Is Cervical Traction Effective in Chronic Nonspecific Neck Pain Patients With Unsatisfactory NSAID Control? A Nomogram to Predict Effectiveness

Abstract

In this retrospective study, our objective was to establish a nomogram to predict the effectiveness of cervical traction in young and middle-aged chronic nonspecific neck pain (CNNP) patients with unsatisfactory nonsteroidal anti-inflammatory drug (NSAID) control. For CNNP patients with unsatisfactory NSAID control, the effectiveness of cervical traction varies. Neck muscle fat infiltration and clinical features may associate with the effectiveness. A total of 186 suitable patients were classified into a training data set (from August 2015 to July 2018, n= 118) and a validation data set (from August 2018 to June 2019, n= 68) with time sequence. All patients were included to receive magnetic resonance imaging scan to calculate posterior cervical fat and muscle features, then undergoing unified cervical traction in an outpatient clinic. The least absolute shrinkage and selection operator (LASSO) regression model was used to select potentially relevant features to predict effectiveness possibility of cervical traction. Multivariable logistic regression analysis was used to develop the predicting model, presenting with a nomogram. The performance of the nomogram was assessed based on its calibration, discrimination, and clinical utility. Through the LASSO regression model, we identified 4 predictors including sex, good exercise compliance, the ratio of the cross-sectional area (CSA) between fat and muscle on C5 level (C5 fat CSA/muscle CSA), the ratio of CSA between fat and centrum on C5 level (C5 fat CSA/centrum muscle CSA). The nomogram provided good calibration and discrimination in the training cohort, showing an area under the curve (AUC) of 0.704 (95% CI, 0.608-0.799) and good concordance between the predicted and actual probabilities with Spiegelhalter's Z-test (P= 0.835). Discrimination of the model in the validation data set was acceptable, with AUC of 0.691 (95% CI, 0.564-0.817). Decision curve analysis revealed the nomogram to be clinically useful. Male sex, good exercise compliance, lower C5 fat CSA/centrum CSA, and and lower C5 fat CSA/muscle CSA could be favorable features to predict the effectiveness of cervical traction in CNNP patients with unsatisfactory NSAID control.