Scientists Say HHS Is Spinning Their Data to Push Medicaid Work Requirements

The Trump administration is using a short policy brief with minimal citations to justify work requirements for Medicaid coverage. Some of the cited researchers say their work is being misinterpreted.

The brief represents a stark departure from the usual quality of work from the Department of Health and Human Services’ top policy office, experts and former federal employees told NOTUS.

One former Health Department official called the policy document “very disappointing.”

“Frankly, I was appalled,” said Richard Frank, who served as assistant secretary for planning and evaluation at HHS during the Obama administration. “The concern is that the political narrative was what took over.”

The brief, titled “Medicaid Work Requirements Incentivize Employment and Are Estimated to Reduce Poverty,” was written by political appointees and career employees within the Office of the Assistant Secretary for Planning and Evaluation.

It was published in conjunction with the Centers for Medicare & Medicaid Services’ interim final rule on the implementation of an 80-hours-per-month work requirement for able-bodied adult Medicaid enrollees and applicants. CMS touted the study in its early June press release announcing the new rule.

ASPE, though little known outside of HHS, is meant to serve as a principal adviser to the Health Secretary; the office generates research reports that play a critical role in developing department policy. (The assistant secretary role is currently vacant.)

Over the course of about four and a half pages, the Trump administration argues that Medicaid work requirements could reduce poverty for millions of people and increase employment by over 4 percentage points in the first five years of implementation, compared with unconditional cash transfers.

“Recent studies also show positive impacts on employment and earnings,” it states.

But the brief’s conclusions are “grounded in unrealistic assumptions instead of evidence,” said Michael Karpman, a researcher with the Urban Institute whose work was cited.

The ASPE brief cites a study Karpman co-authored titled, “Many Working People Would Be Shut Out of Medicaid under Proposed Work Requirements: Findings from the Survey of Income and Program Participation.”

The brief does not describe the actual findings of that study. Instead, it uses it as a source of “factors that could impede the potential reach of this policy” without specifying what those factors are. Had the brief fully cited his work, it might have drawn very different conclusions about work requirements, Karpman said.

“This study really helps explain why opportunities to increase employment through work requirements are so limited,” Karpman said.

Another cited researcher told NOTUS he was concerned that the report drew overly broad conclusions using his relatively narrow research: Michael Webb researched the effectiveness of work requirements for the public housing program in Charlotte, North Carolina, and the ASPE brief uses his study to help bolster the claim that work requirements could increase employment among Medicaid participants nationwide.

Webb said that while the brief didn’t mischaracterize his study’s findings, he noted “a lot of crucial differences” between how the Charlotte Housing Authority implemented the program and how each state will likely implement these new Medicaid requirements. His findings attributed much of the public housing program’s success to the availability of hands-on case managers and the city’s healthy economy, which enabled participants to find jobs.

“Having on-site support was really crucial,” Webb said in a statement. “It’s also important to note that the goal of the program was to get residents to participate in services — not to terminate them from the program.”

The brief cites only 10 outside sources to back up its findings — a small sampling of the research on the impacts of work requirements, which generally comes to a very different conclusion, Adrianna McIntyre, an assistant professor of health policy and politics at Harvard University, told NOTUS.

“The most relevant and most recent research is being left out of their discussion,” said McIntyre, who, along with another researcher, wrote a lengthy analysis of the brief’s shortcomings.

An HHS spokesperson pointed to the brief’s mention of an ASPE-commissioned review of the scientific literature on “work promotion strategies” as evidence that the brief’s authors considered more than the studies cited.

The brief links to a “technical appendix” that lists 132 studies on work-promotion strategies, though only seven of the listed programs involved work requirements and none mention Medicaid participants in the “population studied” column.

The brief itself does not seem to cite the studies listed in the appendix. The HHS spokesperson said via email that ASPE research briefs are intended to be “concise and accessible to readers,” with supporting materials in appendices rather than in the body of the brief. (Other ASPE briefs have been significantly longer.)

Some of the studies included in the technical appendix seem to directly contradict the brief’s findings. Among the studies about Supplemental Nutrition Assistance Program work requirements listed in the technical appendix — studies that the HHS spokesperson specifically pointed to, but which were not cited in the brief — one showed “no effects on employment,” while the other found that they “may increase the risk of disadvantaged individuals failing to receive the assistance they need.”

Experts and former staff told NOTUS that while ASPE’s work can be influenced by the political leanings of each administration, the office’s many career staff generally ensure that its work remains grounded in evidence.

This brief, they said, is different.

“I do think that it is a shame to use the office to push research that is this shoddy,” McIntyre said.

Karpman told NOTUS the brief leaves out research that doesn’t support the conclusion that work requirements raise employment rates.

“They cite selected studies on work requirements in [Temporary Assistance for Needy Families] and housing programs, which were implemented in a much different context,” Karpman said. “They don’t cite studies on SNAP and other literature showing no employment effects from work requirements.”

Notably, multiple experts said, there’s an Arkansas-shaped hole in the brief: While several of the cited sources mention Arkansas, the first state to fully implement Medicaid work requirements, the state goes unmentioned by the ASPE brief.

The Arkansas work requirements implemented during the first Trump administration were widely considered a failure. More than 18,000 people — about 25% of the population subject to the requirements — lost coverage while they were in effect. The requirement was discontinued in 2019 after less than a year.

“It’s an enormous gap in this analysis, and it makes it very hard to view it at face value as anything other than a political effort,” Benjamin Sommers, who was the senior official overseeing health policy at ASPE during the Biden administration, said of the brief’s omission of Arkansas’ implementation.

Sommers, who is now a professor of health care economics at Harvard University, noted that the brief cited one study by his team on Medicaid participation rates — but conspicuously omitted two papers he worked on that showed Arkansas saw no change in employment as a result of work requirements.

“The assumptions here are just really implausible and outright ignore the evidence on Medicaid work requirements, and so I just don’t think you can really put much stock in this at all,” Sommers said.

The HHS spokesperson said via email that the review “focused on studies that received moderate or high ratings on methodological rigor” based on a government research clearinghouse. They added that Sommers’ 2020 study on work requirements in Arkansas used a study design that was “assessed as not sufficient,” and was therefore not included. (Sommers strongly disputed this assessment.)

Two of the cited authors told NOTUS that their research was correctly characterized, but otherwise didn’t comment on the brief’s conclusions.

The brief also models the potential effects of the new Medicaid work requirements — one with perfect implementation and one with less-than-perfect enrollment, employment rate and coverage. But those scenarios don’t reflect reality, said McIntyre, the Harvard University health policy expert.

“This is what I call a ‘sunshine and rainbows’ approach to analysis,” McIntyre said. “It’s basically ignoring everything we know about how work requirements operate in practice to say, ‘Well, what could they do if we ignore all of the evidence?’”

The brief acknowledges that the modeling scenarios are limited in their usefulness.

“Our results should be interpreted as approximate estimates rather than precise point estimates,” it states.

The HHS spokesperson said via email, “The purpose of the simulations is to illustrate the full potential of the policy when implemented as intended and to highlight poverty reduction as a key objective. The Trump Administration and HHS are committed to helping individuals move toward greater economic stability and self-sufficiency.”

This is not the first time an HHS report produced during the Trump administration has used questionable methods to back up scientifically dubious claims. A report from the Make America Healthy Again Commission published last year included fabricated citations, as NOTUS first reported.

There has been some indication that the administration is trying to improve its research standards. A recent advisory on the dangers of screen time was transparent that generative AI was used in its production.

But several of the former HHS officials and experts who spoke to NOTUS said they were concerned that this brief indicates a loosening of standards at a federal research office known for producing work grounded in evidence.

“Going back to the Bush administration, the Clinton administration, both Bushes — there was always a basic credibility,” Frank said. “When you kind of make up the assumptions, and you don’t rely on data, and you don’t rely on evidence, and you in fact put out the evidence and then ignore it — that goes to the credibility of the exercise, and that’s very damaging.”