The study, requested in late 2015 by five senior Republican members of Congress, comes at a possible turning point in the 53-year history of Medicaid.

The Trump administration has been flooded with requests from states seeking Medicaid waivers. They want to impose work requirements, charge premiums, suspend coverage for people who fail to pay premiums, and restrict eligibility or benefits in other ways. A few states would like to impose time limits on Medicaid coverage for some beneficiaries.

Concern about waivers is bipartisan. Republicans were worried about the cost of some demonstration projects approved by the Obama administration. Democrats are alarmed at work requirements and other policy changes that the Trump administration has approved.


Seema Verma, now the administrator of the Centers for Medicare and Medicaid Services, guided her home state of Indiana in expanding Medicaid eligibility under the governor at the time, Mike Pence, while emphasizing “personal responsibility.”

Stephen Crowley/The New York Times

The Government Accountability Office found particular problems in the evaluation of a Medicaid waiver granted to Indiana. The state expanded Medicaid, as permitted under the Affordable Care Act, but obtained a federal waiver for policies emphasizing “personal responsibility,” including monthly premiums and incentives for healthy behavior.

The federal government wanted to know how Medicaid beneficiaries were affected by those changes, but it could not get the data it needed from the state, so its evaluation was severely limited, the Government Accountability Office said.

Conservatives see Indiana’s Medicaid program as a model, and other states are following suit. The architects of the Indiana plan include Vice President Mike Pence, who was the governor of Indiana from 2013 to January 2017, and Seema Verma, a consultant to the state who is now the administrator of the federal Centers for Medicare and Medicaid Services.

Similar problems have stymied the evaluation of an important experiment in Arkansas, which is using federal Medicaid funds to help pay premiums for low-income people who buy private coverage in a health insurance exchange established under the Affordable Care Act.

Federal and state officials wanted to know whether consumers would have better access to care under this arrangement than in the traditional Medicaid program. But, the report said, “Arkansas officials told us that the state was not collecting the specific data from exchange plans requested by the federal evaluator, and that collecting those data would result in additional cost.”

In its report, the G.A.O. said federal officials had “no standard policy” for the public release of findings from the evaluation of Medicaid demonstration projects. As a result, it said, the Centers for Medicare and Medicaid Services “is missing an opportunity to inform important federal and state policy discussions.”

Medicaid has spent increasing amounts on demonstration projects in the past decade. Federal spending under Medicaid waivers now exceeds $100 billion a year, up from $29 billion in 2005 and $49 billion in 2010.

Waivers are typically approved for an initial five-year period, but federal officials often renew and extend waivers without waiting for an evaluation to show whether the project was worthwhile, the Government Accountability Office said.

The auditors made three recommendations and the Trump administration accepted all of them, saying it recognized the need to conduct rigorous independent evaluations and release the results in a more timely way.

In granting Medicaid waivers, federal officials typically require states to measure the effects of a policy change on costs and the health of beneficiaries. But, the report said, it is often difficult to isolate those consequences because the overall health care system is changing so fast.

Continue reading the main story