Quantifying the mortality and morbidity impact of medicaid work requirements: a modeling study

Summary

Background

Federal Medicaid work requirements, recently enacted under national legislation, are expected to cause substantial disenrollment among low-income adults. However, the public health consequences of these policies remain poorly quantified.

Methods

Extending our previously developed modeling framework, we projected national and state-level excess mortality and uncontrolled morbidity attributable to Medicaid coverage loss among expansion enrollees. Our three coverage loss scenarios were based on Congressional Budget Office projections and disenrollment patterns observed in Arkansas and New Hampshire. We incorporated administrative variation in automatic exemption capacity and reporting compliance. Mortality estimates used hazard ratios comparing insured and uninsured adults, while morbidity estimates focused on uncontrolled diabetes, hypertension, and high cholesterol.

Findings

We estimate between 7049 and 9252 excess deaths annually under the three national scenarios. We also project up to 113,607 additional cases of uncontrolled diabetes, 135,135 of hypertension, and 37,800 of high cholesterol. State-level excess deaths range from under 20 to over 2,100, with per-capita mortality highest in the District of Columbia and states such as New York and New Mexico. State performance varies widely based on administrative capacity. For example, North Carolina and Rhode Island would avert over 90 percent of deaths that could be avoided through stronger exemption systems, while Pennsylvania and South Dakota may only avert fewer than 30 percent.

Interpretation

Medicaid work requirements are projected to increase mortality and chronic disease burden across the United States. These findings highlight the role of state administrative capacity and underscore the public health cost of restricting insurance access.

Funding

Notsew Orm Sands Foundation.

Research in context.

Evidence before this study

Prior research has indicated that Medicaid work requirements would lead to significant coverage losses and have economic implications. These studies also show that most affected individuals are either already employed or face significant obstacles to finding work. Early implementations in Arkansas and New Hampshire resulted in considerable disenrollment, largely due to administrative challenges rather than failure to meet work activity requirements.

We searched PubMed and Embase (Ovid) from inception to May 2025 using the terms “work requirements” AND Medicaid AND (mortality OR morbidity), with no language restrictions. Of the studies identified, only our previous work estimated excess mortality under a proposed scenario modeled by the Congressional Budget Office. However, no prior studies have evaluated the health consequences of the version of Medicaid work requirements that was ultimately enacted into federal law. Additionally, beyond mortality, the excess burden of uncontrolled chronic diseases such as diabetes, hypertension, and high cholesterol resulting from coverage loss under Medicaid work requirements has not been previously quantified.

Added value of this study

This study provides the first comprehensive national and state-level estimates of both mortality and morbidity associated with Medicaid disenrollment under federal work requirements. We introduce a novel, state-specific modeling framework that captures real-world administrative variation in exemption capacity and reporting compliance. By integrating state-level data on Medicaid enrollment, administrative capacity, and historical disenrollment patterns, we generate granular projections of public health impact across diverse state contexts.

Implications of all the available evidence

Our findings suggest that Medicaid work requirements are likely to cause widespread health harm, including increased mortality and deteriorating chronic disease control. These effects are especially pronounced in states with limited administrative capacity to exempt eligible individuals or support manual compliance. As states prepare to implement these policies under newly enacted federal law, our analysis provides timely insight into how administrative differences may shape public health outcomes. Policymakers and state officials should consider these risks when designing and managing implementation strategies. Future research should assess the real-world effects of implementation and identify strategies that can mitigate harm, particularly in states with constrained administrative capacity.

Introduction

The recent passage of the One Big Beautiful Bill marked a significant shift in United States (US) health policy by codifying Medicaid work requirements into federal law.¹ These requirements, previously the subject of debate and limited state-level pilots, will now be implemented nationally, with profound implications for millions of Americans who rely on Medicaid for health coverage.^2,3 Medicaid, a joint federal and state program, provides health insurance to low-income individuals and families. The Affordable Care Act (ACA), through its Medicaid expansion provision, enabled states to broaden eligibility, significantly decreasing uninsured rates in participating states.

The new federal policy imposes work-related conditions on Medicaid eligibility that apply to individuals enrolled through the expansion program. Many of those who will be affected already face significant socioeconomic challenges and a disproportionate disease burden, raising concerns that coverage losses would exacerbate health issues for those most in need. Although these requirements are often framed as initiatives to promote self-sufficiency and reduce dependence on public assistance, they mandate that beneficiaries engage in a specified number of work hours, job training, or other qualifying activities to maintain their Medicaid coverage. Yet the majority of Medicaid enrollees already work,⁴ and the new policy introduces additional administrative burdens and overhead costs. Furthermore, the retractions will interrupt chronic care for many, potentially leading to more severe and costly health outcomes. Onerous reporting processes and administrative hurdles risk disenrollment even for those meeting work requirements or qualifying for exemptions. These challenges are compounded for enrollees with limited digital literacy, unstable work schedules, or difficulty navigating bureaucratic systems.⁵

Prior attempts at implementing work requirements in Arkansas and New Hampshire have proven to be counterproductive. In Arkansas, 25% of Medicaid enrollees lost coverage within the first six months. Of those, 89% remained unenrolled after a year, often citing confusion over reporting rules and difficulties accessing the online portal.⁶ Similarly, New Hampshire suspended work requirements when it became apparent that thousands of individuals were at risk of losing coverage due to reporting bureaucracy.⁷ These examples highlight the potential for rapid, widespread disenrollment with lasting health repercussions.

While earlier evaluations have examined the economic and employment effects of Medicaid work requirements, the public health consequences remain insufficiently understood. In this study, we estimate the national and state-level mortality consequences of Medicaid disenrollment under work requirements. Our analysis incorporates state-level variation in administrative capacity to identify individuals eligible for automatic exemptions and to support manual reporting or exemption processes, offering a granular view of how implementation differences may influence health outcomes under the new federal mandate. In addition to estimating excess deaths, we quantify the additional burden of uncontrolled chronic conditions—including diabetes, hypertension, and high cholesterol—that would result from loss of insurance coverage triggered specifically by Medicaid work requirements nationally.

Methods

We estimated the national and state-level mortality impacts, as well as national-level morbidity impacts, of Medicaid disenrollment resulting from the implementation of federal work requirements using an extension of our previously developed modeling framework.⁸ Coverage loss at the national level was modeled using three empirically grounded scenarios. The first scenario used the Congressional Budget Office (CBO) estimate of 4.8 million individuals losing Medicaid coverage under increased administrative barriers.⁹ We assumed a steady-state enrollment loss consistent with the CBO’s 2034 projection, reflecting full policy implementation. The second and third scenarios were based on Urban Institute projections informed by observed disenrollment patterns in Arkansas (5.5 million) and New Hampshire (6.3 million), two states that briefly implemented work requirements.¹⁰ These Urban Institute estimates account for automatic exemption of all enrollees who qualify based on existing data sources—such as indicators of caregiving status, Supplemental Nutrition Assistance Program (SNAP) participation, or earnings consistent with regular employment—as well as varying levels of non-compliance among individuals required to report or seek an exemption (72% in Arkansas, 82% in New Hampshire).

For each scenario, we updated the current insurance rate for adults aged 19–64¹¹ based on projected coverage loss and estimated annual excess mortality by applying a hazard ratio of 1.40, derived from a nationally representative cohort study comparing mortality risk between privately insured and uninsured adults¹² (Supplementary Appendix). To reflect uncertainty in our projections, we conducted a Monte Carlo simulation incorporating empirical variability in both the hazard ratio and insurance coverage. We generated 1000 samples to produce 95 percent uncertainty intervals around our estimates of excess deaths.

To estimate excess uncontrolled morbidity, we applied the same national coverage loss scenarios to three prevalent chronic conditions: diabetes, hypertension, and high cholesterol (Supplementary Appendix). For each condition, we applied disease-specific prevalence13, 14, 15 and baseline control rates^13,14,16 to the population losing insurance. We then estimated the number of additional cases of uncontrolled disease attributable to loss of Medicaid coverage by applying estimates of the change in disease control associated with losing health insurance. For diabetes, we used a relative risk of poor glycemic control among the uninsured compared to insured¹⁷; for hypertension and high cholesterol, we used adjusted differences in control rates from national survey data.¹⁸

Building on the Arkansas-based disenrollment scenario, we extended our analysis to estimate state-level variation in mortality burden. To capture differences in administrative capacity and policy design across states, we modeled three exemption scenarios under a consistent 72% non-compliance assumption: (i) a full exemption scenario, aligned with national assumptions in which all eligible individuals are automatically exempted; (ii) a limited exemption scenario, in which only caregivers are exempted; and (iii) an efficiency-based exemption scenario, where the share of enrollees automatically exempted varies by state based on its administrative capacity.

To implement the full and limited exemption scenarios, we used data from the Center on Budget and Policy Priorities¹⁹ to estimate the number of Medicaid expansion enrollees in each state and the proportion required to manually report (Supplementary Appendix). The efficiency-based exemption was then calculated by interpolating between these two scenarios. We used Medicaid ex parte renewal rates as a proxy for each state’s administrative capacity to identify exemptions and interpolated insurance loss accordingly.²⁰ Excess deaths were estimated using the same hazard ratio applied in the national model. Results are expressed in both absolute terms and per-capita rates. We also quantify each state’s performance in averting preventable mortality—defined as the gap between limited and full exemption scenarios—under its observed exemption efficiency (see Supplementary Appendix for details).

We conducted additional sensitivity analyses to explore a range of assumptions for automatic exemption capacity and manual reporting compliance at both national and state levels. We also performed a secondary analysis using a hazard ratio of 1.27, based on a recent quasi-experimental study²¹ that estimated the causal effect of Medicaid coverage on mortality among low-income adults, in order to assess the robustness of our national mortality estimates.

All data inputs, including population estimates, insurance coverage rates, mortality rates, and disease prevalence, were drawn from sources published between 2023 and 2025. Analyses were conducted using Python (version 3.9) with NumPy, pandas, geopandas and matplotlib libraries.

This study used only publicly available, de-identified data and did not require institutional review board (IRB) approval.

Role of the funding source

The funder of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report.

Results

Under newly enacted federal work requirements, we estimate that between 7049 and 9252 excess deaths could occur annually in the US, depending on the scale of Medicaid disenrollment. Using the CBO estimate of 4.8 million individuals losing coverage, we project a median of 7049 excess deaths (95% uncertainty interval [UI]: 1443–14,526). Under higher coverage loss scenarios informed by observed disenrollment patterns in Arkansas and New Hampshire, excess mortality increases to 8077 (95% UI: 1654–16,644) and 9252 (95% UI: 1894–19,065) deaths, respectively (Fig. 1A).

Fig. 1.

National estimates of excess mortality under Medicaid work requirements. A) Annual excess deaths among adults aged 19–64 under three disenrollment scenarios: CBO-based (4.8 M), Arkansas-based (5.5 M), and New Hampshire-based (6.3 M). B) Contour plot of excess mortality as a function of auto-exemption efficiency (the proportion of eligible individuals automatically exempted using administrative data) and reporting non-compliance (the proportion of individuals required to report who fail to do so). Real-world implementation points from Arkansas and New Hampshire are shown.

Excess mortality also increases substantially with lower automatic exemption efficiency and higher non-compliance rates. For example, with 60% exemption efficiency and a 75% non-compliance rate, projected deaths reach 10,120 (95% UI: 2072–20,855) (Fig. 1B).

To assess the sensitivity of our findings to the assumed relationship between insurance status and mortality, we also conducted a secondary analysis using a lower hazard ratio of 1.27, based on a recent quasi-experimental study of Medicaid expansion.²¹ Under this assumption, the median number of excess deaths ranges from 4777 under the CBO-based scenario to 6269 under the New Hampshire-based scenario.

We also estimated the burden of uncontrolled chronic disease associated with coverage loss. Across the three national scenarios, we project between 86,558 and 113,607 additional cases of uncontrolled diabetes, 102,960–135,135 cases of uncontrolled hypertension, and 28,800–37,800 cases of uncontrolled high cholesterol (Fig. 2). These increases reflect the expected rise in unmanaged chronic conditions resulting from reduced access to medications and clinical care.

Fig. 2.

National estimates of excess uncontrolled chronic disease cases under Medicaid work requirements. Projected increases in uncontrolled high cholesterol, diabetes, and hypertension among adults aged 19–64 under three disenrollment scenarios: CBO-based (4.8 M), Arkansas-based (5.5 M), and New Hampshire-based (6.3 M). Estimates reflect additional unmanaged cases attributable to insurance loss resulting from policy implementation.

At the state level, we estimated coverage loss and resulting mortality based on each state’s capacity to automatically exempt eligible individuals. We combined these with a fixed 72 percent non-compliance rate, informed by outcomes observed in Arkansas, to generate state-specific estimates of excess deaths. Under the efficiency-based scenario, which uses ex parte renewal rates as a proxy for exemption efficiency, the estimated number of excess deaths varied widely across states, ranging from fewer than 20 in smaller states such as North Dakota and South Dakota to more than 2100 in California (Fig. 3A). When standardized by population, excess deaths ranged from 2.3 to 16.7 per 100,000, with the highest per-capita burdens observed in the District of Columbia and in states such as New York and New Mexico (Fig. 3B).

Fig. 3.

State-level variation in excess mortality under Medicaid work requirements. A) Annual excess deaths among adults aged 19–64 by state under the efficiency-based scenario, which reflects current auto-exemption capacity using ex parte renewal rates. B) Excess deaths per 100,000 population under the same scenario. C) Proportion of preventable excess mortality averted in each state, calculated as the share of the difference between limited and full exemption scenarios that is averted under each state’s exemption efficiency. See Supplementary Table S1 and S2 for 95% uncertainty intervals corresponding to A) and B).

To assess how effectively states are mitigating preventable mortality, we calculated the proportion of excess deaths averted under each state’s current exemption performance. Preventable mortality was defined as the difference between excess deaths under the limited exemption scenario and the full exemption scenario. We then measured each state's performance as the share of this preventable mortality that is avoided under its observed efficiency. For example, Pennsylvania and South Dakota will avert fewer than 30 percent of preventable deaths, while North Carolina and Rhode Island avert more than 90 percent (Fig. 3C). These differences reflect substantial variation in administrative capacity to implement exemptions and limit unnecessary coverage loss. Additional state-level results across a wider range of exemption and compliance assumptions are presented in the Supplementary Appendix (Supplementary Figs. S1–S7).

Discussion

Our findings suggest that implementation of federal Medicaid work requirements could lead to substantial increases in both mortality and morbidity across the US. Depending on the scale of disenrollment, we project between 7000 and 9200 excess deaths annually among Medicaid expansion enrollees, along with tens of thousands of additional cases of uncontrolled diabetes, hypertension, and high cholesterol. These impacts represent significant avoidable harm, particularly for a population that already faces elevated rates of chronic illness and barriers to care.

The CBO estimates that 4.8 million individuals would lose Medicaid coverage under the enacted work requirement provisions. While this estimate reflects net coverage loss accounting for individuals who may gain private insurance, it does not incorporate the administrative challenges that typically accompany such policies.⁹ Specifically, the CBO assumes that all eligible individuals are automatically exempted using existing data systems and that reporting compliance proceeds without major disruption. This diverges sharply from observed experiences in Arkansas and New Hampshire.^4,22,23 As a result, the CBO projection likely reflects a conservative estimate of actual disenrollment.

Moreover, the final legislation imposes stricter criteria than the policy modeled by CBO. Whereas CBO’s analysis assumed an exemption for all parents of dependent children, the enacted law limits this exemption to parents or caregivers of children aged 13 or younger.²⁴ This change significantly expands the population subject to work requirements, particularly among parents with older children. In combination with the well-documented administrative burdens and procedural hurdles that amplify disenrollment beyond formal eligibility criteria, this suggests that our Arkansas- and New Hampshire-based scenarios, along with the accompanying sensitivity analyses, may more accurately reflect the potential scale of coverage loss. Therefore, our mortality and morbidity projections tied to these scenarios may offer a more realistic estimate of the public health consequences.

While our analysis uses the CBO’s 2034 projection of Medicaid disenrollment to estimate excess mortality and morbidity, we pair this with current demographic and epidemiological data, including baseline insurance coverage, mortality rates, and chronic disease prevalence. Given that forward-looking projections for these health indicators are limited, incorporating speculative assumptions about future trends could obscure the relationship between disenrollment and health outcomes. By using well-characterized, contemporary data, our estimates reflect the steady-state annual public health burden under full policy implementation, conditioned on current population health.

Depending on how the policy is implemented, the mortality impact could vary widely. We examined outcomes across states based on two key administrative factors: the capacity to automatically exempt eligible individuals using existing data sources, and the likelihood that individuals required to report will successfully comply. Our efficiency-based scenario incorporates state-specific ex parte renewal rates as a proxy for exemption capacity. While ex parte rates are not designed specifically for work requirements, they provide a reasonable, policy-relevant proxy for a state’s ability to match data and reduce paperwork burdens. States with lower exemption performance and high non-compliance are projected to experience substantially greater mortality. We also quantified preventable deaths by comparing limited and full exemption scenarios, and found that states like Pennsylvania and South Dakota currently avert fewer than 30 percent of preventable deaths, while North Carolina and Rhode Island avert over 90 percent. These findings highlight important administrative disparities. States with lower performance can reduce mortality significantly by improving their exemption systems and investing in outreach and support for compliance. Supplementary analyses show the expected mortality impact across a wide range of administrative assumptions and may help states prioritize areas for improvement.

We do not yet have individual-level data on the clinical characteristics of those who will lose Medicaid coverage under the new federal policy. Therefore, we assume an average health risk across the affected population. However, evidence from Arkansas and other early implementations suggests this assumption is conservative. Individuals who lost coverage were often not healthier, and were more likely to report unstable employment, limited digital access, and chronic health conditions—factors associated with higher mortality risk.^5,25,26

In our primary mortality model, we apply a hazard ratio of 1.40,¹² which compares mortality among uninsured adults to those with private insurance. While this comparison may overstate the risk associated with losing Medicaid, we apply the hazard ratio to a baseline insured mortality rate that includes both private and public coverage. This approach helps offset potential overestimation by grounding projections in a more representative insured population. To assess the robustness of our findings, we also conducted a secondary analysis using a lower hazard ratio of 1.27 from a quasi-experimental study.²¹ Although mortality estimates are lower under this assumption, the harm remains substantial, reinforcing the public health risks associated with coverage loss.

Our estimates may also understate the full health impact of work requirements. In addition to assuming average baseline risk, our projections do not account for spillover effects such as delays in care, reduced medication adherence, or disruptions to preventive services for individuals who face administrative friction but do not lose coverage. For morbidity estimates, we use age-stratified national prevalence and apply published differences in disease control between insured and uninsured adults. However, if those losing coverage have higher baseline prevalence of chronic conditions—as evidence from Arkansas suggests—our projections may understate the true burden of uncontrolled disease. Taken together, these estimates likely represent conservative lower bounds on the total public health burden.

While our morbidity estimates are based on nationally representative observational data demonstrating significant associations between insurance status and chronic disease control, the Oregon Health Insurance Experiment, a randomized evaluation of Medicaid expansion, did not find statistically significant improvements in hypertension or cholesterol control over a two-year follow-up period.²⁷ This discrepancy highlights the uncertainty in the causal relationship between insurance coverage and chronic disease outcomes. Compared to the Oregon study, which focused on newly insured adults in a single state with relatively short follow-up, our estimates reflect broader, population-level patterns and potential long-term effects. Nonetheless, these divergent findings are an important limitation and suggest that morbidity impacts may vary depending on the condition, population, and implementation context.

Compared to the Limit, Save, Grow Act, passed in 2023, which would have applied work requirements to adults aged 19–55,²⁸ the legislation enacted through the One Big Beautiful Bill substantially widens the affected population by extending these requirements to adults aged 55–64. While our results estimate mortality and morbidity across the full 19–64 age range, the health implications may be especially pronounced for adults aged 55–64, who face higher baseline mortality and a greater prevalence of chronic conditions.^29,30

Our findings also align with broader concerns about the cost-effectiveness of work requirement policies. For example, Georgia’s Pathways to Coverage program, launched in 2023, allocated over 90 percent of its $26 million budget to administrative overhead while enrolling fewer than 5 percent of its projected beneficiaries.^31,32 Such examples highlight the administrative complexity and inefficiency of these programs, which may undermine their stated objectives while introducing new health risks.

The passage of the One Big Beautiful Bill has transformed Medicaid work requirements from a policy proposal into a binding federal mandate. This shift underscores the urgency and relevance of our findings. As states prepare for implementation, our modeling framework provides timely insight into the likely mortality and chronic disease burden associated with different administrative strategies. By incorporating state-level data on exemption capacity and reporting compliance, this study offers a tool to guide implementation decisions, inform mitigation efforts, and support real-time evaluation of policy effects.

Contributors

AP and APG conceived the study. AP and YY developed the modeling framework. AP, YY, and CB conducted literature searches and collected data. AP conducted the analyses and drafted the manuscript. AP, YY, CB, BHS, and APG contributed to data analysis and interpretation. All authors reviewed and revised the manuscript for submission. AP and YY had full access to all the data in the study and verified the underlying data. APG had final responsibility for the decision to submit for publication.

Data sharing statement

All data used in this study are publicly available and documented in the Supplementary Appendix. The Python code used for the modeling analysis, along with all input datasets and results, is available at: https://github.com/abhiganit/wr4medicaid.

Editor note

The Lancet Group takes a neutral position with respect to territorial claims in published maps and institutional affiliations.

Declaration of interests

We declare no competing interests.