Abstract：
Traumatic subdural effusion (TSDE) may increase progressively or evolve into chronic subdural hematoma. These events, defined as deterioration of the effusion, often require close observation or even surgical treatment. The aim of our study was to develop and validate a nomogram for predicting the possibility of an effusion deteriorating in patients with TSDE based on the available clinical characteristics.
Clinical data from 78 patients with TSDE were retrospectively analyzed. All patients were admitted from January 2019 to May 2022. Logistic regression was applied to the data to screen for independent predictors of effusion deterioration within six months; then, a predictive nomogram model was established in R language. The consistency, predictive accuracy and clinical utility of the model were evaluated with the C-index, calibration plots, ROC curves and decision curve analysis (DCA). Furthermore, we performed internal validation using a bootstrap approach to assess the effectiveness of the model.
Time of effusion after trauma, maximum thickness of the effusion, CT value of the effusion as well as the use of atorvastatin were identified as predictors in the nomogram. The predictive model was well calibrated and demonstrated good discrimination (C-index: 0.893). The AUC of the model was 0.893 (95% CI: 0.824-0.962), and the modified C-index (0.865) indicated excellent performance in the internal validation. In addition, DCA revealed that the nomogram had clinical value.
This predictive model can effectively assess the risk of effusion deterioration in TSDE patients within six months and identify high-risk patients early.
Clinical data from 78 patients with TSDE were retrospectively analyzed. All patients were admitted from January 2019 to May 2022. Logistic regression was applied to the data to screen for independent predictors of effusion deterioration within six months; then, a predictive nomogram model was established in R language. The consistency, predictive accuracy and clinical utility of the model were evaluated with the C-index, calibration plots, ROC curves and decision curve analysis (DCA). Furthermore, we performed internal validation using a bootstrap approach to assess the effectiveness of the model.
Time of effusion after trauma, maximum thickness of the effusion, CT value of the effusion as well as the use of atorvastatin were identified as predictors in the nomogram. The predictive model was well calibrated and demonstrated good discrimination (C-index: 0.893). The AUC of the model was 0.893 (95% CI: 0.824-0.962), and the modified C-index (0.865) indicated excellent performance in the internal validation. In addition, DCA revealed that the nomogram had clinical value.
This predictive model can effectively assess the risk of effusion deterioration in TSDE patients within six months and identify high-risk patients early.
摘要：
背景：外伤性硬膜下积液（TSDE）可能逐渐增加或演变为慢性硬膜下血肿。这些事件,定义为积液恶化，通常需要密切观察甚至手术治疗。我们研究的目的是根据可用的临床特征开发和验证用于预测TSDE患者积液恶化的可能性的列线图。
方法：对78例TSDE患者的临床资料进行回顾性分析。所有患者均于2019年1月至2022年5月入院。对数据进行Logistic回归,筛选出6个月内积液恶化的独立预测因子;用R语言建立了预测列线图模型。一致性,用C指数评估模型的预测准确性和临床实用性，校准图,ROC曲线和决策曲线分析(DCA)。此外，我们使用Bootstrap方法进行了内部验证,以评估模型的有效性.
结果：创伤后积液时间，积液的最大厚度，在列线图中，积液的CT值以及阿托伐他汀的使用被确定为预测因子。预测模型被很好地校准并且表现出良好的辨别(C指数:0.893)。模型的AUC为0.893（95％CI：0.824-0.962），修改后的C指数(0.865)在内部验证中表现优异。此外,DCA显示列线图具有临床价值。
结论：该预测模型可以有效评估TSDE患者6个月内积液恶化的风险，并早期识别高危患者。
方法：对78例TSDE患者的临床资料进行回顾性分析。所有患者均于2019年1月至2022年5月入院。对数据进行Logistic回归,筛选出6个月内积液恶化的独立预测因子;用R语言建立了预测列线图模型。一致性,用C指数评估模型的预测准确性和临床实用性，校准图,ROC曲线和决策曲线分析(DCA)。此外，我们使用Bootstrap方法进行了内部验证,以评估模型的有效性.
结果：创伤后积液时间，积液的最大厚度，在列线图中，积液的CT值以及阿托伐他汀的使用被确定为预测因子。预测模型被很好地校准并且表现出良好的辨别(C指数:0.893)。模型的AUC为0.893（95％CI：0.824-0.962），修改后的C指数(0.865)在内部验证中表现优异。此外,DCA显示列线图具有临床价值。
结论：该预测模型可以有效评估TSDE患者6个月内积液恶化的风险，并早期识别高危患者。
