Predictors of long-term myelopathy relief in patients with moderate to severe cervical spondylotic myelopathy following surgery: a Spine CORe™ analysis of QOD data

OBJECTIVE: Surgery for cervical spondylotic myelopathy (CSM) aims to halt disease progression. However, some patients will also ultimately note relief from their myelopathic symptoms. This study aimed to identify factors that predict improvement from moderate or severe myelopathy to mild or no myelopathy through an analysis of preoperative clinical and demographic data. Predictive models were developed to identify patients likely to achieve durable relief from myelopathy 24 months after surgery for CSM. METHODS: Data were obtained from the Quality Outcomes Database (QOD) CSM dataset, a prospective registry cohort of 1085 patients from the 14 Spine CORe™ sites. Patients were excluded if they were younger than 18 years, were missing baseline or 24-month modified Japanese Orthopaedic Association scale (mJOA) scores, or had baseline mJOA score > 14. The remaining patients were partitioned into a training (n = 467) or test (n = 117) set. Logistic regression and random forest models, with and without principal component analysis, were trained to predict whether patients achieved mild/no myelopathy at 24 months, which was defined as an mJOA score of 15 or greater. RESULTS: Overall, 584 patients (47.1% female) with moderate to severe myelopathy met the prespecified inclusion criteria, with a mean ± SD age of 61.7 ± 11.2 years, body mass index (BMI) 30.3 ± 6.3 kg/m2, and preoperative mJOA score 11.2 ± 2.4. At 24 months, 45% of patients (n = 263) had mJOA score > 14. There were no significant differences in performance between models with area under the receiver operating characteristic curve (AUROC) near 0.63 and area under the precision-recall curve (AUPRC) near 0.56. Significant positive predictors of postoperative relief from myelopathy in both the models and sensitivity analysis were symptom duration ≤ 12 months (OR 1.88, 95% CI 1.20-2.94, p < 0.01) and participation in activities outside the home (OR 2.19, 95% CI 1.12-4.27, p = 0.02). CONCLUSIONS: In a large prospective registry cohort of patients operated on for CSM, consistent factors associated with myelopathic symptom relief at 24 months were symptom duration ≤ 12 months and participation in activities outside the home. The finding that longer symptom duration was associated with a decreased propensity for relief from myelopathic symptoms suggests that earlier surgery may be beneficial for patients with moderate to severe myelopathy.