疾病预测模型在慢性肾脏病并发造影剂肾病中的应用

Application of disease prediction model in chronic kidney disease complicated with contrast induced nephropathy

目的构建慢性肾脏病(CKD)并发造影剂肾病(contrast-induced nephropathy,CIN)的疾病预测模型,并检验该模型在临床的应用价值。方法选择2012年2月至2018年5月在南京医科大学附属淮安第一医院肾内科住院的CKD并发心血管疾病(cardiovascular disease,CVD)并行造影剂检查的患者作为研究对象。采用回顾性分析研究,记录所有患者在应用造影剂前后的临床资料,包括性别、年龄、CKD病因、左室射血分数(LVEF)、空腹血糖、血压、24 h尿量、血红蛋白(Hb)、血清肌酐(Scr)、肾小球滤过率(eGFR)和造影剂剂量,研究终点为CIN的发生。从中筛选出和CIN发生有明确关联的指标,并将这些指标作为变量,按照一定数值进行打分,统计每个患者的具体分值总和,然后根据总和观察患者发生CIN的风险并构建疾病预测模型。结果共有311例患者纳入本次研究,对照设定的CIN标准,我们发现其中80例在原有CKD基础上并发了CIN,并且年龄≥65岁、高血糖、低LVEF和高Scr的患者罹患CIN的风险更高。我们进一步将年龄、Scr、LVEF和血糖4个指标进行分值量化处理,将总分划分为3个级别,构建疾病预测模型。研究发现总分越高,发生CIN的风险也越大。结论利用4种变量构建疾病预测模型对CKD并发CIN有较好的预测作用,能指导临床进行早期防范和干预。

Objective To develop a disease prediction model of chronic kidney disease(CKD)with contrast induced Nephropathy(CIN),and to test its clinical value.Methods Patients with CKD complicated with cardiovascular disease(CVD)receiving contrast-based examination who were hospitalized in the department of nephropathy,the affiliated Huai’an No.1 people’s hospital of Nanjing medical university from February 2012 to May 2018,were selected as study subjects.A retrospective study was conducted to record the clinical data of all patients before and after the use of contrast medium,including gender,age,etiology of CKD,left ventricular ejection fraction(LVEF),fasting blood glucose,blood pressure,24-hour urine volume,hemoglobin(HGB),Serum creatinine(Scr),estimated glomerular filtration rate(eGFR)and contrast dose.The end point of this study was the occurrence of CIN.The variables that were tightly related to CIN were screened and scored,the total score of each patient was summarized in order to evaluate the risk of CIN,then the disease prediction model was developed based on the total score and risk for CIN in the patients was assessed by using the model.Results A total of 311 patients were included in the study,according to the established CIN standard,80 patients were found to have developed contrast induced nephropathy on the basis of original CKD.And the patients at the age of over 65,with high blood glucose,low LVEF,and high Scr had higher risk to develop CIN.After the four variables of scoring age,Scr,LVEF and blood glucose,and the total scores were divided into three grades to develop the disease prediction model.It was found that with higher the total score,CKD patients had higher risk for CIN.Conclusions Using four variables to build the corresponding risk prediction model has a significant predictive effect on the occurrence of CKD complicated with CIN,and can guide early prevention and intervention in clinical treatment.