Intra-aortic balloon pump use and in-hospital mortality in acute myocardial infarction with Killip class III or IV: a nationwide cohort study

Figures

• 

  Fig. 1. Study cohort. This flow diagram illustrates the derivation of the analytic cohort from adult patients with acute myocardial infarction recorded in the China Chest Pain Center Registry from 2019 through 2023. Patients treated in hospitals with no recorded IABP use were excluded to restrict the analysis to hospitals in which the IABP strategy was observed in routine practice. Numbers above the weighting step represent unweighted patient counts; numbers below the weighting step represent weighted counts after sIPTW-BART. AMI indicates acute myocardial infarction; CPC, Chest Pain Center; ECMO, extracorporeal membrane oxygenation; IABP, intra-aortic balloon pump; LVAD, left ventricular assist device; PCI, percutaneous coronary intervention; sIPTW-BART, stabilized inverse probability of treatment weighting based on Bayesian additive regression trees–estimated propensity scores

• 

  Fig. 2. Weighted cumulative risk of in-hospital all-cause mortality by treatment group. The curve shows the cumulative incidence of in-hospital all-cause mortality among patients with acute myocardial infarction and Killip class III or IV presentation. Weighted cumulative incidence curves were generated within each of the 40 RF-MICE completed datasets using sIPTW-BART weights and then averaged across imputations for display. Hazard ratios and 95% confidence intervals were estimated using weighted Cox proportional hazards models with hospital-clustered robust sandwich standard errors, fitted separately within each imputed dataset and pooled using Rubin’s rules, with the control group as the reference. Numbers at risk are shown below the panel. Proportional hazards diagnostics and treatment-by-time interaction sensitivity analyses are reported in Table S5 (Additional file 1)

• 

  Fig. 3. Subgroup analyses of in-hospital all-cause mortality by treatment group. Hazard ratios and 95% confidence intervals for in-hospital all-cause mortality associated with intra-aortic balloon pump use versus control were estimated using sIPTW-BART–weighted Cox proportional hazards models fitted separately within each of the 40 RF-MICE completed datasets and pooled using Rubin’s rules. Robust sandwich standard errors were clustered at the hospital level. Subgroup analyses were performed according to age, sex, systolic and diastolic blood pressure on admission, consciousness status, chest pain presentation, type of myocardial infarction, troponin status, creatinine level, B-type natriuretic peptide or N-terminal pro–B-type natriuretic peptide elevation, and hospital level. P values for interaction were calculated to assess association modification across subgroups. Subgroup analyses were exploratory