Children’s Health Insurance Coverage and Parental Immigration Status: 2015–2019

We used the American Community Survey (ACS), a nationally representative survey administered by the Census Bureau. The data, provided by the Integrated Public Use Microdata Series, includes 3.5 million respondents annually, and includes information about self-reported immigration status, health insurance status, family structure, and acculturation-related factors (eg, English proficiency or nativity).¹⁹  Our study sample consisted of children (age 0 to 18) in the United States between 2015 and 2019.

ACS asks respondents about immigration status for each person in the household, and individuals can report US citizenship by birth (born abroad by US citizen parents or born in the United States, Puerto Rico, Guam, the US Virgins, or Northern Marianas), US citizenship by naturalization, or not being a US citizen (which can include lawfully present and undocumented immigrants) (Supplemental Table 4). Because the ACS includes identifiers for intrafamilial relationships, we were able to identify parental immigration status for each child in the dataset. Building upon prior work, we created 4 mutually exclusive categories based on parental immigration status: citizen children with US-born parents, citizen children with a naturalized citizen parent, citizen children with a noncitizen parent (or children from mixed-status families), and noncitizen children.¹ 

We fitted linear probability models to estimate changes in uninsurance and Medicaid enrollment by parental immigration status. We divided our study period into 2 parts: before the Trump administration (2015–2016) versus during the Trump administration (2017–2019). We specified our model as an event-study to compare annual changes in outcomes to pooled baseline estimates (2015–2016), where all changes were measured in percentage point terms. Though data before 2015 was available, we selected the baseline years because 2014 coincides with the initial implementation of the ACA. Models included an indicator for parental immigration status, indicators for each year, and their interaction. Consistent with prior work, the models adjusted for child age, gender, and race and ethnicity; parental educational attainment; household employment status (2 full-time workers, 1 full time worker, only part-time workers, and only nonworkers), family structure (1 parent with children, 2 parents with children, multigenerational, and other), income measured as a percentage of the federal poverty level (FPL), speaking a language other than English at home, and state Medicaid expansion status.^16,20,21  Models also include state fixed effects. All analyses used Stata Version 15 (StataCorp, College Station, TX) and apply ACS survey weights.

First, we limited our main analyses to children in low-income households (<200% FPL), as these children would more likely be eligible for Medicaid coverage. We also stratified by residence in a Medicaid expansion state and presented changes in the 5 states with the largest populations of immigrants (California, Texas, Florida, New York, and New Jersey).²²  We tested whether the changes in health insurance coverage by parental immigration status differed for children from racial and ethnic minority groups, for children residing in states that expanded Medicaid, or for children residing in states that were democratically-led (where, in a given year, the democratic party held the governorship and majorities in the state senate and state house).²³  To do so, we reran our main analyses and included a 3-way-interaction between parental immigration status, period, and indicators for each characteristic (eg, parental immigration status x period x child race and ethnicity). We also conducted multiple sensitivity analyses: first, we pooled 2017 to 2019 estimates rather than modeling them as an event-study. Second, we assessed whether there were changes in private health insurance, as the study period coincided with economic growth nationwide. Third, we present changes in health insurance coverage for parents by immigration status. Fourth, we present nationwide estimates of our outcomes between 2011 and 2019 to further contextualize trends.

This study follows the Strengthening the Reporting of Observational Studies in Epidemiology reporting guidelines for cross-sectional studies. This study followed a prespecified analysis plan, and some posthoc stratified and sensitivity analyses were conducted to test the validity of our results.

Our study sample consisted of an annual average of 692 757 (standard deviation [SD] 6602) observations (representing approximately 64 million children each year), of which 76.7% were citizen children with US-born parents, 10.0% were citizen children with a naturalized citizen parent, 11.2% were citizen children with a noncitizen parent (or children from mixed-status families), and 2.1% were noncitizen children (Table 1). There was wide variation in sociodemographic characteristics of children (eg, age, race and ethnicity) and households (eg, parental educational attainment, household structure, language other than English spoken at home, and household income) by parental immigration status. We also present changes between 2015 and 2016 and 2017 and 2019, by parental citizenship status (Supplemental Tables 5–8).

Children from mixed-status families and noncitizen children had the highest uninsurance rates compared with citizen children with US-born parents and citizen children with a naturalized parent (Fig 1, Panel A). From 2015 to 2016, rates of uninsurance were stable for citizen children with US-born parents, citizen children with a naturalized parent, and children with a noncitizen parent. However, there were sharper declines in uninsurance among noncitizen children during this period. Beginning in 2017, there were statistically nonsignificant increases in uninsurance for citizen children with US-born parents (2015–2016: 3.6%, 2017–2019: 3.9%, difference = 0.3 percentage points [PP]) and citizen children with a naturalized citizen parent (2015–2016: 4.1%, 2017–2019: 4.3%, difference = 0.2 PP), and greater increases in uninsurance for citizen children with a noncitizen parent (2015–2016: 6.8%, 2017–2019: 7.9%, difference = 1.1 PP) (Table 2). For children from mixed-status families, this reflects an approximate increase of 95 700 uninsured children.

Rates of Medicaid enrollment were consistently higher for children from mixed-status families and noncitizen children compared with citizen children with US-born parents and citizen children with a naturalized parent (Fig 1, Panel B). Comparing the periods from 2015 to 2016 to 2017 to 2019, Medicaid enrollment rates were stable for citizen children with US-born parents (2015–2016: 31.8%, 2017–2019: 31.4%, difference = −0.4 PP) and citizen children with a naturalized parent (2015–2016: 30.0%, 2017–2019: 29.8%, difference = −0.2 PP) (Table 2); however, there were decreases in Medicaid enrollment among children from mixed-status families (2015–2016: 60.0%, 2017–2019: 57.2, difference = −2.8 PP) and noncitizen children (2015–2016: 38.8%, 2017–2019: 37.8%, difference = −1.0 PP). Although statistically nonsignificant, these estimates represent approximate decreases of 243 600 and 16 000 children, respectively, enrolled in Medicaid.

In adjusted models, there were significant increases in uninsurance among children from mixed-status families beginning in 2017 (0.48 PP, 95% confidence interval [CI]: 0.06 to 0.91) that increased to 1.48 PP (95% CI: 0.98 to 1.99) by 2019 when compared with concurrent trends among citizen children of US-born parents (Table 3). These changes were accompanied by significant decreases in Medicaid enrollment by 2019 (−0.89 PP, 95% CI: −1.62 to −0.16). Adjusted estimates in uninsurance among noncitizen children were comparable to unadjusted estimates, albeit attenuated in magnitude. Changes in uninsurance and Medicaid enrollment were not statistically significant for citizen children with a naturalized parent. Unadjusted estimates are presented in Supplemental Table 9.

When limiting our analysis to children living in low-income households, statistically significant changes in health insurance coverage and type were limited to 2019 for children from mixed-status families (Supplemental Table 10). Among children from mixed-status families and noncitizen children, rates of uninsurance were lower in Medicaid expansion states, though changes in health insurance coverage were similar in expansion and nonexpansion states (Supplemental Tables 11 and 12). There was wide state-level variation in health insurance coverage changes in states with larger populations of immigrants (Supplemental Table 13). Changes in health insurance coverage by parental immigration status differed for some characteristics (results not shown). There were statistically larger declines in Medicaid enrollment for Asian noncitizen children (P = .002 for 3-way interaction) and statistically larger increases in Medicaid enrollment for noncitizen children residing in Medicaid expansion states (P = .026 for 3-way interaction). Rates of uninsurance for noncitizen children increased in nondemocratically-led states, but decreased in democratically-led states (P <.001 for 3-way interaction). In nondemocratically-led states, rates of Medicaid enrollment declined for both noncitizen children (P = .002 for 3-way interaction) and children with a noncitizen parent (P = .002 for 3-way interaction), whereas remaining stable or increasing in democratically-led states. Estimates pooling 2017–2019 were consistent with our main findings in statistical significance and direction (Supplemental Table 9). In comparing changes in private health insurance coverage, we did not detect statistically significant differences in trends by parental immigration status (Supplemental Table 14). Analyses focusing on changes in health insurance coverage among parents indicate that, compared with US-born parents, uninsurance increased among parents who were naturalized citizens (Supplemental Tables 15 and Supplemental Tables 16). Lastly, to contextualize our findings and trends in our outcomes, we present trends beyond our study period both nationwide and by parental immigration status (Supplemental Fig 2).

Using a nationally representative sample of US children, we examined changes in uninsurance and Medicaid enrollment between 2015 and 2019, both overall and by parental immigration status. Throughout our study period, inequities in children’s health insurance coverage by parental immigration status persisted: although citizen children of US-born and naturalized parents reported lower rates of uninsurance (approximately 4%), rates for citizen children from mixed-status families were nearly twice as high (approximately 8%) and rates for noncitizen children were more than 5 times as high (approximately 22%), consistent with prior estimates.¹  Beginning in 2017, rates of uninsurance among children nationwide increased for the first time in nearly a decade. Relative to citizen children with US-born parents, rates of uninsurance among children from mixed-status families significantly increased between 2017 and 2019, with the magnitude of disparity widening over time. These changes coincided with statistically significant declines in Medicaid enrollment and no significant changes in private health insurance coverage. Although modest in percentage point terms, these changes indicate that thousands of children from mixed-status families became uninsured starting in 2017.

Several studies suggested that uninsurance rates were declining among both children and adults irrespective of immigration status before 2017, particularly in states that had expanded Medicaid eligibility as part of the ACA.^{16–18,20,21}  Our results are comparable to recent estimates for children’s uninsurance rates by parental immigration status,¹  though our estimates of uninsurance rates for noncitizen children are lower than 2008 to 2018 estimates using the National Health Interview Survey.³  Our findings build upon these studies first by extending these analyses into the Trump administration, during which several policy decisions could have eroded coverage gains based on immigration status. Although several studies have estimated increases in uninsurance for nonelderly adults, fewer have focused on children by parental immigration status.²⁴  Our findings support previously reported increases in children’s uninsurance, and specifically identified that the children with the greatest increases in uninsurance are US citizens with a noncitizen parent.^14,25 

There is extensive evidence that Trump administration immigration policies had negative effects on enrollment in public programs among immigrant families.^10,12,13,26  For example, by some estimates, between 20% to 25% of low-income immigrants reported avoiding enrollment in a public program following the announcement of the “public charge” rule proposal.^27,28  There were concerns that confusion about immigration policies, as well as immigration-related concerns, would have negative spillover effects to individuals who were eligible.^12,13,15  Our findings suggest that citizen children with noncitizen parents, who are eligible for Medicaid coverage, experienced the largest declines in Medicaid enrollment and increases in uninsurance. Loss of insurance and immigration-related concerns can impact utilization of care for children of immigrants: for example, in a study examining changes in appointment cancellations among uninsured Hispanic children in North Carolina, there was a sudden and sustained increase in cancellations after immigration-related executive orders were proposed and enacted.^{10,15,29–31}  Importantly, because our study period aligned with efforts to destabilize the ACA, our results cannot be attributed causally to immigration-related policies and rhetoric alone, and may also reflect concurrent changes in health policies. Another possible interpretation is that some of the changes in uninsurance were relatively small, and efforts from government and community-based workers perhaps mitigated more substantial losses of insurance coverage.

Further, the increases in uninsurance reported in our study period occurred during a time of economic growth. The coronavirus disease 2019 (COVID-19) pandemic and the subsequent economic recession led to widespread unemployment and loss of employer-sponsored health insurance. It is well documented that racial and ethnic minorities and immigrants, particularly Hispanic families with at least 1 noncitizen member, have been disproportionately affected by the pandemic, experiencing higher rates economic instability.³²  Many immigrant parents reported disenrolling themselves and their children in Medicaid because they feared that participation would jeopardize future immigration status.²⁸  Immigrant families reported avoiding necessary COVID-19 testing and treatment, even in the presence of symptoms, because of lack of insurance or immigration-related concerns.³³ 

Immigrant families face substantial structural barriers to accessing health insurance coverage and health care (eg, language barriers, cultural differences, and affordability).²¹  Our findings, if primarily driven by immigration-related policies and rhetoric, suggest the need for federal and state policymakers to identify strategies that prevent further erosion in health insurance coverage among immigrant families, particularly children from mixed-status families. For example, some states, such as Massachusetts and California, have instituted policies to make coverage options available for people regardless of immigration status.³⁰  In addition to extensions of eligibility, targeted outreach, multilingual and culturally inclusive enrollment assistance, and ongoing federal funding also facilitate enrollment in Medicaid.^21,28,31  Although the Biden administration reversed the Trump administration’s changes in March 2021, it has been reported that fear, mistrust in the health care system, and misinformation may remain among immigrant communities,^34,35  thereby impacting health insurance coverage and access for children of immigrants. Several reports indicate that engagement with communities to dispel misinformation, mitigate immigration-related fears, and communicate this reversal is needed to counteract erosion in Medicaid enrollment.^34,36  Beyond health insurance, advancing children’s health equity requires prioritization of other social factors (eg, wealth inequality, racism, xenophobia, neighborhood safety, and early childhood opportunities).

First, our analysis relies on self-reported health insurance coverage and immigration status, all of which may be subject to misreporting error, particularly for Medicaid coverage.³⁷  Second, concerns of providing identifying information may have reduced survey response rates among immigrants. Third, ACS is a repeated cross-sectional survey and does not longitudinally follow individuals over time, which would provide more individual-level information about history of health insurance coverage; however, it remains 1 of the few nationally representative surveys with information about immigration status and health insurance coverage. Fourth, our results cannot be attributed to Trump administration-related immigration policies, nor can our results be interpreted causally. There were concurrent changes in national health policy which could partially explain changes in our outcomes: for example, our study period coincided with several attempts to overturn the ACA and repeal the individual mandate; elimination of funding for ACA navigators; and other state-level policy decisions that may have influenced outreach and enrollment.³⁸  Fifth, variation in state-level changes suggests further investigating state-level Medicaid policies, immigration policies, and immigration-related rhetoric to disentangle the mechanisms influencing our findings.²  Sixth, there is likely heterogeneity by nationality, particularly among Hispanic and Asian American and Pacific Islander immigrants,^39,40  that we were unable to examine because of sample size. Lastly, 2019 is the most recent year of ACS data available, and it is possible that more pronounced effects were observed in 2020, when the “public charge” rule was enacted. The Census Bureau did not release its standard 2020 1-year estimates because of the impacts of the COVID-19 pandemic on data collection and we were unable to include 2020 data in our study.⁴¹ 

Children from mixed-status families and noncitizen children continue to face worse access to health insurance coverage compared with other children. Despite years of substantial reductions in children’s uninsurance across parental immigration statuses, we find that nationwide uninsurance rates increased beginning in 2017. Children from mixed-status families in particular experienced sharper increases in uninsurance, which coincided with decreases in Medicaid enrollment. Federal and state policymakers should consider health and immigration-related policies to expand access to care for immigrant children and prevent further erosion in health insurance coverage.

ACA

Affordable Care Act

ACS

American Community Survey

FPL

federal poverty level

PP

percentage points