Early Impact of Medicare Accountable Care Organizations on Inpatient Surgical Spending.

objectiveTo evaluate whether hospital participation in accountable care organizations (ACOs) is associated with reduced Medicare spending for inpatient surgery.

backgroundACOs have proliferated rapidly and now cover more than 32 million Americans. Medicare Shared Savings Program (MSSP) ACOs have shown modest success in reducing medical spending. Whether they have reduced surgical spending remains unknown.

methodsWe used 100% Medicare claims from 2010 to 2014 for patients aged 65 to 99 years undergoing 6 common elective surgical procedures [abdominal aortic aneurysm (AAA) repair, colectomy, coronary artery bypass grafting (CABG), hip or knee replacement, or lung resection]. We compared total Medicare payments for 30-day surgical episodes, payments for individual components of care (index hospitalization, readmissions, physician services, and postacute care), and clinical outcomes for patients treated at MSSP ACO hospitals versus matched controls at non-ACO hospitals. We accounted for preexisting trends independent of ACO participation using a difference-in-differences approach.

resultsAmong 341,675 patients at 427 ACO hospitals and 1,024,090 matched controls at 1531 non-ACO hospitals, patient and hospital characteristics were well-balanced. Average baseline payments were similar at ACO versus non-ACO hospitals. ACO participation was not associated with reductions in total Medicare payments [difference-in-differences estimate=-$72, confidence interval (CI95%): -$228 to +$84] or individual components of payments. ACO participation was also not associated with clinical outcomes. Duration of ACO participation did not affect our estimates.

conclusionAlthough Medicare ACOs have had success reducing spending for medical care, they have not had similar success with surgical spending. Given that surgical care accounts for 30% of total health care costs, ACOs and policymakers must pay greater attention to reducing surgical expenditures.