States are mandated through the One Big Beautiful Bill Act to implement Medicaid work requirements by January 1, 2027. As states plan next steps to guide Medicaid enrollees through the new eligibility requirements, experts in substance use disorder, public policy, and health services research are raising concerns about the potential impact on people with opioid use disorder (OUD).
Medicaid covers 2 in 5 people with OUD and provides access to life-saving and affordable medications for OUD (MOUD). While individuals with substance use disorder initially qualified for a work exemption, the interim final rule reveals the condition must also “significantly impair their ability” to work. This shift in policy has elevated concerns that people with OUD will be at higher risk for becoming uninsured and cut off from treatment for their chronic medical condition.
Simulating New Work Requirements
Speaking at a virtual conference co-hosted with the University of Pennsylvania Leonard Davis Institute in March of 2026, CHERISH Population Data & Modeling Core Director Benjamin Linas warned that over 400,000 individuals with substance use disorder will likely lose coverage as a result of the mandated work requirement. At the same time, while Medicaid changes may reduce federal spending, overall healthcare spending will likely rise and shift the cost burden to hospital systems and funding pools that support free healthcare services.
Linas, who is an infectious diseases physician, professor of medicine and epidemiology, and a lead investigator at the Syndemics Lab at Boston Medical Center, arrived at the estimate by simulating different implementation strategies of the work requirements through the Researching Effective Strategies to Prevent Opioid Death (RESPOND) model.
Even after considering the best- and worst-case scenarios, the analysis projected significant declines in Medicaid coverage and, subsequently, increase in overdose deaths and costs. Administrative hurdles, such as self-reporting one’s chronic substance use or work activities, were a key driver of Medicaid disenrollment.
The modeling analysis prompted a lively conversation where conference attendees inquired about the simulation parameters and future research considerations. Linas’ responses to the questions, based on the modeling projections, are summarized below.
Questions About the Simulation Model
Could you give us rough estimates of the insurance losses in terms of millions of people affected, not just percentages? A group of us at the Brookings Institution estimated that about four million last year were at risk of being affected, using cruder methods and based on earlier interpretations of how the work requirements would be implemented. It would be interesting to compare.
We predict that about 435,450 individuals with substance use disorder on Medicaid in the U.S. will lose coverage. This number is devastating. The difference reflects that we are estimating coverage losses only among people who use opioids.
The policy implementation will increase the cycle of insurance enrollment and disenrollment and cause significant disruptions for people receiving treatment. Shelter-based providers already experience difficulties with enrollment verification. By requiring individuals to re-enroll every six months, this policy will increase the workload for providers, barriers to care, the risk of disenrollment, and the likelihood of overdose for patients.
Do you have a sense of the distributional impacts by age and income? By race?
Not yet. Our model can answer some of these questions, but we wanted to answer one big question quickly and then drill down.
For the RESPOND model, did you consider a status of receiving psychosocial/behavioral outpatient treatment (e.g., receiving group or intensive outpatient therapy but not on MOUD)?
No. The ONLY mechanism we modeled was loss of access to MOUD. We did not model loss of access to naloxone, lack of access to behavioral therapies, or loss of access to all other health care. So, this is a modest projection that likely underestimates impact.
Questions About Cost-Effectiveness Research
Given the anticipated Medicaid eligibility churn and work requirements, what evidence do we have, or need, on the cost-effectiveness of low-barrier substance use disorder treatment models (e.g., MOUD, telehealth, overdose prevention) in maintaining continuity of care and preventing downstream high-cost utilization?
There are a few great examples of studies of the cost-effectiveness of low-barrier SUD treatment models such as Avik Chatterjee’s paper on the outcomes of offering MOUD in homeless shelters or Joella Adams’ paper on the impact of MOUD initiation at syringe service programs. We would be happy to discuss further as is helpful.
We have known for a long time now that MOUD treatment is cost-effective, often cost saving even, and saves lives. Those data alone, however, cannot overcome the stigma related to substance use disorders. The economic case for providing low barrier treatment for people with OUD is necessary, but not sufficient for moving policy.
Given that formal cost-effectiveness evidence can be inconsistent across the SUD service menu, what frameworks are you using (or suggest we use) to operationalize what 'high value' is? How do we guard against the concept that everything does HAVE to be cost effective in a short/measurable time?
It is important to note that there is a difference between saying that something is “cost-effective” and saying that it is “cost-saving.” Cost-effectiveness analysis does not seek to save money. Quite the opposite, cost-effectiveness research begins with the presumption that we are going to spend all the money available. The question is how to spend that money so that we get the biggest possible health benefit.
Compare this to saying something is “cost-saving.” First, saying something is cost-saving nearly always requires the context of “for whom?” For example, H.R.1 itself is cost-saving from the perspective of Medicaid programs. Looking at the larger picture from the societal perspective, however, total spending on opioid use disorder will likely go up under H.R.1, not down.
At the bottom line, I would argue that everything we do should be cost-effective. If it is not, that implies that the money can be spent elsewhere to have greater impact on health outcomes. It’s important to challenge the assumption that everything needs to be cost-saving. We do not expect chemotherapy for colon cancer to save money; we expect it to provide health.
Questions About Policy Impact
Will this data inform policy makers to reconsider the cost of life that's going to happen with Medicaid reductions?
We can't answer that for certain, but we want to make sure these findings are integrated into the narrative of policy making. In a democracy, policy makers can and will make policies which we may not agree with. While we cannot control the outcome of their decision making, it is important that we clearly articulate the trade-offs. We can make the decision to reduce Medicaid spending, but we cannot pretend that no one will be affected by the change.
Have the findings from these simulations been made public?
We disseminated an early version of the findings on the LDI website. We scheduled an extension of today’s presentation at several meetings in the spring and are working hard on a peer-reviewed paper. We are eager to share the findings with real world policy makers and advocates.
Remaining Thoughts
People have not been talking about the role of stigma, mistrust--it's a lot to expect that people will reveal this [substance use disorder] information to the government if not already in claims histories. Lots of child welfare concerns. People may fear information leaking and affecting their employment (ironically), housing, child custody and other aspects of their lives.
We agree. We have received critiques from some colleagues that our estimate of the proportion who will be unable to navigate the burden of self-attestation, and therefore lose coverage, is too high, but we think it is an accurate number for this reason. Stigma deters individuals from disclosing information about their substance use, especially to the government. However, there are ways states can implement self-attestation to mitigate stigma by allowing for medical frailty to be a broader category that includes substance use disorder.