Predictive Value Of Glycosylated Hemoglobin For Postoperative Acute Kidney Injury In Patients Undergoing Non-cardiac Surgery

Objective: Acute kidney injury(AKI)is one of the common complications after surgery,glycosylated hemoglobin(Hb A1c)can reflect the average blood glucose level in the past 2-3 months.Patients with elevated preoperative Hb A1 c were at increased risk of postoperative adverse events.This study aims to explore relationship between pre-operative Hb A1 c and postoperative AKI in patients undergoing non-cardiac surgery.Methods: We reviewed the medical records of 2785 patients(≥18years old)who underwent non-cardiac surgery in the Third Xiangya Hospital of Central South University between October 2011 and December 2020.The occurrence of postoperative AKI was diagnosed by comparing the changes in serum creatinine levels before and after surgery according to KDIGO criteria,and the patients were divided into postoperative AKI group(94 patients)and non-postoperative AKI group(2691 patients).Univariate,multivariate logistic regression and trend test were used to analyze the relationship between preoperative Hb A1 c and postoperative AKI.The variables in the univariate logistic analysis(P<0.1)and the risk factors of AKI reported in the previous literature were included in the multivariate logistic regression analysis,which included age,preoperative glomerular filtration rate,albumin,hemoglobin,Hb A1 c,ASA,liquid amount,blood loss,hypertension,coronary heart disease.These 10 variables constructed 2 models respectively,one model was model 1,it contained 9 variables(without Hb A1c),and the other model was model 2,it contained 10 variables(with Hb A1c).Hb A1 c was also added to the previously reported postoperative AKI models—SPARK model and GSAKI model,to form SPARK+Hb A1 c model and GSAKI+Hb A1 c model.Area under the receiver operating characteristic curve(AUC),Net reclassification index(NRI),Integrated discrimination improvement index(IDI)were used to evaluate the predictive ability for different models,clinical decision curve analysis(DCA)was used to evaluate the net benefit of patients with different models after adding the Hb A1 c variable.Finally,in the form of nomogram,the clinical utility of Model 2containing the variable "preoperative Hb A1c" was visualized.Results: 3.3% of patients(94 of 2785)developed acute kidney injury within one week after surgery.In univariate logistic regression analysis,Hb A1 c was associated with postoperative AKI(P=0.002);in multivariate logistic regression analysis,the study concluded that Hb A1 c was an independent risk factor for postoperative AKI(P=0.001).The trend test of Hb A1 c also showed that Hb A1 c was independently associated with postoperative AKI(OR 5.02,95%CI,1.90-13.24,P=0.001).Compared to the model 1,Hb A1 c improved predictive power(AUC value from 0.7387 to 0.7543,continuous NRI 0.2767,95%CI,0.0715-0.4818,IDI 0.0048,95%CI,-5e-04-0.0101).Compared with the SPARK model,the predictive ability of the SPARK+Hb A1 c model was also improved,(AUC value from 0.7024 to 0.7158,continuous NRI was 0.2136,95% CI,0.0094-0.4177,IDI was 0.0015,95%CI,-9e-04-0.0039).Compared with the GSAKI model,the predictive ability of the GSAKI+Hb A1 c model was significantly improved(AUC value from 0.6593 to 0.6780,continuous NRI was 0.2371,95%CI,0.0325-0.4417,IDI was 0.001,95%CI,-0.0011-0.0032).In the DCA analysis,the model with Hb A1 c significantly improved the net benefit of patients compared with the model without Hb A1 c.Model 2 with the addition of "preoperative Hb A1 c level" can be used clinically to predict postoperative AKI risk in non-cardiac surgery patients.Conclusions: Preoperative Hb A1 c level is an independent risk factor for postoperative AKI,it can effectively predict the risk of postoperative AKI in non-cardiac surgery patients and improved the predictive power in model 1,SPARK model,GSAKI model.