynamic instability and/or hypoperfusion fails to respond to initial interventions
Stage E (“extremis”)Patients with CS and overt or impending circulatory collapse, including CA with ongoing resuscitation

CA = cardiac arrest; CS = cardiogenic shock; SCAI = Society for Cardiovascular Angiography and Intervention.

Statistical analysis

The primary endpoint was all-cause hospital mortality; secondary endpoints included CICU mortality. Hospital disposition and all-cause mortality were determined using electronic review of medical records. Categorical variables are reported as number and percentage and the Pearson chi-square test was used to compare groups. Continuous variables are reported as mean ± SD. Trends across the SCAI shock stages were determined using linear regression. Logistic regression was used to determine the association between the SCAI shock stages and hospital mortality before and after adjusting for age, sex, CCI, APACHE-IV predicted hospital mortality, admission diagnosis of CA, and the use of vasoactive medications, intra-aortic balloon pump, coronary angiography, percutaneous coronary intervention, and mechanical ventilation. Discrimination was assessed using the area under the receiver-operating characteristic curve (C-statistic) value. Two-tailed p values <0.05 were considered statistically significant. Statistical analyses were performed using JMP Pro version 14.1.0 (SAS Institute, Cary, North Carolina).

Results

Study population

We screened 12,904 adult admissions to the CICU during the study period and excluded 2,900 patients (1,877 readmissions, 755 patients without Minnesota Research Authorization, and 268 patients whose admission did not occur entirely within the study period), as demonstrated in Figure 1 (13–15). The final study population of 10,004 unique patients had a mean age of 67.4 ± 15.2 years, including 3,746 (37.4%) women. The mean CCI was 2.4 ± 2.6 and the mean APACHE-IV predicted hospital mortality was 16.9 ± 20.0% overall. Admission diagnoses (not mutually exclusive) included ACS in 4,267 (43.1%) patients, HF in 4,564 (46.1%) patients, and CA in 1,193 (12.1%) patients; 2,704 (27.3%) patients had neither ACS nor HF as an admission diagnosis.

Figure 1.
Figure 1.

Study Inclusion and Exclusion Criteria and Distribution of SCAI Shock Stages

The final study population included 10,004 unique patients. CICU = cardiac intensive care unit; SCAI = Society for Cardiovascular Angiography and Intervention.

A total of 2,468 (24.7%) patients received vasoactive drugs during the CICU stay, including vasopressors in 2,090 (20.9%) and inotropes in 928 (9.3%). Among patients receiving vasoactive drugs, 1,182 (47.9%) received >1 vasoactive drug. An intra-aortic balloon pump was placed during the CICU stay in 865 (8.6%) patients and Impella (Abiomed, Danvers, Massachusetts) or ECMO support was used during the hospitalization in 21 (0.2%) and 72 (0.7%) patients, respectively.

Hypotension or tachycardia during the first hour in the CICU was present in 4,367 (43.7%) patients, including 2,545 (25.4%) who met criteria for hypotension and 2,956 (29.5%) who met criteria for tachycardia; 1,134 (11.3%) patients met criteria for both hypotension and tachycardia. Hypoperfusion was present in 2,404 (24.0%) patients, including an admission lactate level >2 mmol/l in 888 (41.6%) of 2,135 patients with available data. Deterioration within 24 h of admission occurred in 2,075 (20.7%) patients, and refractory shock was identified in 153 (1.5%) patients. Late deterioration after 24 h occurred in 708 (7.1%) patients.

The proportion of patients with SCAI shock stages A through E were 46.0%, 30.0%, 15.7%, 7.3%, and 1.0%, respectively (Figure 1). Baseline demographics, comorbidities, admission diagnoses, and critical care therapies varied significantly across the SCAI shock stages (Table 3). The prevalence of CA increased across the SCAI shock stages, from 7.3% in stage A to 55.8% in stage E. As the SCAI shock stage increased, there were more extensive vital sign and laboratory abnormalities, higher severity of illness scores, and more frequent AKI (Table 4). The use and dosage of vasoactive medications and supportive therapies including mechanical ventilation, MCS, and dialysis increased across the SCAI shock stages (Table 4). The prevalence of late deterioration increased as a function of SCAI shock stage, being highest in SCAI shock stage D (Table 4).

Table 3. Baseline Characteristics, Comorbidities, Admission Diagnoses, and Therapies of Patients According to SCAI Shock Stage

With Data, %Stage A (n = 4,602)Stage B (n = 2,998)Stage C (n = 1,575)Stage D (n = 732)Stage E (n = 97)p Value
Demographics
Age, yrs100.067.1 ± 14.766.4 ± 15.769.9 ± 16.068.7 ± 14.168.3 ± 14.7<0.001
Female100.01,562 (33.9)1,216 (40.6)654 (41.5)278 (38.0)36 (37.1)<0.001
White100.04,289 (93.2)2,779 (92.7)1,430 (90.8)659 (90.0)79 (81.4)<0.001
Comorbidities
Charlson Comorbidity Index99.82.1 ± 2.52.5 ± 2.62.8 ± 2.83.0 ± 2.82.1 ± 2.6<0.001
History of MI99.8914 (19.9)581 (19.4)311 (19.8)160 (21.9)14 (14.4)0.79
History of HF99.8746 (16.2)638 (21.3)339 (21.6)212 (29.0)18 (18.6)<0.001
History of diabetes mellitus99.81,222 (26.6)851 (28.5)483 (30.8)257 (35.2)24 (24.7)<0.001
History of CKD99.8770 (16.8)604 (20.2)418 (26.6)221 (30.2)18 (20.4)<0.001
Prior dialysis100.0131 (2.8)130 (4.3)192 (12.2)107 (14.6)11 (11.3)<0.001
Admission ICD-9 diagnoses
ACS99.02,111 (46.4)1,172 (39.4)634 (40.7)300 (41.3)50 (52.6)<0.001
HF99.01,675 (36.8)1,562 (52.5)758 (48.7)513 (70.7)56 (59.0)<0.001
Cardiac arrest99.0330 (7.3)311 (10.5)219 (14.1)280 (38.6)53 (55.8)<0.001
Shock99.0217 (4.8)449 (15.1)166 (10.7)429 (59.1)88 (92.6)<0.001
Respiratory failure99.0506 (11.1)660 (22.2)389 (25.0)460 (63.4)64 (67.4)<0.001
Sepsis99.0102 (2.2)205 (6.9)100 (6.4)163 (22.4)35 (36.8)<0.001
AF/SVT99.01,165 (25.6)1,150 (38.7)571 (36.7)304 (41.9)30 (31.6)<0.001
VT/VF99.0665 (14.6)516 (17.4)247 (15.9)158 (21.8)22 (23.2)<0.001
Therapies and procedures
Vasoactives first 1 h100.0126 (2.7)339 (11.3)90 (5.7)249 (34.0)81 (83.5)<0.001
Number of vasoactives first 1 h100.00.0 ± 0.20.1 ± 0.40.1 ± 0.30.4 ± 0.61.4 ± 1.0<0.001
VIS first 1 h99.10.2 ± 1.61.3 ± 10.51.4 ± 12.15.4 ± 15.330.1 ± 46.7<0.001
NEE first 1 h100.00.00 ± 0.010.01 ± 0.100.01 ± 0.120.05 ± 0.150.29 ± 0.47<0.001
Invasive ventilator first 24 h100.0257 (5.6)419 (14.0)232 (14.7)417 (57.0)73 (75.3)<0.001
Dialysis100.0119 (2.6)138 (4.6)72 (4.6)137 (18.7)21 (21.6)<0.001
CRRT100.09 (0.2)30 (1.0)23 (1.5)94 (12.8)11 (11.3)<0.001
IABP during hospitalization100.0266 (5.8)330 (11.0)69 (4.4)162 (22.1)38 (39.2)<0.001
Impella100.05 (0.1)7 (0.2)4 (0.2)3 (0.4)2 (2.1)0.004
ECMO100.015 (0.3)28 (0.9)8 (0.5)16 (2.2)5 (5.2)0.02
PAC100.0239 (5.2)245 (8.2)49 (3.1)163 (22.3)25 (25.8)<0.001
Coronary angiogram100.02,694 (58.5)1,471 (49.1)720 (45.7)349 (47.7)50 (51.6)<0.001
PCI100.01,834 (39.8)932 (31.1)430 (27.3)205 (28.0)26 (26.8)<0.001
RBC transfusion100.0308 (6.7)403 (13.4)197 (12.5)228 (31.2)37 (28.1)<0.001

Values are mean ± SD or n (%), unless otherwise indicated. The p value is for the trend across SCAI shock stages A to E.

ACS = acute coronary syndrome; AF = atrial fibrillation; CICU = cardiac intensive care unit; CKD = chronic kidney disease; CRRT = continuous renal-replacement therapy; ECMO = extracorporeal membrane oxygenation; HF = heart failure; ICD-9 = International Classification of Diseases-9th Revision; MI = myocardial infarction; PAC = pulmonary artery catheter; PCI = percutaneous coronary intervention; RBC = red blood cell; SVT = supraventricular tachycardia; VF = ventricular fibrillation; VT = ventricular tachycardia; other abbreviations as in Tables 1 and 2.

∗ Admission diagnoses are not mutually exclusive and sum to >100%.

Table 4. Severity of Illness Scores, Vital Signs, and Laboratory Data of Patients According to SCAI Shock Stage

With Data, %Stage A (n = 4,602)Stage B (n = 2,998)Stage C (n = 1,575)Stage D (n = 732)Stage E (n = 97)p Value
Severity of illness
APACHE-III score100.051.3 ± 18.260.8 ± 20.469.5 ± 24.497.4 ± 31.5118.5 ± 38.8<0.001
APACHE-IV predicted hospital mortality, %100.09.9 ± 11.615.8 ± 16.722.0 ± 20.748.4 ± 28.164.8 ± 27.8<0.001
Day 1 SOFA score99.92.3 ± 2.03.4 ± 2.74.4 ± 2.99.1 ± 3.911.4 ± 3.9<0.001
Severe AKI89.2257 (6.1)333 (12.4)233 (17.8)269 (44.2)40 (49.4)<0.001
Late deterioration100.0188 (4.1)190 (6.3)108 (6.9)205 (28.0)17 (17.5)<0.001
Admission vital sign data
Systolic blood pressure, mm Hg99.4130.8 ± 22.9114.4 ± 26.1123.0 ± 27.7113.8 ± 27.999.8 ± 25.3<0.001
Diastolic blood pressure, mm Hg96.272.1 ± 14.567.1 ± 18.668.8 ± 17.865.7 ± 19.557.6 ± 19.4<0.001
Mean arterial pressure, mm Hg96.287.2 ± 15.179.6 ± 19.683.1 ± 18.979.9 ± 20.970.1 ± 20.9<0.001
Heart rate, beats/min99.472.4 ± 13.993.1 ± 26.884.8 ± 25.589.4 ± 24.795.5 ± 27.9<0.001
Shock index99.40.57 ± 0.150.84 ± 0.290.72 ± 0.280.83 ± 0.291.04 ± 0.38<0.001
Respiratory rate, breaths/min95.917.3 ± 5.219.1 ± 6.019.3 ± 5.920.2 ± 5.921.8 ± 6.5<0.001
Pulse oximetry, %99.496.6 ± 4.295.6 ± 5.695.2 ± 7.393.2 ± 9.086.5 ± 16.8<0.001
Glasgow Coma Scale97.314.5 ± 2.013.9 ± 2.913.6 ± 3.39.8 ± 5.18.3 ± 5.1<0.001
Urine output first 24 h, l97.02.16 ± 1.102.26 ± 1.421.02 ± 1.121.25 ± 1.331.41 ± 2.41<0.001
Admission laboratory data
Creatinine, mg/dl96.31.2 ± 0.81.3 ± 1.01.7 ± 1.71.9 ± 1.41.9 ± 1.2<0.001
BUN, mg/dl96.023.5 ± 16.427.0 ± 18.930.0 ± 20.436.0 ± 23.134.6 ± 20.5<0.001
ALT, U/ml46.551.9 ± 139.576.2 ± 222.2127.3 ± 529.8270.8 ± 675.0644.5 ± 1211.5<0.001
Peak troponin T, mg/dl63.31.8 ± 3.31.7 ± 3.21.8 ± 3.43.3 ± 6.94.0 ± 6.5<0.001
Hemoglobin, g/l96.412.5 ± 2.011.9 ± 2.211.9 ± 2.311.8 ± 2.511.6 ± 2.5<0.001
Arterial pH32.37.39 ± 0.087.36 ± 0.107.36 ± 0.107.30 ± 0.117.20 ± 0.15<0.001
Bicarbonate, mEq/l96.924.6 ± 3.723.9 ± 4.423.4 ± 4.621.2 ± 5.316.7 ± 6.5<0.001
Anion gap, mEq/l89.311.0 ± 2.911.5 ± 3.212.6 ± 3.814.4 ± 4.320.6 ± 8.7<0.001
Lactate, mmol/l21.31.2 ± 0.41.3 ± 0.43.0 ± 2.33.6 ± 2.310.6 ± 5.1<0.001

Values are mean ± SD or n (%), unless otherwise indicated. The p value is for the trend across SCAI shock stages A to E.

AKI = acute kidney injury; ALT = alanine aminotransferase; APACHE = Acute Physiology and Chronic Health Evaluation; BUN = blood urea nitrogen; SCAI = Society for Cardiovascular Angiography and Intervention; SOFA = Sequential Organ Failure Assessment.

∗ Shock index is defined as the ratio of heart rate to systolic blood pressure.

CICU and hospital mortality

There was a stepwise increase in unadjusted CICU and hospital mortality with each higher SCAI shock stage in the overall population, with hospital mortality rising from 3.0% in SCAI shock stage A to 67.0% in SCAI shock stage E (p < 0.001) (Central Illustration). Unadjusted hospital mortality was higher among patients with CA at each SCAI shock stage (Figure 2) (all p < 0.001). The same stepwise increase in hospital mortality was seen in patients with ACS and HF and in patients without a diagnosis of either ACS or HF (Figure 3). Patients with late deterioration had higher mortality overall (31.4% vs. 7.4%; p < 0.001), and in each SCAI shock stage except stage E (Figure 4).

Figure 2.
Figure 2.

Hospital Mortality as a Function of SCAI Shock Stage Among Patients With and Without an Admission Diagnosis of CA

Hospital mortality was higher among patients with an admission diagnosis of cardiac arrest (CA) in each Society for Cardiovascular Angiography and Intervention (SCAI) shock stage (all p < 0.001).

Figure 3.
Figure 3.

Hospital Mortality as a Function of SCAI Shock Stage

Hospital mortality as a function of Society for Cardiovascular Angiography and Intervention (SCAI) shock stage among patients with acute coronary syndrome (ACS) (left), heart failure (HF) (middle), or neither ACS nor HF (right). Hospital mortality increased as a function of higher SCAI shock stage in patients with ACS or HF.

Figure 4.
Figure 4.

Hospital Mortality and SCAI Shock Stage in Patients With and Without Late Deterioration, Defined as Rising Vasopressor Requirements After 24 h

Hospital mortality was higher among patients with late deterioration in each Society for Cardiovascular Angiography and Intervention (SCAI) shock stage, except stage E (all other p < 0.001).

Compared with SCAI shock stage A, the unadjusted odds ratio (OR) values for hospital mortality in SCAI shock stages B through E were 2.44, 4.56, 21.80, and 65.22, respectively. The unadjusted OR value for hospital mortality was 12.17 (95% confidence interval [CI]: 10.34 to 14.31; p < 0.001) in patients meeting criteria for SCAI shock stage D or E, compared with SCAI shock stages A through C. The SCAI shock classification itself had an area under the receiver-operating characteristic curve value of 0.765 for hospital mortality overall, 0.775 among patients with ACS, and 0.732 among patients with HF.

After multivariable adjustment, each higher SCAI shock stage was associated with increased hospital mortality compared with SCAI shock stage A (all p < 0.001), as was CA (adjusted OR: 3.99; 95% CI: 3.27 to 4.86, 95% CI; p < 0.001); the final multivariable model area under the receiver-operating characteristic curve value was 0.883 in the overall population. Compared with SCAI shock stage A, the adjusted OR values for hospital mortality in SCAI shock stages B through E were 1.53, 2.62, 3.07, and 6.80, respectively (Figure 5). Each higher SCAI shock stage was associated with higher adjusted hospital mortality compared with SCAI shock stage A among patients with ACS (all p < 0.001), HF (all p < 0.001), and neither ACS nor HF (all p < 0.001). Likewise, CA was associated with higher adjusted hospital mortality in each of these subgroups (all p < 0.001).

Figure 5.
Figure 5.

Adjusted OR Plot for Hospital Mortality

Adjusted odds ratios (ORs) and 95% confidence intervals for hospital mortality with each Society for Cardiovascular Angiography and Intervention (SCAI) shock stage derived from multivariable logistic regression, using stage A as referent. Higher SCAI shock stages had incrementally higher adjusted odds for hospital mortality.

Discussion

Using a large cohort of unselected CICU patients, we validated the association between the recently described SCAI shock classification and hospital mortality. We stratified patients into 5 SCAI shock stages at the time of CICU admission, reflecting a continuum of increasing shock severity using a simplified definition based on hypotension or tachycardia, hypoperfusion, deterioration, and refractory shock, which can be easily applied in clinical practice. This functional SCAI shock stages classification effectively stratified mortality risk and performed similarly in patients with admission diagnoses of ACS and HF, even when adjusting for the higher illness severity and greater use of hemodynamic support at higher shock stages. Patients with refractory shock (SCAI shock stage E) had >20-fold higher crude hospital mortality than hemodynamically stable patients without shock (SCAI shock stage A). An admission diagnosis of CA increased the risk of hospital mortality among patients with each SCAI shock stage, supporting its inclusion as an effect modifier in the SCAI shock classification schema. These data support the validity of the recent SCAI classification of CS stages for mortality risk stratification as a framework for future CS clinical practice and research. Our examination of a mixed CICU cohort allowed us to demonstrate the predictive ability of this classification system in patients with diagnoses of ACS and HF, which are the dominant causes of CS, as well as in patients without these diagnoses. The strong association between SCAI shock stages and mortality in a heterogeneous CICU population, even after adjustment for known predictors of mortality, emphasizes the robustness of this classification system and its potential to be applied in other critically ill patient cohorts.

The SCAI shock classification was developed using expert consensus for the purpose of describing CS severity to clarify communication of patient status between providers in different care settings to facilitate patient triage and selection for advanced therapies (12). In addition, the SCAI classification of CS stages was designed to facilitate clinical research by simplifying the heterogeneity inherent to CS populations and help determine whether treatment interactions exist as a function of CS severity. A similar classification problem was addressed by the development of the INTERMACS (Interagency Registry for Mechanically Assisted Circulatory Support) profiles, which subdivided patients with advanced HF into clinically relevant groups to determine their need for durable MCS (24). In essence, most patients with CS would typically be classified as INTERMACS profile 1 (“crash and burn”) or 2 (“sliding on inotropes”), and the SCAI classification provides further granularity by dividing these patients into stages C, D, and E (12,24). In addition, the INTERMACS profiles are intended for application at the single time point of implantation of a durable MCS device and do not have a construct to assess deterioration of status.

Nearly one-half of this CICU population was classified as SCAI shock stage A (“at risk”) and the 3% observed hospital mortality in this group suggests that patients without hypotension, tachycardia, or hypoperfusion at the time of CICU admission have a favorable prognosis. Crude hospital mortality more than doubled among patients with evidence of hemodynamic instability (SCAI shock stage B [“beginning”]), and nearly doubled again among patients with hypoperfusion (SCAI shock stage C [“classic shock”]). Crude hospital mortality rose to 40% among patients with deterioration (SCAI shock stage D [“deteriorating”]), similar to that observed in recent randomized controlled trials and observational studies of CS (2–4,6–9). This suggests that patients with CS who respond to initial stabilization measures (SCAI shock stage C) have a relatively favorable prognosis, while the majority of patients included in published studies of CS likely meet criteria for SCAI shock stage D. The marked step-up in short-term mortality risk among patients with SCAI shock stage D and E (“extremis”) suggests a potential role for advanced hemodynamic support options including MCS in patients demonstrating evidence of deterioration. More than two-thirds of patients classified as SCAI shock stage E died in the hospital, emphasizing the need to identify improved therapies for these highest-risk patients. The prevalence of hemodynamic deterioration after 24 h increased with higher SCAI shock stages and was associated with higher hospital mortality.

CA is common in CS populations and has been associated with an increased risk of death, yet the influence of this major risk modifier on therapeutic responses has not been well studied (7,9,25,26). The SCAI statement authors clearly emphasize the added hazard posed by the presence of CA occurring in patients with or at risk of CS (12). In this cohort, the prevalence of CA increased substantially with increasing shock stage, highlighting the correlation between CA and severe shock in CICU patients. In our analysis, we clearly demonstrate the added mortality hazard posed by CA at all levels of shock severity, validating CA as a prognostically important modifier in the SCAI shock classification. Shock severity demonstrated a stepwise association with mortality in patients with CA, emphasizing the synergistic mortality effects of concomitant CS and CA in CICU patients, as previously demonstrated in patients with acute MI (25). The relative effect of CA on mortality appeared to be greater among patients with mild or no shock (SCAI shock stages A through C). Although it remains clear that the presence of CA among CS patients is associated with worse outcomes, it is unlikely that all such events carry the same hazard and future studies should address whether brief CA episodes have any prognostic importance and whether the presence of brain injury from CA modifies the response to CS therapies (26).

Van Diepen et al. (1) have proposed a “hub-and-spoke” care model involving transfer to tertiary centers for patients with CS, as has been instituted for other high-acuity medical conditions such as trauma. Uncertainty remains regarding when transfer to a higher level of care is warranted, and ideally this should be determined early in the course if a patient is not responding as expected to initial therapy. Based on the high risk of mortality after the onset of hemodynamic deterioration (SCAI shock stage D), we propose that patients with hypoperfusion (SCAI shock stage C) who do not rapidly stabilize (i.e., progression to SCAI shock stage D) should be considered for transfer to a higher level of care before development of overt deterioration. The simple functional definitions used in this study could be leveraged by an electronic medical record system to identify patients with new onset or increasing severity of shock or clinical deterioration, to highlight their escalating mortality risk in real time and facilitate early transfer or involvement of palliative care services if indicated (1).

We propose that the relative efficacy of various therapeutic interventions at each CS stage should be further explored, as CS severity might determine the need for and clinical response to specific therapies. For example, temporary MCS devices can effectively increase cardiac output in CS, yet none of these temporary MCS devices has resulted in a proven improvement in survival in published randomized clinical trials of CS patients (1,8,9,27). Notwithstanding their established hemodynamic benefits, the invasive nature and acquisition costs of temporary MCS devices emphasize the need to evaluate when and in whom these devices may be most effective for improving patient-centered outcomes (1,8,27). We suspect that each temporary MCS device will have a different risk-benefit profile at a given CS stage, but this hypothesis will need to be tested in future studies.

Study limitations

Despite its large sample size and granular data, this study has a number of limitations that are inherent to all retrospective cohort studies, including the need for prospective validation and the potential for unmeasured confounders and missing data to have influenced the results. Owing to lack of available invasive hemodynamic data, we cannot be sure to what extent the shock states in this cohort were cardiogenic in nature. The inclusion of a mixed CICU population implies that some patients meeting criteria for shock had noncardiogenic or mixed shock states, similar to a recent multicenter CICU registry (4). To focus on the presence of shock on admission, we defined the shock stages using variables from within 1 to 24 h of CICU admission, recognizing that many patients develop CS after hospital admission; owing to data availability, our limited definition of late deterioration only included vasopressor dosage and not worsening SCAI shock stage (3). We were unable to include prognostically relevant physical examination findings such as cool or clammy extremities or altered mental status in our definition of hypoperfusion; the inclusion of oliguria and rising creatinine in the definition of hypoperfusion is less relevant among patients with end-stage renal disease (28). In addition, the definition for SCAI shock stage C includes the requirement for an intervention to treat hypoperfusion, a criterion we did not include in our definition of stage C for ease of application (12). The infrequent use of advanced temporary MCS devices (e.g., Impella or ECMO) in this cohort could have influenced the observed mortality, particularly among patients with higher shock severity; nonetheless, the overall utilization of MCS devices in this population is consistent with national utilization among patients with CS (29). By using International Classification of Diseases-Ninth Revision codes to define admission diagnoses, we were unable to define the primary admission diagnoses and could not distinguish in-hospital CA from out-of-hospital CA. Data regarding timing, arrest rhythm, and neurologic status of patients with CA were not available. We grouped patients based on the presence of ACS without distinguishing between ACS subtypes, limiting our ability to draw conclusions about CS caused by acute MI. Data regarding resuscitation status and limitations of therapies were not available.

Conclusions

In a large, heterogeneous CICU cohort, we demonstrated the feasibility of classifying patients into 5 shock stages (SCAI shock stages A to E) reflecting progressively increasing levels of illness severity. This pragmatic SCAI shock stages classification provided robust mortality risk stratification in the overall cohort including patients with ACS and HF, in a manner that was amplified by the presence of CA. Despite its limitations, the functional adaptation of the SCAI shock classification described herein had very good discrimination for mortality, emphasizing its validity and practical utility with the potential for improved risk stratification when rigorously applied. The simple, intuitive SCAI shock stage definitions we report herein could be easily applied in clinical practice by providers with different levels of expertise. We suggest that future CS clinical trials consider stratifying patients according to SCAI shock stage and the presence of CA, to ensure consistent outcomes reporting and to assess whether the effects of the tested intervention vary by CS stage.

Perspectives

COMPETENCY IN PATIENT CARE AND PROCEDURAL SKILLS: A CS classification system developed by the SCAI can effectively stratify patients in a CICU, including those with an ACS or HF, for risk of mortality. Patients with SCAI cardiogenic shock stages D and E are at higher risk and may benefit from early transfer to specialized centers offering advanced modalities for circulatory support.

TRANSLATIONAL OUTLOOK: Prospective studies using a systematic approach to shock assessment and management are needed to determine if the efficacy of various advanced modalities is related to disease severity.

  • 1. van Diepen S., Katz J.N., Albert N.M.et al. : "Contemporary management of cardiogenic shock: a scientific statement from the American Heart Association". Circulation 2017; 136: e232.

    CrossrefMedlineGoogle Scholar
  • 2. Kolte D., Khera S., Aronow W.S.et al. : "Trends in incidence, management, and outcomes of cardiogenic shock complicating ST-elevation myocardial infarction in the United States". J Am Heart Assoc 2014; 3: e000590.

    CrossrefMedlineGoogle Scholar
  • 3. Hunziker L., Radovanovic D., Jeger R.et al. : "Twenty-year trends in the incidence and outcome of cardiogenic shock in AMIS Plus registry". Circ Cardiovasc Interv 2019; 12: e007293.

    CrossrefMedlineGoogle Scholar
  • 4. Berg D.D., Bohula E.A., van Diepen S.et al. : "Epidemiology of shock in contemporary cardiac intensive care units". Circ Cardiovasc Qual Outcomes 2019; 12: e005618.

    CrossrefMedlineGoogle Scholar
  • 5. Hochman J.S., Sleeper L.A., Webb J.G.et al. : "Early revascularization in acute myocardial infarction complicated by cardiogenic shock. SHOCK Investigators. Should We Emergently Revascularize Occluded Coronaries for Cardiogenic Shock". N Engl J Med 1999; 341: 625.

    CrossrefMedlineGoogle Scholar
  • 6. Investigators T., Alexander J.H., Reynolds H.R.et al. : "Effect of tilarginine acetate in patients with acute myocardial infarction and cardiogenic shock: the TRIUMPH randomized controlled trial". JAMA 2007; 297: 1657.

    CrossrefMedlineGoogle Scholar
  • 7. Thiele H., Akin I., Sandri M.et al. : "PCI strategies in patients with acute myocardial infarction and cardiogenic shock". N Engl J Med 2017; 377: 2419.

    CrossrefMedlineGoogle Scholar
  • 8. Thiele H., Jobs A., Ouweneel D.M.et al. : "Percutaneous short-term active mechanical support devices in cardiogenic shock: a systematic review and collaborative meta-analysis of randomized trials". Eur Heart J 2017; 38: 3523.

    CrossrefMedlineGoogle Scholar
  • 9. Thiele H., Zeymer U., Neumann F.J.et al. : "Intraaortic balloon support for myocardial infarction with cardiogenic shock". N Engl J Med 2012; 367: 1287.

    CrossrefMedlineGoogle Scholar
  • 10. Harjola V.P., Lassus J., Sionis A.et al. : "Clinical picture and risk prediction of short-term mortality in cardiogenic shock". Eur J Heart Fail 2015; 17: 501.

    CrossrefMedlineGoogle Scholar
  • 11. Poss J., Koster J., Fuernau G.et al. : "Risk stratification for patients in cardiogenic shock after acute myocardial infarction". J Am Coll Cardiol 2017; 69: 1913.

    View ArticleGoogle Scholar
  • 12. Baran D.A., Grines C.L., Bailey S.et al. : "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 in April 2019". Catheter Cardiovasc Interv 2019; 94: 29.

    MedlineGoogle Scholar
  • 13. Jentzer J.C., Bennett C., Wiley B.M.et al. : "Predictive value of the sequential organ failure assessment score for mortality in a contemporary cardiac intensive care unit population". J Am Heart Assoc 2018; 7: e008169.

    CrossrefMedlineGoogle Scholar
  • 14. Jentzer J.C., Murphree D.H., Wiley B.et al. : "Comparison of mortality risk prediction among patients >/=70 versus <70 years of age in a cardiac intensive care unit". Am J Cardiol 2018; 122: 1773.

    CrossrefMedlineGoogle Scholar
  • 15. Bennett C.E., Wright R.S., Jentzer J.et al. : "Severity of illness assessment with application of the APACHE IV predicted mortality and outcome trends analysis in an academic cardiac intensive care unit". J Crit Care 2018; 50: 242.

    CrossrefMedlineGoogle Scholar
  • 16. Herasevich V., Pickering B.W., Dong Y., Peters S.G. and Gajic O. : "Informatics infrastructure for syndrome surveillance, decision support, reporting, and modeling of critical illness". Mayo Clin Proc 2010; 85: 247.

    CrossrefMedlineGoogle Scholar
  • 17. Nguyen H.V., Havalad V., Aponte-Patel L.et al. : "Temporary biventricular pacing decreases the vasoactive-inotropic score after cardiac surgery: a substudy of a randomized clinical trial". J Thorac Cardiovasc Surg 2013; 146: 296.

    CrossrefMedlineGoogle Scholar
  • 18. Jentzer J.C., Vallabhajosyula S., Khanna A.K., Chawla L.S., Busse L.W. and Kashani K.B. : "Management of refractory vasodilatory shock". Chest 2018; 154: 416.

    CrossrefMedlineGoogle Scholar
  • 19. Jentzer J.C., Wiley B., Bennett C.et al. : "Temporal trends and clinical outcomes associated with vasopressor and inotrope use in the cardiac intensive care unit". Shock 2019 Jun 4; . [E-pub ahead of print].

    Google Scholar
  • 20. Chandra S., Kashyap R., Trillo-Alvarez C.A.et al. : "Mapping physicians' admission diagnoses to structured concepts towards fully automatic calculation of acute physiology and chronic health evaluation score". BMJ Open 2011; 1: e000216.

    CrossrefMedlineGoogle Scholar
  • 21. Aakre C., Franco P.M., Ferreyra M., Kitson J., Li M. and Herasevich V. : "Prospective validation of a near real-time EHR-integrated automated SOFA score calculator". Int J Med Inform 2017; 103: 1.

    CrossrefMedlineGoogle Scholar
  • 22. Harrison A.M., Yadav H., Pickering B.W., Cartin-Ceba R. and Herasevich V. : "Validation of computerized automatic calculation of the sequential organ failure assessment score". Crit Care Res Pract 2013; 2013: 975672.

    MedlineGoogle Scholar
  • 23. Singh B., Singh A., Ahmed A.et al. : "Derivation and validation of automated electronic search strategies to extract Charlson comorbidities from electronic medical records". Mayo Clin Proc 2012; 87: 817.

    CrossrefMedlineGoogle Scholar
  • 24. Stevenson L.W., Pagani F.D., Young J.B.et al. : "INTERMACS profiles of advanced heart failure: the current picture". J Heart Lung Transplant 2009; 28: 535.

    CrossrefMedlineGoogle Scholar
  • 25. Gupta N., Kontos M.C., Gupta A.et al. : "Characteristics and outcomes in patients undergoing percutaneous coronary intervention following cardiac arrest (from the NCDR)". Am J Cardiol 2014; 113: 1087.

    CrossrefMedlineGoogle Scholar
  • 26. Jentzer J.C., Herrmann J., Prasad A., Barsness G.W. and Bell M.R. : "Utility and challenges of an early invasive strategy in patients resuscitated from out-of-hospital cardiac arrest". J Am Coll Cardiol Intv 2019; 12: 697.

    View ArticleGoogle Scholar
  • 27. Rihal C.S., Naidu S.S., Givertz M.M.et al. : "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 Assocation, the Cardiological Society of India, and Sociedad Latino Americana de Cardiologia Intervencion; Affirmation of Value by the Canadian Association of Interventional Cardiology-Association Canadienne de Cardiologie d'intervention". J Am Coll Cardiol 2015; 65: e7.

    View ArticleGoogle Scholar
  • 28. Hasdai D., Holmes D.R., Califf R.M.et al. : "Cardiogenic shock complicating acute myocardial infarction: predictors of death. GUSTO Investigators. Global Utilization of Streptokinase and Tissue-Plasminogen Activator for Occluded Coronary Arteries". Am Heart J 1999; 138: 21.

    CrossrefMedlineGoogle Scholar
  • 29. Stretch R., Sauer C.M., Yuh D.D. and Bonde P. : "National trends in the utilization of short-term mechanical circulatory support: incidence, outcomes, and cost analysis". J Am Coll Cardiol 2014; 64: 1407.

    View ArticleGoogle Scholar

Abbreviations and Acronyms

ACS

acute coronary syndrome

AKI

acute kidney injury

APACHE

Acute Physiology and Chronic Health Evaluation

BP

blood pressure

CA

cardiac arrest

CCI

Charlson Comorbidity Index

CI

confidence interval

CICU

cardiac intensive care unit

CS

cardiogenic shock

ECMO

extracorporeal membrane oxygenation

HF

heart failure

MCS

mechanical circulatory support

MI

myocardial infarction

OR

odds ratio

SCAI

Society for Cardiovascular Angiography and Intervention

VIS

vasoactive-inotropic score

Footnotes

Dr. Baran has served as a consultant for Abiomed, Abbott, Getinge, and LivaNova; and has served as a speaker for Novartis. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.

Listen to this manuscript's audio summary by Editor-in-Chief Dr. Valentin Fuster on JACC.org.