Variations in 30-day readmissions and length of stay among spine surgeons: a national study of elective spine surgery among US Medicare beneficiaries

OBJECTIVE Pay-for-performance programs are targeting hospital readmissions. These programs have an underlying assumption that readmissions are due to provider practice patterns that can be modified by a reduction in reimbursement. However, there are limited data to support the role of providers in influencing readmissions. To study this, the authors examined variations in readmission rates by spine surgeon within 30 days among Medicare beneficiaries undergoing elective lumbar spine surgery for degenerative conditions. METHODS The authors applied validated ICD-9-CM algorithms to 2003-2007 Medicare data to select beneficiaries undergoing elective inpatient lumbar spine surgery for degenerative conditions. Mixed models, adjusting for patient demographics, comorbidities, and surgery type, were used to estimate risk of 30-day readmission by the surgeon. Length of stay (LOS) was also studied using these same models. RESULTS A total of 39,884 beneficiaries were operated on by 3987 spine surgeons. The mean readmission rate was 7.2%. The mean LOS was 3.1 days. After adjusting for patient characteristics and surgery type, 1 surgeon had readmission rates significantly below the mean, and only 5 surgeons had readmission rates significantly above the mean. In contrast, for LOS, the patients of 288 surgeons (7.2%) had LOS significantly lower than the mean, and the patients of 397 surgeons (10.0%) had LOS significantly above the mean. These findings were robust to adjustments for surgeon characteristics and clustering by hospital. Similarly, hospital characteristics were not significantly associated with readmission rates, but LOS was associated with hospital for-profit status and size. CONCLUSIONS The authors found almost no variations in readmission rates by surgeon. These findings suggest that surgeon practice patterns do not affect the risk of readmission. Likewise, no significant variation in readmission rates by hospital characteristics were found. Strategies to reduce readmissions would be better targeted at factors other than providers.