What are minimal clinically important differences for patient-reported outcome measures after surgery for cervical spondylotic myelopathy? A 5-year Spine CORe™ analysis of QOD data.

Abstract

The minimal clinically important difference (MCID) is widely used to interpret patient-reported outcome measures (PROMs) in cervical spondylotic myelopathy (CSM). However, consensus on its definition is lacking, and its long-term consistency remains unknown. The objective of this study was to determine if MCID thresholds for PROMs at 5 years after surgery for CSM remained consistent when compared to previously established 2-year values. The Spine CORe™ study group performed a post hoc analysis of the prospective Quality Outcomes Database. Eight established anchor- and distribution-based methods were applied to define MCID thresholds for the following PROMs: Neck Disability Index (NDI), neck pain numeric rating scale (NP-NRS), arm pain numeric rating scale (AP-NRS), 5-dimension EuroQol health utility questionnaire (EQ-5D) for quality-adjusted life years, and modified Japanese Orthopaedic Association (mJOA) scores. Predictive validity was evaluated using area under the curve (AUC) analysis with North American Spine Society satisfaction as the anchor, and results were compared with calculated 2-year values from the same cohort using DeLong's test. A total of 1085 patients were originally enrolled, with ≥ 80% follow-up for all PROMs except the mJOA score (79%). At 5 years, optimal percentage change and ≥ 30% improvement methods were consistently highest performing for the NDI (AUC 0.71 and 0.68, respectively), NP-NRS (AUC 0.65 for both), and AP-NRS (AUC 0.73 and 0.72, respectively) scores. For the EQ-5D score, both the optimal numeric cutoff and one-half standard deviation methods performed best, yielding a consistent MCID threshold of 0.11 (AUC 0.64 for both). For the mJOA score, the severity-adjusted method provided the strongest discrimination, with an AUC of 0.74 at 5 years. MCID thresholds were consistent between 2 and 5 years, except for the severity-adjusted MCID for the mJOA score (0.74 at 5 years vs 0.65 at 2 years, p = 0.026). The 30% improvement threshold corresponded to absolute changes of 11.3 points for the NDI score, 1.5 points for the NP-NRS score, and 1.4 points for the AP-NRS score based on mean baseline scores. To the authors' knowledge, this represents the largest cohort of patients with CSM in the United States with validated, long-term 5-year MCID thresholds. This study establishes practical MCID definitions for NDI (≥ 30% improvement threshold of 11.3 points), NP-NRS and AP-NRS (≥ 30% improvement thresholds of 1.5 and 1.4 points, respectively), EQ-5D (optimal numeric cutoff of 0.11), and mJOA (severity-adjusted: ≥ 3 points for severe, ≥ 2 for moderate, ≥ 1 for mild) scores that can serve as benchmarks for evaluating improvement after CSM surgery in both research and routine clinical practice.