The Impact of Alternative Payment Models on Care Coordination

Each year, poorly coordinated care in the US healthcare system results in $25-45 billion in unnecessary spending due to avoidable complications and readmissions. Coordinating care between healthcare organizations is particularly challenging since there are many opportunities for communication breakdowns when information travels across organizational boundaries. Recent healthcare reform programs have developed several innovative models of paying providers, collectively known as Alternative Payment Models (APMs), which create incentives for providers to reduce costs. Policy discussions surrounding APMs are usually focused on their potential impact on care coordination and its implications for cost reduction. However, these models reward providers for reducing costs through any means, not only care coordination. Prior studies have found mixed evidence on the impact of APMs on cost and quality; but little is known about the impact of APMs on care coordination directly. Using national datasets on acute-care hospitals in the US, my dissertation investigates the impact of hospital participation in APMs on 3 measures of care coordination: patient engagement during hospital care transitions, health information exchange, and hospital-SNF referral patterns. In Chapter 1, I ask whether patient engagement during hospital care transitions is associated with reduced readmission rates, and whether APM-participation is associated with improved patient engagement. I use panel analysis techniques on retrospective observational data on 2,685 hospitals from 2013-2017. I found that patient engagement was associated with reduced readmission rates, but APM-participating hospitals were not improving patient engagement compared to non-APM hospitals. These findings suggest that APM-participating hospitals may be overlooking patient engagement as an important patient-centered strategy for reducing healthcare costs. In Chapter 2, using cross-sectional analysis on retrospective observational data on 798-1730 hospitals from 2014-2015, I evaluate the association between APM participation and four measures of health information exchange (HIE). Together, these measures paint a detailed picture of how well hospitals are sending information for patients that are discharged from a hospital setting. I found that APM-participating hospitals sent data for fewer patients, but routinely sent more data types, to more partner types, using a greater variety of exchange approaches. Findings from this chapter suggest that APM-participating hospitals may be investing in HIE infrastructure with a selective group of partners, rather than engaging in more community-based approaches to HIE. Finally, in Chapter 3, I assess the impact of Medicare’s Bundled Payment for Care Improvement (BPCI) program on hospital-SNF referral patterns. Social network theory posits that hospitals with fewer SNF partners, or more concentrated SNF referral patterns may have an easier time coordinating care for their patients. I conduct a panel analysis on 2,354 hospitals from 2009-2015 to assess whether BPCI participation leads to reductions in SNF referrals, a pruning of SNF partners, or more concentrated SNF referrals; and whether the quality of SNF partners improves after hospitals join the BPCI program. I found that participation in the BPCI program did not lead to significant changes in hospital-SNF referral patterns, and the quality of SNF partners improved after hospitals joined the BPCI program. These findings suggest that BPCI participating hospitals are not changing referral patterns to favor higher-quality SNFs; instead, the quality of SNFs in BPCI-participating hospital referral networks may be improving. As a whole, this dissertation contributes to understanding whether APMs improve care coordination and suggests that policymakers may need to consider more explicit methods of incentivizing care coordination.PHDHealth Services Organization & PolicyUniversity of Michigan, Horace H. Rackham School of Graduate Studieshttps://deepblue.lib.umich.edu/bitstream/2027.42/151662/1/sunnylin_1.pd