Predicting Clinical Improvement for Patients With Low Back Pain: Keeping It Simple for Patients Seeking Physical Therapy Care.

Abstract

This study sought to develop and validate an original prediction formula that estimated the probability of success for patients with low back pain (LBP) to achieve a minimal clinically important difference (MCID) on the Modified Low Back Disability Questionnaire (MDQ). Patients were 10 to 90years old in this retrospective cohort study. Data were extracted from Intermountain Healthcare's registry, Rehabilitation Outcomes Management System: 62,858 patients admitted to physical therapy from 2002 to 2013 formed the training dataset, and 15,128 patients admitted 2015 to 2016 formed the verification dataset. Predicted probability to achieve MCID was compared with the actual percentage who succeeded. Two models were developed: 6-point improvement and 30% improvement. MDQ assessed disability, and numeric pain score assessed pain intensity. Predictive models used restricted cubic splines on age, initial pain, and disability scores for non-linear effects. Sex, symptom duration, and payer type were included as indicator variables. Predicted chance of success was compared with the actual percentage of patients that succeeded. Relative change in R-squared was calculated to assess variable importance in predicting success. Odds ratios for duration of injury and payer were calculated. A positive trend was observed in both models between predicted and actual success achieved. Both "verification" models appear accurate and closely approximate the "training dataset." Baseline MDQ score was the most important factor to predict a 6-point improvement. Payer type and injury duration were important factors to predict 30% improvement. Best odds to achieve an MCID was having a workers compensation insurance payer and seeking care within 14days. The 2 models demonstrated an accurate visualization of the chance of patients achieving significant improvement compared with the usual representation of the average rate of improvement for all patients. Enhancing physical therapists' understanding of the probability of a patient achieving significant clinical improvement can enhance decision-making processes and help physical therapists manage a patient's care more effectively.