Flawed evaluation of Medicaid demonstrations limits policy impact, GAO says
- A new Government Accountability Office (GAO) report finds major shortcomings in state evaluations of Medicaid demonstration projects, limiting their ability to inform policy decisions.
- The watchdog agency looked at demonstrations in Arizona, Arkansas and Massachusetts and found significant limitations in each — from lack of information on quality measures to failure to discuss rationales and conclusions.
- CMS-led evaluations have also been wanting due to data challenges that have hampered progress and limited the scope of those evaluations, according to GAO.
States are required to evaluate their demonstrations, which accounted for a third of the more than $300 billion in federal Medicaid dollars in 2015. CMS can also conduct its own reviews. The demonstrations allow new coverage and care delivery approaches to be tested before they are launched more broadly.
The report raises concern because providers will struggle to improve care or lower costs without reliable data and demonstrations. Lately, some states have been taking matters into their own hands with more conservative Medicaid waivers and health plans that don’t follow Affordable Care Act requirements. If states don't have a reliable way to gather the proper data on provisions such as work requirements and mandatory premiums, the impact of these major shifts will not be known.
In Arizona, the demonstration sought to determine whether providing long-term services and support via a managed care delivery model boosted access and quality of care. However, the evaluation report lacked information on key measures of quality and access, according to GAO.
A review of Arkansas’ evaluation of a demonstration using Medicaid funds to purchase private insurance for a select group of Medicaid beneficiaries also came up short — failing to address the hypothesis that doing so would improve continuity of coverage. The aim was to reduce coverage gaps for beneficiaries who experienced frequent changes in income and Medicaid eligibility.
In a third case, Massachusetts failed to provide any conclusions on whether providing millions in incentive payments to seven hospitals improved care quality and reduced per capita costs.
GAO blamed the flaws, in part, on CMS requiring final evaluation reports after the demonstrations concluded rather than following each three- to five-year demonstration cycle. While the agency has claimed it now plans to require more frequent reports, it has not published procedures for implementing the change, the report says. Plans to let states do scaled-down assessments of some demonstrations also lack criteria on when that would be allowed.
The report also calls out CMS for failings of federal evaluations. For example, problems getting data from states resulted in CMS narrowing the scope of a multi-state evaluation of the impact of various demonstrations on four priority policy issues. And though much of the data was eventually obtained, CMS has no policy for publishing the results — reducing the chance to impact Medicaid policy.
For example, Iowa obtained a Medicaid waiver that allows the state to stop retroactive Medicaid coverage, a move the state said would save $36.8 million annually and reduce monthly enrollment by more than 3,300 enrollees.
Kansas, meanwhile, got the Trump administration to okay a one-year extension of its Medicaid program, KanCare, which allows most of its Medicaid population to be covered in a managed care plan. The waiver lets Kansas maintain its safety net care pool, which helps pay for uncompensated care and incentive payments for delivery system reforms.