This commentary is by Julie Wasserman of Burlington, a health policy consultant working independently. She worked for Vermont state government for over 25 years; her jobs included director of Vermont’s dual eligible initiative at the Agency of Human Services, division director of policy and planning for the Department of Aging and Disabilities, and legislative staff to the Vermont Senate Health and Welfare Committee. She participated in Vermont’s three-year state innovation models initiative, the precursor to the all-payer model.
Authors from the UVM College of Medicine, UVM Medical Center, and UVM Health Network published a piece in the well-regarded Health Affairs journal on Vermont’s health care reform efforts. The authors discuss Vermont’s all-payer accountable care organization model and the practical issues that arise when a switch is made from volume-based payments to value-based payments. They expertly underscore fundamental challenges and pose important questions to consider.
However, the article perpetuates a longstanding distortion of the value of the all-payer model in controlling Vermont’s health care costs.
The blog mistakenly implies that Vermont’s all-payer model had savings of $97 million. To support this, the authors cite results from an external evaluation performed by RTI International. To date, there has been no external evaluation of Vermont’s all-payer model.
The authors write, “And an external evaluation of the effect of the APM …” However, the external evaluation was performed on the “state innovation models,” not Vermont’s all-payer model.
Vermont’s state innovation model occurred from 2014 to 2016; the all-payer model did not exist at that time, having begun in 2017. Moreover, the UVM authors conclude their article by asking “what are the lessons learned from the Vermont APM.” Again, they erroneously apply the results of Vermont’s state innovation model to the all-payer model, citing the $97 million in savings.
Compounding this error, the $97 million in Medicaid savings referenced in RTI’s evaluation of Vermont’s state innovation model is clearly questionable. RTI researchers were unable to identify the source of these Medicaid savings, nor which interventions led to the savings. The $97 million turned out not to be actual “savings” but rather the relative difference in spending between the Vermont Medicaid accountable care organizations and a comparison group. Curiously, the $97 million is mentioned only once in RTI’s 658-page evaluation (p. ES-1), and does not appear anywhere else in RTI’s extensive research findings, not even in the 80+ page Vermont chapter.
RTI’s findings on the state innovation models were subsequently detailed in a Milbank Quarterly peer-reviewed journal article. The Milbank Quarterly article makes no mention of Vermont’s $97 million in Medicaid savings. Instead, it states that Vermont accountable care organization enrollees experienced an increase in total Medicaid expenditures, although less than the growth in expenditures for the comparison group. (p.27 hardcopy.) Of note, RTI’s evaluation similarly described Vermont’s increase in Medicaid expenditures while simultaneously claiming $97 million in savings.
Furthermore, the RTI evaluation failed to subtract out the cost of Vermont’s state innovation model intervention ($45 million) as well as the accountable care organization administrative costs. The Milbank Quarterly article concurs, stating “Additionally, the implementation costs are significant, none of which are factored into the savings calculations.” (p.30 hardcopy.)
During the 2019 legislative session, the then-secretary of the Agency of Human Services testified in Vermont’s key legislative committees touting the $97 million in Medicaid savings, implicitly crediting OneCare. Yet, there were no real “savings.” As described above, Medicaid expenditures actually rose during the three-year state innovation model initiative. Furthermore, CHAC (the accountable care organization comprised of Federally Qualified Health Centers), not OneCare, was responsible for the slower rise in Medicaid spending. CHAC had cost reductions in all three years whereas OneCare had cost reductions only in the first year but was over budget in years two and three.
Vermont’s all-payer accountable care organization model is officially in year four of its five-year agreement with Centers for Medicare and Medicaid Services. Contrary to the UVM authors’ contention, the all-payer model has found OneCare’s cumulative cost overruns to be significant. OneCare has been over budget by $26 million, offset by $13 million in savings for the period 2017-2019 (most recent available data, p.7-8). When factoring in OneCare’s administrative costs of $36 million for the same period, it becomes challenging to conclude that the all-payer ACO model is generating savings.
The Centers for Medicare and Medicaid Services Innovation Center writes in this week’s New England Journal of Medicine that the vast majority of the center’s value-based models have not saved money. Both “state innovation models” and “Next Generation ACO models” (Vermont’s current all-payer model) are among those models deemed to have net losses.
The governor, the secretary of the Agency of Human Services, and the Legislature would be well-advised to call for an honest assessment of OneCare’s financial performance in the all-payer model.
