Defining levels of care in cardiogenic shock

Abstract

Background: Expert opinion and professional society statements have called for multi-tier care systems for the management of cardiogenic shock (CS). However, little is known about how to pragmatically define centers with different levels of care (LOC) for CS.

Methods: Eleven of 23 hospitals within our healthcare system sharing a common electronic health record were classified as different LOC according to their highest mechanical circulatory support (MCS) capabilities: Level 1 (L-1)-durable left ventricular assist device, Level 1A (L-1A)-extracorporeal membrane oxygenation, Level 2 (L-2)-intra-aortic balloon pump and percutaneous ventricular assist device; and Level 3 (L-3)-no MCS. All adult patients treated for CS (International Classification of Diseases, ICD-10 code R57.0) between 2016 and 2022 were included. Etiologies of CS were identified using associated diagnostic codes. Management strategies and outcomes across LOC were compared.

Results: Higher LOC centers had higher volumes: L-1 (n = 1): 2,831 patients, L-1A (n = 4): 3,452, L-2 (n = 1): 340, and L-3 (n = 5): 780. Emergency room admissions were more common in lower LOC (96% at L-3 vs. 46% L-1; p < 0.001), while hospital transfers were predominant at higher LOC (40% at L-1 vs. 2.7% at L-3; p < 0.001). Men comprised 61% of the cohort. Patients were younger in the higher LOC [69 (60-78) years at L-1 vs. 77 (67-85) years at L-3; p < 0.001]. Patients with acute myocardial infarction (AMI)-CS and acute heart failure (AHF)-CS were concentrated in higher LOC centers while other etiologies of CS were more common in L-2 and L-3 (p < 0.001). Cardiac arrest on admission was more prevalent in lower LOC centers (L-1: 2.8% vs. L-3: 12.1%; p < 0.001). Patients with AMI-CS received more percutaneous coronary intervention in lower LOC (51% L-2 vs. 29% L-1; p < 0.01) but more coronary arterial bypass graft surgery at higher LOC (L-1: 42% vs. L-1A: 23%; p < 0.001). MCS use was consistent across levels for AMI-CS but was more frequent in higher LOC for AHF-CS patients (L-1: 28% vs. L-2: 10%; p < 0.001). Despite increasing in-hospital mortality with decreasing LOC, no significant difference was seen after multivariable adjustment.

Conclusion: This is the first report describing a pragmatic classification of LOC for CS which, based on MCS capabilities, can discriminate between centers with distinct demographics, practice patterns, and outcomes. This classification may serve as the basis for future research and the creation of CS systems of care.

Keywords: acute heart failure; acute myocardial infarction; cardiogenic shock; cardiogenic shock centers; levels of care; mechanical circulatory support.

Figure 1

Median age and length of stay for cardiogenic shock patients across levels of care. (A) Median age was higher in lower LOC. (B) Median length of stay was longer in higher LOC. The p-values shown represent trends across LOC.

Figure 2

Prevalence of cardiogenic shock by levels of care over time. Average number of patients with either a principal or secondary diagnosis of cardiogenic shock (ICD-10 code R57.0) in each year from 2016 up until August 2022.

Figure 3

Relative prevalence of each cardiogenic shock etiology across levels of care. AMI-CS and AHF-CS were more prevalent in the higher LOC while “other” etiologies of CS were more prevalent in the lower LOC. The p-value shown is for trend across LOC.

Figure 4

Inter-hospital transfer map. Sankey diagram displaying the transfer of CS patients from their original level of care to their destination level.

Figure 5

Frequency of mechanical circulatory support and pulmonary artery catheter use over time across levels of care. (A) Trends in overall MCS use per LOC. (B) Trends in pulmonary artery catheter use over time by LOC. The p-values shown represent overall difference in MCS or PAC usage across LOC.

Figure 6

Management procedures by cardiogenic shock etiology in each level of care. (A) Level 1 LVAD-capable center. (B) Level 1A ECMO-capable centers. (C) Level 2 pVAD-capable center. Level 3 not depicted since no interventional or surgical procedures are performed at this level.

Figure 7

Observed mortality rate among cardiogenic shock patients with and without cardiac arrest on admission by level of care. Mortality increased in a stepwise fashion with decreasing LOC and was higher among patients with cardiac arrest on admission in all LOC (all p < 0.001).

Figure 8

Forest plot of factors associated with hospital mortality for cardiogenic shock patients. After multivariable adjustment, level of care was not associated with hospital mortality. Level 1 was used as reference for LOC, male as a reference for sex category, AMI-CS as reference for etiology, and Medicare/Medicaid as reference for insurance category.

Figure 9

Hospital outcomes by cardiogenic shock etiology: (A) level 1 LVAD-capable center, (B) level 1A ECMO-capable centers, (C) level 2 pVAD-capable centers, and (D) level 3 no MCS capability.

Figure 10

Levels of care for cardiogenic shock defined by highest MCS capability. Important differences in demographics, management strategies and outcomes were seen across different levels. Main characteristics of each level are summarized here.