儿童慢性扁桃体炎发病因素的列线图预测模型构建与评价

Construction and evaluation of nomogram prediction model for pathogenic factors of chronic tonsillitis in children

目的探讨儿童慢性扁桃体炎的发病因素并构建列线图预测模型。方法选取2022年2月至2023年2月新余市中医院收治的因慢性扁桃体炎行手术治疗的80例患儿作为研究组,选取同期新余市城北幼儿园及新余市长青小学80例健康儿童作为对照组,记录两组临床资料并进行单因素分析,采用多因素logistic回归模型分析儿童出现扁桃体炎的影响因素,采用R语言软件4.0“rms”包构建儿童扁桃体炎危险因素的预测模型,通过校正及决策曲线对列线图预测模型进行内部验证及临床预测效能评估。结果两组是否为剖宫产儿、家族史、过敏史、清淡饮食、父母有无吸烟史、服用免疫抑制剂比较,差异有统计学意义(P<0.05)。多因素logistic回归模型分析结果显示,剖宫产儿(β=2.456,OR=11.759,95%CI:2.351~58.821)、有家族史(β=1.860,OR=3.142,95%CI:1.267~6.914)、有过敏史(β=1.817,OR=6.421,95%CI:1.315~31.344)、父母有吸烟史(β=1.801,OR=15.000,95%CI:1.287~28.484)、服用免疫抑制剂(β=3.876,OR=48.302,95%CI:5.175~450.844)是儿童发生慢性扁桃体炎的独立危险因素,清淡饮食(β=-3.711,OR=0.690,95%CI:0.213~0.898)为儿童发生慢性扁桃体炎的保护因素。列线图预测模型的校正曲线与原始曲线及理想曲线接近,C-index为0.983(0.965~1.000),模型拟合度较高。列线图预测模型的阈值>0.17,可提供临床净收益,且临床净收益均高于剖宫产儿、家族史、过敏史、清淡饮食、父母有无吸烟史、免疫抑制剂(P<0.05)。结论儿童慢性扁桃体炎发病的影响因素与剖宫产、家族史、过敏史、清淡饮食、父母吸烟史、免疫抑制剂有关,以其作为预测因子构建个性化列线图预测模型有助于对儿童发生慢性扁桃体炎的预测评估,以期尽早预防。

Objective To investigate the pathogenic factors of chronic tonsillitis in children and establish a nomogram prediction model.Methods A total of 80 children treated in Xinyu Hospital of Traditional Chinese Medicine for chronic tonsillitis from February 2022 to February 2023 were selected as the study group,and 80 healthy children from Chengbei Kindergarten and Changqing Primary School in Xinyu were selected as the control group.Clinical data of the two groups were recorded and univariate analysis was performed.Multivariate logistic regression model was used to analyze the influencing factors of tonsillitis in children.The R language software 4.0"rms"package was used to construct prediction models of risk factors for childhood tonsillitis,and the calibration and decision curve were used to intemally verify the nomogram prediction model and evaluate the clinical prediction efficiency.Results There were statistically significant differences between the two groups in whether the children were delivered by caesarean section,family history,allergic history,light diet,parents'smoking history and taking immunosuppressants(P<0.05).The results of multivariate logistic regression model analysis showed that the children were born by cesarean section(β=2.456,OR=11.759,95%CI:2.351-58.821),had family history(β=1.860,OR=3.142,95%CI:1.267-6.914),had allergic history(β=1.817,OR=6.421,95%CI:1.315-31.344),parents had smoking history(β=1.801,OR=15.000,95%CI:1.287-28.484),taking immunosuppressants(β=3.876,OR=48.302,95%CI:5.175-450.844)were independent risk factor for chronic tonsillitis in children.Light diet(β=-3.711,OR=0.690,95%CI:0.213-0.898)was a protective factor for chronic tonsillitis in children.The correction curve of the nomogram prediction model was close to the original curve and the ideal curve,and the C-index was 0.983(0.965-1.000),indicating a high degree of fit.The threshold value of the nomogram predicted model was>0.17,which could provide clinical net benefit,and the clinical net benefit was higher than those of C-section infants,family history,allergic history,light diet,parents'smoking history,and immunosuppressants(P<0.05).Conclusion The factors affecting the incidence of chronic tonsillitis in children are related to caesarean section,family history,allergic history,light diet,parents'smoking history and immunosuppressants.Using these factors as predictors,the construction of a personalized columbaric prediction model is helpful for the prediction and evaluation of chronic tonsillitis in children,so as to prevent it as early as possible.