We conducted a retrospective cohort study of adults with myocardial infarction (MI) and cardiogenic shock during the index hospitalization after coronary artery bypass graft (CABG) surgery using the United States TriNetX (2006 to 2025) database, comparing those who underwent percutaneous coronary intervention (PCI: ≤24 h) to those who did not. Outcomes were reported as unadjusted and adjusted hazard ratios (HRs) or subdistribution HRs (sHRs) using propensity score matching and Fine–Gray models. The primary endpoint was 30-day all-cause mortality. Secondary endpoints included 30-day hospitalizations for MI, stroke, heart failure (HF), and all-cause hospitalization. Sensitivity analyses included a falsification endpoint (bone fracture) and landmark analyses by PCI timing (≤12 h vs 12 to 24 h). Among 218,827 patients, 22,762 underwent PCI. Compared with those who did not undergo PCI, PCI was associated with lower rates of all-cause mortality (4.5% vs 5.7%), MI (5.0% vs 7.0%), stroke (2.5% vs 5.0%), HF (15.0% vs 28.0%), and all-cause hospitalization (20.0% vs 28.0%). These estimates remained consistent after adjustment for all-cause mortality (HR, 0.88 [95% CI, 0.80 to 0.96]), MI (sHR 0.86 [0.79 to 0.94]), stroke (sHR 0.41 [0.37 to 0.46]), HF (sHR 0.52, 0.50 to 0.55), and all-cause hospitalization (sHR 0.70 [0.67 to 0.73]). PCI showed no association with hospitalization for bone fracture (HR, 0.82 [0.62 to 1.10). Landmark analyses revealed similar trends for mortality at both ≤12 h and 12 to 24 h intervals. In conclusion, PCI was associated with lower 30-day mortality and cardiovascular outcomes after CABG-related cardiogenic shock. These findings are hypothesis-generating and warrant confirmation in prospective studies.
Coronary artery bypass grafting (CABG) remains a cornerstone of revascularization for patients with complex coronary artery disease, offering durable symptom relief and survival benefits for select individuals. However, postoperative complications, including cardiogenic shock, occur in approximately 1% to 3% of cases, often triggered by acute myocardial ischemia due to graft failure, incomplete revascularization, or native vessel progression. , This scenario carries a high in-hospital mortality ranging from 18% to 50%. , In this setting, percutaneous coronary intervention (PCI) has been proposed as a rescue strategy to address acute ischemia, especially when graft occlusion or native vessel compromise is suspected. Current guidelines recommend urgent angiography and PCI for postoperative myocardial ischemia manifesting as hemodynamic instability or shock. The SHOCK trial demonstrated similar 30-day and 1-year survival rates between PCI and CABG in patients with acute myocardial infarction (MI) developing shock. Observational data further indicate that PCI outcomes in patients with prior CABG are comparable to those without prior CABG when targeting native vessels, but worse when intervening on grafts (adjusted hazard ratio [HR] 1.33 for mortality), highlighting procedural challenges in this cohort. Despite these insights, dedicated outcome data among those who undergo PCI in cardiogenic shock arising during index hospitalization after CABG remain scarce. To address this evidence gap, the present study aims to evaluate all-cause mortality and cardiovascular outcomes among patients who received PCI within 24 hours for cardiogenic shock due to MI, after isolated CABG.
Methods
This study was exempt from Institutional Board Review approval due to the use of deidentified publicly available data. This study follows the Reporting of studies Conducted using Observational Routinely-collected health data (RECORD) guidelines. This retrospective study was conducted utilizing the United States (US) Collaborative Network of TriNetX database, a comprehensive federated health research platform that provides access to electronic health records (EHRs), encompassing demographic details, socioeconomic indicators, diagnoses, and procedures [recorded using ICD-10 (international classification of diseases, 10th revision, and CPT (Current Procedure Terminology) codes)], medications, laboratory values, and genomic data for more than 90 million patients across >60 healthcare organizations (HCOs) across 50 US states (22% Northeast, 16% Midwest, 39% South, 13% West, 10% Unspecified).
TriNetX employs stringent deidentification protocols compliant with Section 164.514(a) of the HIPAA Privacy Rule. Demographic attributes such as self-reported sex (men, women), and standardized race/ethnicity categories (Asian, American Indian/Alaskian Native, Black/African American, Native Hawaiian/Other Pacific Islander, White, Hispanic or Latino, Not Hispanic or Latino, and Unknown) are derived from the EHRs of participating institutions.
TriNetX conducts thorough data preprocessing to reduce missing values and standardizes the data into a uniform clinical model, ensuring consistent query results across different data sources. All variables are structured as binary, categorical (converted to multiple binary columns), or continuous. Age data is always available. In cases where sex data is missing, it is labeled as “Unknown Sex.” Absences in race/ethnicity data are categorized as “Unknown Race” or “Unknown Ethnicity.” For other data types, such as medical conditions, procedures, lab tests, and social determinants of health, the data are either available or not; we excluded cases with missing values.
This retrospective cohort compared the strategy of PCI (within 24 hours) versus no PCI among patients who developed MI and cardiogenic shock during the index hospitalization after CABG. Eligible patients were identified from the TriNetX database spanning January 2006 to June 2025. Inclusion criteria focused on adults (≥18 years) who underwent isolated CABG and subsequently developed MI and cardiogenic shock during the same index hospitalization ( Supplementary Table 1 for code lists). Exclusion criteria included prior CABG, PCI before CABG within 3 months, mechanical complications, missing mortality data or follow-up, missing procedure dates, prior CABG, hybrid procedures, or isolated valve surgeries. Follow-up commenced at shock diagnosis for both groups and ended at 30 days, with PCI treated as a time-dependent exposure to mitigate immortal time bias.
The primary endpoint was 30-day all-cause mortality. This was chosen over cardiovascular mortality due to limitations in cause-of-death adjudication. Secondary endpoints included MI, stroke, HF, and all-cause hospitalization at 30 days. We also identified in-hospital complications: stroke, major bleeding, initiation of renal replacement therapy (RRT), nosocomial infection, and vascular complications. A falsification endpoint—30-day bone fracture hospitalization—was included to assess residual confounding.
Statistical analysis
All analyses were performed between July 1 and July 10, 2025. Continuous data were reported as medians (Q1 to Q3) and categorical variables as numbers (%). We used the Rao-Scott Chi-square test for categorical variables and the t-test or Wilcoxon rank-sum test for continuous variables. We reported both unadjusted estimates and adjusted estimates after propensity score matching and multivariable modeling.
Propensity score matching was performed using a set of covariates, including demographics (age, sex, ethnicity/race), comorbidities (hypertension, diabetes mellitus, hyperlipidemia, smoking, prior stroke, prior MI, peripheral vascular disease, atrial fibrillation/flutter, prior heart failure (HF), drug or alcohol abuse, chronic obstructive pulmonary disease, renal disease, neuropathy, cancer), operative details (on-pump vs off-pump CABG), clinical presentation (ST-elevation MI, Non-ST-elevation MI), use of mechanical circulatory support, in-hospital pressors or inotropes (norepinephrine, dopamine, epinephrine, dobutamine, milrinone), laboratory values (left ventricular ejection fraction [LVEF], arterial lactate, low density lipoprotein cholesterol [LDL-C], hemoglobin, troponin), and discharge medications (aspirin, P2Y12 inhibitors, angiotensin-converting enzyme inhibitor; angiotensin II receptor blocker; statins, fibrates, ezetimibe, proprotein convertase subtilisin/kexin type 9 inhibitors, diuretics, calcium channel blockers, nitrates, spironolactone, angiotensin receptor-neprilysin inhibitor, sodium-glucose cotransporter-2 inhibitors, ranolazine) ( Table 1 ).
Table 1
Baseline characteristics of the study population before propensity matching
| Characteristics |
PCI
( n = 22,762) |
No PCI
( n = 196,065) |
p-value |
|---|---|---|---|
| Age, years | 74.1 (66.6-81.6) | 75.4 (67.9-82.9) | <0.001 |
| Women | 5,879 (25.9%) | 46,801 (25.7%) | 0.59 |
| Ethnicity/race | |||
| White adults | 17,112 (75.3%) | 133,526 (73.3%) | <0.001 |
| Black adults | 2,126 (9.4%) | 17,168 (9.4%) | 0.72 |
| Hispanic adults | 1,023 (4.5%) | 9,544 (5.2%) | <0.001 |
| Asian adults | 818 (3.6%) | 6,823 (3.7%) | 0.26 |
| Native American adults | 104 (0.5%) | 782 (0.4%) | 0.54 |
| Native Hawaiian or Other Pacific Islander | 236 (1.0%) | 3,519 (1.9%) | <0.001 |
| Other/Unknown | 2,337 (10.3%) | 20,317 (11.1%) | <0.001 |
| Comorbidities | |||
| Hypertension | 12,078 (53.1%) | 87,856 (48.2%) | <0.001 |
| Diabetes mellitus | 9,506 (41.8%) | 79,556 (43.7%) | <0.001 |
| Hyperlipidemia | 12,668 (55.7%) | 91,359 (50.2%) | <0.001 |
| Smoking | 3,954 (17.4%) | 24,085 (13.2%) | <0.001 |
| Prior stroke | 1,623 (7.1%) | 15,157 (8.3%) | <0.001 |
| Prior MI | 5,525 (24.3%) | 30,731 (16.9%) | <0.001 |
| PVD | 2,002 (8.8%) | 16,588 (9.1%) | 0.13 |
| Atrial fib/flutter | 646 (2.8%) | 8,544 (4.7%) | <0.001 |
| Prior heart failure | 8,116 (35.7%) | 80,897 (44.4%) | <0.001 |
| Drug abuse | 50 (0.2%) | 422 (0.2%) | 0.73 |
| Alcohol | 325 (1.4%) | 2,952 (1.6%) | 0.03 |
| COPD | 2,850 (12.5%) | 28,462 (15.6%) | <0.001 |
| Renal disease | 570 (2.5%) | 7,374 (4.0%) | <0.001 |
| Neuropathy | 252 (1.1%) | 2,915 (1.6%) | <0.001 |
| Cancer | 26 (0.1%) | 683 (0.4%) | <0.001 |
| Off-pump CABG | 3,414 (15.0%) | 23,528 (12.0%) | <0.001 |
| On-pump CABG | 19,348 (85.0%) | 172,537 (88.0%) | <0.001 |
| Clinical presentation | |||
| ST-segment elevation MI | 6,557 (28.8%) | 39,184 (20.0%) | <0.001 |
| Non-ST-segment elevation MI | 16,205 (71.2%) | 156,881 (80.0%) | <0.001 |
| Vitals | |||
| Systolic BP, mm Hg | 85 (75-95) | 92 (80-105) | <0.001 |
| Diastolic BP, mm Hg | 50 (40-60) | 55 (45-65) | <0.001 |
| HR, beats/min | 110 (92-125) | 100 (85-115) | <0.001 |
| Mechanical circulatory support | |||
| IABP | 7,967 (35.0%) | 39,213 (20.0%) | <0.001 |
| Impella | 2,731 (12.0%) | 7,843 (4.0%) | <0.001 |
| Mechanical ventilation | 1,373 (6.0%) | 13,174 (7.2%) | <0.001 |
| In-hospital pressors/inotropes | |||
| Norepinephrine | 1,527 (6.7%) | 11,192 (6.1%) | 0.001 |
| Dopamine | 733 (3.2%) | 2,442 (1.3%) | <0.001 |
| Epinephrine | 1,716 (7.5%) | 10,945 (6.0%) | <0.001 |
| Dobutamine | 450 (2.0%) | 3,770 (2.1%) | 0.36 |
| Milrinone | 324 (1.4%) | 3,785 (2.1%) | <0.001 |
| Laboratory values, median (Q1-Q3) | |||
| LVEF, % | 46.7 (36.6-56.8) | 46.5 (35.8-57.2) | 0.54 |
| Arterial lactate level, mmol/L | 2.5 (0.7-4.3) | 2.4 (0.9-3.9) | 0.04 |
| LDL-C, mg/dL | 93.6 (63.2-124.0) | 93.4 (62.6-124.2) | 0.79 |
| Hgb, g/dL | 12.5 (10.8-14.2) | 11.9 (10.0-13.8) | <0.001 |
| Troponin, ng/L | 7.0 (0.0-22.2) | 3.2 (0.0-22.2) | <0.001 |
| Medications at discharge | |||
| Aspirin | 16,757 (73.7%) | 79,211 (43.5%) | <0.001 |
| P2Y12 inhibitor | 14,954 (65.7%) | 27,755 (15.2%) | <0.001 |
| ACEI/ARB | 4,757 (20.9%) | 20,647 (11.3%) | <0.001 |
| Statin | 15,958 (70.2%) | 79,201 (43.5%) | <0.001 |
| Fibrates | 599 (2.6%) | 2,510 (1.4%) | <0.001 |
| Ezetimibe | 1,124 (4.9%) | 3,981 (2.2%) | <0.001 |
| PCSK9 inhibitor | 46 (0.2%) | 248 (0.1%) | 0.01 |
| Diuretics agent | 7,308 (32.1%) | 49,787 (27.3%) | <0.001 |
| Calcium-channel blocker | 8,482 (37.3%) | 29,330 (16.1%) | <0.001 |
| Nitrates | 15,491 (68.1%) | 57,401 (31.5%) | <0.001 |
| Spironolactone | 689 (3.0%) | 4,535 (2.5%) | <0.001 |
| ARNI | 323 (1.4%) | 2,002 (1.1%) | <0.001 |
| SGLT2 inhibitor | 454 (2.0%) | 2,199 (1.2%) | <0.001 |
| Ranolazine | 957 (4.2%) | 2,895 (1.6%) | <0.001 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
