Page 72 - 南京医科大学自然版

南京医科大学学报（自然科学版） 第46卷第6期
·854 · Journal of Nanjing Medical University（Natural Sciences） 2026年6月

·临床研究·

基于性别分层构建儿童免疫性血小板减少症慢性化风险预测

模型

薛园园，张荣荣，田兆方，徐 蔚，赵继鸥，梁冬梅，袁玉芳 *

南京医科大学附属淮安第一医院儿科，江苏 淮安 223300

［摘 要］ 目的：目前关于性别对儿童免疫性血小板减少症（immune thrombocytopenia，ITP）预后的影响尚存争议，女性是慢性
化高危因素的观点未取得普遍共识。本研究旨在通过性别分层构建慢性ITP的性别特异性预测模型，为个体化预后评估和早
期临床干预提供依据。方法：回顾性纳入2019年1月1日—2023年1月1日于南京医科大学附属淮安第一医院儿科首次确诊
为ITP的224例患儿作为研究对象，其中男128例，女96例，收集人口学资料和临床资料，并规律随访至少1年。多因素Logistic
回归分析建立不同性别慢性ITP的预测模型，通过受试者工作特征（receiver operating characteristic，ROC）曲线、校准曲线和临
床决策曲线评估模型的效能。结果：性别是儿童ITP预后的独立影响因素（P < 0.05）。基于性别分层构建预测模型，男性患儿
中，中性粒细胞计数、中性粒细胞/淋巴细胞比值、治疗第 7 天血小板计数、补体 C4 及出血分级是 ITP 慢性化的独立预测因
素（P 均< 0.05），5项指标联合预测的ROC 曲线下面积为0.879（95% CI：0.819~0.938），而女性患儿中，淋巴细胞计数、补体C3、
治疗第 7 天血小板计数及免疫球蛋白 G 是 ITP 慢性化的独立预测因素（P 均 < 0.05），其联合预测的 ROC 曲线下面积为 0.902
（95% CI：0.842~0.961）。两模型的校准曲线均接近理想曲线，且临床决策曲线显示在0.10~0.70的阈概率范围内均具有良好的
临床净获益。结论：儿童ITP的慢性化预测因素存在显著性别差异，基于性别分层构建的模型具有良好的预测效能。
［关键词］ 免疫性血小板减少症；儿童；性别分层；慢性；预测因素
［中图分类号］ R725.5 ［文献标志码］ A ［文章编号］ 1007⁃4368（2026）06⁃854⁃10
doi：10.7655/NYDXBNSN251270

Development of prediction models for chronic immune thrombocytopenia risk in children
stratified by sex

*
XUE Yuanyuan，ZHANG Rongrong，TIAN Zhaofang，XU Wei，ZHAO Ji’ou，LIANG Dongmei，YUAN Yufang
Department of Pediatrics，the Affiliated Huai’an No.1 People’s Hospital of Nanjing Medical University，Huai’an
223300，China

［Abstract］ Objective：The impact of sex on the prognosis of childhood immune thrombocytopenia（ITP）remains controversial，and
the view that female sex is a high⁃risk factor for chronicity has not reached a universal consensus. This study aimed to construct sex⁃
specific prediction models for chronic ITP through sex stratification，providing a basis for individualized prognosis assessment and
early clinical intervention. Methods：This retrospective study enrolled 224 children initially diagnosed with ITP who were hospitalized
in the Department of Pediatrics at The Affiliated Huai’an No.1 People’s Hospital of Nanjing Medical University，between 1 January
2019 and 1 January 2023. The cohort included 128 males and 96 females. Demographic and clinical data were collected，with a follow⁃
up period of at least one year. Multivariate logistic regression analysis was used to develop prediction models for chronic ITP separately
for males and females. The performance of the models was evaluated using receiver operating characteristic（ROC）curves，calibration
curves，and clinical decision curve analysis. Results：This study found that sex was an independent factor influencing the prognosis of
childhood ITP（P < 0.05）. Sex⁃stratified prediction models were constructed. In the male model，absolute neutrophil count，neutrophil⁃
to⁃lymphocyte ratio，platelet count on day 7 of treatment，complement C4，and bleeding grade were independent predictors of chronicity
（P < 0.05）. The area under the ROC curve（AUC）for the combination of these five indicators was 0.879（95% CI：0.819-0.938）.

［基金项目］ 淮安市科技计划项目（HAB2025043）
通信作者（Corresponding author），E⁃mail：yyf92992@njmu.edu.cn（ORCID：0009⁃0002⁃4738⁃8551）
∗