Introduction: It is often difficult to predict clinical outcomes in patients with degenerative cervical myelopathy due to the heterogeneity of this population. The objective of this study is to develop a clinical prediction rule relating the best combination of clinical and imaging variables to surgical outcome, based on international data from two multi-center prospective studies.
Methods: Two hundred and seventy eight patients diagnosed with CSM were enrolled in the CSM-North American study at twelve different sites. An additional 493 patients participated in the CSM-International study from sixteen global sites in four continents. Univariate analyses were performed to evaluate the relationship between outcome, assessed by a dichotomized modified Japanese Orthopaedic Association (mJOA) score, and various clinical and imaging predictors. A set of important variables for the final model was selected based on author consensus, literature support and statistical findings. Logistic regression was used to formulate the final prediction model.
Results: Univariate analyses demonstrated that the odds of a successful outcome decreased with the presence of certain symptoms, including clumsy hands (OR=1.48, p=0.045), impaired gait (OR=4.10, p<0.001) and limb weakness (OR=2.19, p=0.0007); the presence of certain signs, including corticospinal distribution motor deficits (OR=2.26, p<0.0001), upgoing plantar responses (OR=1.61, p=0.0041) and lower limb spasticity (OR=1.79, p=0.0004); smoking (OR=0.67, p=0.028); the presence of cardiovascular co-morbidities (OR=0.45, p<0.0001); a lower baseline mJOA (OR=1.37, p<0.0001); and older age (OR=0.96, p<0.0001). The final clinical prediction rule included age (OR=0.97, p=0.0045), duration of symptoms (OR=0.88, p=0.070), smoking status (OR=0.66, p=0.093), impairment of gait (OR=2.48, p=0.0011) cardiovascular co-morbidities (OR=0.62, p=0.030) and baseline severity score (OR=1.26, p<0.0001). The area under the receiver operator curve was 0.77, indicating good model prediction.
Conclusions: We have identified a list of the most important imaging and clinical predictors of surgical outcome. This will allow surgeons to quantify a patient’s likely outcome and appropriately manage their expectations.
Patient Care: Predicting a patient's likely surgical outcome will enable a surgeon to appropriately manage expectations.
Learning Objectives: To determine the most important imaging and clinical predictors of outcome and formulate a clinical prediction rule based on these findings.