川崎病患儿并发心脏损害的风险模型构建及pro-BNP的预测价值

许辉; 刘敬; 张彤如

doi:10.12125/j.chj.202101061

川崎病患儿并发心脏损害的风险模型构建及pro-BNP的预测价值

Construction of risk model for heart damage in children with Kawasaki disease and predictive value of pro-BNP

摘要

摘要:
目的探究川崎病（KD）患儿并发心脏损害的风险及脑利钠肽前体（pro-BNP）的预测价值，建立预测心脏损害发生的危险因素列线图模型并验证。
方法采用随机数表法选取郴州市第一人民医院2016年1月～2020年7月收治的KD患儿450例，根据患儿是否发生心脏损害将其分为未发生心脏损害组(n=394) 和发生心脏损害组(n=56)。回顾性分析患者的临床信息，采用Logistic回归分析法筛选出KD患儿发生心脏损害的独立风险因素。利用R软件对筛选出的独立风险因素建立列线图预测模型，并对模型的预测性及准确度进行验证。
结果通过Logistic回归分析患者的风险因素数据可知，性别（OR=2.856，95%CI：1.493-5.461）、发热天数（OR=3.636，95%CI：1.636-8.083）、CRP（OR=2.519，95%CI：1.332-4.763）、治疗时机（OR=6.057，95%CI：2.289-16.024）、HGB（OR=4.705，95%CI：2.474-8.947）为KD患儿并发心脏损害的独立危险因素。基于该5项风险因素，建立列线图模型，Calibration验证结果显示，预测值与实测值曲线拟合度较高，表明该预测模型较实测准确度较好；同时Bootstrap内部验证结果显示，C-index指数高达0.837（95%CI：0.786～0.888），提示该模型精准度和区分度较好。同时，本研究发现，发生心脏损害组患儿的pro-BNP含量明显高于未发生组（P<0.05）。
结论通过建立风险模型筛选出的风险因素显示：性别、发热天数、CRP、治疗时机、血色素指标（HGB）为KD患儿发生心脏损害的风险因素，针对性干预措施将减少KD患儿心脏损害的发生率，pro-BNP指标可应用于KD疾病临床诊断和心脏损害的预测。

Abstract:
AIM To explore the risk of complicated heart damage in children with Kawasaki disease (KD) and the predictive value of forebrain natriuretic peptide (pro-BNP) and to establish and verify a line graph model to predict the risk factors of heart damage.
METHODS A total of 450 KD children admitted to our hospital from January 2016 to July 2020 were randomly selected by using the random number table, among whom 56 had heart damage and 394 had no heart damage. Clinical data of the patients was analyzed retrospectively and independent risk factors for heart damage of KD children were screened out by logistic regression analysis. R software was used to establish the nomogram prediction model for the screened independent risk factors, and the prediction and accuracy of the model were verified.
RESULTS By logistic binary regression analysis of the risk factor data of the patients, we could know that gender (OR=2.856, 95%CI: 1.493 - 5.461), days of fever (OR=3.636, 95%CI: 1.636 - 8.083), CRP (OR=2.519, 95%CI: 1.332 - 4.763), treatment timing (OR=6.057, 95%CI: 2.289 - 16.024) and course of disease during proglobulin treatment (OR=3.963, 95%CI: 2.065 - 7.607) were the independent risk factors for heart damage in children with KD. Based on the above five risk factors, a calibration graph model was established and the calibration result showed that the predicted value matched the measured value curve more closely, indicating that the predicted model had better accuracy than the actual measured value. Meanwhile, the internal validation results of Bootstrap showed that the C-index index was as high as (OR=4.705, 95%CI: 2.474-8.947), indicating that the model had good accuracy and differentiation.
CONCLUSION Through the establishment of a risk model, gender, fever days, CRP, treatment timing and HGB are screened out as the risks of heart damage in KD children. Effective prevention and treatment measures aimed at these risks could reduce the incidence of heart damage in KD children and are of significance for the prognosis of patients.

HTML全文

参考文献(21)

施引文献

资源附件(0)