非哺乳期乳腺炎的危险因素分析及预测模型建立

Analysis of risk factors of non-puerperal mastitis and establishment of prediction model

目的探讨非哺乳期乳腺炎发生的相关危险因素,并建立列线图预测模型。方法回顾性分析2011年9月1日至2019年9月1日首都医科大学附属北京天坛医院乳腺外科收治的116例非哺乳期乳腺炎患者(研究组)的临床资料,均经病理学检查确诊;随机选取同期在该院进行体检的未罹患乳腺相关疾病的女性116例作为对照组。采用单因素分析筛选出非哺乳期乳腺炎发生的危险因素,将有统计学意义的变量应用R软件进行多因素logistic回归分析,建立预测非哺乳期乳腺炎发生风险的列线图模型,采用Boostrap法进行模型验证,并绘制校准曲线,同时应用受试者工作特征(receiver operating characteristic,ROC)曲线评估列线图模型对非哺乳期乳腺炎的预测价值。结果研究组平均年龄为(34.0±7.9)岁,对照组平均年龄为(40.2±10.8)岁,两组比较差异无显著性(P>0.05)。单因素分析显示两组在超重/肥胖、初潮年龄、哺乳时间、乳头内陷、催乳素,水平升高方面差异有显著性(P<0.05)。多因素logistic回归分析结果显示,超重/肥胖、初潮年龄、哺乳时间、乳头内陷以及催乳素水平升高为非哺乳期乳腺炎的独立危险因素。建立的列线图模型预测非哺乳期乳腺炎发生风险的C-index为0.777,采用Boostrap法进行验证,校正C-index为0.760,校准曲线显示该模型的性能良好,ROC曲线显示该模型的曲线下面积为0.732。结论超重/肥胖、初潮年龄低、哺乳时间长、乳头内陷以及催乳素水平升高是发生非哺乳期乳腺炎的独立危险因素。建立的列线图模型具有较高的区分度和准确度,对甄别高风险人群及制定相应干预策略有一定指导意义。

Objective To explore the risk factors of non-puerperal mastitis,and establish a nomogram prediction model.Method A total of 116 NPM patients(case group)diagnosed by pathology in the department of breast surgery,Beijing Tiantan Hospital,Capital Medical University from September 1st 2011 to September 1st 2019 were retrospectively analyzed.116 people without breast related diseases in the physical examination center of Beijing Tiantan Hospital,Capital Medical University at the same time period were randomly selected as the control group.The risk factors of NPM were screened out by χ^(2) test.The variables with statistical significance were analyzed by multivariate logistic regression with R software,and the nomogram model was established to predict the risk of NPM.The model was validated by Boostrap method and the calibration curve was drawn.At the same time,the ROC curve was used to evaluate the prediction efficiency of nomogram model for NPM occurrence risk.Result The average age of the case group was(34.0±7.9)years old and that of the control group was(40.2±10.8)years old,there was no significant difference in age between the two groups(P=0.11).There were significant differences in overweight/obesity,menarche age,lactation time,nipple invagination and prolactin elevation between the two groups(P<0.05).Multivariate logistic regression analysis showed overweight/obesity,menarche age),lactation time,inverted nipple and prolactin were independent risk factors of NPM.The nomogram model used to predict the risk of NPM C-index=0.777,and its high prediction accuracy was further verified by bootstrap verification(corrected C-index=0.760).The calibration curve shows the good performance of the model.ROC curve showed high sensitivity and specificity(AUC=0.732).Conclusion Overweight/obesity,young menarche age,long lactation time,nipple invagination and elevated prolactin were independent risk factors for NPM.The nomogram model has the characteristics of high differentiation and accuracy,and has certain clinical application value.It has certain guiding significance for screening high-risk population and formulating corresponding intervention strategies.