Risk factors of acute kidney injury in severe COVID-19 patients and their predictive values

Objective  To screen the risk factors of acute kidney injury （AKI）in severe COVID-19 patients, construct a predictive model and explore its predictive value.

Methods  From December 1,2022 to July 1, 2023, the relevant clinical data were retrospectively reviewed for 150 patients of severe COVID-19. Based upon the presence or absence of AKI, they were assigned into two groups of AKI（n=54）and non-AKI （n=96）. Multivariate Logistic regression analysis was utilized for screening the risk factors for AKI in severe COVID-19 patients. A prediction model was constructed. And predictive value of model was verified by receiver operating characteristic curve（ROC）.

Results  Age was significantly lower in AKI group than that in non-AKI group74.50（63.00, 81.50） vs 75.00（70.00, 83.00） year. Proportion of AKI plus chronic obstructive pulmonary disease（1.85%）and proportion on high-flow nasal cannula（HFNC）（38.89%）were significantly lower than those in non-AKI group（13.54%, 58.33%） （both P＜0.05）. Scores of APACHE Ⅱ20.50（15.00, 25.00） and sequential organ failure assessment（SOFA）8.00（6.00, 10.00） were significantly higher in AKI group than those in non-AKI group16.00（13.00, 18.25）, 5.00（4.00, 6.00）. Proportion of patients with bacterial infection（61.11%） and receiving invasive ventilation therapy（57.41%）were significantly higher in AKI group than those in non-AKI group（40.62%, 25.00%） （both P＜0.05）. Oxygenation index109.55（75.99, 149.50）mmHg (1 mm Hg=0.133 kPa) and hemoglobin（109.61±24.46）g/L were significantly lower in AKI group than those in non-AKI group134.50（105.36, 156.57）mmHg and （121.24±21.77）g/L. N-terminal brain natriuretic peptide 3148.28（446.42, 13572.58）ng/L, procalcitonin1.48（0.56, 8.42）ng/L, partially activated thromboplastin time39.80（34.42, 45.32）second and aspartate aminotransferase38.00（21.00, 66.50）U/L were significantly higher in AKI group than those in non-AKI group502.41（171.50, 1703.00）ng/L, 370（110, 1040）ng/L, 36.95（32.70, 42.20）second, 34.00（20.50, 53.00）U/L. Blood urea nitrogen13.86（8.68, 18.70）mmol/L, serum creatinine177.00（90.30, 375.00）μmol/L and cystatin C2.67（1.77, 4.08）mg/L were significantly higher than those in non-AKI group7.31（5.51, 9.67）mmol/L, 62.95（51.75, 78.10）μmol/L and 1.34（1.10, 1.61）mg/L（all P＜0.05）. Multivariate Logistic regression analysis revealed that high SOFA score, concurrent bacterial infection and elevated serum creatinine were independent risk factors for AKI（all P＜0.05）. A prediction model was constructed with the above risk factors and ROC curve analysis performed. The results indicated that area under ROC curve（AUC） of model for predicting AKI was 0.916（95%CI:0.8687-0.9633） with a sensitivity of 77.36% and a specificity of 92.63%.

Conclusion  Elevated serum creatinine, high SOFA score and concurrent bacterial infection are independent risk factors for severe COVID-19 AKI. The prediction model constructed by the above risk factors has some value in predicting the occurrence of AKI.