Association of Maternal Citizenship and State-Level Immigrant Policies With Health Insurance Coverage Among US-Born Latino Youths

This cross-sectional study uses data from the American Community Survey to examine the association of state-level immigrant integration and criminalization policies and maternal citizenship status with health insurance coverage among US-born Latino youths.

The association of immigration policy with the overall health of Latino immigrant populations in the US has been documented. Arrests, deportations, and raids by US Immigration and Customs Enforcement 35 and racist rhetoric during and after the 2016 presidential election have been associated with preterm births and low-birth-weight infants born to Latina mothers in the US. 36,37 Using the Andersen Behavioral Model of Health Services 38 framework encompassing demographic and income factors, we examined variation in uninsurance by state-level integration and criminalization immigrant policies for US-born Latino youths with citizen and noncitizen mothers.
Given that state-level immigrant integration and criminalization policies are forms of legalized discrimination based on legal status, 29 it is important to examine variation by maternal citizenship status. We hypothesized that uninsurance would be more common among youths with noncitizen mothers and among youths who reside in states that have the fewest immigrant integration policies and the most immigrant criminalization policies.

Data Source
This cross-sectional study used pooled American Community Survey (ACS) data from January 1, 2016, to December 31, 2018, from the Integrated Public Use Microdata Series. 39 These data are publicly available and de-identified, and thus, the Drexel University institutional review board exempted this study from human subjects protocol review. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline. 40 The ACS is a yearly nationwide survey administered by the US Census Bureau and collects information on demographic, social, and economic characteristics. It is mailed to approximately 295 000 US addresses per month, or 3.5 million households per year, in all 50 US states, the District of Columbia, and Puerto Rico. Response rates were 94.7% in 2016, 93.7% in 2017, and 92% in 2018. 41 The ACS contains state-level Federal Information Processing Standard codes, and Latino ethnicity is self-reported in the ACS. 42 The study sample consisted of Latino youths (17 years or younger) for whom linked Latina maternal characteristics (eg, citizenship status) were available. We only included youths of Latina mothers who were younger than 65 years at the time of the survey and who reported that their youngest child was no older than 17 years at the time of the survey. The sample size after applying the inclusion criteria was 226 691.

Measures
Youth uninsurance was the outcome measure of interest and was assessed based on the mothers' reports of whether their youths had any health insurance coverage at the time of the survey (yes or no). The main independent variables were maternal citizenship status (US citizen or noncitizen), 1 binary measure of state-level immigrant integration policy (high integration or low integration), and 1 binary measure of state-level immigrant criminalization policy (low criminalization or high criminalization). We classified state-level immigrant integration and criminalization policies using the Young index of state-level immigrant integration and criminalization policy context. 29 The index was created through a systematic review of implemented immigrant policies up to December 31, 2015, and policies were classified as integration or criminalization. Fourteen policies were classified as integration because they either facilitated or contributed to barriers for immigrant access to state institutions and programs (eg, access to health insurance coverage for children regardless of legal status, extension of wage and hour protections for agricultural workers and domestic workers, and laws protecting noncitizen workers from employer retaliation related to their legal status). 29 The criminalization category included 6 policies that could influence activities of noncitizens by increasing or decreasing their exposure to law enforcement (eg, complying with Real ID, mandating employer use of E-Verify, and requiring or allowing law enforcement to verify an individual's legal status during an arrest or traffic stop). A table listing all of these policies can be found in eTable 1 in the Supplement.
The total numbers of integration and criminalization policies were summed separately, and median and mean scores of immigrant policy contexts were calculated. We used the median score to classify states and the District of Columbia as high integration or low integration and as low criminalization or high criminalization (based on whether they were at or above the median vs below the median). In general, states were in the high-integration category if they had 4 or more integration policies and in the low-integration category if they had 3 or fewer integration policies. The highcriminalization category included states with 4 or more criminalizing policies, and the low-criminalization category included states with 3 or fewer criminalizing policies. 29 Summaries of integration and criminalization policies and integration and criminalization categorizations by state are presented in eTables 2 and 3 in the Supplement.
We used the Andersen Behavioral Model of Health Services 38 to select control variables in the models. Factors included youth age (0-4, 5-9, 10-13, or 14-17 years) and sex (male or female) and mother's language (English or Spanish), marital status (married, divorced or separated, or never married), age (18-29, 30-39, 40-49, or 50-64 years), number of birth children (1-3 or Ն4), employment status (working or with a job, unemployed, or not in the labor force), and educational level (less than high school diploma, high school diploma, or college degree or more). We adjusted for income as a percentage of the federal poverty level (Ն400%, 300%-399%, 200%-299%, 100%-199%, or Յ99%). We controlled for contextual characteristics from the 2016 presidential election by creating a binary indicator of states whose electoral college voted for Hillary Clinton or voted for Donald Trump. 43 We also included survey year (2016, 2017, and 2018) and state (all 50 states and the District of Columbia) fixed effects.

Statistical Analysis
We conducted a secondary data analysis of ACS data using Stata, version 15 (StataCorp LLC). 44 First, we used descriptive statistics to examine differences in youth uninsurance and among the aforementioned demographic and socioeconomic factors by maternal citizenship status using χ 2 tests. Second, we used χ 2 tests to examine variation in youth uninsurance by maternal citizenship (citizen or noncitizen), state immigrant integration policy context (low integration or high integration), and state immigrant criminalization policy context (high criminalization or low criminalization). Third, initial multivariable logistic regressions were used to assess whether associations existed among maternal citizenship status and youth uninsurance while controlling for the demographic and socioeconomic variables. We added state integration and criminalization policy variables into the models and then tested interactions between maternal citizenship and state integration and criminalization policy when assessing youth uninsurance. Comparisons and associations were considered significant at 2-sided P < .05. In addition, we calculated the probability of youth uninsurance by running a margins command to examine the mean marginal association of maternal citizenship and of state integration and criminalization policy with youth uninsurance. We also used sensitivity models to exclude states that covered prenatal care regardless of documentation status; states that used a 4-level, state-level immigrant policy measure that combined integration and criminalization; and states with continuous measures of integration and criminalization (eTables 7-10 in the Supplement). All analyses were conducted with weighted survey data to produce estimates of the noninstitutionalized population of US-born Latino youths.

Results
Of 226  Differences in the proportion of youth uninsurance by state immigrant integration and criminalization policy contexts and maternal citizenship are presented in Table 2. Of youths in high integration states, 5.39% (95% CI, 5.22%-5.57%) were uninsured compared with 6.60% (95% CI, 6.11%-7.14%) in low integration states. The percentage of uninsured youth with noncitizen mothers was 6.69% (95% CI, 6.36%-7.03%) in high integration states compared with 9.10% (95% CI, 8.22%-10.06%) in low integration states. In high criminalization states, the percentage of uninsured youths with citizen mothers was 5.91% (95% CI, 5.64%-6.20%). Of youths with noncitizen mothers, 4.15% (95% CI, 3.80%-4.54%) in low criminalization states were uninsured compared with 9.37% (95% CI, 8.90%-9.87%) in high criminalization states. Table 3 presents the probabilities and pairwise comparisons of youth uninsurance by maternal citizenship and state immigrant policy environments from multivariate logistic regression. The probability of youth uninsurance in the high level of integration immigrant state-level policy context was 1.3% (95% CI, 0.3%-2.2%; P < .05) higher for youths with noncitizen mothers compared with youths with citizen mothers. Compared with youths with citizen mothers in low integration statelevel policy contexts, the probability of youth uninsurance was 3.3% (95% CI, 2.3%-4.4%; P < .001) higher for youths with noncitizen mothers. The youth uninsurance rate was 2.1% (95% CI, 0.6%-3.6%; P = .006) higher among youths with noncitizen mothers in low integration states compared with youths with noncitizen mothers in high integration states. There was no difference between youths with noncitizen and citizen mothers in the probability of youth uninsurance in states with low levels of criminalization policies. Conversely, in high criminalization states, there was a 2.6% (95% CI, 1.9%-3.0%; P < .001) higher probability of uninsurance among youths with noncitizen mothers compared with youths with citizen mothers. The probability of youth uninsurance among youths with noncitizen mothers in states with high criminalization policy contexts was 1.7% (95% CI, 0.7%-2.7%; P < .001) higher than that in states with low criminalization policy contexts. The full regression and probabilities for all citizenship and policy context results are presented in eTables 4-5 in the Supplement. Results were similar to those of the sensitivity analyses performed (eTables 7-10 in the Supplement).

Discussion
Latino youths who are US born and have immigrant parents face persistent barriers in access to health care compared with youths whose parents are US natives. 4-6 Some of these disparities are associated with parental uninsurance disparities owing to differences in documentation and citizenship status that can impact eligibility for programs and access to affordable benefits offered by employers. 7,8,12,15,16,18 State-level immigrant policies may be associated with a greater ability for immigrants to access health care. For instance, California has allowed young adult undocumented immigrants to access the state's Medicaid program. 45 We sought to understand the associations among maternal citizenship, state-level immigrant integration and criminalization policies, and US-born Latino youth uninsurance. We observed that immigrant integration policies were not associated with uninsurance among youths with citizen mothers but that a higher proportion of uninsurance among youths with noncitizen mothers was associated with a context of low levels of immigrant integration policy. High levels of immigrant criminalization policy were associated with uninsurance among US-born Latino youths; in this study, almost 6% of youths who had citizen mothers and lived in high criminalization states were uninsured.

JAMA Network Open | Health Policy
State-Level Immigrant Policies and Health Insurance Coverage for Latino Youths

JAMA Network Open | Health Policy
State-Level Immigrant Policies and Health Insurance Coverage for Latino Youths Disparities in insurance coverage among US-born Latino youths may limit the ability of this group, the largest ethnic minority youth group in the US, to receive annual well-child visits and preventive care. 20 We observed that in high integration states, uninsurance rates were 1.3% higher for youths with noncitizen mothers compared with youths whose mothers were citizens. In contrast, in low integration states, uninsurance rates were 3.3% higher for youths with noncitizen mothers compared with youths whose mothers were citizens. The uninsurance rate of 1.3% equated to approximately 232 700 uninsured youths in the US. In contrast, in low integration states, uninsurance rates were 3.3% higher for youths with noncitizen mothers compared with youth with citizen mothers, equating to approximately 590 700 youths in the US. On average, US-born Latino youths with noncitizen mothers had a 2.1% higher probability of being uninsured in a low integration state compared with living in a high integration state. Translating this 2.1% probability into quantities, an estimated 375 900 US-born Latino youths with noncitizen mothers in the US were uninsured, and this was associated with state-level integration policies. 46 High levels of immigrant integration policies at the state level were associated with lower uninsurance for youths with noncitizen mothers; thus, these policies may be associated with reduced disparities in access to care for US-born Latino youths. Low levels of immigrant integration policies may be associated with less ability of  immigrants, including parents of US-born Latino youths, to integrate into programs and benefits, which by extension may have consequences for US-born youths who may be entitled to these programs and benefits. From a health policy and cost-benefit perspective, providing health insurance coverage for uninsured youths, specifically youths eligible for Medicaid or the Children's Health Insurance Program, could save the US $8 billion to $10 billion annually. 47 High levels of immigrant criminalization policies at the state level were significantly associated with a 2.6% higher probability of uninsurance for youths with noncitizen mothers compared with youths whose mothers were citizens. These findings suggest that noncitizen parents, particularly mothers, may face structural barriers to seeking health care access, programs, and benefits for their US-born youths. Fear of detention and/or deportation that arises from residing in a state with a high level of immigrant criminalization policies may be associated with uninsurance disparities for US-born Latino youths and also has been found to discourage immigrant mothers from seeking well-child visits for their US-born infants. 48

Limitations
This study has limitations. We relied on a multidimensional index that measured whether certain state-level policies were enacted as of December 31, 2015, which limited our ability to use a preimplementation-postimplementation study design. We relied on ACS data from 2016 to 2018 to examine the association between state policies and uninsurance of Latino youths, assuming that state-level classifications in the index did not drastically shift, such as from high to low integration or low to high criminalization. Although this is a limitation, using this index was important because it measured multiple immigrant policies that capture state-level immigrant integration and criminalization policy contexts that might be associated with uninsurance among US-born Latino youth. To our knowledge, this approach has not been done previously. Second, we could not assess whether there have been any changes to the probability of uninsurance for the same youths over time. Third, the estimates in this study may be conservative given that participating in the ACS may be discouraging for noncitizen mothers in the current sociopolitical climate. Fourth, there may be other factors across states that could be associated with health insurance but were not measured in this study and were assumed to be held constant by the state and year fixed-effects analysis. In