Determinants of responders in a dose-response trial of spinal manipulation for the care of chronic low back pain

Background: The aim of this secondary analysis is to identify determinants of success of spinal manipulation (SMT) for the treatment of chronic low back pain (cLBP). Methods: We randomized 400 patients with cLBP to receive 18 sessions of lumbar SMT or a light massage control scheduled over 6-weeks; with SMT at 0, 6, 12, or 18 of the visits. Patients were followed for 52-weeks. Successful response to treatment is defined as a 50%-improvement in pain-score measured with a modified Von-Korff (MVK) 100-point pain-scale. Determinants of successful response are baseline measures of pain, disability, outside care, health status, age, gender, relative confidence in SMT/massage, any previous SMT/massage care, and time-point. Three-quarters of the data are randomly allocated into a data-set used to develop the predictive models. Models built by stepwise logistic regression are validated on the remaining data. Sensitivity and specificity for the predictive model will be reported. Results: Preliminary results of univariate models of the entire data-set show that 50%-improvement in the MVK pain-scale was predicted best by lower number of comorbidities followed by more baseline pain, younger age, more baseline disability, and better health status; with an increased count of the number of comorbidities preventing recovery at an OR per 1 comorbidity of 0.84 95% CI[0.72,0.97; p=0.02]. Prediction models developed using 75% of the data and their predictive performance on the remaining data will be presented. Conclusions: Findings from this analysis will assist with developing models for predicting which patients would especially benefit from SMT for their cLBP.