Study Finds Fault With Ark. Medicaid Work Requirement

When the state of Arkansas implemented a “community engagement” requirement in June 2018 for Medicaid recipients ages 30-49 — colloquially called a work requirement — officials said it would encourage unemployed, able-bodied recipients to get jobs, while protecting the disabled, those already working, and other populations needing public assistance.

“If you want this coverage you have to be able to put some skin in the game,” an Arkansas official told MedPage Today in 2017. Around the same time, the Centers for Medicare & Medicaid Services (CMS) Administrator Seema Verma argued that able-bodied adults shouldn’t receive Medicaid at all.

As Verma said at another venue, work requirements would help Medicaid recipients “make changes … to move to a better life.”

Now, a survey of low-income adults in Arkansas suggests that didn’t happen, while predictions by the policy’s critics largely came true. (A federal judge ordered a halt to the policy in March, finding that it was contrary to the Medicaid program’s statutory purpose.)

“We found no significant changes in employment associated with the policy, and more than 95% of persons who were targeted by the policy already met the requirement or should have been exempt,” wrote Benjamin Sommers, MD, PhD, of the Harvard T.H. Chan School of Public Health in Boston, and colleagues in the New England Journal of Medicine.

But many people did lose Medicaid coverage after the policy went into effect — who mostly “were unaware of the policy or were confused about how to report their status to the state,” the researchers said.

Sommers and colleagues surveyed a total of 5,955 low-income individuals, ages 19-64, about half in Arkansas and the rest in three nearby states where no work requirements were implemented, in late 2016 and late 2018 (in a new set of respondents) after the Arkansas work requirement went into effect. The policy exempted students, the disabled, persons responsible for full-time care of a child or other family member, and pregnant women. Qualifying “community engagement” activities included employment (at least 80 hours per month), job training, school enrollment, or volunteer work.

Respondents — contacted by telephone — were asked about current insurance, community engagement activities, and access to healthcare. Arkansas respondents were also queried about their knowledge of the work requirement policy and whether they thought it applied to them. Because different individuals were surveyed in the 2016 and 2018 iterations, Sommers and colleagues employed modeling techniques to estimate different groups’ gains or losses in coverage from the survey data.

In the 30-49 age group in Arkansas, rates of Medicaid or Obamacare “marketplace” coverage declined by 10.4 percentage points (95% CI -18.5 to -2.4) from 2016 to 2018 relative to this age group in the control states, after adjusting for demographic variables. Rates of uninsurance, meanwhile, rose by 5.9 percentage points (95% CI 0.4-11.4) among 30- to 49-year-olds in Arkansas versus control states.

Rates of “community engagement” increased by about 4 percentage points in Arkansas between the two surveys in the 30-49 group, but this was no different from what was seen in control states in the same age group (adjusted difference 1.6 percentage points, P=0.63). In fact, community engagement increased across the board among all respondent groups, including those younger and older than 30-49, in other states as well as in Arkansas.

Sommers and colleagues said their findings “are consistent with the official report from Arkansas that nearly 17,000 adults were removed from Medicaid between October and December 2018.” Moreover, they wrote, “[w]e did not find any significant change in employment … or in the related secondary outcomes of hours worked or overall rates of community engagement activities.”

It appeared, they continued, that “nearly everyone who was targeted by the policy already met the requirements, so there was little margin for the program to increase community engagement.”

So why did Medicaid enrollment drop? The survey found that 33% of Arkansas respondents had not heard of the policy, and nearly half of respondents were unsure whether the policy applied to them.

“Among the respondents who had been told by the state that they needed to report community engagement activities, only 49.3% were doing so regularly,” Sommers and colleagues noted. “The most common reason for not reporting was a belief that they were not meeting the requirement” — but the researchers found that all of these respondents (n=22) did meet the requirements, based on their responses to other survey questions. Also, one-third of those who failed to report their activities said they lacked internet access, which was required for the reporting system until December 2018, when a telephone option was added.

Under the law, those who meet the requirements but don’t file the required documentation are removed from Medicaid, although they can reapply in the next quarter. But Sommers and colleagues noted that the lack of continuity in coverage is problematic, and the Arkansas system appeared to exacerbate it.

The researchers acknowledged a number of limitations to the survey-based study, including a low response rate, possible inaccuracies in participants’ responses, and the possibility that unmeasured confounders could have influenced results.