Objective:
Spending for acute myocardial infarction (AMI) episodes varies widely across hospitals, driven primarily by payments made more than 30 days after discharge. Through collective incentives and an emphasis on care coordination, Medicare accountable care organizations (ACOs) may help reduce this variation. To test this hypothesis, we analyzed national Medicare data.
Methods:
Using a 20% random sample, we identified Medicare beneficiaries admitted for AMI from January 2010 to December 2013. We distinguished admissions to hospitals affiliated with a Medicare ACO from those that were not. We then calculated 90-day, price-standardized, risk-adjusted episode payments made on behalf of beneficiaries, differentiating between early (index admission to 30 days post-discharge) and late payments (31 to 90 days). We also calculated component payments, including those for the index hospitalization, readmissions, physician services, and post-acute care. Finally, we used difference-in-differences estimation to measure the effect of admission to an ACO-affiliated hospital on early and late episode payments.
Results:
Over the study period, 15,219 beneficiaries were admitted to 299 eventual ACO-affiliated hospitals and 73,910 were admitted to 1,685 never ACO-affiliated hospitals (
p
<0.001). While beneficiaries admitted to eventual ACO-affiliated hospitals tended to be younger than those admitted to never ACO-affiliated hospitals (mean age: 79.2 ± 8.6 versus 80.0 ± 8.5, respectively;
p
=.003), they had similar levels of comorbidity (mean Elixhauser score: 2.7 ± 1.4 versus 2.7 ± 1.4, respectively;
p
=0.526). Mean 90-day episode payments were greater for ACO-affiliated hospitals [$24,887 versus $23,966;
p
<0.001]. In the period after ACO implementation (2012 and 2013), total payments for AMI episodes fell by $1259 (Figure;
p
<0.001). Most of this savings was attributable to decreases in early ($1118) versus late ($141) episode payments. However, none of these savings differed based on admission to an ACO-affiliated hospital (
p
=0.363 for the difference).
Conclusions:
Early Medicare ACOs have not affected 90-day episode payments for AMI admissions. Future studies should explore the possibility of heterogeneity in effect based on ACO structure.