RESUMO
Objective To evaluate the accuracy of high-aensitivity C-reactive protein (hs-CRP) in the prediction of perioperative cardiovascular accidents (CVAS) in patients with diabetes mellitus. Methods Ninetyone type 2 diabetic patients undergoing elective non-cardiac surgery were divided into 3 groups according to their blood concentrations of hs-CRP: low-concentration group ＜ 1.0 mg/L ( group Ⅰ , n = 9); median-concentration group 1.0-2.9 mg/L (group Ⅱ ,n= 16) and high-concentration group 3.0-10.0 mg/L (group Ⅲ ,n =66). Risk factors for perioperative CVA were analyzed using Spearman rank correlation analysis. Various risk factors including sex, age, ASA physical status, BMI, NYHA claasification, smoking history ,preoperative ECG changes, duration,treatment and complications of diabetes mellitus, hypertension and blood lipids were correlated with perioperative CVA. The risk factors of which P values were less than 0.05 would enter the binary logistic regression analysis to stratify CVA-related risk factors. Area under the ROC curve was used to analyze the accuracy of the risk factors in prediction of perioperative CVA. Results The incidences of perioperative CVAs were 0, 31% and 35 % in Ⅰ ,Ⅱ and Ⅲ groups respectively. It was determined by Spearman rank correlation analysis that preoperative ECG,NYHA classification, age, preoperative cardiovascular color doppler ultrasound, blood hs-CRP concentration and ASA classification of physical status were important risk factors ( P ＜ 0.05 or 0.01 ). Binary logistic regression .malysis indicated that preoperative ECG, NYHA classification and age were highly related to perioperative CVA (P ＜0.05 or 0.01).The area under ROC curve of preoperative ECG, NYHA classification, age and blood hsCRP concentration was 0.855, 0.755, 0.702 and 0.605 respectively. Conclusion The accuracy of blood concentration of hs-CRP in prediction of perioperative CVA in patients with diabetes mellitus is lower than that of the preoperative ECG, NYHA classification or age.