Including Undocumented Immigrants in Health Research—A Narrative Review of Barriers, Effective Approaches, and Best Practices in the United States
ABSTRACT
Background and Aims
Undocumented immigrants (UDIs) in the United States are an understudied population with unique health needs. Half are uninsured, with state-to-state variation. Creative methodologies have been used to approximate UDIs in health data by geography and scope. However, no review exists of UDI health coverage or methods used to study this population. We conducted a narrative review defining health coverage options by state and research methodologies for UDIs.
Methods
We conducted a narrative review of the literature to answer two research objectives: (i) what health coverage is available to UDIs at the federal and state level, and (ii) what methodologies have been used by health services researchers to study UDI health. First, we reviewed Medicaid websites, gray literature, and legislative briefs to summarize federal and state-level health coverage for UDIs. We categorized states by care availability into three groups: “available,” “limited,” and “restricted.” We then conducted a formal literature search on health research among UDIs for every state, using gray literature and articles from PubMed and Google Scholar. Data was extracted to describe study characteristics, data type, methods for approximating or identifying UDI data, and scope. Total number of health studies by state was grouped according to their care availability status.
Results
UDI health coverage was determined for all 50 states and Washington, DC: three states and Washington, DC were considered “available,” 28 were “limited,” providing care to special UDI populations, and 19 were “restricted” with no coverage options. Thirty-seven articles on UDI health coverage were included in this study. Methodologies to study UDI patients were unstandardized. Most studies were single-center, retrospective, or qualitative. Creative methods were used to approximate UDI health data, including linking social services datasets and using Emergency Medicaid claims. Large-scale data set studies were rare, but California's restricted Medi-Cal demonstrated successful use of claims data for UDI research. Included research articles were categorized by state coverage and an average number of studies produced per state in that category: available (M = 3.0), limited (M = 0.9), and restricted (M = 0.4).
Conclusions
From this narrative review on health coverage and health studies on UDI patients, we found variability in health coverage, unstandardized methods to approximate UDI patients, and creative and diverse methodologies to study UDI populations, with varied degrees of accessibility and accuracy for approximating this population. Understanding these approaches can better inform health researchers when studying the health of UDIs to apply these methodologies to their own contexts and research questions.
Abbreviations
-
- ACA
-
- Affordable Care Act
-
- CHIP
-
- Children's Health Insurance Program
-
- CHIS
-
- California Health Inpatient Survey
-
- EMTALA
-
- Emergency Medical Treatment and Active Labor Act
-
- FQHC
-
- Federally Qualified Health Center
-
- PRISMA
-
- Preferred Reporting Items for Systematic reviews and Meta-Analyses
-
- RMA
-
- Refugee Medical Assistance
-
- UDI
-
- undocumented immigrants
-
- UNOS
-
- United Network for Organ Sharing
-
- US
-
- United States
1 Introduction
Undocumented immigrants (UDIs) in the United States are individuals who lack documentation to permit legal entry, work, or residence where they reside, and often gain entry through unregulated crossings at ports of entry and overstaying visas [1]. There are an estimated 12 million UDIs in the United States, comprising 4% of the population and growing annually [2]. UDIs face challenges in accessing care in the United States [3]. They are the largest group of uninsured individuals in the country, with an estimated 46%–71% without coverage, double that of lawfully present immigrants and 6-times greater than citizens [4]. Challenges in accessing healthcare for UDIs extend beyond insurance and include low socioeconomic status, fear of deportation, lack of health system knowledge, language barriers, and discrimination [3]. While some states allocate state-specific funding to offer select health services, including through expanded Medicaid, policies are often changing, difficult to navigate, and state dependent [5]. Understanding how state by state health coverage differences may impact care for UDIs is important not only for undocumented persons' care but also promoting the three aims of US health care: improving the patient experience of care, improving the health of populations, and reducing the per capita cost of healthcare [6].
Health services researchers may have more limited tools to accurately study the health needs of UDIs [3]. A key approach to studying population health utilizes claims databases: records of patient encounter bills from public and private insurance providers [7]. These data document demographic, clinical encounter, diagnostic, and treatment variables, which are critical to understanding health delivery at the population level. Given that many UDIs lack insurance and claims databases do not distinguish UDI patients, claims data have not been used as a reliable source to study epidemiologic and health access trends in UDI patients, as it can be for other populations with distinguishable health characteristics [7]. To overcome the major barriers which prevent the systematic study of UDI patients, researchers have utilized innovative methodologies to identify and approximate undocumented patients. However, these various methods have not been critically reviewed, nor studied in relation to state insurance policies.
We conducted a narrative review of health coverage options and health studies among UDIs by state to highlight successful methods for studying UDIs. Health services researchers may benefit from studying the varied research approaches presented in this narrative review, to know more possible methodologies to study UDI populations, particularly where data availability is scarce. This study does not evaluate whether variations in insurance or state policies influence the number of published studies on UDI health. Application of these creative methodologies could be useful to better understand the health needs, disease burdens, and system barriers affecting this population.
2 Methods
2.1 Determining Health Coverage by State
To determine specific health coverage for UDIs by state, we analyzed official Medicaid Handbooks and websites for all 50 states and Washington, DC to determine eligibility and coverage offered for UDIs. Additionally, we conducted analyzed state websites for legislation on health coverage for UDIs. When indeterminate, we searched Google Scholar and gray literature to understand state-based health coverage for UDIs, anticipatory legislation, and briefs not otherwise available (Supporting Information 1). All health coverage options were differentiated by state, year enacted, and service offered.
2.2 Categorization
States were categorized according to health coverage options into three groups: “available,” “limited,” and “restricted.” These three categories are novel constructs developed solely for this study, as no previous framework differentiated levels of healthcare access for this population. In state categorized as “available,” individuals have access to a similar scope of health coverage as beneficiaries of Medicaid-funded plans in that state, which—while varying by jurisdiction—generally encompasses annual preventive health appointments, routine immunizations, and insurance coverage for HIV antiretrovirals as well as medications for noncommunicable diseases. Since coverage specifics differ from state to state, we do not catalog every service; rather, we anchor our “available” designation in the principle that the suite of covered services mirrors Medicaid eligibility. States in the “limited” tier provide publicly funded coverage only to defined subgroups, most commonly children and pregnant people, and generally restrict benefits to treatments or programs tied to those populations' specific health needs. Finally, “restricted” states offer no state-level coverage beyond federally mandated emergency services such as Emergency Medical Treatment and Active Labor Act (EMTALA).
2.3 Search Strategy and Study Selection
For the second portion of our review, we conducted a literature review search to gather information on different methodologies used in the study of UDI health. Our search included studies from January 1995 to October 2023 on UDI health, with this timeframe of the last nearly 30 years being selected to provide both robust and actualized evidence on UDI research. We searched for relevant articles on PubMed and Google Scholar with the key search terms of: “undocumented immigrant,” “health research,” “unlawfully present,” “health access,” “refugee,” “immigrant,” and “Deferred Action for Childhood Arrivals (DACA),” individually for each state and Washington, DC. Our initial search revealed 83 articles (Figure 1). The search strategy and list of articles reviewed at the full-text level are available in Supporting Information 2.
These articles underwent independent assessment by two reviewers at the abstract/title, followed by full-text level. Inclusion criteria included original research articles from the United States with complete or partial UDI samples, written in English or Spanish, and with clear definition for how UDI patients were defined or approximated. Articles were excluded if they lacked UDIs or health data. Disagreements which arose between the reviewers were mediated by discussion and, when necessary, independent decision by a third reviewer [8].
2.4 Data Extraction and Analysis
Data were extracted from included papers by two independent reviewers using a data extraction tool developed by the research team (Supporting Information 3). This extraction included study details on data type (quantitative vs. qualitative), state, scope including institution and participant number, and method for identifying or approximating UDI patients undertaken by study investigators. Studies were grouped at the first level by whether they identified or approximated UDI samples, and second by type of data being presented (quantitative, qualitative, or machine learning predictions). Finally, state availability of health coverage for UDIs was compared with the number of research studies from those states. This metric was evaluated by calculating a state-quantity-weighted average for each state category by dividing the number of studies conducted within states in that category by the total number of states in that category, to evaluate a distribution of which states were publishing studies on UDI health.
3 Results
3.1 Unique UDI Health Coverage Barriers Compared to Other Immigrant Populations
With regard to healthcare access, UDIs are more limited in their health options than other immigrant populations (i.e., refugees, asylum seekers). Refugees are eligible for Refugee Medical Assistance (RMA), which provides Medicaid-like coverage for up to 8 months upon arrival to the United States [9]. For refugees living in the 41 states with expanded Medicaid, they are also Medicaid-eligible through state funds [10]. Provisions exist for special populations, including Afghan refugees, who were deemed eligible for Medicaid or Children's Health Insurance Program (CHIP) without a 5-year waiting period [11]. CHIP is available to most asylum seekers and refugees and provides low-cost health coverage to children in families that earn too much to qualify for Medicaid but not enough for private insurance. In some states, CHIP also covers pregnant women [11]. Further, the Patient Protection and Affordable Care Act (ACA) of 2010 increased health insurance access for asylum seekers and refugees, but explicitly excluded UDIs [12]. UDIs are unable to access most federally-provided aid due largely to the Personal Responsibility and Work Opportunity Act of 1996, including Medicare, Medicaid, CHIP, or ACA marketplace plans [4]. Barriers exist for private insurance, as most employee-sponsored healthcare plans require documentation [13]. UDIs may purchase low cost-benefit private insurance if reliable healthcare coverage is desired, but many rely on charity care, safety net hospitals, or paying out-of-pocket [14]. Finally, patients qualifying as DACA status, though considered undocumented, have specific national-level health options due to the ACA. As of 2024, DACA recipients can enroll in the ACA marketplace, including the health plans, subsidies, and financial assistance which it affords [15].
3.2 Federal-Level Health Coverage
3.2.1 Emergent Coverage
Beginning in 1986, the EMTALA required hospitals to address emergency health needs of all individuals, regardless of immigration status or ability to pay [16]. EMTALA is a foundational, nondiscriminatory law that is federally enforced, requiring hospitals to accept and stabilize patients in crisis [16]. Related, Emergency Medicaid provides hospital reimbursement for centers caring for patients with medical emergencies who would qualify for Medicaid if not for their immigration status [17]. The program accounts for approximately $2 billion dollars in spending annually among 100,000 individuals [17]. However, “emergency” classification is discretionary by state, with wide-ranging interpretations including chemotherapy and radiation in New York to solely emergency stabilization in Florida [17].
3.2.2 Federally Qualified Health Centers (FQHCs)
FQHCs are a valuable healthcare option for UDIs. Under Section 330 of the Public Health Service Act, FQHCs were created to be sliding-fee medical centers that operate in underserved areas by providing comprehensive services, regardless of residency or insurance [18]. Today, over 1400 centers serve more than 20 million patients, 38% of whom are uninsured [19]. However, differences exist at the state level that complicate FQHC access. For example, among 69 FQHCs in Texas, only one offers care to noncitizen, uninsured patients [20]. Further, specialty care for UDIs at FQHCs varies by state, with some being labeled as nonexistent [21].
3.3 State-Level Health Coverage
3.3.1 “Available” State Coverage
As of 2024, California, Massachusetts, Oregon, and Washington, DC provided comprehensive (Medicaid-like) health coverage to UDIs (Table 1, Supporting Information 4).
| State | Policy | Year | Explanation |
|---|---|---|---|
| Available (n = 4) | |||
| California | Adult coverage | 2024 | State-funded Medi-Cal coverage to all income-eligible UDI individuals regardless of age |
| Pediatric coverage | 2016 | State-funded coverage to all income-eligible children up to 26, CHIP | |
| Pregnancy and postpartum coverage | 2021 | Prenatal care via CHIP unborn child option and state funded postpartum coverage to 12 months | |
| End stage renal disease (ESRD) | 1996 | Maintenance dialysis and transplant | |
| Massachusetts | Adult coverage | 1997 | MassHealth Limited for > 18-year-old, Health Safety Net. Not comprehensive, emergency services, and some outpatient coverage |
| Pediatric coverage | 1993, 1997 | Children's Health Insurance Program (CHIP), children's medical security plan, primary, and preventative services | |
| Pregnancy and postpartum coverage | 2022 | Prenatal via CHIP unborn child option 12 months postpartum coverage MassHealth Standard via CHIP extension and state funding |
|
| ESRD | 1997 | Maintenance dialysis Kidney transplantation |
|
| Oregon | Adult coverage | 2023 | State funded Cover all People Act, Medicaid-like coverage |
| Pediatric coverage | 2018 | State funded, Cover All Kids, Medicaid-like coverage to all low-income children | |
| Pregnancy and postpartum coverage | — | Prenatal coverage via CHIP unborn child option 12 months postpartum, state funded |
|
| ESRD | 2021 | Maintenance dialysis | |
| Washington, DC | Adult coverage | 1999 | State funded, Healthcare Alliance program equivalent to federal Medicaid |
| Pediatric coverage | 2021 | State funded Immigrant Children's Program, coverage to all income-eligible children age 20 and under | |
| Pregnancy and postpartum coverage | 2022 | Prenatal coverage via CHIP unborn child option 12 months postpartum, state funded |
|
| ESRD | Before 2019 | Maintenance dialysis | |
| Limited (n = 28) | |||
| Colorado | Adult coverage | 2023 | May purchase individual coverage with state subsidies, discounted health services through the Colorado Indigent Care Program |
| Pediatric coverage | 2025 | Children < 19 eligible for Medicaid coverage, CHIP equivalent | |
| Pregnancy and postpartum coverage | 2025 | Prenatal via state funded Medicaid coverage 12 months postpartum, state funded | |
| ESRD | 2019 | Maintenance dialysis | |
| Illinois | Adult coverage | 2020 | Coverage for all low-income adults ≥ 42 years, though the program is paused (ages 42–64) and enrollment capped (age 65+) |
| Pediatric coverage | 2022 | State-funded, All Kids Program, coverage to all income-eligible children | |
| Pregnancy and postpartum coverage | 2021 | Prenatal care via CHIP unborn child 12 months postpartum coverage through CHIP health services initiative amendments |
|
| ESRD | 2014 | Maintenance dialysis and transplant | |
| Minnesota | Adult coverage | 2025 | Plan to provide full coverage to all low-income individuals regardless of immigration status- MinnesotaCare |
| Pregnancy and postpartum coverage | 2011 | Prenatal coverage via CHIP unborn child option 12 months postpartum coverage, state funded |
|
| ESRD | Before 2017 | Maintenance dialysis Kidney transplant |
|
| New York | Adults | 2024 | Adults 65+ Medicaid equivalent coverage |
| Pediatric coverage | 1990 | State funded Child Health Plus Program | |
| Pregnancy and postpartum coverage | 2001 | Prenatal coverage, state funded 12 months postpartum, state-funded medicaid expansion |
|
| ESRD | 2011 | Maintenance dialysis | |
| Vermont | Pediatric coverage | 2021 | State funded full coverage Immigrant Health Insurance Plan |
| Pregnancy and postpartum coverage | 2021 | Prenatal and 12 months postpartum via State funded Immigrant Health Insurance Plan | |
| Washington | Adult coverage | 2024 | Alien Emergency Medical Program and eligible to purchase health insurance on Washington Healthplanfinder with subsidies, via state innovation waiver |
| Pediatric coverage | 2007 | State funded Apple Health, Medicaid-like coverage for all low-income children | |
| Pregnancy and postpartum coverage | 2021 | Prenatal care Apple Health via CHIP unborn child option 12 months postpartum, state funded After Pregnancy Care |
|
| ESRD | 2015 | Maintenance dialysis | |
| Restricted (n = 19) | |||
| Includes all remaining states, see Supplemental material 2 | |||
- Note: This table demonstrates selected examples of health coverage insurance options for undocumented immigrants in the United States by state. A complete list including all 50 states and Washington, DC is available in Supplemental material 2. Health coverage policy dates which were not identifiable were indicated with a hyphen.
In California, UDIs can access various forms of state-funded Medicaid (Medi-Cal) depending on age. Restricted scope Medi-Cal is the traditional health coverage option for UDIs, allowing access to primary care services, pregnancy coverage, and emergency care [22]. Since 2020, California has expanded full scope Medi-Cal to UDIs aged 50 and older, and in January 2024, California expanded full-scope Medicaid to all income-eligible residents, regardless of immigration status.
Massachusetts offers MassHealth Limited and Health Safety Net to all UDIs. With restricted options compared to Medicaid, these policies cover inpatient hospitalizations, emergency room visits, pharmacy services, and preventive services when provided by a community center or acute care hospital [23]. Beginning July 1, 2023, Oregon enacted the Cover All People Act through a program called “Healthier Oregon,” which extends health plan benefits to all income-eligible adults regardless of immigration status [24]. Since 1999, Washington, DC's Healthcare Alliance program has provided primary care coverage to all low-income individuals regardless of immigration status [25]. A locally funded program, Healthcare Alliance, provides access to primary, dental, and emergency and urgent care to all Washington, DC residents who otherwise do not qualify for Medicaid.
3.3.2 “Limited” State Coverage
Twenty-eight states provide coverage for specific UDI populations, particularly children and pregnant patients. Connecticut, Illinois, Maine, New Jersey, New York, Oregon, Vermont, and Washington all offered CHIP or CHIP-like coverage to children of a certain age, regardless of immigration status [22]. Colorado and Washington residents who do not qualify for ACA coverage, including UDIs, can apply for state-funded subsidies to purchase private insurance, though it is unclear how often these waivers are granted [26]. Colorado and Washington both offer marketplace like private health insurance programs that allow UDIs to sign up for private health insurance off the marketplace exchange [27]. Twenty states in this category offered outpatient hemodialysis and for 11 of these states, this was the only health service option offered to UDIs [28]. A select number of states in the limited category have indicated intent to expand coverage options, which could move states to the available category if enacted. For example, in 2025, Colorado implemented Medicaid coverage to undocumented pregnant people and children who would otherwise qualify.
3.3.3 “Restricted” State Coverage
Nineteen states fell into the restricted category, offering no health coverage to UDIs. Undocumented persons living in these areas would only have access to health coverage through private or federal options including FQHCs and Emergency Medicaid. Geographically, these states were evenly distributed across the country, including two in the Northeast, seven in the South, four in the Midwest, and six in the West (Figure 2).
3.4 Research Approaches to UDI Studies
Despite a lack of traditional avenues for researching healthcare outcomes of UDIs, several creative approaches have been developed to directly study or approximate UDI patients. We outlined the most effective approaches, categorized by data type (qualitative vs. quantitative) and whether researchers directly studied UDIs or used a proxy measure to approximate the population (Table 2).
| Research perfectly defining undocumented immigrants | ||||||
|---|---|---|---|---|---|---|
| Authors | Study | State/region | Study type | Scope | Participant number | Identification method |
| Quantitative (n = 10) | ||||||
| Ro et al. | Severity of inpatient hospitalizations among undocumented immigrants and Medi-Cal patients in a Los Angeles, California, Hospital: 2019 | California | Retrospective review | Single institution | 8000 | Restricted scope Medi-Cal insurance |
| Ro et al. | Emergency department utilization among undocumented Latino patients during the COVID-19 pandemic | California | Retrospective review | Single institution | 77,283 | Restricted scope Medi-Cal insurance |
| Cervantes et al. | Association of emergency-only versus standard hemodialysis with mortality and health care use among undocumented immigrants with end-stage renal disease | Colorado, Texas, California | Retrospective review | Multicenter | 211 | Specified in chart |
| Siddharthan et al. | Inpatient utilization by undocumented immigrants without insurance | Florida | Retrospective review | Single institution | 3211 | Specified in chart |
| Guadamuz et al. | Immigration status and disparities in the treatment of cardiovascular disease risk factors in the Hispanic Community Health Study/Study of Latinos | New York, Illinois, California, Florida | Prospective cohort (HCHS/SOL) | Multicenter | 6415 | Confirmation of being UDI is part of the study intake questionnaire |
| Inglesby et al. | Hand and upper extremity trauma in the undocumented immigrant population in the United States | New York | Retrospective review | Single institution | 176 | Specified in chart |
| Roblyer | Stress and alcohol, cigarette, and marijuana use among Latino adolescents in families with undocumented immigrants | Oklahoma | Population-based survey | Single community | 102 | Specified in questionnaire |
| Korinek et al. | Prenatal care among immigrant and racial-ethnic minority women in a new immigrant destination: Exploring the impact of immigrant legal status | Utah | Retrospective review | Single state | 3000 | Queried births from a the Utah Population Database and linked maternal birth records to maternal driver license division for immigration status |
| Wen et al. | Neighborhood socioeconomic status and BMI differences by immigrant and legal status: Evidence from Utah | Utah | Retrospective cross-sectional review | Single state | 8987 | Driver license records from Utah Population Database |
| Mehta et al. | Cervical cancer prevention: Screening among undocumented Hispanic women compared with documented Hispanic women | Rhode Island | Cross-sectional survey | Single institution | 73 | Specified in questionnaire |
| Qualitative (n = 3) | ||||||
| Adu-Boahene et al. | Health-needs assessment for West African immigrants in Greater Providence, RI | Rhode Island | Cross-sectional interview | Single community | 10 | Asked participants directly about immigration status |
| Gomez et al. | On edge all the time': Mixed-status households navigating health care post Arizona's most stringent anti-immigrant law | Arizona | Cross-sectional interview | Single community | 29 | Asked participants directly about immigration status |
| Armenta et al. | Receptionists, doctors, and social workers: Examining undocumented immigrant women's perceptions of health services | Pennsylvania | Cross-sectional interview | Single community | 59 | Asked participants directly about immigration status |
| Research using UDI proxies | ||||||
| Quantitative (n = 16) | ||||||
| Atkins et al. | The impact of expanded health insurance coverage for unauthorized pregnant women on prenatal care utilization | Nebraska, South Carolina | Natural experiment design | Multistate | State-wide 20,876 participants |
Public birth certificate records with absence of social security number |
| Bustamante et al. | Variations in healthcare access and utilization among Mexican immigrants: The role of documentation status | California | Cross-sectional survey | Single state | 1038 | Used the California Health Interview Survey (CHIS), which asks if respondents have permanent residency with green card. If answer is “no,” there is a 90%–95% chance of being undocumented |
| Ortega et al. | Health care access and physical and behavioral health among undocumented Latinos in California | California | Retrospective review | Single state | 3053 | CHIS survey questions |
| Lopez Mercado et al. | Undocumented Latino immigrants and the Latino health paradox | California | Retrospective review | Single state | 2972 | CHIS survey questions |
| Bustamante et al. | Health Policy challenges posed by shifting demographics and health trends among immigrants to the United States | California | Retrospective review | Single state | Not specified | CHIS and National Health Inpatient Survey specific survey questions |
| Sohn et al. | Geographic variation in COVID-19 vulnerability by legal immigration status in California: A prepandemic cross-sectional study | California | Cross-sectional study | Single state | 1117 | CHIS survey questions |
| Bustamente et al. | Identifying health insurance predictors and the main reported reasons for being uninsured among US immigrants by legal authorization status | California | Retrospective cohort | Single state | 1471 | CHIS survey questions |
| Swartz et al. | Expanding prenatal care to unauthorized immigrant women and the effects on infant health | Oregon | Natural experiment with difference-in-difference design | Single state | 12,344 | State-specific prenatal care option (CAWEM) with stepwise natural experiment |
| DuBard et al. | Trends in emergency Medicaid expenditures for recent and undocumented immigrants | North Carolina | Cross-sectional study | Single state | 48,391 | North Carolina Emergency Medicaid data (97%–99% UDI) |
| Rodriguez et al. | Association of expanded prenatal care coverage for immigrant women with postpartum contraception and short interpregnancy interval births | Oregon, South Carolina | Cohort study with difference-in-difference design | Single state | 26,586 | Emergency Medicaid claims |
| Hainmueller et al. | Protecting unauthorized immigrant mothers improves their children's mental health | Oregon | Cross-sectional | Single state | 8610 | Medicaid claims data with DACA age qualification cutoff for mothers with quasi-random assignment |
| Nwadiuko et al. | Changes in health care use among undocumented patients | Maryland | Retrospective cohort study | Single health system | 1501 | Health services only accessible to Medicaid ineligible patients (90% UDI) |
| Nandi et al. | Access to and use of health services among undocumented Mexican immigrants in a US urban area | New York | Cross-sectional survey | Single community | 505 | Recruited from communities with large Mexican populations |
| White et al. | Changes in use of county public health services following implementation of Alabama's immigration law | Alabama | Retrospective cohort study | Single health system | 20,524 | Latino ethnicity |
| Nguyen et al. | Association of scheduled versus emergency-only dialysis with health outcomes and costs in undocumented immigrants with end-stage renal disease | Texas | Retrospective cohort study | Single institution | 181 | Patients receiving emergency-only hemodialysis in a large safety net hospital |
| Palacio et al. | The mechanism and pattern of injuries of undocumented immigrants crossing the Texas–Mexico border along the Rio Grande Valley | Texas | Retrospective cross-sectional study | Single institution | 178 | Chart review for patients without a social security number |
| Qualitative (n = 3) | ||||||
| Cavazos-Rehg et al. | Legal status, emotional well-being and subject health status of Latino immigrants | Missouri | Cross-sectional interview | Single community | 143 | Answering “yes” to the survey question “fear of deportation?” |
| Flynn et al. | Undocumented status as a social determinant of occupational safety and health: The workers' perspective | Ohio, New Mexico | Cross-sectional interview | Two communities | 103 | Latino ethnicity |
| Panikkar et al. | Precarious essential work, immigrant dairy farmworkers, and occupational health experiences in vermont | Vermont | Cross-sectional survey | Single community | 117 | Immigrant dairy farmworker (estimates of 90% UDI in this community) |
| Machine learning predictions (n = 5) | ||||||
| Ruhnke et al. | A healthy migrant effect? Estimating health outcomes of the undocumented immigrant population in the United States using machine learning | National | Nonparametric machine learning model | Entire United States | 572,339 | Pooled 20 years of NHIS survey and Survey of Income and Program Participation (SIPP) to determine donor sample legal status |
| Ro et al. | Undocumented older Latino immigrants in the United States: Population projections and share of older undocumented Latinos by health insurance coverage and chronic health conditions 2018–2038 | National | Cohort component method | Entire United States | 491,000 | Used 5 years of the American Community Survey to project growth of Latino adults aged > 55 population over the next 20 years |
| Heintzman et al. | Using electronic health record data to study Latino immigrant populations in health services research | Oregon | Electronic health record-based algorithm, cross-sectional interviews | Single institution | 441 | Combined electronic health record variables including: FHQC patient, low-income, Latino, primary Spanish language, age > 18, and uninsured |
| Wilson et al. | Comparison of use of health care service and spending for unauthorized immigrants versus authorized immigrants or US citizens using a machine learning model | California | Forest classifier machine learning model | Single city | 47,199 | Used data from Los Angeles Family and Neighborhood Survey to develop random forest classifier machine learning model |
| Pourat et al. | Accessing health care services used by California's undocumented immigrant population in 2010 | California | Machine learning model with Poisson regression and Taylor series approximation | Statewide | 59,848 | Trained a model with CHIS data to estimate adult and child UDI population |
- Note: Published studies derived from the authors' literature review of studies on undocumented immigrants' health research from 1995–2023.
3.4.1 Research Identifying UDIs
Most research exclusive to UDIs consisted of retrospective chart reviews at single centers, qualitative interviews with small samples, and case reports [29, 30]. The majority were qualitative, as researchers can more directly ascertain legal status through direct demographic questions [31]. Qualitative methods allow for in-depth inquiry on experiences, health access, and challenges faced by UDIs and have made major contributions to understanding health access and perceptions towards restrictive policies [32]. However, qualitative studies are limited in generalizability and can be affected by inadvertent power dynamics or selection bias. Most of the quantitative studies that precisely defined UDIs were conducted retrospectively at single centers through querying electronic health data [30, 33, 34]. Less common are multicenter retrospective studies or prospective recruitment studies, though these have the benefit of asking patient immigration status as inclusion criteria [35]. Few states including Utah and Maryland have unique ways to link patients and other social services to confirm immigration status. For example, the Utah Population Database can be linked to driver license information to identify UDIs within this database [36].
Dialysis and kidney transplant research stand as a unique example for studying undocumented patients with high population accuracy. Given unique registration numbers in the United Network for Organ Sharing (UNOS) system, transplant researchers can identify UDIs from a specific social security number designation: beginning with 9FN [37]. This differentiation has facilitated dozens of studies on undocumented patients' access to and outcomes from dialysis and transplant, including cost-effectiveness findings, and may be a major contributor to the nationwide push to offer UDI outpatient maintenance dialysis at the state level [29].
Much rarer are quantitative, large-scale studies exclusive to UDIs. Given the uniqueness of California's restricted scope Medi-Cal program—eligible persons who meet the income level threshold for Medi-Cal but do not meet immigration status requirements (not US nationals, citizens, or lawful permanent residents)—claims data from this source has been used to conduct health services research specific to UDIs [33, 35]. Massachusetts' MassHealth Limited data has been used similarly, albeit less frequently as a proxy, which may be because asylum seekers and other immigrant populations can also access MassHealth Limited, making it an imperfect measure for UDIs. With these rare and regional exceptions, researchers using claims data cannot differentiate UDIs from other immigrant and low socioeconomic populations.
3.4.2 Research Using Proxies for UDIs
Health service researchers have approximated UDI populations with varying degrees of accuracy in diverse research approaches. Both quantitative and qualitative studies demonstrate a wide range of approximation accuracy. Some of the most accurate proxy measures given extensive social research are shown to be age > 18 with Medicaid-ineligibility (> 90% undocumented) and Emergency Medicaid use (97%–99% undocumented) [38, 39]. Less accurate variables have included Latino ethnicity, lack of social security number, non-English speaking, or occupation such as farmer [40, 41]. More innovative solutions have included linking of state-specific databases with proxy variables and designing quasi-experimental natural studies to align with state policies that have particular impact on UDI populations [42, 43].
Researchers have also approximated UDIs through disease conditions, particularly renal failure. This is made possible through policies regarding immigration status and access to emergency versus scheduled hemodialysis, which vary by state [37]. Finally, certain states conduct regular, census-like surveys of their population, which provides data to help approximate UDIs [44, 45]. While there are no state or nationwide health surveys, which explicitly ask documentation status, researchers have defined proxies through these tools to reasonably estimate which respondents are likely undocumented. For example, the California Health Inpatient Survey (CHIS) uses the question, “Are you a permanent resident with a green card?” which estimates that approximately 95% of those answering “no” are undocumented [46].
More recently, machine learning has been used to predict the likelihood of UDI status using electronic health records and database linkage. While still approximations, sophisticated analytical techniques have allowed for the development of algorithms using national-level data that could be deployed on larger scales [47, 48]. These models incorporate variables not always solicited in state claims databases, including citizenship, permanent resident status, residency change, use of a FQHC, and insurance history [49]. Given the novelty of these approaches and advanced statistical expertise required to use them, such algorithms are not yet widely used in UDI research.
4 Discussion
In this narrative review on health coverage availability and research methodologies to study undocumented patients by state, we found: (1) wide variability in health coverage for UDIs, (2) diverse and unstandardized methods to study UDI patients, and (3) creative approaches to approximate UDI patients, particularly to conduct large-scale, quantitative research. In terms of health coverage, a complex, and dynamic scheme exists for UDIs depending on region, disease condition, and age. Many states offered no health coverage option to UDIs. However, 32 offered some form of coverage, which could provide effective study sites for states considering expansion options. Health services research on UDIs relies on varied data access and diverse methodologies. Most studies consisted of single-center retrospective chart reviews, qualitative interviews with small sample sizes, and case reports. While these studies provide valuable in-depth information on individual perceptions, overreliance on these methods risks an incomplete understanding of UDIs, particularly for population-level disease burden and care utilization. With rare exceptions, it was impossible to identify UDI status through large databases. Exceptions included prospective cohort studies with intentional recruitment and nuanced approaches to link social services databases with immigration status to health records, as with the Utah Population Database. We also found a wide range in accuracy when approximating undocumented populations and no standard approach for doing so. Certain proxies estimated UDIs with high accuracy, including Emergency Medicaid claims, age > 18 with Medicaid ineligibility, and machine learning algorithms. However, machine learning approaches have yet to be tested against a gold standard and could introduce selection or training data bias, if data utilized during model training and development is inaccurate or misrepresentative [50]. Less accurate proxies included being Latino, living in a zip code with a high undocumented population, and occupation such as farm working. A lack of standardized approaches can lead to inaccurate understanding of UDI populations, further emphasizing the need for precise and validated approaches depending on specific research questions. Our calculation of a state-quantity-weighted-average for each state category demonstrates research output categorized by grouping of health coverage availability. While no relationship inference can be asserted between insurance availability and research output given the multitude of confounding variables which affect research output, this metric can be helpful for health services researchers to see where work is being done with methodologies that could be replicated or collaborators with whom they could work.
There were notable examples where health services researchers within specific states could conduct large-scale, quantitative research studies on UDI health, due to those states offering insurance coverage, and therefore collecting claims data for, UDI populations. Specifically, much of the literature utilizing claims data for UDIs comes from California, where researchers can access individual patient records and state-level claims for UDIs. States which collect large-scale data on UDIs are likely better positioned to understand disease burden, access issues, and policy effects than those that do not. Health services researchers use claims data to investigate health issues and access, inform quality improvement efforts, and conduct quasi-experimental studies due to the high volume, accessibility, and relatively low-cost of claims data [51]. Claims databases are also free from nonresponse bias found in studies utilizing other methods such as surveys or ethnography [52]. Certainly, availability of research opportunities depends on a variety of confounding variables, including total population of UDI patients, institutional support, researcher interests, and funding for such pursuits. Additionally, since health coverage for UDIs is decided by state policy, health services researchers looking to conduct quantitative, population-level research with claims data are required to adjust their methods depending on the state and time period of interest for study. For example, researchers in California could use restricted Medi-Cal claims data and those in Utah could leverage database linking to approximate UDI patients, while researchers studying UDIs in other states would have to use prospective recruitment or institution-specific electronic health record review to do so. This narrative review does not make definitive assertions about a relationship between health coverage and research output related to UDIs. Future research should examine the role of health-insurance coverage using large administrative datasets—such as claims records—which would enable the exploration of how state and national policy could impact UDI health. To our knowledge, this is the first attempt to summarize research methodologies regarding undocumented patients. From our review, significant gaps remain in understanding how to best study this population. Further, current sociopolitical trends and recently implemented immigration policy could affect how UDIs seek and access care in the United States. Examples include recent executive orders in Texas and Florida which require hospitals and clinicians to share patient immigration status with law enforcement, or national level executive actions which rescinded sensitive location exemptions for apprehensions by Immigration and Customs Enforcement [53, 54]. These policies are likely to bring a new series of opportunities and challenges to studying UDI health. Collecting patient variables on documentation status, though harmful to patient protections if disclosed to be used for legal action, may help to stratify disease burden, health costs, and other critical health indicators by immigration status. It was known from restrictive immigration policies implemented in 2018 that immigrants were less likely to seek healthcare due to fear of apprehension [55]. Understanding specific barriers from a personal and systemic level which emerge from these policies, and interventions to mitigate them, presents important areas of study for health services researchers in the coming years.
4.1 Limitations
Despite these contributions, our study had several limitations. There are additional factors which affect coverage eligibility and access which were not the focus of this study. These include income, as most available options required Medicaid-eligible income levels. Second, while types of immigrants have distinct legal definitions (i.e., asylum seeker, refugee, undocumented), individuals often move between them. Third, the landscape of United States health coverage is rapidly fluctuating, shown by multiple states in this review with proposed legislation to increase health coverage for UDIs. Updated methods aimed at understanding the health effects of these quickly changing policies will be important to address health needs and access for UDIs. Fourth, one of the challenges in studying UDIs is that many legal experts explicitly recommend against medical documentation of legal status [56]. This approach is driven by a fear from both patient and provider that noting undocumented status in the medical record could lead to patient harm. Basic coverage, especially that which distinguishes UDIs through claims in a way that is non-identifiable at the individual level, would alleviate that fear and allow better data collection to improve health studies. Due to the sensitive nature of public health research among UDIs, experts largely advise against collecting or recording participants' legal status to limit exposure to legal or social harm. Strict confidentiality measures to include anonymizing data, employing secure storage protocols, and limiting identifiable information are essential to protect participants from risks including deportation, discrimination, or detention. Researchers should also implement transparent, culturally appropriate informed-consent processes to ensure that any study undertaken with UDI information works to address community needs, such as improved access to care, without inadvertently increasing vulnerabilities or undesirable exposure. Ongoing ethical oversight is critical for adapting study procedures in response to emerging risks and community feedback in the evolving geopolitical landscape.
Finally, factors influencing policy implementation can affect how immigrants access health. Cuts to ACA-related policies from 2017 to 2020 likely had an effect on UDIs' decisions to pursue healthcare during that time, including the public charge rule which limited green-card seeking immigrants from accessing health services for fear of denial [57]. Notwithstanding, creating systems where accurate UDI data becomes accessible is critical to promoting UDI and wider population health. Such infrastructure could empower researchers to conduct studies with data to assess variations of disease incidence, healthcare cost and access, and empirical health effects from immigration policy.
5 Conclusions
UDIs in the United States are an understudied population with varied healthcare coverage depending on state, age, and special conditions. Innovative methodologies have been used to access and approximate health data for undocumented patients, particularly when conducting quantitative research with large datasets. Our findings can be helpful to health researchers investigating the health of UDI patients, to better understand available health coverage, research methodologies for approximating these patients, and the geographic areas where this study is being conducted. Better understanding the range of methodological approaches can help researchers working with UDIs to practice successful and rigorous research designs and facilitate efforts to optimize health care delivery among UDIs in the United States.
Author Contributions
Christopher W. Reynolds: conceptualization, investigation, writing – original draft, methodology, validation, writing – review and editing, data curation. Karthik Reddy: visualization, investigation, writing – original draft, methodology, writing – review and editing, data curation. Samantha Peña: data curation, writing – review and editing, writing – original draft, investigation. Priya J. Desai: writing – original draft, writing – review and editing, investigation, data curation. Rachel I. Ekaireb: writing – original draft, writing – review and editing. Sabrina E. Sanchez: writing – review and editing, writing – original draft. Sarah L. Kimball: writing – review and editing, supervision. Megan G. Janeway: supervision, conceptualization, investigation, writing – review and editing, writing – original draft.
Disclosure
All authors have read and approved the final version of the manuscript. Dr. Megan G. Janeway, the principal investigator, had full access to all of the data in this study and takes complete responsibility for the integrity of the data and the accuracy of the data analysis. Dr. Megan G. Janeway affirms that this manuscript is an honest, accurate, and transparent account of the study being reported; that no important aspects of the study have been omitted; and that any discrepancies from the study as planned have been explained.
There are no relevant funding disclosures or financial involvement related to this study or during study design; collection, analysis, and interpretation of data; writing of the report; and the decision to submit the report for publication. All authors certify that they have no affiliations with or involvement in any organization or entity with any financial interest or nonfinancial interest in the subject matter or materials discussed in this manuscript.
Transparency Statement
The lead author Christopher W. Reynolds affirms that this manuscript is an honest, accurate, and transparent account of the study being reported; that no important aspects of the study have been omitted; and that any discrepancies from the study as planned (and, if relevant, registered) have been explained.
Open Research
Data Availability Statement
The data that supports the findings of this study are available in the supporting material of this article. Data sharing is not applicable to this article as no datasets were generated or analyzed during the current study. A list of search terms, screened articles, and inclusion/exclusion criteria is provided in the supplement.
