Abstract：
OBJECTIVE: Tubular microdiskectomy (tMD) is one of the most commonly used for treating lumbar disk herniation. However, there still patients still complain of persistent postoperative residual low back pain (rLBP) postoperatively. This study attempts to develop a nomogram to predict the risk of rLBP after tMD.
METHODS: The patients were divided into non-rLBP (LBP VAS score < 2) and rLBP (LBP VAS score ≥ 2) group. The correlation between rLBP and these factors were analyzed by multivariate logistic analysis. Then, a nomogram prediction model of rLBP was developed based on the risk factors screened by multivariate analysis. The samples in the model are randomly divided into training and validation sets in a 7:3 ratio. The Receiver operating characteristic (ROC) curve, calibration curve, and decision curve analysis (DCA) were used to evaluate the diskrimination, calibration and clinical value of the model, respectively.
RESULTS: A total of 14.3% (47/329) of patients have persistent rLBP. The multivariate analysis suggests that higher preoperative LBP visual analog scale (VAS) score, lower facet orientation (FO), grade 2-3 facet joint degeneration (FJD) and moderate-severe multifidus fat atrophy (MFA) are risk factors for postoperative rLBP. In the training and validation sets, the ROC curves, calibration curves, and DCAs suggested the good diskrimination, predictive accuracy between the predicted probability and actual probability, and clinical value of the model, respectively.
CONCLUSIONS: This nomogram including preoperative LBP VAS score, FO, FJD and MFA can serve a promising prediction model, which will provide a reference for clinicians to predict the rLBP after tMD.
METHODS: The patients were divided into non-rLBP (LBP VAS score < 2) and rLBP (LBP VAS score ≥ 2) group. The correlation between rLBP and these factors were analyzed by multivariate logistic analysis. Then, a nomogram prediction model of rLBP was developed based on the risk factors screened by multivariate analysis. The samples in the model are randomly divided into training and validation sets in a 7:3 ratio. The Receiver operating characteristic (ROC) curve, calibration curve, and decision curve analysis (DCA) were used to evaluate the diskrimination, calibration and clinical value of the model, respectively.
RESULTS: A total of 14.3% (47/329) of patients have persistent rLBP. The multivariate analysis suggests that higher preoperative LBP visual analog scale (VAS) score, lower facet orientation (FO), grade 2-3 facet joint degeneration (FJD) and moderate-severe multifidus fat atrophy (MFA) are risk factors for postoperative rLBP. In the training and validation sets, the ROC curves, calibration curves, and DCAs suggested the good diskrimination, predictive accuracy between the predicted probability and actual probability, and clinical value of the model, respectively.
CONCLUSIONS: This nomogram including preoperative LBP VAS score, FO, FJD and MFA can serve a promising prediction model, which will provide a reference for clinicians to predict the rLBP after tMD.
摘要：
目的：肾小管微盘切除术(tMD)是治疗腰椎间盘突出症最常用的方法之一。然而,仍有患者抱怨持续的术后残余下腰痛（rLBP）。这项研究试图开发一个列线图来预测tMD后rLBP的风险。
方法：将患者分为非rLBP(LBPVAS评分<2)组和rLBP(LBPVAS评分≥2)组。采用多因素logistic回归分析rLBP与这些因素的相关性。然后，基于多变量分析筛选的危险因素,建立了rLBP的列线图预测模型.模型中的样本以7:3的比例随机分为训练集和验证集。接收机工作特性(ROC)曲线,校正曲线,和决策曲线分析(DCA)用于评估结果，模型的校准和临床价值，分别。
结果：总共14.3%(47/329)的患者有持续性rLBP。多因素分析提示术前LBP视觉模拟量表(VAS)评分较高,下刻面取向(FO)，2-3级小关节变性(FJD)和中重度多裂脂肪萎缩(MFA)是术后rLBP的危险因素.在训练集和验证集中，ROC曲线，校正曲线,和DCA建议了良好的数据处理，预测概率和实际概率之间的预测准确性，以及该模型的临床价值，分别。
结论：此列线图包括术前LBPVAS评分，FO,FJD和MFA可以服务于一个有前途的预测模型，这将为临床医生预测tMD后的rLBP提供参考。
方法：将患者分为非rLBP(LBPVAS评分<2)组和rLBP(LBPVAS评分≥2)组。采用多因素logistic回归分析rLBP与这些因素的相关性。然后，基于多变量分析筛选的危险因素,建立了rLBP的列线图预测模型.模型中的样本以7:3的比例随机分为训练集和验证集。接收机工作特性(ROC)曲线,校正曲线,和决策曲线分析(DCA)用于评估结果，模型的校准和临床价值，分别。
结果：总共14.3%(47/329)的患者有持续性rLBP。多因素分析提示术前LBP视觉模拟量表(VAS)评分较高,下刻面取向(FO)，2-3级小关节变性(FJD)和中重度多裂脂肪萎缩(MFA)是术后rLBP的危险因素.在训练集和验证集中，ROC曲线，校正曲线,和DCA建议了良好的数据处理，预测概率和实际概率之间的预测准确性，以及该模型的临床价值，分别。
结论：此列线图包括术前LBPVAS评分，FO,FJD和MFA可以服务于一个有前途的预测模型，这将为临床医生预测tMD后的rLBP提供参考。
