Department of Sports Medicine and Exercise Physiology, Institute of Occupational, Social and Environmental Medicine, Goethe University Frankfurt, Frankfurt am Main, Germany.
Department of Rehabilitation Medicine, Hannover Medical School, Hannover, Germany.
Pain. 2024 Apr 1;165(4):772-784. doi: 10.1097/j.pain.0000000000003087. Epub 2023 Oct 18.
To individually prescribe rehabilitation contents, it is of importance to know and quantify factors for rehabilitation success and the risk for a future healthcare use. The objective of our multivariable prediction model was to determine factors of rehabilitation success and the risk for a future healthcare use in patients with high-grade, chronic low back pain. We included members of the German pension fund who participated from 2012 to 2019 in multimodal medical rehabilitation with physical and psychological treatment strategies because of low back pain (ICD10:M54.5). Candidate prognostic factors for rehabilitation success and for a future healthcare use were identified using Gradient Boosting Machines and Random Forest algorithms in the R-package caret on a 70% training and a 30% test set. We analysed data from 154,167 patients; 8015 with a second medical rehabilitation measure and 5161 who retired because of low back pain within the study period. The root-mean-square errors ranged between 494 (recurrent rehabilitation) and 523 (retirement) days ( R2 = 0.183-0.229), whereas the prediction accuracy ranged between 81.9% for the prediction of the rehabilitation outcome, and 94.8% for the future healthcare use prediction model. Many modifiable prognostic factors (such as duration of the rehabilitation [inverted u-shaped], type of the rehabilitation, and aftercare measure), nonmodifiable prognostic factors (such as sex and age), and disease-specific factors (such as sick leave days before the rehabilitation [linear positive] together with the pain grades) for rehabilitation success were identified. Inpatient medical rehabilitation programmes (3 weeks) may be more effective in preventing a second rehabilitation measure and/or early retirement because of low back pain compared with outpatient rehabilitation programs. Subsequent implementation of additional exercise programmes, cognitive behavioural aftercare treatment, and following scheduled aftercare are likely to be beneficial.
为了进行个性化的康复内容制定，了解和量化康复成功的因素以及未来医疗保健使用的风险非常重要。我们的多变量预测模型的目的是确定患有高级、慢性下腰痛患者的康复成功因素和未来医疗保健使用的风险。我们纳入了德国养老金基金的成员，这些成员在 2012 年至 2019 年期间因腰痛（ICD10：M54.5）接受了多模式医学康复治疗，包括身体和心理治疗策略。使用 R 包 caret 中的梯度提升机和随机森林算法，在 70%的训练集和 30%的测试集上确定康复成功和未来医疗保健使用的候选预后因素。我们分析了 154167 名患者的数据；8015 名患者接受了第二次医疗康复措施，5161 名患者在研究期间因腰痛退休。均方根误差范围在 494 天（再次康复）至 523 天（退休）之间（R2=0.183-0.229），而预测准确性范围在 81.9%（康复结果预测）至 94.8%（未来医疗保健使用预测模型）之间。确定了许多可修改的预后因素（如康复时间[倒 U 形]、康复类型和康复后护理措施）、不可修改的预后因素（如性别和年龄）和疾病特异性因素（如康复前的病假天数[线性正]与疼痛等级一起）与康复成功相关。与门诊康复项目相比，住院医疗康复项目（3 周）可能更有效地预防第二次康复措施和/或因腰痛提前退休。随后实施额外的运动项目、认知行为康复后护理治疗和定期康复后护理可能是有益的。
