164 Should Spine Surgeons Be Held Accountable for 30-Day Readmissions?

INTRODUCTION: Recent efforts to improve health care quality have focused on variations in outcomes such as 30-day readmissions (30d-R). Pay-for-performance programs hold providers accountable to reduce these variations, with the assumption that variations are due to discretionary practices of providers and can be influenced by changes in reimbursement. We examined variations in length of stay (LOS) and 30d-R among the surgeons for elective lumbar spine surgery for degenerative conditions to determine if these outcomes are a valid target for pay for performance programs. METHODS: We applied validated International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) algorithms to 2003 to 2007 Medicare data to select patients undergoing elective lumbar spine surgery (with and without fusion) for degenerative conditions. We then used multilevel models and adjusted for patient demographics, comorbid conditions, and type of surgery to generate 95% estimates for LOS and risk of 30d-R by surgeon. RESULTS: The final sample consisted of 39884 patients operated upon by 3987 spine surgeons. The mean LOS was 3.1 days (median 3.0 days) and the mean readmission rate was 7.2%. After adjusting for patient characteristics and type of surgery, the patients of 288 surgeons (7.2%) had LOS significantly lower than the average, and the patients of 397 surgeons (10.0%) had LOS significantly above average. On the other hand, no surgeon had readmission rates below average, and only 4 surgeons had patients with readmission rates significantly above average. Our findings were robust to adjustments for surgeon characteristics and for clustering by hospital. CONCLUSION: We found significant variations in LOS by surgeon after adjusting for patient characteristics and type of surgery. This suggests that surgeon practice varies and impacts length of stay. Yet, we found no variation in 30d-R by surgeon, implying that surgeon practice variation does not affect 30d-R. Thus, pay-for-performance programs aimed at providers are unlikely to improve 30d-R.