Werneke Mark W, Edmond Susan, Deutscher Daniel, Ward Jason, Grigsby David, Young Michelle, McGill Troy, McClenahan Brian, Weinberg Jon, Davidow Amy L
J Orthop Sports Phys Ther. 2016 Sep;46(9):726-41. doi: 10.2519/jospt.2016.6266. Epub 2016 Jul 31.
Study Design Retrospective cohort. Background Patient-classification subgroupings may be important prognostic factors explaining outcomes. Objectives To determine effects of adding classification variables (McKenzie syndrome and pain patterns, including centralization and directional preference; Symptom Checklist Back Pain Prediction Model [SCL BPPM]; and the Fear-Avoidance Beliefs Questionnaire subscales of work and physical activity) to a baseline risk-adjusted model predicting functional status (FS) outcomes. Methods Consecutive patients completed a battery of questionnaires that gathered information on 11 risk-adjustment variables. Physical therapists trained in Mechanical Diagnosis and Therapy methods classified each patient by McKenzie syndromes and pain pattern. Functional status was assessed at discharge by patient-reported outcomes. Only patients with complete data were included. Risk of selection bias was assessed. Prediction of discharge FS was assessed using linear stepwise regression models, allowing 13 variables to enter the model. Significant variables were retained in subsequent models. Model power (R(2)) and beta coefficients for model variables were estimated. Results Two thousand sixty-six patients with lumbar impairments were evaluated. Of those, 994 (48%), 10 (<1%), and 601 (29%) were excluded due to incomplete psychosocial data, McKenzie classification data, and missing FS at discharge, respectively. The final sample for analyses was 723 (35%). Overall R(2) for the baseline prediction FS model was 0.40. Adding classification variables to the baseline model did not result in significant increases in R(2). McKenzie syndrome or pain pattern explained 2.8% and 3.0% of the variance, respectively. When pain pattern and SCL BPPM were added simultaneously, overall model R(2) increased to 0.44. Although none of these increases in R(2) were significant, some classification variables were stronger predictors compared with some other variables included in the baseline model. Conclusion The small added prognostic capabilities identified when combining McKenzie or pain-pattern classifications with the SCL BPPM classification did not significantly improve prediction of FS outcomes in this study. Additional research is warranted to investigate the importance of classification variables compared with those used in the baseline model to maximize predictive power. Level of Evidence Prognosis, level 4. J Orthop Sports Phys Ther 2016;46(9):726-741. Epub 31 Jul 2016. doi:10.2519/jospt.2016.6266.
回顾性队列研究。背景：患者分类亚组可能是解释预后结果的重要预后因素。目的：确定在预测功能状态（FS）结果的基线风险调整模型中加入分类变量（麦肯齐综合征和疼痛模式，包括向心性和方向偏好；症状清单背痛预测模型[SCL BPPM]；以及工作和体育活动的恐惧回避信念问卷子量表）的效果。方法：连续的患者完成了一系列问卷调查，收集了11个风险调整变量的信息。接受机械诊断和治疗方法培训的物理治疗师根据麦肯齐综合征和疼痛模式对每位患者进行分类。出院时通过患者报告的结果评估功能状态。仅纳入数据完整的患者。评估选择偏倚的风险。使用线性逐步回归模型评估出院FS的预测，允许13个变量进入模型。显著变量保留在后续模型中。估计模型的功效（R²）和模型变量的β系数。结果：对2660例腰椎损伤患者进行了评估。其中，分别有994例（48%）、10例（<1%）和601例（29%）因社会心理数据不完整、麦肯齐分类数据缺失和出院时FS缺失而被排除。最终用于分析的样本为723例（35%）。基线预测FS模型的总体R²为0.40。在基线模型中加入分类变量并未导致R²显著增加。麦肯齐综合征或疼痛模式分别解释了2.8%和3.0%的方差。当同时加入疼痛模式和SCL BPPM时，总体模型R²增加到0.44。尽管这些R²的增加均不显著，但与基线模型中包含的其他一些变量相比，一些分类变量是更强的预测因子。结论：在本研究中，将麦肯齐或疼痛模式分类与SCL BPPM分类相结合时所确定的额外预后能力较小，并未显著改善FS结果的预测。有必要进行更多研究，以调查分类变量与基线模型中使用的变量相比的重要性，以最大限度地提高预测能力。证据水平：预后，4级。《矫形与运动物理治疗杂志》2016年；46(9):726 - 741。2016年7月31日在线发表。doi:10.2519/jospt.2016.6266 。
