Acute kidney injury (AKI) is common in patients undergoing percutaneous coronary intervention (PCI). One risk factor for AKI is periprocedural hemoglobin drop level (> 3 g/dL); however, whether the relationship between hemoglobin drop and AKI is linear or nonlinear remains unknown. We aimed to investigate the relationship between periprocedural hemoglobin drop and AKI after PCI. We evaluated 14,273 consecutive patients undergoing PCI between September 2008 and March 2019. AKI was defined as an absolute or a relative increase in serum creatinine level of 0.3 mg/dL or 50%, respectively. Restricted cubic spline was constructed to assess the association between hemoglobin drop and AKI by logistic regression and machine learning (ML) models, which were used to predict the risk of AKI. The patients' mean age was 68.4 ± 11.6 years; the AKI incidence was 10.5% (N = 1499). An absolute > 3 g/dL or 20% relative decrease in hemoglobin level was an independent predictor of AKI incidence (odds ratio, OR [95% confidence interval, CI]: 2.24 [1.92-2.61], P < 0.001; 2.35 [2.04-2.71], P < 0.001, respectively). An adjusted restricted cubic spline demonstrated that absolute/relative decrease in hemoglobin was linearly associated with AKI. Logistic and ML models with absolute/relative hemoglobin changes were comparable while estimating the risk of AKI (absolute area under the curve [AUC] (logistic):0.826, AUC (ML): 0.820; relative AUC (logistic): 0.818, AUC (ML): 0.816). An absolute/relative decrease in periprocedural hemoglobin after PCI was linearly associated with AKI. Detection of a relative/absolute decrease in hemoglobin may help clinicians identify individuals as high risk for AKI after PCI.
Keywords: Acute kidney injury; Hemoglobin drop; Percutaneous coronary intervention; Periprocedural bleeding.