颈段脊髓损伤后上肢力量恢复的个体化预测:简易预后模型的开发与验证
本研究基于欧洲多中心脊髓损伤研究(EMSCI)423例患者数据,开发并内部验证了预测颈段脊髓损伤后上肢运动评分(UEMS)恢复的个体化预后模型,预测因子包括第1周/第4周UEMS和AIS分级。MAPE=4.77,95%CI 4.28–5.28)和良好校准。仅基于AIS分级和早期UEMS的简易模型可准确预测上肢力量恢复。
要点
- 仅用AIS分级和第1/4周UEMS即可个体化预测颈段脊髓损伤后上肢力量恢复。
- 校准曲线提示对较低预测值存在轻微UEMS低估。
核心结论
- 仅基于AIS分级和早期UEMS的简易个体化预后模型可准确预测颈段脊髓损伤后上肢力量恢复。
研究设计与样本
- 开发与内部验证使用欧洲多中心脊髓损伤研究(EMSCI)423例患者数据,采用bootstrap稳定性图进行内部验证。
- 建立序数回归模型预测第4、12、24、48周UEMS,预测因子为第1周/第4周UEMS和AIS分级。
- 12周模型外部验证使用北美365例患者队列,评估总体拟合与校准。
主要发现
- 内部验证显示所有模型预测稳定、校准曲线良好且稳定、预测误差低。
- 外部验证校准良好(截距=1.10,95%CI -0.08–2.41;斜率=1.01,95%CI 0.97–1.05)。
- 校准曲线提示对较低预测值存在轻微UEMS低估。
临床怎么用
- 若进一步验证,模型可辅助早期临床决策、患者转诊与咨询以及临床试验入组筛选。
局限性
- 12周模型外部验证的校准曲线提示对较低预测值存在轻微UEMS低估。
- 作者指出模型需进一步验证后方可用于临床决策等实践。
以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。
原文
BACKGROUND: Early accurate prediction of upper limb strength recovery after cervical spinal cord injury (SCI) can optimize rehabilitation and research practices. However, no adequately validated, individualized prognostic model exists. OBJECTIVE: To develop and validate individual prognostic models for predicting Upper Extremity Motor Score (UEMS) recovery after cervical SCI. METHODS: We developed and internally validated prognostic models using data from 423 patients from the European Multicenter Study about Spinal Cord Injury (EMSCI) collaboration. Predictors included UEMS at week 1/week 4 and American Spinal Injury Association Impairment Scale (AIS) category. Ordinal regression models to predict UEMS at 4, 12, 24, and 48 weeks were developed and internally validated using bootstrap-based stability plots. External validation of the 12-week model was performed using a North American cohort of 365 patients. Performance metrics included overall fit and calibration. RESULTS: Internal validation showed stable predictions, good and stable calibration curves, and low prediction errors for all models. At external validation, the model demonstrated good overall fit (R2 = .81, 95% confidence interval [CI] [0.77 to 0.85]; MAPE = 4.77, 95% CI [4.28 to 5.28]), and good calibration (intercept = 1.10, 95% CI [-0.08 to 2.41]; slope = 1.01, 95% CI [0.97 to 1.05]). The calibration curve indicated a minor underestimation of UEMS for lower predicted values. CONCLUSIONS: Simple individualized prognostic models based solely on AIS category and early UEMS allow accurate predictions of upper limb strength recovery after cervical SCI. Our models contribute to expanding the existing literature on motor recovery patterns and prognosis/prediction after cervical SCI. If further validated, these models can inform early clinical decision-making, patient referral and counselling, and clinical trial eligibility.
引用
- Furlan L, Russo TL. Individualized Prediction of Upper Limb Strength Recovery After Cervical Spinal Cord Injury: Development and Validation of Simple Prognostic Models. Neurorehabil Neural Repair. 2026. PMID: 42153436. doi:10.1177/15459683261448463
供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。