Presentation (CTI)

OBJECTIVES: Acute kidney injury (AKI) is a common, high-morbidity complication of coronary revascularization. This study compares the risk of AKI following coronary artery bypass grafting (CABG) versus percutaneous coronary intervention (PCI) in patients with coronary artery disease (CAD), using a combined multiple imputation (MI) and propensity score matching (PSM) approach to address selection bias and missing outcome data.
METHODS: We analyzed patients with CAD from the National COVID Cohort Collaborative (N3C) database who underwent CABG or PCI. Exclusions included prior dialysis, end-stage kidney disease, previous AKI, or combined procedures. AKI was defined by KDIGO creatinine criteria. Missingness in AKI status and baseline characteristics was addressed using MI by chained equations (m = 20). PSM was conducted to match baseline variables including age, modified EuroScore II, gender, race, ethnicity, diabetes, AIDS, cancer, hypertension, chronic kidney disease, obesity, and hyperlipidemia. Treatment effect estimates from logistic regressions were combined using Rubin’s rules.
RESULTS: Of the 75,805 patients included in the study, 14,040 (18.5%) had observed AKI, 47,974 (63.3%) had no AKI, and 13,829 (18.2%) had missing AKI outcome. After MI, a representative dataset yielded 17,641 (23.3%) AKI classifications. Covariate balance was achieved across all imputed datasets after PSM. The pooled analysis demonstrated that CABG was associated with significantly higher odds of AKI compared with PCI, yielding an Odds Ratio (OR) of 1.96 (95% CI: 1.83-2.09). The estimates were consistent across imputations, with small variance components (within-imputation variance = 0.00066; between-imputation variance = 0.00047).
CONCLUSIONS: This study, which leverages a combined MI-PSM methodology to address confounding and missing data, found that CABG was associated with nearly twice the odds of AKI compared with PCI. The robust association across 20 imputations and well-balanced matched cohorts supports the internal validity of these findings, providing crucial comparative effectiveness data for revascularization decisions.