Trial of Inserting Prevalence Information Into Lumbar Spine Imaging Reports (LIRE)

• Relative Value Unit (RVU) for spine-related interventions [ Time Frame: 12 months ]
  Composite of diagnostic and treatment interventions related to the low back pain
  
  

• Opioid prescriptions [ Time Frame: 12 months ]
  Opioid prescriptions at 12 months
  
  
• Cross-sectional imaging [ Time Frame: 12 months ]
  CT or MR imaging of the back performed on patients up to 12 months after randomization.
  
  
• Back-pain related estimated payer costs [ Time Frame: 12 months ]
  
• Spine- related surgical interventions [ Time Frame: 12 months ]
  Interventions such as injections or lumbar surgeries
  
  
• Relative Value Unit (RVU) for spine-related interventions at 24 months [ Time Frame: 24 months ]
  
• Opioid Prescriptions at 24 months [ Time Frame: 24 months ]
  
• Cross-sectional Imaging at 24 months [ Time Frame: 24 months ]
  
• Back-pain related estimated payer costs at 24 months [ Time Frame: 24 months ]
  
• Spine- related surgical interventions [ Time Frame: 24 months ]
  

The long-term public health significance is that a simple, inexpensive intervention has the potential to substantially reduce unnecessary and expensive care for back pain patients. Importantly, this approach could be applied to a wide range of other conditions and other diagnostic tests (e.g. other imaging tests, laboratory tests, genetic testing). If this study is positive, adding epidemiologic benchmarks to diagnostic test reporting could become the dominant paradigm for communicating all diagnostic information.

To test this hypothesis, the investigators propose to conduct a pragmatic cluster, randomized controlled trial, randomly assigning primary care clinics at four large health systems to receive either standard lumbar spine imaging reports or reports containing epidemiological benchmarks for common imaging findings. The investigators will use a novel stepped wedge randomization scheme that temporally randomizes sites, allowing within-site before/after comparisons in addition to between-site comparisons, while assuring that all sites will eventually receive the intervention. The primary outcome will be a metric of back-related intervention intensity. The primary analysis will occur at the clinic level and not the patient level.

This pragmatic trial will demonstrate both the feasibility of randomly assigning clinics within large health systems as well as the feasibility of passively collecting outcomes data up to two years after enrollment using the robust electronic medical records systems available at each health system.