| 摘要: |
| ［摘要］ 目的 构建并验证腹部外科手术后伤口延迟愈合列线图预测模型。方法 回顾性分析2023年8月1日至2024年8月1日在江苏大学附属医院接受腹部外科手术治疗的635例患者的临床资料。按7∶3的比例将所有患者分为训练集（444例）和验证集（191例）。根据伤口愈合情况将训练集患者分为正常愈合组（愈合时间≤2周，322例）和延迟愈合组（愈合时间>2周，122例）。通过LASSO回归筛选自变量，采用多因素logistic回归分析腹部外科手术后伤口延迟愈合的影响因素，建立列线图预测模型。分别通过受试者工作特征（ROC）曲线、校准曲线、Hosmer-Lemeshow拟合优度检验和临床决策曲线分析评估模型的区分度、校准度、拟合度和临床适用性。结果 LASSO回归筛选出的8个自变量分别为血红蛋白、总蛋白、伤口面积、糖尿病史、术后2周内有无感染、术前3个月内有无化疗、吸烟史、手术类型。多因素logistic回归分析结果显示，较大的伤口面积、有糖尿病史、术后2周内有感染、术前3个月内有化疗、有吸烟史是腹部外科手术后伤口延迟愈合的独立危险因素（P<0.05），较高的血红蛋白水平、较高的总蛋白是腹部外科手术后伤口延迟愈合的独立保护因素（P<0.05）。基于7个影响因素构建腹部外科手术后伤口延迟愈合的列线图预测模型。ROC曲线分析结果显示，模型在训练集的曲线下面积（AUC）（95%CI）为0.950（0.926～0.975），灵敏度为92.60%，特异度为87.90%；模型在验证集的AUC（95%CI）为0.945（0.900～0.990），灵敏度为96.00%，特异度为87.20%，具有良好的区分度。校准曲线分析结果显示，模型预测腹部外科手术后伤口延迟愈合风险的概率与实际概率基本一致，具有良好的校准度。Hosmer-Lemeshow拟合优度检验结果（训练集： χ2=4.805，P=0.778；验证集： χ2=0.402，P=0.818）证明模型具有较好的拟合度。临床决策曲线分析结果显示模型在临床的适用性好。结论 基于血红蛋白、总蛋白、伤口面积、糖尿病史、术后2周内有无感染、术前3个月内有无化疗、吸烟史等7个影响因素构建的腹部外科手术后伤口延迟愈合的列线图预测模型具有较好的预测价值。 |
| 关键词: 腹部外科手术 伤口延迟愈合 列线图 预测模型 |
| DOI：10.3969/j.issn.1674-3806.2025.04.19 |
| 分类号:R 641 |
| 基金项目: |
|
| Construction and validation of a nomogram model for predicting delayed wound healing after abdominal surgery |
|
ZHAO Qiang, CAO Songmei, BAI Suping
|
|
Nursing Department, Affiliated Hospital of Jiangsu University, Zhenjiang 212000, China
|
| Abstract: |
| ［Abstract］ Objective To construct and validate a nomogram model for predicting delayed wound healing after abdominal surgery. Methods The clinical data of 635 patients who underwent abdominal surgery in Affiliated Hospital of Jiangsu University from August 1, 2023 to August 1, 2024 were retrospectively analyzed. All the patients were divided into a training set(444 cases) and a validation set(191 cases) at a ratio of 7∶3. According to wound healing, the patients in the training set were divided into normal healing group(healing time ≤2 weeks, 322 cases) and delayed healing group(healing time >2 weeks, 122 cases). The independent variables were screened by LASSO regression. The influencing factors of delayed wound healing after abdominal surgery were analyzed by multivariate logistic regression, and a nomogram prediction model was established. The differentiation, calibration, goodness of fit and clinical applicability of the model were evaluated by using receiver operating characteristic(ROC) curve, calibration curve, Hosmer-Lemeshow goodness-of-fit test and clinical decision curve analysis, respectively. Results Eight independent variables which were screened out by LASSO regression were hemoglobin, total protein, wound area, history of diabetes, presence or absence of infection within 2 weeks after surgery, presence or absence of chemotherapy within 3 months before surgery, history of smoking and type of surgery. The results of multivariate logistic regression analysis showed that larger wound area, history of diabetes, infection within 2 weeks after surgery, chemotherapy within 3 months before surgery, and history of smoking were independent risk factors for delayed wound healing after abdominal surgery(P<0.05). Higher hemoglobin level and higher total protein were independent protective factors for delayed wound healing after abdominal surgery(P<0.05). A nomogram model for predicting delayed wound healing after abdominal surgery was constructed based on 7 influencing factors. The results of ROC curve analysis showed that area under the curve(AUC)(95%CI) of the model in the training set was 0.950(0.926-0.975), with a sensitivity of 92.60% and a specificity of 87.90%. The results of ROC curve analysis showed that area under the curve(AUC)(95%CI) of the model in the validation set was 0.945(0.900-0.990), with a sensitivity of 96.00% and a specificity of 87.20%. The results of ROC curve analysis showed good differentiation. The calibration curve analysis showed that the probability of the model predicting the risk of delayed wound healing after abdominal surgery was basically consistent with the actual probability, which had a good calibration. The results of Hosmer-Lemeshow goodness-of-fit test(the training set: χ2=4.805, P=0.778; the validation set: χ2=0.402, P=0.818) proved that the model had satisfactory goodness of fit. The results of clinical decision curve analysis showed that the model had good applicability in clinic. Conclusion The nomogram model constructed based on 7 influencing factors including hemoglobin, total protein, wound area, history of diabetes, presence or absence of infection within 2 weeks after surgery, presence or absence of chemotherapy within 3 months before surgery and history of smoking has a relatively good value in predicting delayed wound healing after abdominal surgery. |
| Key words: Abdominal surgery Delayed wound healing Nomogram Predictive model |
