肝素结合蛋白联合D-二聚体对儿童重症支原体肺炎的早期预测价值及风险模型构建

The Predictive Value of Heparin-Binding Protein Combined with D-Dimer for Early Detection of Severe Mycoplasma pneumoniae in Children and the Development of a Risk Model

摘要: 目的：探讨血清肝素结合蛋白(HBP)联合D-二聚体(D-dimer)对儿童重症肺炎支原体肺炎(SMPP)的早期预测价值，并基于上述指标构建临床风险预测模型。方法：回顾性选取2021年1月至2023年12月于青岛大学附属医院儿科住院治疗的肺炎支原体肺炎(MPP)患儿的临床资料，依据病情分为重症组98例、非重症组214例。对比两组一般临床资料、实验室检验指标、肺部影像学结果；采用多因素Logistic回归筛选SMPP独立危险因素；通过受试者工作特征(ROC)曲线分析HBP、D-二聚体单独及联合的预测效能；基于独立危险因素建立Logistic回归风险模型，使用Hosmer-Lemeshow检验、Bootstrap重复抽样、决策曲线分析(DCA)完成模型校准、内部稳定性验证与临床获益评估。结果：重症组患儿发热持续时间显著长于非重症组(P < 0.001)；重症组HBP、D-二聚体、C-反应蛋白(CRP)、降钙素原(PCT)、乳酸脱氢酶(LDH)水平高于非重症组(P < 0.05)；重症组大叶性实变、胸腔积液、支气管充气征检出占比更高(P < 0.05)。多因素Logistic回归提示HBP升高、D-二聚体升高、LDH升高、发热持续时间延长是儿童SMPP的独立危险因素(P < 0.05)。ROC曲线结果显示HBP、D-二聚体单独预测SMPP的曲线下面积(AUC)分别为0.786 (95% CI: 0.736~0.830)、0.810 (95% CI: 0.762~0.852)，二者联合预测AUC可达0.873 (95% CI: 0.831~0.908)，优于单一指标(P < 0.05)；基于4项独立危险因素构建风险预测模型，模型AUC = 0.923，Hosmer-Lemeshow检验提示模型拟合校准良好(χ² = 7.124, P = 0.523)；1000次Bootstrap内部校正后AUC为0.908；DCA分析提示模型在临床决策中具有较高的净获益。结论：HBP、D-二聚体对儿童SMPP具备较好早期筛查价值，二者联合检测预测效能进一步提升；依托HBP、D-二聚体、LDH、发热持续时间建立的风险预测模型稳定性佳、临床获益可观，可用于临床早期识别重症支原体肺炎高危患儿。

Abstract: Objective: To explore the early predictive value of serum heparin-binding protein (HBP) combined with D-dimer for severe Mycoplasma pneumoniae pneumonia (SMPP) in children, and to construct a clinical risk-prediction model based on the above-mentioned indicators. Methods: The clinical data of children diagnosed with Mycoplasma pneumoniae pneumonia (MPP) hospitalized in the Department of Pediatrics, the Affiliated Hospital of Qingdao University from January 2021 to December 2023 were retrospectively enrolled. The subjects were divided into the severe group (98 cases) and the non-severe group (214 cases) according to disease severity. General clinical data, laboratory test indicators and pulmonary imaging findings were compared between the two groups. Multivariate Logistic regression analysis was adopted to screen the independent risk factors for SMPP. The receiver-operating characteristic (ROC) curve was plotted to analyse the predictive efficiency of HBP and D-dimer when used alone and in combination. A Logistic-regression-based risk-prediction model was established on the basis of independent risk factors. The Hosmer-Lemeshow test, Bootstrap resampling and decision-curve analysis (DCA) were performed for model calibration, internal stability verification and clinical-benefit evaluation. Results: The duration of fever in the severe group was significantly longer than that in the non-severe group (P < 0.001). The levels of HBP, D-dimer, C-reactive protein (CRP), procalcitonin (PCT) and lactate dehydrogenase (LDH) in the severe group were higher than those in the non-severe group (P < 0.05). The detection rates of pulmonary lobar consolidation, pleural effusion and air-bronchogram sign were markedly higher in the severe group (P < 0.05). Multivariate Logistic regression identified elevated HBP, elevated D-dimer, elevated LDH and prolonged fever duration as independent risk factors for paediatric SMPP (P < 0.05). ROC-curve analysis showed that the area under the curve (AUC) of HBP and D-dimer for predicting SMPP stood at 0.786 (95% CI: 0.736~0.830) and 0.810 (95% CI: 0.762~0.852) respectively, while the combined detection yielded an AUC of 0.873 (95% CI: 0.831~0.908), which was superior to single-indicator prediction (P < 0.05). The risk-prediction model constructed from the four independent risk factors achieved an AUC of 0.923. The Hosmer-Lemeshow test indicated favourable model calibration (χ² = 7.124, P = 0.523). The Bootstrap internal-validation with 1000 resampling cycles produced a corrected AUC of 0.908. DCA demonstrated satisfactory net clinical benefit of this model for clinical decision-making. Conclusion: HBP and D-dimer possess favourable early-screening value for paediatric SMPP, and combined detection can further improve the predictive efficiency. The risk-prediction model established by HBP, D-dimer, LDH and fever duration presents good stability and considerable clinical benefit, and can assist clinicians in the early identification of high-risk children with severe Mycoplasma pneumoniae pneumonia.