Inappropriate MRI utilization for low back pain in Texas: Variation across the state and predictive factors

Objectives: Describe the variation in utilization of Magnetic Resonance Imaging (MRI) for low back pain throughout Texas and associated multilevel factors (patient, provider and regional). Study Design: A retrospective cross-sectional analysis of Blue Cross Blue Shield (BCBS) of Texas claims data was carried out, using the Medicare's Hospital Outpatient Quality Report (HOQR) measure OP-8 to evaluate MRIs. This estimates the proportion of MRIs for low back pain (LBP) performed without prior conservative therapy. Weighted means of OP-8 were estimated at the Hospital Referral Region (HRR) and state level. A multilevel multivariate logistic model was used to assess the odds ratio (OR) of being a provider with high OP-8 measures adjusting for area level resources. A second multilevel multivariate logistic regression on whether an individual MRI at the patient level was inappropriate. Population Studied: Adults between 18-64 years of age, enrolled in PPO/PPO+ plans and receiving care in a Texas Hospital referral Region (HRR). Results: There were 50,934 MRIs performed between 2008 and 2010 by 897 providers. Only 324 providers performed more than 42 MRIs (minimum number necessary for precession). Texas weighted mean of inappropriate MRIs was 37.89%, with a 16% variation across HRRs. Harlingen was the HRR with highest OP-8 value (47.54%) and Austin the lowest (33.80%). Hospital providers were more likely to have high OP-8 scores (OR, 1.34; CI, 1.12-1.60). Providers with high volume of imaging tests had smaller odds of high OP-8 values (OR, 0.58; CI, 0.48-0.70). The multilevel model on inappropriate MRI at the patient level found that men were more likely to receive an unnecessary MRI (OR, 1.21; CI, 1.17-1.26) while older patients were less likely than younger ones (OR, 0.92; CI, 0.86-0.99). The pattern of care received six months prior to the imaging test was also significantly associated, with those receiving care by a mix of specialist and primary care practitioners (PCP) or only specialists less likely to have an inappropriate MRI than those seen only by PCPs (Specialist only: OR,0.7; CI, 0.66-0.74; Mix Specialist-PCP OR, 0.32; CI, 0.30-0.34). The regional characteristics were not significant in the adjusted model. Conclusions: The proportion of inappropriate MRIs for LBP performed in Texas was high, particularly in certain regions but variation across the state was small. Odds of performing high number of inappropriate MRIs could be linked to certain provider and patient characteristics.