列线图构建去骨瓣减压术后患者并发硬膜下积液的风险模型

Construction of nomogram risk prediction model for subdural effusion after decompressive craniectomy

摘要:

目的 探讨行去骨瓣减压术（DC）患者术后硬膜下积液发生情况及其影响因素，并以此构建预测模型并验证。方法 回顾性分析2021年1月—2022年12月温州市中西医结合医院收治的行DC手术的117例患者，根据患者术后是否并发硬膜下积液将其分为发生组（33例）和未发生组（84例）。分析患者术后并发硬膜下积液的影响因素，并以此构建Nomogram列线图模型，预测术后硬膜下积液发生风险；用受试者工作特征（ROC）曲线下面积（AUC）分析模型对术后并发硬膜下积液的预测效能。结果 发生组年龄≥ 60岁占比、蛛网膜下腔出血占比、皮层切开占比、中线移位距离≥ 10 mm占比、伴颅内感染占比均高于未发生组（P <0.05），骨瓣边缘至中线距离≥2 cm占比低于未发生组（P <0.05）。多因素逐步Logistic回归分析结果显示，蛛网膜下腔出血［O^R=4.295（95% CI：1.953，9.443）］、中线移位距离［O^R=5.526（95% CI：2.962，10.306）］、伴颅内感染［O^R=6.435（95% CI：2.202，18.800）］是患者术后并发硬膜下积液的危险因素（P <0.05）；骨瓣边缘至中线距离［O^R=0.196（95% CI：0.066，0.578）］是患者术后并发硬膜下积液的保护因素（P <0.05）。基于上述影响因素构建的列线图预测模型经Bootstrap法内部验证显示C-index指数为0.847（95% CI：0.752，0.923），预测术后并发硬膜下积液的校正曲线趋近于理想曲线（P >0.05）。ROC曲线结果显示列线图模型预测患者术后并发硬膜下积液的敏感性为85.70%（95% CI：0.520，0.926）、特异性为86.60%（95% CI：0.651，0.930），曲线下面积为0.872（95% CI：0.783，0.960）。结论 蛛网膜下腔出血、中线移位距离、伴颅内感染是DC患者术后并发硬膜下积液的独立危险因素，骨瓣边缘至中线距离是独立保护因素，基于上述因素构建的列线图风险预测模型可较好地评估患者术后硬膜下积液发生风险。

Abstract:

Objective To investigate the incidence and influencing factors of subdural effusion in patients undergoing decompressive craniectomy (DC), to construct and validate a prediction model based on these findings, and therefore to provide clinical guidance for reducing the risk of subdural effusion following DC.Methods A total of 117 patients who underwent DC surgery in our hospital from January 2021 to December 2022 were retrospectively analyzed. According to the presence of postoperative subdural effusion, patients were divided into the occurrence group (33 cases) and the non-occurrence group (84 cases). The factors affecting the occurrence of postoperative subdural effusion were analyzed, based on which a nomogram model was constructed to predict the risk of postoperative subdural effusion. The area under the receiver operating characteristic (ROC) curve (AUC) was used to analyze the predictive efficacy of the prediction model for postoperative subdural effusion.Results The proportions of patients aged ≥ 60 years, with subarachnoid hemorrhage, undergoing cortical incision, with midline shift ≥10 mm, and with intracranial infection in the occurrence group were all higher than those in the non-occurrence group (P < 0.05), and the proportion of patients with the distance from the edge of the bone flap to the midline ≥ 2 cm in the occurrence group was lower than that in the non-occurrence group (P < 0.05). Multivariable Logistic regression analysis demonstrated that subarachnoid hemorrhage [O^R = 4.295 (95% CI: 1.953, 9.443) ], long midline shift distance [O^R = 5.526 (95% CI: 2.962, 10.306) ], and intracranial infection [O^R = 6.43 (95% CI: 2.202, 18.800) ] were risk factors for postoperative subdural effusion (P < 0.05), and that long distance from the edge of the bone flap to the midline [O^R = 0.196 (95% CI: 0.066, 0.578) ] was a protective factor (P < 0.05). The nomogram prediction model constructed based on the aforementioned influencing factors showed a C-index of 0.847 (95% CI: 0.752, 0.923) after the internal validation using the Bootstrap method, and the calibration curve for predicting postoperative subdural effusion closely approached the ideal curve (P > 0.05). The results of ROC curve indicated that the sensitivity and specificity of the nomogram model for predicting postoperative subdural effusion were 85.70% (95% CI: 0.520, 0.926) and 86.60% (95% CI: 0.651, 0.930), respectively, with the AUC being 0.872 (95% CI: 0.783, 0.960) (P < 0.05).Conclusions Subarachnoid hemorrhage, long midline shift distance, and intracranial infection are independent risk factors for postoperative subdural effusion in patients undergoing DC, and long distance from the edge of the bone flap to the midline is an independent protective factor. The nomogram risk prediction model based on the above factors can better evaluate the risk of postoperative subdural effusion in these patients.