Methods: This retrospective cohort study involved a secondary analysis of data from adult patients with LBP treated in outpatient physical therapy clinics from 2014 to 2016. Patients with complete outcomes data at admission and discharge were included to develop the risk-adjustment model. Clinics with complete outcomes data for at least 50% of patients and at least 10 complete episodes of care per clinician per year were included for ranking assessment. The R² shrinkage and predictive ratio were used to assess overfitting. Agreement between unadjusted and adjusted rankings was assessed with percentile ranking by deciles or 3 distinct quality ranks based on uncertainty assessment.

Results: The primary sample included 414 125 patients (mean ± SD age, 57 ± 17 years; 60% women) treated by 12 569 clinicians from 3048 clinics from all US states; 82% of patients from 2107 clinics were included in the ranking assessment. The R² shrinkage was less than 1%, with a predictive ratio of 1. Risk adjustment impacted ranking for 70% or 31% of clinics, based on deciles or 3 distinct quality levels, respectively.