Outcomes in Non-Ischemic Cardiogenic Shock
Key Takeaways
- In this large U.S. inpatient cohort of non-ischemic cardiogenic shock, most hospitalizations did not involve temporary mechanical circulatory support.
- Intra-aortic balloon pump was the most commonly used single device, and unadjusted in-hospital mortality was lowest with intra-aortic balloon pump and highest with extracorporeal membrane oxygenation.
- The source’s main adjusted analyses report lower observed in-hospital mortality with intra-aortic balloon pump than with no mechanical circulatory support (OR 0.58, 95% CI 0.53–0.64).
- All three device strategies were associated with higher costs than no mechanical circulatory support; the source reports longer hospitalizations for IABP and ECMO, but gives conflicting accounts of whether Impella was also associated with longer length of stay.
A retrospective cross-sectional analysis used the Nationwide Inpatient Sample from 2017 to 2019. Hospitalizations were identified with International Classification of Diseases, Tenth Revision, Clinical Modification code R57.0 for cardiogenic shock, and acute myocardial infarction codes I21.x and I22.x were excluded to isolate non-ischemic cases. Exposure groups were mutually exclusive no mechanical circulatory support, intra-aortic balloon pump (IABP) only, extracorporeal membrane oxygenation (ECMO) only, and Impella only, with combination-device hospitalizations excluded under HCUP cell-size rules. The final analytic cohort included 62,059 hospitalizations, and primary outcomes were in-hospital mortality, hospital length of stay, and total hospital cost. Adjusted models accounted for demographics, hospital characteristics, payer, and comorbidity burden using Elixhauser, with a sensitivity analysis using Charlson.
Most hospitalizations received no device support, and IABP was the most common single-device strategy. Among single-device groups, in-hospital mortality was 19.9% with IABP, 37.4% with Impella, and 44.3% with ECMO.
In the source’s main adjusted analyses, compared with no mechanical circulatory support, IABP was associated with lower observed in-hospital mortality, with OR 0.58 (95% CI 0.53–0.64; p < 0.01), although the source’s concluding text contains a contradictory statement; ECMO and Impella were associated with higher observed mortality, with OR 2.41 (95% CI 2.14–2.472; p < 0.01) and OR 1.39 (95% CI 1.24–1.57; p < 0.01), respectively. All three device strategies were also associated with higher costs; the source reports longer hospitalizations for IABP and ECMO, but gives conflicting accounts of whether Impella was also associated with longer length of stay. ECMO showed the largest adjusted cost increase, with a cost ratio of 3.30 (95% CI 3.12–3.50; p < 0.01).
Because the analysis was observational and cross-sectional, the reported associations do not establish causality, and residual confounding by indication and shock severity remains likely. Administrative data did not capture granular markers such as lactate, vasopressor dose, invasive hemodynamics, Sequential Organ Failure Assessment components, Society for Cardiovascular Angiography and Interventions stage, cardiac power output, or the timing of device implantation. Early planned support also could not be separated from rescue or salvage use, post-discharge outcomes were unavailable, and device-associated secondary complications were not analyzed because event timing could not be determined reliably.
Clinician Questions
How was non-ischemic cardiogenic shock defined in this nationwide inpatient analysis?
In this U.S. Nationwide Inpatient Sample analysis, non-ischemic cardiogenic shock was defined with International Classification of Diseases, Tenth Revision, Clinical Modification code R57.0 for cardiogenic shock, with acute myocardial infarction codes I21.x and I22.x excluded.
Which mechanical circulatory support strategies were compared in non-ischemic cardiogenic shock, and were combination-device cases included?
The comparison used mutually exclusive groups of no mechanical circulatory support, IABP only, ECMO only, and Impella only. Hospitalizations involving two or three support strategies were excluded under HCUP cell-size rules, so the estimates reflect single-device strategies only.
Why do the mortality comparisons in non-ischemic cardiogenic shock remain non-causal even after adjustment?
The models adjusted for demographics, hospital characteristics, payer, and comorbidity burden, but the database lacked granular shock-severity and timing variables such as lactate, vasopressor dose, invasive hemodynamics, SOFA components, SCAI stage, cardiac power output, and timing of support implantation. Early planned support also could not be separated from rescue or salvage use, so residual confounding by indication and illness severity remains likely.
What hospital-level outcomes were measured for mechanical circulatory support in non-ischemic cardiogenic shock?
The primary hospital-level outcomes were in-hospital mortality, hospital length of stay, and total hospital cost during the index hospitalization. Post-discharge survival, functional status, and readmission were not available in this database.