Medicaid expansion for undocumented adults and its association with health insurance coverage among noncitizens in California, 2017-2023

Abstract

Introduction

California's 4.8 million noncitizen adults, half of whom are undocumented, endure substantial exclusions from health care. To address this gap, California policymakers expanded full-scope, state-funded Medicaid without discriminating by immigration status, first extending coverage to undocumented young adults (18-25 years) in 2020 and then to older adults (≥50 years) in 2022.

Methods

Using the American Community Survey (2017-2023), we assessed whether California's Medicaid expansion for young and older undocumented adults was associated with changes in insurance coverage by comparing pre- and post-expansion differences between citizens and noncitizens (difference-in-differences).

Results

Compared to citizens, the expansion was not associated with increased health insurance or Medicaid enrollment among young noncitizens. However, among older adults, the expansion was associated with a modest 1.3% increase in overall insurance coverage for noncitizens, including a 2.4% increase in Medicaid. Following these expansions, noncitizens remain significantly less likely to have insurance: 28% of young noncitizens and 16% of older noncitizens lack coverage, compared to 8% and 3% for young and older citizens, respectively.

Conclusion

Given these persistent inequities—where noncitizens across nearly all sociodemographic factors are less likely to be insured—preserving and strengthening the existing pathways to insurance coverage for noncitizens, including undocumented immigrants, remains critical.

Introduction

In total, 4.5 million adults in California are not US citizens.¹ Noncitizens without federal authorization to reside in the country, or undocumented immigrants (who account for half of all noncitizens),² have limited access to employer-based health insurance because they are often barred from formal employment.³ For nearly three decades, the Personal Responsibility and Work Opportunity Act has categorically excluded almost all undocumented immigrants, and many documented immigrants, from federally funded health insurance such as Medicare or Medicaid.^3,4 Access to federally funded insurance for documented immigrants depends on their specific visa and length of US residency.³ Taken together, it is evident that noncitizens are structurally excluded from health insurance. Consequently, only 65% of noncitizens in the United States are insured compared to 93% of citizens.⁵

States seeking to mitigate these federal exclusions have used their own funds to do so. For example, 14 states and D.C. offer full-scope (ie, comprehensive coverage, beyond solely emergency or pre-/post-partum care)⁶ Medicaid for undocumented children using their own funds.³ In 2020, California became the first state to provide full-scope Medicaid to undocumented adults by extending coverage to adults aged 19-25 regardless of their immigration status. This was followed by an expansion to older adults in 2022.⁷ However, years after these expansions, little is known about their impact on health insurance coverage among noncitizens. Evaluating California's Medicaid expansions for undocumented young and older adults is critically important, as these expansions paved the way for the state's 2024 policy extending coverage to all noncitizens previously excluded due to their immigration status.⁷

Here, we examine whether California's Medicaid expansion for undocumented young and older adults is associated with greater health insurance coverage among noncitizens. Despite policymakers’ concerns about unexpectedly high enrollment rates among noncitizens,⁸ we hypothesized that these expansions would result in modest increases in the number of noncitizens enrolled in Medicaid. Noncitizens have historically shown disproportionately low enrollment in safety net programs,^9,10 a pattern unlikely to reverse given the current climate of anti-immigrant rhetoric and policies.¹¹

Study data and methods

Data and sample

This study used the American Community Survey (ACS), an annual cross-sectional survey conducted by the Census Bureau.¹² The ACS is the largest household survey in the United States and represents 1% of the population.¹² Surveys are collected via the internet, mail, phone, or in person. Mail and internet surveys are available in both English and Spanish; in-person and phone surveys can be conducted in over 30 languages.¹³ We used data from 2017 to 2023 (obtained from IPUMS).¹⁴

We focused on young (19-25 years) and older adults (≥ 50 years). After excluding individuals with missing citizenship status (1.2%), our analytic sample included 226 571 young adults and 1 002 996 older adults, representing approximately 17 million individuals.

Measures

Citizenship status, the primary exposure, was assessed through questions on place of birth and naturalization and categorized as noncitizen or citizen.

We focus on the impact of Medicaid expansion on noncitizens, rather than on undocumented adults alone, due to methodological and policy considerations. Namely, the ACS does not contain self-reported documentation status, and because imputing this variable introduces serious methodological concerns (see eMethods), it is inappropriate for use as a main specification.^15,16

While this Medicaid expansion targets undocumented adults, our primary analysis, which uses citizenship status, is unlikely to miss substantial changes in health insurance coverage, should they truly occur. This is because (1) half of all noncitizens are undocumented, meaning the policy has the potential to impact half of the entire noncitizen population;² (2) the ACS (the largest survey of Californians)¹² grants us ample statistical power to examine differences and heterogeneous policy effects; and (3) noncitizens have low baseline rates of insurance coverage, which indicates this population has a high need for coverage and is not near a coverage ceiling.⁵

Focusing on the broader noncitizen group is also statistically advantageous because it allows us to examine pre-trends in the control and exposure groups (citizens vs noncitizens)—a necessity for our quasi-experimental approach. Before the Medicaid expansion, nonpregnant undocumented adults were categorically excluded from the program. This exclusion makes it impossible to observe pre-policy trends in Medicaid coverage among undocumented adults (as the data would contain only zeros), and, consequently, makes it challenging to assess policy effects.

While California has long offered state-funded Medicaid coverage for federally ineligible noncitizens with a legal status, its recent expansion to undocumented noncitizens may also increase enrollment among documented noncitizens who were previously entitled. Documented noncitizens eligible for California Medicaid report much lower enrollment rates than citizens due to multiple barriers, including logistical hurdles stemming from complex eligibility rules and fears associated with living in mixed-status households.^17,18 Therefore, our focus on all noncitizens, rather than exclusively on undocumented immigrants, is appropriate from a methodological and policy perspective.

Nevertheless, we utilized imputed documentation status to enable a sensitivity analysis evaluating whether the Medicaid expansion in question was associated with changes in health insurance coverage among documented and undocumented noncitizens. Noncitizens were considered documented if they reported:^19,20 arriving in the United States before 1980; being Cuban and arriving in the United States before 2017; being active-duty military personnel or veterans; having access to specific federally funded public programs (eg, Medicare, Supplemental Nutrition Assistance Program, Social Security); working the government sector or in occupations requiring licensing (eg, physicians, air traffic controllers, lawyers); or being the spouse of a US citizen or a person considered to be a documented noncitizen. All other noncitizens were considered undocumented. 48.5% of noncitizen adults in ACS were potentially undocumented (Appendix Table A), similar to the percentage reported by others.² Additional information on our imputation approach is available in our eMethods.

Outcomes included self-reported health insurance coverage, categorized as any coverage, Medicaid, or private insurance. Covariates included demographic (age, gender, race/ethnicity, marital status, and having a child) and economic (education attainment, current enrollment in school, employment status, and poverty-to-income ratio [based on family or personal income, depending on household composition]) factors. Immigration factors (years in the U.S., limited English proficiency, and region of birth) were used to assess inequities among noncitizens.

Statistical analysis

Descriptive statistics were used to characterize the population and evaluate trends and inequities in health insurance coverage. Differences were assessed using Kruskal–Wallis tests for continuous variables and χ² tests for categorical variables.

Difference-in-differences linear probability models were used to assess whether the Medicaid expansion for young and older undocumented adults was associated with changes (interpreted as percent-point changes) in health insurance coverage, by comparing pre- and post-expansion differences between citizens and noncitizens. Specifically, these models examined the outcome of health insurance and included terms for citizenship status, pre- and post-expansion periods, and their interaction (the difference-in-differences estimator). They were adjusted for demographic and economic factors, year of interview, and county of residence. Post-expansion periods were 2020-2023 for young adults and 2022-2023 for older adults.

Difference-in-difference models require pre-expansion trends to be parallel between the exposed (noncitizens) and nonexposed groups (citizens, the reference group). We visually evaluated this assumption by examining trends in health insurance coverage by citizenship status. The observed trends appear parallel, supporting this assumption (Appendix Figures A and B).

We conducted sensitivity analyses to assess the robustness of our findings. First, we restricted our difference-in-differences analysis to persons who were income-eligible for Medicaid (income <138% of the federal poverty line). Then, we replicated our main model after excluding the first year of Medicaid expansion for each age group (2020 for young adults; 2022 for older adults) to account for potential transition effects during the initial implementation. Finally, we used imputed documentation status to evaluate whether the Medicaid expansion increased health insurance coverage among documented and undocumented noncitizens (relative to citizens). We hypothesized that the Medicaid expansion would increase health insurance coverage for both documented and undocumented noncitizens, with a stronger effect on the latter.

All analyses incorporate sample weights that account for the differential probability of selection and nonresponse. P-values were 2-sided. Analyses were completed using R version 4.4.2.

This study relied on de-identified public data. Therefore, Institutional Review Board approval was not required (45 CFR × 46.102[f]).

Strengths and limitations

To the best of our knowledge, this is the first study to examine California's Medicaid expansion for undocumented young and older adults and its potential impact on health insurance coverage among noncitizens. We used the largest representative survey of California adults, which provided sufficient statistical power to evaluate differences across citizenship status. We also employed quasi-experimental methods with multiple robustness checks.

Despite these strengths, our study has limitations. First, the ACS, like all surveys that collect information across citizenship status, faces inherent risks of differential nonresponse and misclassification. Noncitizens, driven by fear and stigma related to their status, may disproportionately refuse to participate in the ACS or overreport naturalization. Second, we could not assess differences across documentation statuses, such as permanent and temporary visa holders and undocumented immigrants with tenuous protections; some may already have existing pathways to coverage through employer-sponsored insurance or earlier expansions of state-funded Medicaid. Together, these limitations may attenuate the observed disparities and reduce our ability to detect a policy impact. Yet, we observed substantial inequities in health insurance coverage between citizens and noncitizens, and our conclusions proved robust to a sensitivity analysis using imputed documentation status. We found similar changes (or lack thereof) in health insurance coverage after the Medicaid expansion among documented and undocumented noncitizens when each group was compared to citizens, mirroring the results from the overall noncitizen comparison group.

Third, our quasi-experimental approach requires parallel pre-trends. While our visual evaluation of pre-expansion insurance coverage trends suggests that the assumption of parallel trends holds, we were limited by the number of pre-expansion time periods available for confirmation. Nevertheless, our findings remain consistent across several sensitivity analyses.

Finally, these findings may not be generalizable to other states with similar expansions or to future contexts, given the increasing frequency of anti-immigrant rhetoric and policies in the U.S.

Results

In this California sample, 10.5% of young adults (Table 1) and 11.8% of older adults (Appendix Table B) were noncitizens. Compared to citizens, noncitizens were more likely to be people of color, have a child, have less than a high school education, and have lower incomes—patterns observed among both age groups (all P < 0.01). Among noncitizens, 20.6% of young adults and 54.2% of older adults had limited English proficiency, and the majority of both young and older noncitizens were born in Latin America.

Table 1.

Characteristics of young adults in California, by citizenship status, 2017-2023.

	% (95% CI)a			P c
	Overall (n = 226 571)	Citizenb (n = 203 920)	Noncitizen (n = 22 651)
Population, no. (%)	3.6 million (100.0)	3.3 million (89.5)	0.4 million (10.5)
Demographic factors
Age (years)d	22.1 (22.1-22.1)	22.0 (22.0-22.0)	22.3 (22.3-22.3)	<0.001
Women	48.4 (48.1-48.6)	48.5 (48.2-48.7)	47.5 (46.7-48.3)	0.03
Race/ethnicity				<0.001
NL-White	27.3 (27.1-27.5)	29.5 (29.2-29.7)	8.9 (8.4-9.4)
NL-Black	5.7 (5.5-5.8)	6.1 (6.0-6.3)	1.7 (1.5-2.0)
Latine	49.0 (48.8-49.3)	48.1 (47.8-48.4)	56.7 (55.9-57.5)
NL-API	13.2 (13.0-13.3)	11.1 (10.9-11.3)	30.6 (29.9-31.4)
NL-Other	4.9 (4.7-5.0)	5.2 (5.1-5.3)	2.0 (1.8-2.2)
Currently/previous married	8.7 (8.5-8.8)	7.8 (7.7-8.0)	16.0 (15.4-16.6)	<0.001
Has a child	6.8 (6.6-6.9)	6.2 (6.1-6.4)	11.4 (10.9-12.0)	<0.001
Economic factors
Education attainment				<0.001
≥High school	55.3 (55.1-55.6)	56.3 (56.0-56.5)	47.4 (46.6-48.2)
High school	40.0 (39.7-40.3)	40.1 (39.8-40.4)	39.1 (38.3-39.9)
<High school	4.7 (4.6-4.8)	3.6 (3.5-3.7)	13.6 (13.0-14.1)
Currently in school	45.1 (44.8-45.3)	45.5 (45.3-45.8)	41.2 (40.4-42.0)	<0.001
Employment status				<0.001
Employed	63.8 (63.5-64.0)	64.2 (64.0-64.5)	59.9 (59.1-60.7)
Unemployed	7.4 (7.2-7.5)	7.5 (7.4-7.7)	5.7 (5.4-6.1)
Not in labor force	28.9 (28.6-29.1)	28.2 (28.0-28.5)	34.4 (33.6-35.1)
Poverty-to-income ratio				<0.001
<138%e	24.7 (24.4-24.9)	23.4 (23.1-23.6)	35.6 (34.8-36.4)
139%-250%	21.3 (21.1-21.5)	20.8 (20.5-21.0)	26.2 (25.5-26.9)
251%-400%	21.8 (21.6-22.1)	22.0 (21.7-22.2)	20.8 (20.1-21.5)
>400%	32.2 (31.9-32.4)	33.9 (33.6-34.2)	17.4 (16.8-18.0)
Immigration factors f
Years in the US				-h
≥10	-h	-h	39.8 (39.0-40.6)
5-9	-h	-h	21.2 (20.5-21.8)
<5	-h	-h	39.1 (38.3-39.9)
Limited English proficiencyg	-h	-h	20.6 (19.9-21.2)	-h
Region of birth				-h
Asia	-h	-h	33.3 (32.6-34.1)
Latin America	-h	-h	57.8 (57.0-58.6)
Other	-h	-h	8.8 (8.4-9.3)

^aAmerican Community Survey data (2017-2023). Limited to young adults (ages 19-25) in California. Weighted estimates

^b95.1% of citizens are US-born, and 4.9% are naturalized.

^cAssessed using Kruskal–Wallis tests for continuous variables and Pearson's χ² tests for categorical variables.

^dMean value.

^eIncome-eligible for Medicaid. Persons may be eligible via other pathways.

^fOnly assessed among noncitizens.

^gSpeaks English less than “well.”

^hNonapplicable.

Abbreviations: API, Asian or Pacific Islander; CI, confidence interval; NL, non-Latine; US, United States.

Health insurance among young adults

Among young adults, noncitizens are less likely to be insured than citizens before (73.6% vs 90.6%) and after (74.5% vs 91.5%) California's Medicaid expansion for undocumented young adults (both P < 0.01 per unadjusted linear regressions) (Table 2). While noncitizens had slightly higher rates of Medicaid coverage in both pre-expansion (27.1% vs 25.3%) and post-expansion (30.1% vs 27.1%) periods, they had significantly lower rates of private insurance coverage (47.3% vs 67.1% and 46.3% vs 66.8%) (all P < 0.01). Medicaid expansion for undocumented young adults was not significantly associated with changes in health insurance coverage for noncitizens, relative to citizens (difference-in-differences estimate [DID] 0.7% [95% confidence interval {CI}: −0.7% to 2.2%]), including Medicaid (DID 1.2% [CI: −0.3% to 2.7%]) or private coverage (DID −0.1% [CI: −1.7% to 1.4%]).

Table 2.

Changes in health insurance coverage associated with California's Medicaid expansion for undocumented young and older adults, 2017-2023.

	% (95% CI)a
	Citizen			Noncitizend			Adjusted difference-in-differencec,e
	Insurance		Unadjusted differencec	Insurancec		Unadjusted differencec
	Pre-expansion	Post-expansionb		Pre-expansion	Post-expansionb
Young adults, no.	89 600	114 320		10 298	12 353
Any insurance	90.6 (90.4-90.9)	91.5 (91.3-91.7)	0.8 (0.5-1.2)**	73.6 (72.5-74.7)	74.5 (73.6-75.5)	0.9 (−0.5 to 2.4)	0.7 (−0.7 to 2.2)
Medicaid	25.3 (24.9-25.6)	27.1 (26.7-27.4)	1.8 (1.3-2.3)**	27.1 (26.1-28.2)	30.1 (29.0-31.1)	2.9 (1.4-4.4)**	1.2 (−0.3 to 2.7)
Private	67.1 (66.7-67.5)	66.8 (66.4-67.1)	−0.3 (−0.8 to 0.3)	47.3 (46.1-48.5)	46.3 (45.3-47.4)	−1.0 (−2.6 to 0.7)	−0.1 (−1.7 to 1.4)
Older adults, no.	627 489	276 951		67 149	31 407
Any insurance	96.8 (96.7-96.8)	97.4 (97.3-97.5)	0.6 (0.5-0.7)**	82.1 (81.8-82.5)	83.9 (83.4-84.4)	1.8 (1.2-2.4)**	1.3 (0.7-2.0)**
Medicaid	18.6 (18.4-18.7)	19.3 (19.1-19.5)	0.7 (0.5-1.0)**	35.3 (34.8-35.7)	37.1 (36.5-37.8)	1.9 (1.1-2.7)**	2.4 (1.6-3.2)**
Private	64.6 (64.4-64.7)	63.2 (62.9-63.4)	−1.4 (−1.7 to −1.2)**	38.6 (38.2-39.1)	38.7 (38.0-39.3)	0.0 (−0.7 to 0.8)	−0.3 (−1.0 to 0.5)

^aAmerican Community Survey data (2017-2023). Limited to young (ages 19-25) and older adults (≥ 50 years) in California.

^bCalifornia's expansion of Medicaid for noncitizens federally excluded from the program occurred in 2020 for young adults and 2022 for older adults.

^cAll reported differences are based on findings from unadjusted and adjusted linear probability model regressions.

^dAcross all insurance types, noncitizens had significantly different coverage rates than citizens in both pre- and post-expansion periods (unadjusted, all P < 0.01).

^eDifference-in-differences models adjusted for age, gender, race/ethnicity, marital status, having a child, education attainment, school enrollment, employment status, poverty-to-income ratio, year, and county of residence.

^** P < 0.01.

In 2023, the most recent year available, noncitizens had lower health coverage rates than citizens across nearly every demographic and economic factor evaluated, though the magnitude of these differences varied (Table 3). For example, among Latine young adults, 60.8% of noncitizens were insured compared to 89.8% of citizens, while the coverage gap was much smaller among Asian or Pacific Islanders (89.8% vs 94.7%) (both P < 0.01). The lowest health insurance coverage rates were observed among noncitizens with less than a high school education (41.5%), limited English proficiency (44.5%), and those born in Latin America (61.0%). 63.6% of potentially undocumented young adults were insured (Appendix Table A).

Table 3.

Health insurance coverage among young adults, by citizenship status and select factors, 2023.

	Health insurance coverage, %
		Citizen status		P b
	Overall	Citizen	Noncitizen
Total a	90.0 (89.6-90.5)	92.2 (91.8-92.6)	71.7 (69.6-73.6)	<0.001
Demographic factors
Gender
Men	88.5 (87.9-89.1)	91.2 (90.6-91.8)	67.6 (64.9-70.3)	<0.001
Women	91.7 (91.1-92.2)	93.3 (92.8-93.8)	76.6 (73.6-79.3)	<0.001
Race/ethnicity
NL-White	94.9 (94.2-95.5)	95.3 (94.6-95.9)	84.5 (77.9-89.5)	<0.001
NL-Black	91.7 (89.2-93.7)	92.0 (89.4-93.9)	82.0 (64.1-92.0)	0.06
Latine	86.1 (85.4-86.8)	89.8 (89.1-90.4)	60.8 (58.1-63.5)	<0.001
NL-API	93.7 (92.7-94.5)	94.7 (93.8-95.6)	89.8 (87.1-92.0)	<0.001
NL-other	92.9 (91.3-94.2)	93.3 (91.8-94.6)	83.4 (65.1-93.1)	0.04
Marital status
Currently/previous married	87.3 (85.4-88.9)	92.3 (90.8-93.7)	67.6 (61.9-72.8)	<0.001
Single/never married	90.3 (89.8-90.7)	92.2 (91.8-92.6)	72.4 (70.3-74.4)	<0.001
Have children
No	90.1 (89.7-90.6)	92.2 (91.7-92.6)	72.0 (69.9-74.0)	<0.001
Yes	88.6 (86.5-90.4)	93.4 (91.4-94.9)	68.1 (61.2-74.2)	<0.001
Economic factors
Education attainment
≥High school	93.6 (93.1-94.0)	94.3 (93.8-94.7)	85.9 (83.4-88.2)	<0.001
High school	88.0 (87.3-88.7)	90.0 (89.3-90.7)	70.1 (66.9-73.1)	<0.001
<High school	68.7 (65.7-71.6)	86.8 (84.0-89.2)	41.5 (36.5-46.7)	<0.001
Currently in school
No	87.4 (86.7-88.0)	90.8 (90.2-91.4)	62.7 (60.0-65.3)	<0.001
Yes	93.4 (92.9-93.9)	93.9 (93.4-94.4)	88.0 (85.4-90.2)	<0.001
Employment status
Employed	90.1 (89.6-90.6)	92.5 (92.0-93.0)	69.0 (66.4-71.5)	<0.001
Unemployed	86.8 (84.8-88.6)	88.9 (86.9-90.6)	66.0 (56.8-74.2)	<0.001
Not in labor force	90.7 (89.9-91.5)	92.5 (91.7-93.2)	78.0 (74.6-81.0)	<0.001
Poverty-to-income ratio
<138%c	89.2 (88.3-90.1)	91.3 (90.4-92.2)	76.8 (73.4-79.8)	<0.001
139%-250%	86.2 (85.1-87.2)	89.9 (88.8-90.8)	64.1 (60.0-68.1)	<0.001
251%-400%	88.3 (87.3-89.2)	91.2 (90.3-92.0)	62.9 (58.4-67.1)	<0.001
>400%	94.2 (93.6-94.8)	94.8 (94.2-95.4)	84.3 (80.2-87.7)	<0.001
Immigration factors d
Years in the US
≥10	-f	-f	78.7 (31.2-35.9)	-f
5-9	-f	-f	76.3 (25.4-29.9)	-f
1-4	-f	-f	64.0 (36.5-41.4)	-f
Limited English proficiency
No	-f	-f	82.1 (80.8-84.6)	-f
Yese	-f	-f	44.5 (15.4-19.2)	-f
Region of birth
Asia	-f	-f	89.2 (34.0-38.9)	-f
Latin America	-f	-f	61.0 (49.9-54.9)	-f
Other	-f	-f	86.8 (9.8-12.8)	-f

^aAmerican Community Survey data (2023). Limited to young adults (19-25 years) in California. Weighted estimates.

^bAssessed using Pearson's X² tests for categorical variables.

^cIncome-eligible for Medicaid. Persons may be eligible via other pathways.

^dOnly assessed among noncitizens.

^eSpeaks English less than “well.”

^fNonapplicable.

Abbreviations: API, Asian or Pacific Islander; CI, confidence interval; NL, non-Latine; US, United States.

Health insurance among older adults

Table 2 shows similar inequities among older adults, with noncitizens being less likely to be insured than citizens (82.1% vs 96.8%) and after the older adult expansion (83.9% vs 97.4%), both differences being statistically significant (P < 0.01). Noncitizens had higher rates of Medicaid coverage in both pre-expansion (35.3% vs 18.6%) and post-expansion (37.1% vs 19.3%) periods, but lower rates of private insurance coverage in both periods (38.6% vs 64.6% and 38.7% vs 63.2%, respectively) (all P < 0.01). For noncitizens relative to citizens, the older adult expansion was associated with modest increases in overall health insurance coverage (DID 1.3% [CI: 0.7%-2.0%]) and Medicaid specifically (DID 2.4% [CI: 1.6%-3.2%]) (both P < 0.01), but no influence on private insurance coverage (DID −0.3% [CI: −1.0% to −0.5%]).

Consistent with findings among young adults, in 2023, noncitizen older adults had lower health coverage rates than citizens across most demographic and economic factors examined (Table 4). 66.4% of potentially undocumented older adults were insured (Appendix Table A).

Table 4.

Health insurance coverage among older adults, by citizenship status and select factors, 2023.

	Health insurance coverage, %
		Citizen status		P b
	Overall	Citizen	Noncitizen
Total a	95.8 (95.7-96.0)	97.4 (97.3-97.5)	84.5 (83.8-85.2)	<0.001
Demographic factors
Gender
Men	95.2 (95.0-95.4)	97.1 (96.9-97.3)	82.2 (81.1-83.3)	<0.001
Women	96.4 (96.2-96.6)	97.7 (97.5-97.8)	86.8 (85.9-87.7)	<0.001
Race/ethnicity
NL-White	98.2 (98.0-98.3)	98.3 (98.2-98.4)	93.2 (91.3-94.7)	<0.001
NL-Black	96.6 (96.1-97.1)	96.7 (96.2-97.2)	92.2 (85.0-96.1)	0.01
Latine	91.2 (90.9-91.6)	95.4 (95.1-95.7)	80.8 (79.8-81.7)	<0.001
NL-API	97.1 (96.8-97.3)	98.0 (97.8-98.3)	91.9 (90.8-92.9)	<0.001
NL-other	96.2 (95.5-96.8)	96.4 (95.7-97.0)	90.5 (83.5-94.7)	0.001
Marital status
Currently/previous married	96.4 (96.2-96.5)	97.8 (97.7-97.9)	85.9 (85.2-86.6)	<0.001
Single/never married	91.9 (91.4-92.4)	94.5 (94.0-95.0)	75.7 (73.3-77.9)	<0.001
Have children
No	96.6 (96.4-96.7)	97.6 (97.5-97.8)	84.1 (82.9-85.3)	<0.001
Yes	94.6 (94.4-94.9)	97.0 (96.8-97.2)	84.8 (83.9-85.7)	<0.001
Economic factors
Education attainment
≥High school	98.0 (97.8-98.1)	98.4 (98.3-98.5)	90.5 (89.3-91.6)	<0.001
High school	95.1 (94.8-95.4)	96.5 (96.2-96.7)	84.6 (83.1-85.9)	<0.001
<High school	89.7 (89.2-90.2)	94.7 (94.2-95.1)	81.5 (80.4-82.6)	<0.001
Currently in school
No	95.8 (95.7-96.0)	97.4 (97.3-97.5)	84.5 (83.7-85.2)	<0.001
Yes	95.7 (94.6-96.5)	96.3 (95.3-97.1)	89.0 (83.1-93.1)	<0.001
Employment status
Employed	94.9 (94.7-95.2)	97.0 (96.8-97.1)	82.7 (81.6-83.7)	<0.001
Unemployed	88.7 (87.1-90.2)	91.4 (89.8-92.8)	76.5 (71.5-80.9)	<0.001
Not in labor force	96.9 (96.7-97.0)	98.0 (97.8-98.1)	87.3 (86.3-88.2)	<0.001
Poverty-to-income ratio
<138%c	93.4 (93.0-93.8)	95.1 (94.7-95.5)	84.6 (83.0-86.1)	<0.001
139%-250%	92.9 (92.5-93.4)	96.3 (95.9-96.6)	80.2 (78.6-81.7)	<0.001
251%-400%	94.5 (94.2-94.9)	96.8 (96.5-97.1)	83.0 (81.4-84.4)	<0.001
>400%	97.9 (97.8-98.1)	98.5 (98.4-98.6)	89.7 (88.6-90.8)	<0.001
Immigration factors d
Years in the US
≥10	-f	-f	84.7 (84.1-85.6)	-f
5-9	-f	-f	88.9 (7.1-8.2)	-f
1-4	-f	-f	78.7 (7.0-8.1)	-f
Limited English proficiency
No	-f	-f	86.8 (44.8-46.9)	-f
Yese	-f	-f	82.7 (53.1-55.2)	-f
Region of birth
Asia	-f	-f	92.0 (25.0-26.8)	-f
Latin America	-f	-f	80.8 (63.7-65.6)	-f
Other	-f	-f	93.2 (8.9-10.0)	-f

^aAmerican Community Survey data (2023). Limited to older adults (≥50 years) in California. Weighted estimates.

^bAssessed using Pearson's X² tests for categorical variables.

^cIncome-eligible for Medicaid. Persons may be eligible via other pathways.

^dOnly assessed among noncitizens.

^eSpeaks English less than “well.”

^fNonapplicable.

Abbreviations: API, Asian or Pacific Islander; CI, confidence interval; NL, non-Latine; US, United States.

Robustness checks

All findings remained consistent in robustness checks, including models restricted to income-eligible adults (Appendix Table C) and analyses excluding the first year of Medicaid expansion for each age group (Appendix Table D).

We also found similar effects across imputed documentation statuses (Appendix Table E). The Medicaid expansion for undocumented adults was not associated with changes in overall health insurance coverage among young noncitizens (relative to citizens), regardless of their documentation status. However, it was associated with a modest increase in coverage among older documented (DID 1.0% [CI: 0.4%-1.6%]) and undocumented noncitizens (DID 2.6% [CI: 1.4%-3.9%]).

Discussion

Despite California's 2020 Medicaid expansion to young undocumented adults, this representative study found no significant increase in Medicaid enrollment or improvement in overall health insurance coverage rates among young adult noncitizens, compared to citizens. Among older adults, who may be more likely to enroll in Medicaid when eligible due to increased health care needs associated with aging,²¹ we found that the 2022 Medicaid expansion was associated with a modest 1.3% increase in health insurance coverage, including a 2.4% increase in Medicaid enrollment. Following these expansions, almost one-third of young noncitizens and one-sixth of older noncitizens lack health insurance. In contrast, the uninsured rates for young and older citizens are only 8% and 3%, respectively. These inequities were observed across nearly every sociodemographic factor examined.

There were near-universal inequities among lower-income persons who were income-eligible for Medicaid: noncitizens, regardless of age, were less likely to have any health insurance, including both Medicaid and private coverage. Despite this, we found that the expansion for young adults was not associated with increased health insurance coverage for income-eligible noncitizens. The expansion for older adults was only modestly associated with increased coverage among income-eligible noncitizens. These findings suggest that eligible noncitizens may be unaware of or avoid enrolling in Medicaid.

The limited impact of California's Medicaid expansion is unsurprising given the challenging federal policy context. For example, in 2019, the Trump administration expanded the Public Charge Rule—which allows the government to deny visa applications to noncitizens who might become dependent on public benefits—to include enrollment in publicly-funded health insurance as grounds for such denials.⁹ Although the expanded Public Charge Rule was rescinded in 2022, research suggests it created a persistent chilling effect, with noncitizens continuing to forgo safety-net programs due to fears of immigration repercussions.⁹ Given planned expansions to the Public Charge Rule—which introduce ambiguity and grant officials greater discretion over safety net program considerations—these chilling effects are unlikely to subside.²²

Increasing federal threats to leverage public health insurance data for immigration enforcement and increased immigration enforcement in local communities are expected to deter more noncitizens from enrolling in Medicaid.^10,23 In 2025, the Department of Health and Human Services took the unprecedented step of sharing sensitive personal data of Medicaid beneficiaries with Immigration and Customs Enforcement (ICE) that could be used for deportation purposes.²⁴ Although a federal judge temporarily blocked the move,²⁵ the action will likely deepen fear and prevent noncitizens from using public insurance.²³

Recent immigration policies that strip tenuous immigration protections may make state-funded Medicaid one of the narrowing pathways through which undocumented immigrants can obtain health insurance in California. For example, the Trump administration has taken steps to end Temporary Protected Status (TPS) for individuals from seven countries, in addition to ending humanitarian parole protections for people covered by the Cuban, Haitian, Nicaraguan, and Venezuelan (CHNV) program.^26,27 Rescinding these protections, which provide temporary residency and work permits, will likely result in revoked work permits and decreased access to employer-sponsored insurance, significantly impacting health insurance coverage for noncitizens. Importantly, 46% of young and 39% of older noncitizens rely on private insurance.

The Deferred Action for Childhood Arrivals (DACA) program, which protects undocumented immigrants brought to the United States as children from deportation, is also threatened by ongoing litigation.²⁸ This litigation has halted the approval of new DACA applications since 2021.²⁸ Furthermore, outdated eligibility rules (mandating US arrival before 2007) mean that most undocumented high school graduates (including ∼27 000 per year in California) cannot obtain DACA protections.^28-30 This limited eligibility, coupled with the Trump administration's rescission of DACA recipients’ access to Affordable Care Act (ACA) marketplace subsidies,³¹ underscores the importance of state-funded Medicaid for young noncitizens.

The One Big Beautiful Bill Act (OBBBA), signed into law in 2025, will bring about sweeping changes to the eligibility of noncitizens for federally funded health insurance, placing the burden on states to ensure access to coverage.³² For example, in 2026, resettled refugees and persons granted asylum will be excluded from federally funded Medicaid, Medicare, and ACA marketplaces.³² Along with DACA recipients discussed earlier, noncitizens with TPS, valid temporary visas, and many other valid or pending statuses are now excluded from ACA marketplaces.³² The OBBA is expected to cause 1.4 million noncitizens to lose coverage by 2034.³²

As the federal government slashes pathways to immigration protections and further excludes those with lawful statuses from health insurance, states have followed suit by pausing insurance programs for federally excluded noncitizens. For instance, California, seen as a pioneer for immigrant-inclusive health policies, recently reversed its expansion of Medicaid for undocumented immigrants. Beginning in 2026, undocumented adults will be ineligible to enroll in California's Medicaid.²³ While those currently enrolled can keep their coverage (if they renew successfully), they will be required to pay a $30 monthly premium,²³ making them the only population subject to a Medicaid premium.

California policymakers attribute the cut to essential health care for undocumented immigrants to unexpectedly high enrollment. Yet, these anti-immigrant health policies are enacted as the state faces a projected $28.4 billion cut in federal Medicaid funds by 2034.³³ Therefore, it should be acknowledged that during budget shortfalls, the most politically vulnerable groups are often further marginalized to balance the state's finances. Illinois, Minnesota, and D.C. have also announced plans to scale back their full-scope Medicaid coverage for undocumented immigrants, leaving just four states (Colorado, New York, Oregon, and Washington) with such programs in place.³⁴

The fact that California's Medicaid expansion for undocumented immigrants had, at best, a modest impact on health insurance coverage rates among noncitizens means the program may have failed to address existing inequities fully. Noncitizens have poor health care access,^5,35 while simultaneously enduring worsening mortality inequities.^36,37 Therefore, policymakers interested in protecting the health of noncitizens should focus on preserving the existing Medicaid coverage for undocumented individuals and increasing overall enrollment. To achieve this, policymakers should consider introducing privacy protections that limit how states share information with the federal government and explore other options, such as state-funded tax subsidies for noncitizens excluded from the ACA marketplace.

Renewed local efforts are also necessary to ensure access to health care, even if it does not involve health insurance. For example, My Health LA, funded by Los Angeles County, previously provided comprehensive primary and specialty care to 140 000 patients who were ineligible for California Medicaid, primarily undocumented immigrants. The program was discontinued due to California's Medicaid expansion for undocumented immigrants, which itself is also ending.³⁸ At least 15 counties had similar programs before the statewide expansion.³⁹

California's temporary Medicaid expansion may have left many undocumented immigrants worse off, especially those who did not enroll before 2026 or are unable to renew their enrollment.²³ Several counties wound down established programs in anticipation of sustained state Medicaid coverage; now, with the enrollment pause, undocumented immigrants will have even fewer health care options than before. Although California law requires every county to have health care programs for indigent populations (Section 17000 of the California Welfare and Institutions Code), the decision of whether and how to provide care for excluded undocumented immigrant populations is up to each county's discretion.⁴⁰ State law does not mandate that counties cover undocumented immigrants, although many did so before 2024.³⁹ However, in an era of federal funding cuts affecting both citizens and noncitizens,³³ counties may find it difficult to allocate their limited funds to this increasingly marginalized population.

Conclusion

Years after California expanded Medicaid for young and older adults previously excluded due to their immigration status, more than one-in-five noncitizens remain uninsured. While California policymakers plan to pause Medicaid coverage for undocumented immigrants due to purportedly high enrollment rates, our findings suggest the opposite. These expansions had, at best, a modest impact on Medicaid enrollment. In an era of increasingly punitive federal policies, state policymakers should protect the few extant programs that provide health care without discriminating by citizenship and documentation status.

Supplementary material

Supplementary material is available at Health Affairs Scholar online.

Funding

This work was supported by the Robert Wood Johnson Foundation (RWJF), grant DOI https://doi.org/10.62700/qmhCqlQeHQ. Dr. Guadamuz also reports funding from the RWJF's Health Policy Scholars for Action program, separate from the study here. The funding sources had no role in the design and conduct of the study, analysis, or interpretation of the data; and preparation or final approval of the article before publication.