Expert Commentary

The expanding role of Medicaid in US health care: A research roundup

To help journalists report on Medicaid, we’ve summarized a few studies that can inform reporters about key debates happening about this program.

Photo by National Cancer Institute on Unsplash

This piece was originally published in July 2020. In light of the Republicans’ debt ceiling negotiations with the Biden administration in 2023, which involve discussions of a federal Medicaid work requirement, we updated it on May 24 with some updated statistics and a list of additional studies published as recently as April 2023.

The two largest U.S. health plans share a birthday, July 30, 1965, but they have different roles and public images.

A law signed by President Lyndon B. Johnson created Medicare, which serves people age 65 and older, and Medicaid, which covers people considered to be poor by government standards. Both programs also cover people with disabilities, contributing to overlap between Medicaid and Medicare. About 12.2 million people of the about 60 million people enrolled in Medicare in 2018 also had Medicaid coverage.

People tend to remain enrolled in Medicare. In 2022, 55.5 million of the the 63.8 million participants were age 65 and older, according to the 2022 Medicare trustees report. The rest of the enrollees qualified due to disabilities.

Not so with Medicaid, where there is more churn.

For example, about 86.7 million people were covered by the state-federal program at some point during fiscal 2018, according to a December 2019 report from Medicaid and CHIP Payment and Access Commission (MACPAC). But fewer might be covered by the program at any given point in the year, as can be tracked through Medicaid’s website.  (The program posts a monthly snapshot of recent enrollment as well as releasing more extensive data.)

People gain Medicaid coverage when they lose jobs — for instance, during the recession stemming from the COVID-19 pandemic — and drop it when they become employed again. Some people with disabilities also rely on Medicaid coverage while waiting to qualify for Medicare.

While Medicaid is a safety-net program for many Americans, Medicare is more of an aspiration, which enjoys a significant base of bipartisan support.

“You couldn’t move my mother out of Medicare with a bulldozer,” then House Energy and Commerce Chairman Billy Tauzin, a Louisiana Republican, said in 2003, while working on the last major expansion of the federal health program. “She trusts in it, believes in it. It’s served her well.”

But there’s a sharp partisan divide about Medicaid. There were no Republican votes for the 2010 Affordable Care Act (ACA), which set the stage for a major expansion of Medicaid that’s still unfolding. Instead Republicans have since tried repeatedly to repeal the ACA, while also reviving in recent years attempts to convert federal funding of Medicaid from an open-ended commitment based on formulas to more limited support though block grants.

GOP’s unsuccessful ACA repeal bids in 2017 foundered, though, in part due to growing support for Medicaid, according to Richard Sorian, a former assistant secretary for public affairs at the Department of Health and Human Services in the Obama administration. Sorian looked at how the 2014 expansion had allowed more people to get access to health care in several states with Republican governors, such as Ohio, as well as in those dominated by Democrats.

“For most of its history, Medicaid took a back seat to Medicare, the health benefits program for seniors and others,” Sorian writes, adding that some politicians still seek Medicaid budget cuts. “But many more are leery of touching the program and facing the wrath of the people who elected them.”

A few statistics show how Medicaid underpins much of U.S. health care.

  • About 1 in 5 Americans get health insurance through Medicaid, which is run by the states with federal financial support and oversight. As of January 2023, more than 85.9 million people in the U.S. were enrolled in Medicaid, according to the latest data posted by the Centers for Medicare and Medicaid Services (CMS). There were about  7 million participants in the Children’s Health Insurance Program (CHIP), considered a sister initiative to Medicaid.
  • Medicaid paid for 43% of all births in the U.S. in 2018, while private insurance plans paid for 49%. Policymakers now are looking to expand Medicaid coverage to try to lower the high rate of maternal mortality in the United States. There are bills pending in Congress that would require Medicaid coverage of new mothers from a 60-day period to the entire year following giving birth.
  • Medicaid is the largest U.S. purchaser of what it calls behavioral health services, which include mental health treatment and services to treat addiction and substance abuse.
  • Medicaid is the primary tool through which the Affordable Care Act (ACA) of 2010 expands the public’s access to health care.

News coverage of the federal government’s implementation of the ACA in 2014 focused heavily on hitches with the startup of the online state and federal exchanges through which people who did not get health plans from their employers can buy medical coverage. These were primarily intended to help people whose employers do not offer health plans. Many people get subsidies to purchase their insurance on these exchanges. Without this help, they might not be able to afford insurance.

About 11.4 million people were covered by these health plans sold on the exchanges for 2020, according to a report from the Centers for Medicare and Medicaid Services (CMS).

These include plans sold by for-profit companies as well as ones from nonprofit insurers.

But, by late 2018, about 15.1 million people were enrolled in Medicaid due to expansion created by the ACA, CMS reported.

States have varied eligibility criteria, including income cutoffs, for Medicaid. Before the ACA, many states largely excluded adults who do not have disabilities,  no matter how little they earned.

President Barack Obama and congressional Democrats intended for all states to raise their Medicaid eligibility requirements to allow adults who work but have incomes just above the federal poverty level to get health insurance. CMS last year authorized Utah’s Medicaid expansion, which will allow coverage for single people with annual income of as much as 138% of the federal poverty level ($17,608). For a family of four, this income cutoff would be $36,156.

Under the ACA, states initially were required to set their cutoff for Medicaid eligibility at a level that allowed people with household incomes as high as 138% above the federal poverty level to enroll. It’s important to note that while the actual text of the ACA sets this level at 133%, other provisions of the law have effectively nudged the cap to 138%.

In 2012, the U.S. Supreme Court decided that states could choose whether they wanted to raise the threshold for Medicaid eligibility under ACA. In an effort to encourage states to raise their income thresholds to allow more people to qualify for Medicaid, the federal government offered to cover the majority of the cost of covering these new enrollees.

Many states led by Republican governors initially balked at the offer and GOP political candidates campaigned on pledges to repeal the ACA.

Several Republican leaders, including Gov. Gary Herbert of Utah and then Indiana Gov. Mike Pence, later took the federal government up on its offer. The U.S. opioid epidemic helped persuade GOP holdouts to expand Medicaid. As of May 2023, 10 states have not taken formal action to expand Medicaid, according to a tally kept by the Kaiser Family Foundation.

In some states that have expanded their Medicaid coverage, Republican governors have added conditions for people who are able to enroll thanks to increased income thresholds, including payments of premiums. Congressional and state Democrats have objected to Republican attempts to add work requirements to Medicaid participation for adults who do not have disabilities. They have argued that many people added as a result of the expansion already work and documenting employment is a significant administrative burden.

To help journalists report on Medicaid, we’ve summarized a few studies below to help reporters understand the key debates happening about this program.

At the heart of Medicaid research are persistent questions about how well the massive state-federal health program works. Studies published to date show mixed results on questions of whether having Medicaid coverage helps participants improve or maintain their health.

For example, one study found middle-aged people who live in states that expanded Medicaid under the ACA are less likely to die of heart disease.

But discussions about Medicaid often quickly loop back to the somewhat surprising findings of a 2008 experiment in Oregon involving Medicaid, which is discussed more fully in a section below.

After Oregon officials found they had funds for a limited expansion of the state’s Medicaid program, they used a lottery to select about 30,000 people from a waiting list of almost 90,000. This approach allowed economists a rare opportunity to study the effects of Medicaid coverage in a group of people randomly selected to enroll in the state-federal health plan.

Some findings from studies of the Oregon experiment disappointed advocates for Medicaid expansion. These findings include research that seems to contradict a common theory that people newly enrolled in Medicaid  would use emergency rooms less often for basic health care needs if they could afford to see a primary care doctor by participating in Medicaid. Visits to a primary care clinic cost significantly less than emergency room visits.

In recent years, many researchers have sought to assess the early impacts of the ACA’s Medicaid expansion. The results of the Oregon experiment cast doubt about the possible benefits of Medicaid coverage for a group that had been randomly selected to receive coverage, writes Sarah Miller, an assistant professor at the University of Michigan’s Ross School of Business, and her co-authors in a 2019 paper.

“The inconclusive nature of these results has led to skepticism among some researchers, policymakers, and members of the media as to whether Medicaid has any positive health impacts for this group,” write Miller and her co-authors.

In their paper, though, Miller and her co-authors estimate Medicaid expansion may have prevented 4,800 deaths in their sample population among people ages 55 to 64, or roughly 19,200 fewer deaths over the first four years alone. Miller and her co-authors also offer an estimate of what many states’ decisions against Medicaid expansion meant for their citizens.

“Our estimates suggest that approximately 15,600 deaths would have been averted had the ACA expansions been adopted nationwide as originally intended by the ACA,” Miller and her co-authors write.

Below are the summaries of the Miller paper and other research on Medicaid, including a section on the Oregon Medicaid experiment.

Medicaid and Mortality: New Evidence from Linked Survey and Administrative Data
Sarah Miller, et al. National Bureau of Economic Research working paper No. 26081, Revised August 2019.

Broader access to Medicaid appeared to lower the mortality rate in a study focused on people ages 55 to 64, Miller and her co-authors find.

They analyzed the potential effect of the Medicaid expansion by using data from the federal American Community Survey (ACS). This national survey draws information from about 4 million respondents a year. The survey asks for such details as income level and citizenship status, which allow researchers to make observations about people’s potential Medicaid eligibility.

In their study, Miller and her co-authors focused on adults aged 55 to 64 who appeared likely to qualify for the Medicaid due to the expansion. The researchers  excluded adults living in Delaware, Massachusetts, New York, Vermont and the District of Columbia from the study. Medicaid programs in those areas earlier had allowed coverage of adults living in poverty.

Miller and her co-authors then compared data for the people in the selected expansion states  to ACS data for a similar group of people from states that had not expanded Medicaid. They find that the probability of dying in the first year of the expansion declined by about 0.089 percentage points for the Medicaid expansion group. In the second and third year, the probability of death dropped a little over 0.1 percentage points, the researchers explain. By the fourth year, it declined 0.2 percentage points.

Limitations to this research include the possible effect of instances where people’s income may have risen between 2008 and 2014, the years studied, making them ineligible for the expansion.

Some participants in the ACS also may have moved to different states, causing misclassification about whether those people were eligible for the expansion, Miller and her co-authors write.

The nonpartisan, nonprofit Center on Budget and Policy Priorities released a report that discusses the findings of the paper in simple terms. The University of Michigan produced a press release on Miller’s findings.

Medications for Opioid Use Disorder Among Pregnant Women Referred by Criminal Justice Agencies Before and After Medicaid Expansion: A Retrospective Study of Admissions to Treatment Centers in the United States
Tyler Winkelman, et al. PLOS Med, May 18, 2020.

Pregnant women referred by courts and other criminal justice agencies for opioid abuse treatment were more likely to get medications to help them manage their condition if they lived in states that had expanded their Medicaid eligibility, wrote Tyler Winkelman of the Hennepin Healthcare Research Institute of Minneapolis, and his co-authors.

The rate at which medication for opioid use disorder was prescribed for pregnant women rose from 21.4% to 36% in the states studied that had expanded their Medicaid eligibility. In the states that did not expand Medicaid, the rate increased from 7.0% to 9.6%.

For the study, Winkelman and co-authors analyzed data collected from 1992 to 2017, identifying records for cases of 131,838 pregnant women with opioid use disorder. They drew these from the Treatment Episode Data Set-Admissions (TEDS-A) program, an annual national survey conducted by the federal Substance Abuse and Mental Health Services Administration (SAMHSA). More than half — 63.3% — percent of the women in the sample studied were between the ages aged 18–29.

They acknowledge limits to their paper, including gaps in the TEDS-A data.

Although TEDS-A is the most comprehensive survey of treatment admissions in the U.S., some states only report cases of people whose care was publicly funded. Omitting data about pregnant women who used private insurance to pay for their treatment could potentially alter the reported results, the authors write.

Examination of Changes in Health Status Among Michigan Medicaid Expansion Enrollees From 2016 to 2017
Minal R. Patel, et al. JAMA Network Open, July 10, 2020.

People in Michigan who enrolled in Medicaid through the ACA expansion were less likely to report themselves as being in fair or poor health over time, finds a study by Minal R. Patel, an associate professor at the University of Michigan School of Public Health, and her co-authors.

In 2016, 30.7% of study participants reported being in fair or poor health, while 27.0% did the following year, Patel and co-authors find. They used a longitudinal panel survey, an approach that allows for repeated observations over time, to assess changes in health status for people enrolled in the Medicaid expansion in Michigan. The researchers analyzed the answers that  more than 3,000 people gave during their telephone survey in both 2016 and 2017.

Patel and her co-authors did not find any statistically significant differences in the other aspects of health they studied — for example, the number of days participants reported being in poor mental health or the number of days of usual activities they reported missed owing to poor physical or mental health over time.

Effects Of Medicaid Expansion On Postpartum Coverage And Outpatient Utilization
Sarah H. Gordon, et al. Health Affairs, Jan. 6, 2020.

A comparison of medical claims for women living in Colorado and Utah suggests expanding Medicaid may help new mothers get needed medical care. Sarah H. Gordon, an assistant professor at the Boston University School of Public Health, and her co-authors studied how Medicaid eligibility rules affect women’s access to care after delivery, noting concerns about the high maternal mortality rates in the U.S. compared to other rich nations.

In their paper, Gordon and her co-authors describe this as the first study to examine the impact of the ACA’s Medicaid expansion on postpartum coverage and outpatient medical care among pregnant women. They obtained information about Medicaid claims through agreements with state agencies in Colorado and Utah.

In Colorado in 2013, pregnant women with incomes below 185% of federal poverty level and parents with incomes below 105% of the poverty level were eligible for Medicaid. The state in 2014 expanded its Medicaid program to cover all adults with incomes below 138% of the federal poverty threshold. Utah did not expand its Medicaid eligibility until 2019.

The mean (average) number of months of Medicaid coverage in the six months after delivery rose from 5.3 months in the January-June 2013 period to 5.4 months in the January 2014-June 2015, but dropped in Utah from 4.6 months to 3.6 months.

Gordon and her co-authors compared the treatment that women in the two states received between January and June 2013 and between January 2014 and June 2015. At the time, Utah’s Medicaid eligibility cap was 33% above the poverty line for pregnant women and 44% below the poverty line for parents.

Among the key findings: Mothers in Colorado, on average, had more outpatient medical visits during the first six months after delivery. The average number of outpatient visits rose from 3.0 to 3.3. In Utah, mothers had fewer outpatient visits – an average of 1.8 in the six months post-delivery, down from 2.0.

There was an even larger gain in Colorado for new mothers who had experienced difficulty during pregnancy and delivery, termed maternal morbidities, such as hemorrhage. For these women, the average number of outpatient visits in that state in the six months after delivery rose from 2.7 to 3.4. In Utah, the average number of visits for women in this category slipped from 1.8 to 1.6.

Gordon and her co-authors note in their paper that their findings apply only to women in the two states. They also note that the claims databases used in their work lacked detailed information on patient race, ethnicity and socioeconomic status.

Medicaid Work Requirements — Results from the First Year in Arkansas
Benjamin D. Sommers, et al. New England Journal of Medicine, Sept. 12, 2019. Published online June 19, 2019.

An Arkansas initiative linking work requirements to Medicaid eligibility is associated with a rise in the percentage of uninsured people in the state but no significant change in employment status, finds this study, led by Benjamin D. Sommers, a professor at the Harvard T.H. Chan School of Public Health.

Sommers and his co-authors looked at what happened after Arkansas, in 2018, became the first state to implement a work requirement in connection with the Medicaid expansion. (Subsequent federal court decisions put a halt to this Arkansas requirement, as detailed by the Kaiser Family Foundation on a page where it tracks Medicaid policy changes approved by federal waivers.)

Many Republicans, including Arkansas Gov. Asa Hutchinson, have argued in favor of tying coverage through the Medicaid expansion to work requirements. His state received permission from the federal government to ask Medicaid beneficiaries who were 30 to 49 years old to meet requirements such as working 80 hours per month or participating in another so-called “community engagement activity” such as job training or community service. Pregnant women and people with disabilities could get exemptions from these rules.

Sommers and his co-authors used telephone surveys to assess changes in work patterns before and after the Arkansas work requirements took effect. They studied changes among four groups: people in Arkansas aged 30 to 49, Arkansans 19 to 29 years of age, Arkansans aged 50 to 64, and adults in three comparison states — Kentucky, Louisiana and Texas.

The researchers find that the percentage of people in the group representing Arkansans aged 30 to 49 who worked 20 hours or more a week dropped from 42.4% in 2017 to 38.9% in 2018.

Similar results were seen in control groups used in the study, including the three comparison states.

The percentage of uninsured respondents among Arkansans 30 to 49 years of age increased

from 10.5% in 2016 to 14.5% in 2018, with smaller or no changes in the other groups. The percentage of Arkansans 30 to 49 years of age with employer-sponsored coverage increased slightly, from 10.6% to 12.2%

Sommers and his co-authors note that more than 95% of people in Arkansas who were targeted by the new policy already met the work requirement or should have been exempt.

The researchers note that the study did not determine whether individuals’ loss of coverage was a direct result of the new work requirements or prompted by other factors such as income changes or failure to complete renewal paperwork.

The nonprofit, nonpartisan Commonwealth Fund has a brief about Sommers’ paper, summarizing its findings in simple terms.

The Oregon experiment

The Oregon Experiment — Effects of Medicaid on Clinical Outcomes
Katherine Baicker, et al. New England Journal of Medicine, May 2, 2013.

The findings of this paper often are cited by critics of Medicaid. Katherine Baicker and her co-authors find that expanding Medicaid coverage in Oregon in 2008 had no statistically significant effect on the prevalence or diagnosis of hypertension or high cholesterol levels or on the use of medication for these conditions.

While Medicaid coverage increased the probability of a diagnosis of diabetes and the use of diabetes medication, Baicker and co-authors observed no statistically significant effect on a measure of control of blood-sugar for diabetes.

But Baicker and her co-authors did find increased use of health care services, lower rates of depression and reduced financial strain among people who were able to get Medicaid coverage through the lottery.

“We found a pretty mixed bag there, which I think was surprising to a lot of people,” Baicker said during a panel discussion about the ACA at the Harvard Kennedy School of Politics in 2017.

The Oregon experiment offered a rare opportunity for researchers like Baicker and her co-authors on this paper, including Amy N. Finkelstein of the Massachusetts Institute of Technology.

Economists studying the effects of different insurance policies and plans often work to create valid comparison groups by pulling data from claims records. In contrast, researchers studying the effects of medicines see how these treatments work on people randomly selected to get a drug or placebo.

Oregon officials in 2008 prepared to allow new enrollment in the state’s Medicaid program after capping it. They knew there would be more demand for coverage than the state had budgeted, so they used a lottery to determine who could participate.

This created randomized groups that Baicker and her co-authors could study. The results of their research make clear how tough it is to make decisions about whether and how to use Medicaid to help more people obtain medical care. Policymakers have to weigh the benefits of expanding Medicaid against the expense, as the state-federal health plan competes with other priorities in government budgets.

“You cannot expect to save money in the healthcare system by expanding Medicaid. It costs money because people use more care, so you’ve got to find a way to finance it,” Baicker said during the 2017 Harvard Kennedy School panel discussion.

Baicker and her co-authors have continued to examine the results of the Oregon Medicaid experiment. Recent papers include:

The Effect of Medicaid on Management of Depression: Evidence From the Oregon Health Insurance Experiment
Katherine Baicker, et al. Milbank Quarterly, March 5, 2018.

Having Medicaid coverage appeared to reduce the prevalence of undiagnosed depression by almost 50% and untreated depression by more than 60% in a study comparing adults who enrolled in Oregon’s state health program to fellow applicants who were not selected in a lottery, Baicker and her co-authors find. They write that Medicaid coverage increased use of medications frequently prescribed to treat depression and related mental health conditions and reduced the share of people reporting unmet mental health care needs by almost 40%.

Between September 2009 and December 2010, Baicker and her colleagues collected information on 12,229 people via interviews and questionnaires. They selected people living in the Portland area, of whom 6,387 had been able to enroll in Medicaid through the lottery. The remaining respondents served as a control group.

Effect of Medicaid Coverage on ED [Emergency Department] Use — Further Evidence from Oregon’s Experiment
Amy N. Finkelstein, et al. New England Journal of Medicine, Oct. 20, 2016.

This paper is a follow-up to one published in Science in 2014. In that first paper, Sarah L. Taubman of the National Bureau of Economic Research and co-authors, including Baicker and Finkelstein, reported a 40% increase in emergency department (ED) visits among people who were chosen by lottery to receive Medicaid coverage.

That was a surprising finding, as there had been widespread belief that expanding Medicaid coverage would encourage more use of primary care and thus reduce ED visits, Finkelstein and her co-authors wrote.

In this 2016 paper, Finkelstein and her co-authors present an analysis of additional data to determine whether the increase in emergency department visits was short term.

There had been speculation that people who obtained Medicaid through the lottery would reduce use of emergency departments as they established relationships with primary care physicians.

At the 720-day mark, though, about 21.1% of the Medicaid group had visited an emergency department, while only 15.7% of the control group had, Finkelstein and her co-authors find.

Finkelstein and her co-authors found no such decline in use of emergency departments by people whose Medicaid enrollment should have made primary care visits more affordable.

Instead, “if anything, Medicaid made them more likely to use both,” Finkelstein and co-authors write in a summary of their results posted on the National Bureau of Economic Research (NBER) website.

The NBER’s Oregon Health Insurance Experiment website offers a list of publications based on this work. The MIT  Technology Review provides a history of this project in an April 2020 article, “A healthy understanding: Amy Finkelstein has changed what we know about Medicaid, Medicare, the economics of health care — and, increasingly, medical care itself.”

Additional reading

For more information, see our 6 reporting tips for covering Medicaid during the COVID-19 pandemic

Additional reporting by Naseem S. Miller

About The Author