Mycoplasma pneumoniae pneumonia (MPP) is a common respiratory disease in children, which can easily develop into refractory Mycoplasma pneumoniae pneumonia (RMPP). Children with RMPP have complex conditions, are difficult to treat, and are associated with multiple serious complications. A risk prediction model for RMPP in children can help healthcare professionals to identify children at high risk of RMPP and provide timely therapeutic countermeasures.

To systematically review the risk prediction models for RMPP in children.

CNKI, Wanfang Data, VIP, CBM, PubMed, Embase, Web of Science and Cochrane Library were searched to collect the related studies on risk prediction models for RMPP in Children to February 20th, 2024. Two reviewers independently screened the literature, extracted data and evaluated the quality of the included studies. Meta-analysis of the incidence and predictors of RMPP in children was performed using Stata 18.0 software.

A total of 14 papers were included, containing 17 risk prediction models. Ten studies were internal validated, and only four studies were external validated, all of which had AUC>0.7. PROBAST results showed that 14 included papers were all at high risk of bias. The results of Meta-analysis showed that the incidence of RMPP in children was 28.2% (95%CI=21.2%-35.1%), hyperthermia, lactate dehydrogenase, C-reactive protein, age, neutrophil ratio, procalcitonin and D-dimer were independent influences on the occurrence of RMPP in children (P<0.05) .

Risk prediction models for RMPP in children have good predictive performance, but the overall risk of bias was high and external validation was lacking. The future risk prediction model should focus on hyperthermia, lactate dehydrogenase, C-reactive protein, age, neutrophil ratio, procalcitonin and D-dimer.

肺炎支原体肺炎（MPP）是儿童常见的呼吸道疾病，容易发展为难治性肺炎支原体肺炎（RMPP）。RMPP患儿病情复杂、治疗难度大且伴有多种严重并发症。儿童RMPP风险预测模型有助于医务人员早期识别RMPP高危患儿并及时提供治疗对策。

系统评价儿童RMPP风险预测模型的研究。

在中国知网、万方数据知识服务平台、维普网、中国生物医学文献数据库、PubMed、Embase、Web of Science和Cochrane Library数据库中检索2024-02-20以前发表的有关儿童RMPP风险预测模型的研究。由2位研究员独立筛选文献、提取数据并评价纳入文献的质量。使用Stata 18.0软件对儿童RMPP发生率和预测因子进行Meta分析。

共纳入14篇文献，包括17个风险预测模型。有10篇文献进行了内部验证，仅4篇文献进行了外部验证，其中风险预测模型的受试者工作特征曲线下面积（AUC）均>0.7。偏倚风险评估工具（PROBAST）评估结果显示，纳入的14篇文献均为高偏倚风险。Meta分析结果显示，儿童RMPP发生率为28.2%（95%CI=21.2%~35.1%），高热、乳酸脱氢酶、C反应蛋白、年龄、中性粒细胞比例、降钙素原和D-二聚体是儿童RMPP发生的独立影响因素（P<0.05）。

儿童RMPP风险预测模型表现出良好的预测性能，但整体偏倚风险较高，且缺乏外部验证，未来风险预测模型研究可重点关注高热、乳酸脱氢酶、C反应蛋白、年龄、中性粒细胞比例、降钙素原和D-二聚体等预测因子。