| 摘要: |
| 目的:针对儿童四肢创伤Ⅱ类切口并发症缺乏个体化风险评估标准的现状,识别相关危险因素,构建一个可用于临床的 Nomogram 风险预测模型。 方法:回顾性分析 2021 年 8 月-2024 年 12 月安徽省儿童医院骨科 240 例四肢创伤Ⅱ类切口手术患 儿的临床资料。 按 7 ∶ 3 随机分为训练集与验证集。 经 Lasso 分析、多因素 logistic 回归确定独立危险因素,构建 Nomogram 预测 模型。 使用验证集数据,通过 ROC 曲线、校准曲线和决策曲线分析对模型性能进行综合评价。 结果:术后切口并发症发生率为 16. 67%。 确立切口长度(≥5 cm)、手术时长(≥1 h)、高能量损伤机制及切口手术史是该类切口并发症的 4 个独立危险因素 (均 P<0. 05)。 基于此构建的 Nomogram 模型在训练集与验证集中的 AUC 分别为 81. 4%和 76. 6%。 校准曲线与 H-L 检验提示 模型预测概率准确,决策曲线分析显示在训练集中,净获益的风险阈值在 0% ~80%;在验证集中,该获益区间为 0% ~60%,表明 模型方案总体在与“全治疗”或“全不治疗”的方案相比仍有明显获益。 结论:本研究构建并内部验证了一个包含 4 个易获临床 因素的 Nomogram 预测模型,该模型具有良好的区分度、准确度及临床实用性,为早期识别儿童四肢创伤Ⅱ类切口术后并发症 高危患儿、实施个体化干预提供了直观的风险评估手段。 但该模型是基于一个单中心的回顾性研究建立,外推性受限,下一步 需开展更广泛的外部试验论证和更新。 |
| 关键词: Ⅱ类切口 四肢创伤 术后切口并发症 儿童 风险预测 |
| DOI：10.13407/j.cnki.jpp.1672-108X.2026.09.002 |
|
| 基金项目:安徽省科技厅临床医学研究转化专项子课题,编号 S202527c100210464。 |
|
| Construction of Nomogram Risk Prediction Model for Postoperative Complications of Type Ⅱ Incisions for Pediatric Limb Trauma Based on Lasso Regression |
| Yao Qi1 , Ruan Wenyi1 , Ding Zhen1 , Zhou Chengyue1 , Zhang Yapeng1 , Cai Heping1 , Wang Xiaoling2 |
| (1. Anhui Provincial Children’s Hospital, Hefei 230022, China; 2. Beijing Children’ s Hospital, Capital Medical University, Beijing 100045, China) |
| Abstract: |
| Objective: To identify relevant risk factors and construct Nomogram risk prediction model suitable for clinical practice according to the current lack of individualized risk assessment criteria for postoperative complications of type Ⅱ incisions in pediatric limb trauma. Methods: A retrospective analysis was conducted on clinical data of 240 children undergoing type Ⅱ incision surgeries for limb trauma in the Pediatric Orthopedics Department of Anhui Provincial Children’ s Hospital from Aug. 2021 to Dec. 2024. All children were randomly divided into the training set and validation set at a ratio of 7 ∶ 3. The independent risk factors were determined through Lasso analysis and multivariate logistic regression. A Nomogram prediction model was constructed. Validation set data was used to comprehensively evaluate the model performance through the area under the receiver operating characteristic curve, calibration curve, and decision curve analysis. Results: The incidence of postoperative incision complications was 16. 67%. A combined analysis of Lasso regression and multivariate logistic regression identified that the incision length of ⩾5 cm, surgery duration of ⩾1 h, high-energy injury mechanism, and history of incision surgery were four independent risk factors for postoperative incision complications (P<0. 05). The AUC values of the constructed Nomogram model in the training set and validation set were 81. 4% and 76. 6%, respectively. The calibration curve and Hosmer-Lemeshow test indicated that the predicted probabilities of the model were accurate. Decision curve analysis revealed that the risk threshold for net benefit in the training set ranged from 0% to 80%, while in the validation set, the benefit range was from 0% to 60%, indicating that the model scheme still showed significant benefits when compared with the “full treatment” or “no treatment” schemes. Conclusion: In this study, a Nomogram prediction model containing four obtainable clinical factors is constructed and internally validated. This model has good discrimination, accuracy, and clinical practicability, providing an intuitive risk assessment method for early identification of high-risk postoperative complications of children with type Ⅱ incisions in limb trauma and implementation of individualized interventions. However, this model is constructed based on a single-center retrospective study, which has limited generalizability. Further extensive external trials and updates are required in the next research. |
| Key words: type Ⅱ incision limb trauma postoperative incision complications children risk prediction |
