Reduced health care utilization among young children of immigrants after Donald Trump's election and proposed public charge rule

Abstract

Widespread fear among immigrants from hostile 2016 presidential campaign rhetoric decreased social and health care service enrollment (chilling effect). Health care utilization effects among immigrant families with young children are unknown. We examined whether former President Trump's election had chilling effects on well-child visit (WCV) schedule adherence, hospitalizations, and emergency department (ED) visits among children of immigrant vs US-born mothers in 3 US cities. Cross-sectional surveys of children <4 years receiving care in hospitals were linked to 2015–2018 electronic health records. We applied difference-in-difference analysis with a 12-month pre/post-election study period. Trump's election was associated with a 5-percentage-point decrease (−0.05; 95% CI: −0.08, −0.02) in WCV adherence for children of immigrant vs US-born mothers with no difference in hospitalizations or ED visits. Secondary analyses extending the treatment period to a leaked draft of proposed changes to public charge rules also showed significantly decreased WCV adherence among children of immigrant vs US-born mothers. Findings indicate likely missed opportunities for American Academy of Pediatrics–recommended early childhood vaccinations, health and developmental screenings, and family support. Policies and rhetoric promoting immigrant inclusion create a more just and equitable society for all US children.

Introduction

One in 4 children in the United States has at least 1 immigrant parent.¹ These immigrant and mixed-status families face challenges in accessing health care, ranging from language and cultural alignment, to coverage gaps and cost, to racism and discrimination.^2–4

Threats to immigrant safety and immigration status can affect access to health care, care-seeking, and utilization.⁵ Both state and federal policies can influence the social atmosphere and willingness of immigrants to seek assistance.^6–8 Depending on the state, restrictive and punitive policies may determine whether they qualify for health care coverage⁹ and whether accessing social services puts their immigration status at risk.¹⁰ Complex rules create confusion and stress for health care staff and immigrant patients.¹¹ Thus, immigrants who are affected by rules governing eligibility for Medicaid or other programs as well as those who are not directly affected may avoid seeking assistance or utilizing health care rather than risk potential immigration sanctions, including deportation for themselves or family members.

This phenomenon—avoiding social services or health care due to immigration status–related fear—is known as the “chilling effect.” Policymakers, researchers, and advocates have long been aware of the potential for policies to create chilling effects.¹² For instance, in the 1990s, the federal government enacted sweeping policy changes known as “welfare reform,” including restrictions on immigrant eligibility, which ultimately barred otherwise eligible legal permanent residents (LPRs) from Medicaid and food stamps (now called the Supplemental Nutrition Assistance Program [SNAP]) for their first 5 years in the United States. A study at the time demonstrated that the law had no effect on single, US-born mothers’ Medicaid enrollment, but increased uninsurance among single immigrant mothers and their children.¹³

Renewed attention to chilling effects arose during the 2015–2016 presidential campaign, when Donald Trump used dehumanizing and racist language about immigrants, which was amplified in the press.¹⁴ News reports indicated increased concern about social and health care service access and enrollment among immigrant and mixed-status households, especially after Trump won the election.^15–19 Following through on threats, after the inauguration, the Trump Administration dramatically increased anti-immigrant migration policies^20,21 and immigration enforcement, especially in the interior of the country.^22,23

In the same period, late January 2017, a draft Executive Order was leaked to the press that indicated a dramatic expansion to longstanding policies defining “public charge,” a concept that guides the consideration of public benefit participation and other factors for immigrants applying for LPR status.^24,25 A person is deemed a public charge if judged to be primarily dependent on the government, which can be grounds for denying the LPR application. Although not ultimately adopted in its leaked form, the leak of the proposed public charge rule change received substantial media coverage, generating concerns about chilling effects and health consequences associated with fear of accessing public health services.^24,26–28

One potential serious concern attending to the likely chilling effects produced by the 2016 presidential election is that they might lead to decreased utilization of health care and other critical services among immigrant families. If the strident anti-immigrant rhetoric of the campaign, which was manifest early in the Trump presidency in the leak of the proposed public charge rule change, caused families not to participate in Medicaid or other programs or to eschew services if already enrolled, 1 serious consequence might be a decline in health care utilization.

Indeed, a study of immigrant adults in California demonstrated decreased adult emergency department (ED) usage linked to negative rhetoric common during the campaign and following Trump's election.¹⁸ Among Latina women, the 2016 election was also associated with higher than expected numbers of preterm births.²⁹ For school-age children in Baltimore, Trump's presidential campaign was followed by a decrease in primary care visits and increased ED visits for undocumented, Medicaid-ineligible children compared with Medicaid-eligible children.¹⁶ Finally, the leaked public charge draft was associated with delayed prenatal Medicaid enrollment among pregnant immigrant women in New York and lower infant birth weight.³⁰

To date, however, no study has examined whether the chilling effects produced by the Trump election have affected the health care utilization of young children. Indeed, the study of Baltimore children specifically excluded younger children because of the complexity of accounting for the changing frequency of well-child visits (WCVs) in the first 3 years of life. Studying health care among young children is important for several reasons. Families with young children, particularly families of color and with immigrant members, are more likely than those with older children to have low incomes and, consequently, have difficulty accessing and affording health care, food, and housing.^2,31–33 Birth to age 4 years is a uniquely sensitive developmental period of rapid body and brain growth that establishes the foundation for a child's future physical, socioemotional, and cognitive health, and school readiness.³⁴ Frequent contact with the health care system is expected and encouraged in early childhood. The American Academy of Pediatrics (AAP) recommends 12 WCVs (Appendix Exhibit A27), with specified preventive screenings and assessments, in the first 3 years of life and annual checkups thereafter.³⁵

Several studies have explored WCV schedule adherence, vital to optimal early childhood health and development.^36–39 Among young, commercially insured children in Hawaii, missed WCVs and low continuity of care were associated with increased risk of ambulatory-care–sensitive hospitalizations.³⁶ A 20-state study of families with low incomes found that of all WCVs, toddler and preschool WCVs were most commonly missed.³⁹ Yet, there is little research on key national policy events and associations with WCVs among immigrant and US-born parents with the youngest children. It is important to document whether and to what degree the election differentially affected WCV adherence among children of immigrant and US-born parents to guide practices to ensure health care for all young children.

Our aims were to examine whether Trump's election had a chilling effect on (1) adherence to the WCV schedule and (2) the number of ED visits and hospitalizations among young children of immigrant compared with US-born parents. We hypothesized that, due to a chilling effect, following the election, immigrant parents might be afraid to continue WCVs and thus adherence to the overall schedule and receipt of specific WCVs would drop. Concurrently, rates of pediatric ED visits and hospitalizations would rise as families avoided preventive care and sought care only when children were very sick. We hypothesized that no such change would be detectable for children of US-born parents. Additionally, and acknowledging the proximity of the leak of the proposed public charge rule change soon after the election, we conducted secondary analyses to test the combined impact of the election and leaked draft of the public charge rule on these same outcomes.

Data and methods

Data source and population

This study's sample was identified from children enrolled in the ongoing Children's HealthWatch Study, a cross-sectional, sentinel surveillance, survey-based study of children younger than 4 years of age whose caregivers sought care for them in hospital settings. Surveys were conducted in English or Spanish in either pediatric EDs or hospital-based primary care clinics in 3 US cities (Boston, MA; Minneapolis, MN; and Little Rock, AR). Informed consent included consent to access the child's electronic health record (EHR). Survey responses were linked to the young child's EHR repository data covering the period 2015–2018, which included inpatient, ED, and outpatient visits and corresponding child ages, visit dates, and diagnoses (International Classification of Diseases, Ninth and Tenth Revisions [ICD-9 and 10, respectively]).⁴⁰ The study received institutional review board approval.

Measures

The Children's HealthWatch survey included information on children's health insurance, caregivers’ educational attainment, and marital status. The birth mothers’ race and ethnicity were self-identified using questions from the US Census, asking separately about Latina/Hispanic heritage and race, and categorized as Latina (all races), Black (non-Latina), White (non-Latina), and other/multiple races (non-Latina). Maternal nativity was identified by country of birth (US/territories or other) to form 2 groups—US-born and immigrant—following previous research.^19,31 Household employment was defined by report of 1 or more employed household members.

Health care utilization pattern

We developed a measure to represent children's outpatient health care utilization pattern by first creating variables that determined whether the child had any visits in 6-month age ranges and another to ascertain for how many months in the specific range children returned for a visit. Together, these allowed us to examine how many visits the children had over time in relation to their first visit (sample entry). Five categories resulted, as follows: (1) consistent utilization; (2) inconsistent utilization; (3) early utilization only (seen again within 1 y but never thereafter); (4) sample dropouts who entered but were never seen again; and (5) early utilization, left, and returned. Exploration of children's health care utilization pattern and secondary analyses excluding dropouts helped to examine whether children's data were missing at random (Appendix Exhibits A12-3&4 and A20-3&4).⁴¹

Outcome variables

Well-child visit adherence measures

A visit was coded a WCV if any of the outpatient diagnosis codes for that visit matched standard WCV ICD-9 or ICD-10 codes (V20.2, V20.31, V20.32, Z00.110, Z00.111, Z00.121, Z00.129). Building on Tom et al.'s methods,³⁶ we created a “strict adherence” variable that accounted for the differing entry ages of children with any outpatient visit (WCV or not) into the EHR sample by comparing the recommended number of WCVs for their age during the study period with their actual number of WCVs. Children with the appropriate number of visits for their age were coded as adherent (Appendix Exhibit A27). For example, a child entering the sample at age 2 years and 3 months (27 mo) would be expected to have 2 WCVs in the following year. That child would be adherent if visits occurred during the child's 30th and 36th month (indicated WCV ages).

We also assessed whether results reported below were sensitive to our definition of WCV adherence with a more flexible and real-life adherence measure. “Flexible adherence” was calculated similarly to “strict adherence” but with a buffer period around each visit's timing, such that a WCV received in that buffer period (1 mo in the first year of life, 2 mo in the second year, and 3 mo in the third year) would still count as the child being adherent to the WCV schedule for that age. We also tested a third approach to WCV adherence. “Detailed adherence” was a binary indicator for whether patients attended each individual WCV rather than examining whether they attended all visits.

Hospitalizations

This measure counted all hospitalizations for each child during periods before and after the 2016 election, coded as a dichotomous indicator of any vs no hospitalizations for each of these 2 time periods.

Emergency department visits

We used a dichotomous indicator for whether a child used an ED for each of the 2 time periods.

Statistical methods

We described child and family characteristics comparing children of immigrant vs US-born mothers, using chi-square and t-tests, as appropriate. Our primary analyses were difference-in-difference (DiD) models, which examined whether any changes in WCV adherence, hospitalizations, or ED visits after the 2016 election were more pronounced among children of immigrant compared with US-born mothers. For the DiD analysis, we compared health care utilization in the year prior to the November 8, 2016, election with the year following. Following a standard approach,⁴² we specified our DiD model by including in a regression an interaction term (DiD estimator) between maternal nativity (the treatment variable) and a variable (time, measured in days) indicating whether the observation was in the pre- or postelection period (ie, maternal nativity × time).

A primary assumption of DiD models is that trends in outcome measures are parallel prior to treatment. In our case, we assumed that parents’ desire to access health care did not vary by immigration status. Thus, to test the parallel trend assumption that health care utilization slopes between immigrant (treated) and US-born (controls) were the same before November 8, 2016, we applied linear regression in the year before election day, including an interaction term between the date of the visit and maternal nativity, and adjusting for covariates (below). Interaction term coefficients were not significant, suggesting no violation of the parallel trends assumption (Figure 1).

Figure 1.

Effect of former President Trump's election on well-child visits (flexible adherence) among immigrant compared with US-born families in the post-period, 2015–2018. Source: Authors’ analysis of Children's HealthWatch data, 2015–2018.

Analyses were adjusted for research site (state of residence), child insurance status, maternal race/ethnicity, caregiver marital status and educational attainment, household employment, and child age at the beginning of the pre-period. Analyses of strict and flexible adherence also included child health care utilization pattern as a covariate. For detailed adherence, an indicator allowing multiple observations for each child in the pre and post periods, we adjusted for covariates enumerated above plus fixed effects for each WCV, and the interaction between the post-period and treatment. Last, as a second complementary set of analyses, we explored all measures detailed above with an extended treatment period from election day through to the leak date of the draft public charge rule change in January 2017 (November 8, 2016–January 23, 2017), acting as an additional “intervention.” Because the inauguration was a confirmation of the election results, we chose not to consider the inauguration as a separate treatment but note that this expanded treatment period also contains President Trump's January 20th, 2017, inauguration.

We conducted several sensitivity analyses: (1) stratifying DiD analyses by age groups at sample entry (children <2 y, 2–4 y, ≥5 y), (2) excluding children who dropped out of the sample after a single health care visit (dropouts), (3) stratifying by immigrant mothers’ duration of residence in the United States, and (4) testing a triple DiD to explore differences by race/ethnicity with the interaction term: time × nativity × race/ethnicity (Appendix Exhibits A9-3&4; A10-3&4; A11-3&4; A12-3&4; A13-3&4; A16-3&4; A17-3&4; A18-3&4; A20-3&4; A21-3&4).

Study strengths and limitations

This study's strengths include its large, longitudinal, multistate design and focus on a racially and geographically diverse sample of families with a difficult-to-reach population of young children who have access to health care. The study uses sentinel sampling for the initial survey. The sentinel sample is both a strength and a limitation as a dynamic form of data collection designed to signal early trends and identify and monitor policy effects and disease burdens before they become widely prevalent. Thus, it helps identify emerging health impacts promptly and as they develop over time, so timely interventions can be developed.^43,44 Linked longitudinal data available through the EHR provide a unique opportunity to track children's health care utilization over time in combination with a rich understanding of their families’ social context.

Study limitations include the potential for sample selection bias, as participants were caregivers of young children seeking health care in EDs or primary care clinics in 3 cities, which limits the generalizability of our findings. In addition, while our sample reflected the populations served by the particular EDs and other primary care settings in the 3 Children's HealthWatch cities, it was not representative of the population of families in the United States. For example, Asian mothers are underrepresented, particularly among the subgroup of immigrant families, where they represent a large share of the national population⁴⁵ and those with public health insurance coverage are overrepresented in our sample (86% vs 35.9% of children in the US population).⁴⁶ Arkansas uniquely expanded Medicaid by using Medicaid dollars to support private insurance; therefore, some children with private insurance may have Medicaid-funded coverage. Our analyses are, therefore, likely a conservative estimate of the true impact of the election and the leaked public charge rule.

Lack of generalizability from the sampling is offset by the availability of detailed data. Further, as noted above, a key advantage of the sentinel study design is the ability to track emerging trends in public health, which made it possible to investigate possible chilling effects among a group of families that are often considered hard to reach and are underrepresented in many national studies. Thus, while future research should seek to replicate these findings in a national sample, our findings—which speak to chilling effects among a vulnerable group of American families—are important.

In addition, a critical element of our study design is the comparison of outcomes between children from immigrant and nonimmigrant families. However, the paper considers only mothers’ nativity for classifying immigrant families, and thus might misclassify families where other adults are immigrants. The inclusion of these families in our comparison group would attenuate any estimate of chilling effects, and thus we consider the results reported below to be underestimates.

Last, the EHR dataset was not collected for research but rather in clinical care and thus data missing from the dataset may or may not be missing at random; a related concern is data on children's health care use may be missing if they receive care in other locations. However, following recommendations on missing data in EHR data analysis,⁴¹ we examined children's health care utilization pattern by maternal nativity, race/ethnicity, and study period, and conducted sensitivity analyses to ensure robust results even when excluding children who dropped out. Critically, these results do not differ substantively from our main findings, lessening concern about bias due to children receiving care elsewhere.

Results

There were 50 119 visits included in the sample, representing 10 974 children (Table 1). Of those, 40.9% had immigrant mothers. A total of 13.7% of US-born vs 70.8% of immigrant mothers were Latina and 51.6% of US-born vs 25.1% of immigrant mothers were Black, non-Latina. The mean age at sample entry was 37.7 compared to 38.8 months for children of US-born and immigrant mothers, respectively. The majority of children were publicly insured: 82.4% vs 89.5% for children of US-born and immigrant mothers, respectively.

Table 1.

Sample characteristics of children of immigrant and US-born mothers, 2015–2018.

	Overall, n (%)	US-born mothers, n (%)	Immigrant mothers, n (%)	P
Total sample
Pediatric visits	50 119	27 458 (54.79)	22 661 (45.21)
Unique children	10 974	6427 (58.57)	4493 (40.94)
Site				<.0001
Boston	16 545 (33.01)	8411 (30.63)	8134 (35.89)
Little Rock	19 240 (38.39)	15 401 (56.09)	3839 (16.94)
Minneapolis	14 334 (28.60)	3646 (13.28)	10 688 (47.16)
Child characteristics
Mean (SD) age at sample entry, mo	38.19 (33.78)	37.65 (33.51)	38.84 (33.10)	<.0001
Mean (SD) age, mo	50.01 (34.03)	49.53 (33.81)	50.58 (34.29)	.0003
Health insurance				<.0001
Public	42 760 (85.65)	22 577 (82.44)	20 183 (89.54)
No insurance	2795 (5.60)	1443 (5.27)	1352 (6.00)
Private	4370 (8.75)	3365 (12.29)	1005 (4.46)
Outpatient health care utilization pattern
Consistent utilization	702 (2.89)	702 (2.56)	748 (3.30)	<.0001
Inconsistent utilization	45 776 (91.33)	24 889 (90.64)	20 887 (92.17)
Early utilization only	1679 (3.35)	1065 (3.88)	614 (2.71)
Sample entry only (dropouts)	755 (1.51)	513 (1.87)	242 (1.07)
Early utilization, left, and returned	459 (0.92)	289 (1.05)	170 (0.75)
Maternal characteristics
Mean (SD) age, y	28.90 (6.33)	27.42 (6.01)	30.69 (6.24)	<.0001
Education				<.0001
Less than high school diploma	13 395 (26.79)	3967 (14.47)	9428 (41.75)
High school diploma	17 453 (34.90)	9649 (34.19)	7804 (34.56)
Education beyond high school	19 157 (38.31)	13 805 (50.34)	5352 (23.70)
Marital status				<.0001
Single	18 419 (36.79)	13 521 (49.28)	4898 (21.65)
Married/partnered	23 064 (46.07)	9982 (36.38)	13 082 (57.82)
Separated/divorced	8577 (17.13)	3932 (14.33)	4645 (20.53)
Race/ethnicity				<.0001
Latina	19 691 (39.60)	3737 (13.73)	15 954 (70.83)
Black, non-Latina	19 691 (39.60)	14 040 (51.60)	5651 (25.09)
White, non-Latina	8491 (17.07)	8214 (30.19)	277 (1.23)
Other/multiple races	1858 (3.74)	1217 (4.47)	641 (2.85)
Household characteristics
One or more employed in household	20 740 (41.46)	12 635 (46.11)	8105 (35.83)	<.0001

Column percentages are shown. Source: Authors’ analysis of Children's HealthWatch data, 2015–2018.

Well-child visits

Pre-election trends were parallel (Figure 1). DiD analyses showed that Trump's election was associated with a 5-percentage-point decrease for both strict adherence (−0.05; 95% CI: −0.08, −0.02) and flexible adherence (−0.05; 95% CI: −0.08, −0.02) for children of immigrant compared with US-born mothers (Table 2). In the pre-election period, 53.6% of children of immigrant mothers were adherent. Thus, a 5-percentage-point decrease would equate to the percentage dropping to 48.6% (a 9% relative decrease).

Table 2.

Effect of former President Trump's election and leaked public charge draft: difference-in-difference results for well-child visit adherence, hospitalizations, and emergency department visits among children of immigrant and US-born mothers, 2015–2018.

	Difference-in-difference: 2016 election				Difference-in-difference: alternative treatment period 2016 election and leaked public charge draft
Well-child visits	n	Coefficient	95% CI		n	Coefficient	95% CI
Strict adherence
Immigrant × postelection	10 291	−0.05***	−0.08	−0.02	10 922	−0.09***	−0.12	−0.05
Flexible adherence
Immigrant × postelection	10 291	−0.05***	−0.08	−0.02	10 122	−0.08***	−0.12	−0.05
Detailed adherence
Immigrant × postelection	7728	−0.07***	−0.09	−0.05	7632	−0.07***	−0.09	−0.05
Hospitalizations
Immigrant × postelection	13 706	−0.009	−0.03	0.01	12 871	−0.01	−0.03	0.01
Emergency department visits
Immigrant × postelection	13 085	0.009	−0.009	0.03	12 871	0.01	−0.005	0.03

Models for strict adherence and flexible adherence adjusted for research site, maternal race/ethnicity, education, and marital status, any household employment, child health insurance, health care utilization pattern, and age at sample entry. Models for detailed adherence adjusted for research site, maternal race/ethnicity, education, and marital status, any household employment, child health insurance, age at sample entry, and fixed effects for each well-child visit. Models for hospitalizations and emergency department visits adjusted for research site, maternal race/ethnicity, education, and marital status, any household employment, child health insurance and age at sample entry. Source: Authors’ analysis of Children's HealthWatch data, 2015–2018.

***P ≤ 0.001.

Using the indicator for detailed adherence, we found that the election was associated with a 7-percentage-point (−0.07; 95% CI: −0.09, −0.05) decrease in WCVs for children of immigrant compared with US-born mothers (Table 2). Coefficients were negative and statistically significant for the 4-month, 12-month, 15-month, 24-month, and 48-month WCVs (Table 3).

Table 3.

Effect of former President Trump's election: difference-in-difference results for detailed adherence by well-child visit schedule among children of immigrant and US-born mothers, 2015–2018.

	Difference-in-difference: 2016 election				Difference-in-difference: alternative treatment period 2016 election and leaked public charge draft
Well-child visit schedule	n	Coefficient	95% CI		n	Coefficient	95% CI
Newborn	465	0.02	−0.14	0.18	418	0.02	−0.15	0.19
1 Month	948	−0.05	−0.15	0.04	894	−0.06	−0.16	0.04
2 Months	878	0.10	−0.003	0.19	809	0.03	−0.01	0.06
4 Months	924	−0.11*	−0.21	−0.008	858	−0.15**	−0.26	−0.05
6 Months	1067	−0.06	−0.15	0.03	1003	−0.06	−0.15	0.03
9 Months	1076	−0.06	−0.16	0.04	1028	−0.04	−0.15	0.07
12 Months	1158	−0.10*	−0.20	−0.005	1099	−0.08	−0.18	0.03
15 Months	1102	−0.12*	−0.22	−0.03	1025	−0.08	−0.19	0.03
18 Months	1258	−0.06	−0.16	0.04	1199	−0.09	−0.19	0.01
24 Months	1307	−0.17***	−0.26	−0.07	1262	−0.16**	−0.26	−0.06
30 Months	1045	0.05	−0.05	0.16	1007	0.08	−0.03	0.18
36 Months	1675	−0.01	−0.10	0.07	1642	−0.03	−0.12	0.05
48 Months/4 years	1756	−0.17***	−0.25	−0.10	1717	−0.17**	−0.25	−0.09
5 Years	1765	−0.07	−0.15	0.02	1715	−0.09*	−0.18	0.01
6 Years	1675	−0.09	−0.18	0.004	1597	−0.08	−0.18	0.01
7 Years	1499	−0.08	−0.18	0.02	1456	−0.06	−0.15	0.04
8 Years	1172	−0.02	−0.14	0.10	1144	−0.04	−0.15	0.08
9 Years	660	−0.17	−0.36	0.02	639	−0.16	−0.34	0.02
10 Years	246	−0.05	−0.45	0.34	240	−0.11	−0.47	0.26
11 Years	58	Insufficient sample size			55	Insufficient sample size
12 Years	5	Insufficient sample size			5	Insufficient sample size

Models for detailed adherence adjusted for research site, maternal race/ethnicity, education, and marital status, any household employment, and child health insurance. Source: Authors’ analysis of Children's HealthWatch data, 2015–2018.

*P = 0.05, **P = ≤0.01, ***P ≤ 0.001.

Hospitalizations and emergency department visits

DiD interaction terms for children's hospitalizations and ED visits were not significant (hospitalizations: −0.01 [95% CI: −0.03, 0.01]; ED visits: 0.009 [95% CI: −0.009, 0.03]) (Table 2).

Extended treatment period

In secondary analyses using the alternative treatment period extending from the election to the leaked draft, results for all measures of adherence were strengthened (Table 2). DiD analyses showed that, for children of immigrant compared with US-born mothers, Trump's election and the leaked draft were associated with decreases of 9 percentage points (−0.09; 95% CI: −0.12, −0.05) for strict adherence, 8 percentage points (−0.08; 95% CI: −0.12, −0.05) for flexible adherence, and 7 percentage points (−0.07; 95% CI: −0.09, −0.05) for detailed adherence. Similarly, the election and leaked draft were associated with decreases of 15 percentage points for the 4-month WCV (−0.15; 95% CI: −0.26, −0.05), 16 percentage points for the 24-month WCV (−0.16; 95% CI: −0.26, −0.06), 17 percentage points for the 48-month WCV (−0.17; 95% CI: −0.25, −0.09), and 9 percentage points for the 60-month WCV (−0.09; 95% CI: −0.18, −0.01) (Table 3). Among children in the relevant age range, 83.2% of children of immigrants in the pre-period received their 4-month WCV; thus, a drop of 15 percentage points equates to the percentage dropping to 68.2%. DiD coefficients for children's hospitalizations and ED visits were not significant (Table 2). Sensitivity analyses were consistent with main and secondary analyses for children younger than 2 years and with exclusion of dropouts; other analyses were not significant (Appendix Exhibits A16-3&4 and A20-3&4).

Discussion

We found a significant decrease in adherence to the WCV schedule across all 3 measures of WCVs for children of immigrant compared with US-born mothers after Trump's election. Secondary analyses extending the treatment period to the date of the leaked draft of the public charge rule showed greater decreases in adherence for all measures. In contrast to other studies’ findings of increased pediatric ED visits,¹⁶ we found no differences in hospitalizations and ED visits among children of immigrant compared with US-born mothers.

Trump's election followed a period of intensely negative and sometimes violent language about immigrants.^47,48 This study's findings provide further evidence that this atmosphere may have precipitated a chilling effect among immigrant and mixed-status families, leading them to avoid accessing WCVs. The consequences of missing WCVs could be profound, as children receive vaccines for both personal and community well-being, and are screened for physical well-being and developmental milestones, and parents have opportunities to ask questions and receive guidance on caring for their children.^39,49 Children screening positive for developmental delays or other health problems can be referred for further evaluation and intervention to resolve or mitigate future health problems. Moreover, other referrals for families can be identified, from behavioral health to basic needs supports. Thus, if children of immigrants missed WCVs, health and development impacts of pre-existing inequities in income and access to resources could be compounded.

Null findings for changes in pediatric hospitalization and ED visits were counter to our expectation of increased acute health care utilization among immigrant families. One possible explanation may be that the relative infrequency of hospitalizations and ED visits, compared with WCVs, required a change in behavior too extreme to detect in this age range. It is also possible that there was a chilling effect, such that, even if need increased, parents still may not feel safe accessing EDs. Unlike other studies, we did not have an immigration status measure and thus could not identify parents who were undocumented or had temporary status, groups most likely to experience the chilling effect. However, other studies have shown that even immigrants with permanent status (eg, LPRs), who theoretically need not worry about accessing acute health care, fear accessing health care for immigration reasons.^7,18 Last, since we could not track where else parents received care, it may be that parents sought acute health care elsewhere, although our results did not change when we excluded sample dropouts.

Policy implications

After the 2016 election, anti-immigrant, restrictive, and often punitive policies made it more difficult for immigrants to migrate to the United States, access needed supports and services, adjust their status, and pursue a path to citizenship and their daily lives without arrest or deportation fears.^19,50–54 In August 2019, the Trump Administration finalized and enacted a significant expansion to the public charge rule.⁵⁵ The subsequent administration vacated the revised definition of public charge and released a final rule,^56,57 clarifying a narrow public charge definition and making it more difficult for future administrations to swiftly change course,⁵⁸ although recent efforts by Congress have attempted to undercut this.⁵⁹ Work remains to regain immigrant communities’ trust.

Our study found that Trump's rhetoric and election were associated with household health care–seeking behavior, even without later concrete policy changes.^15,26,50,60 Other studies examining Trump's candidacy declaration and early rhetoric found a similar relationship with declining adult and child health care utilization and public-assistance enrollment among immigrant and mixed-status families in SNAP, the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC), school meals, and Medicaid.^{8,15,16,18,19} It is in our national interest for all children and their families to be able to meet their basic needs, including for health care. Thus, it is important for candidates for public office, executive branch, legislative, health systems, and public health leaders to be mindful of the far-reaching physical and mental health consequences of the language and policy proposals our leaders choose to emphasize as well as the implementation of restrictive laws and policies.^47,61

Conclusion

Words matter and have real-life consequences in campaigns and governance. After a presidential campaign of derogatory and often inflammatory language about immigrants and threats of policy change, Trump's election was associated with widespread negative impacts for immigrant communities, including decreasing the number of immigrant parents securing well-child care for their young children in 3 cities. Examining future changes to social service and health care law, regulation, and rhetoric through the lens of immigration policy and immigrant community impacts will create a more just and equitable society for all US children, regardless of their parents’ nativity.

Supplementary material

Supplementary material is available at Health Affairs Scholar online.

Funding

This analysis had no specific funding.

Conflicts of interest

Please see ICMJE form(s) for author conflicts of interest. These have been provided as supplementary materials.