Mapping the Diffusion of Technology in Orthopaedic Surgery: Understanding the Spread of Arthroscopic Rotator Cuff Repair in the United States

The mechanism by which surgical innovation is spread in orthopaedic surgery is not well studied. The recent widespread transition from open to arthroscopic rotator cuff repair techniques provides us with the opportunity to study the spread of new technology; doing so would be important because it is unclear how novel orthopaedic techniques disseminate across time and geography, and previous studies of innovation in healthcare may not apply to the orthopaedic community. (1) How much regional variation was associated with the adoption of arthroscopic rotator cuff repair in the United States Medicare population between 2006 and 2014 and how did this change over time? (2) In which regions of the United States was arthroscopic rotator cuff repair first adopted and how did it spread geographically? (3) Which regional factors were associated with the adoption of this new technology? We divided the United States into 306 hospital referral regions based upon referral patterns observed in the Centers for Medicare & Medicaid Services MedPAR database, which records all Medicare hospital admissions; this has been done in numerous previous studies using methodology introduced by the Dartmouth Atlas. The proportion of arthroscopic rotator cuff repairs versus open rotator cuff repairs in each hospital referral region was calculated using adjusted procedural rates from the Medicare Part B Carrier File from 2006 to 2014, as it provided a nationwide sample of patients, and was used as a measure of adoption. A population-weighted, multivariable linear regression analysis was used to identify regional characteristics independently associated with adoption. There was substantial regional variation associated with the adoption of arthroscopy for rotator cuff repair as the percentage of rotator cuff repair completed arthroscopically in 2006 ranged widely among hospital referral regions with a high of 85.3% in Provo, UT, USA, and a low of 16.7% in Seattle, WA, USA (OR 30, 95% CI 17.6 to 52.2; p < 0.001). In 2006, regions in the top quartiles for Medicare spending (+9.1%; p = 0.008) independently had higher adoption rates than those in the bottom quartile, as did regions with a greater proportion of college-educated residents (+12.0%; p = 0.009). The Northwest region (-14.4%; p = 0.009) and the presence of an academic medical center (-5.8%; p = 0.026) independently had lower adoption than other regions and those without academic medical centers. In 2014, regions in the top quartile