Scheer Justin K, Osorio Joseph A, Smith Justin S, Schwab Frank, Hart Robert A, Hostin Richard, Lafage Virginie, Jain Amit, Burton Douglas C, Bess Shay, Ailon Tamir, Protopsaltis Themistocles S, Klineberg Eric O, Shaffrey Christopher I, Ames Christopher P
Department of Neurosurgery, University of Illinois at Chicago, 912 South Wood St., Chicago, IL 60612, USA.
Department of Neurological Surgery, University of California, San Francisco, San Francisco, CA 94131, USA.
Spine Deform. 2018 Sep-Oct;6(5):593-599. doi: 10.1016/j.jspd.2018.02.010.
Retrospective review of prospective multicenter adult spinal deformity (ASD) database.
To create a model based on baseline demographic, radiographic, health-related quality of life (HRQOL), and surgical factors that can predict patients meeting the Oswestry Disability Index (ODI) minimal clinically important difference (MCID) at the two-year postoperative follow-up.
Surgical correction of ASD can result in significant improvement in disability as measured by ODI, with the goal of reaching at least one MCID. However, a predictive model for reaching MCID following ASD correction does not exist.
ASD patients ≥18 years and baseline ODI ≥ 30 were included. Initial training of the model comprised forty-three variables including demographic data, comorbidities, modifiable surgical variables, baseline HRQOL, and coronal/sagittal radiographic parameters. Patients were grouped by whether or not they reached at least one ODI MCID at two-year follow-up. Decision trees were constructed using the C5.0 algorithm with five different bootstrapped models. Internal validation was accomplished via a 70:30 data split for training and testing each model, respectively. Final predictions from the models were chosen by voting with random selection for tied votes. Overall accuracy, and the area under a receiver operating characteristic curve (AUC) were calculated.
198 patients were included (MCID: 109, No-MCID: 89). Overall model accuracy was 86.0%, with an AUC of 0.94. The top 11 predictors of reaching MCID were gender, Scoliosis Research Society (SRS) activity subscore, back pain, sagittal vertical axis (SVA), pelvic incidence-lumbar lordosis mismatch (PI-LL), primary version revision, T1 spinopelvic inclination angle (T1SPI), American Society of Anesthesiologists (ASA) grade, T1 pelvic angle (T1PA), SRS pain, SRS total.
A successful model was built predicting ODI MCID. Most important predictors were not modifiable surgical parameters, indicating that baseline clinical and radiographic status is a critical factor for reaching ODI MCID.
Level II.
对前瞻性多中心成人脊柱畸形（ASD）数据库进行回顾性分析。
基于基线人口统计学、影像学、健康相关生活质量（HRQOL）和手术因素创建一个模型，以预测患者在术后两年随访时达到奥斯威斯利残疾指数（ODI）最小临床重要差异（MCID）的情况。
ASD的手术矫正可使ODI所衡量的残疾状况得到显著改善，目标是至少达到一个MCID。然而，目前不存在用于预测ASD矫正后达到MCID的模型。
纳入年龄≥18岁且基线ODI≥30的ASD患者。模型的初始训练包括43个变量，涵盖人口统计学数据、合并症、可改变的手术变量、基线HRQOL以及冠状面/矢状面影像学参数。根据患者在两年随访时是否达到至少一个ODI MCID进行分组。使用C5.0算法构建决策树，并采用五种不同的自举模型。通过70:30的数据划分分别对每个模型进行训练和测试来完成内部验证。模型的最终预测通过随机选择对平局投票进行投票来确定。计算总体准确率以及受试者工作特征曲线（AUC）下的面积。
共纳入198例患者（达到MCID者109例，未达到MCID者89例）。总体模型准确率为86.0%，AUC为0.94。达到MCID的前11个预测因素为性别、脊柱侧凸研究学会（SRS）活动亚评分、背痛、矢状垂直轴（SVA）、骨盆入射角-腰椎前凸不匹配（PI-LL）、初次翻修、T1脊柱骨盆倾斜角（T1SPI）、美国麻醉医师协会（ASA）分级、T1骨盆角（T1PA）、SRS疼痛、SRS总分。
构建了一个成功预测ODI MCID的模型。最重要的预测因素并非可改变的手术参数，这表明基线临床和影像学状态是达到ODI MCID的关键因素。
二级。
