Study sample and recruitment
This paper uses survey data collected between September 2020 to February 2021 from the COVID-19 BRAVE (Building community Raising All immigrant Voices for health Equity) Study, which aimed to examine the social, economic, and health impacts of COVID-19 among undocumented young adult immigrants in California. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines. Our study was approved by the Institutional Review Board at the University of California, Los Angeles.
We used a community-engaged approach by consulting a Community Advisory Board (CAB) composed of undocumented individuals and experts in health policy, education, and immigrant advocacy throughout survey development, recruitment, and data interpretation processes. Participants were recruited through community and school-based organizations with established histories of serving undocumented immigrants. Community partners recruited through their email list serves and flyers; additionally, to minimize potential selection bias for immigrants embedded in partner networks, instructions on accessing password-protected surveys were also advertised on social media. Each community partner provided a unique password for the survey to minimize fraudulent participation. Participants received a $10 gift card for participation. Emails were delinked from study data. Screening and surveys were administered online through Qualtrics. Eligible participants were: 1) undocumented; 2) Asian and/or Latinx; 3) aged 18–39 years; 4) live in California; and 5) able to take a 15-min online survey in English or Spanish. An original goal of the overall BRAVE Study was to examine how Deferred Action for Childhood Arrivals (DACA) influences immigrant health. DACA is a major immigration policy, an Executive Order signed by the Obama Administration in 2012, which granted certain undocumented young adults up to age 30 years the right to work and live in the US. As 2022 marks the 10th anniversary of DACA, we include up to age 39 years to include the earliest DACA recipients. All participants provided informed consent and all participants chose to take the survey in English.
A non-random sample of 438 participants were emailed a link to the survey, and 366 respondents completed the survey (83.5% participation rate). An additional 163 were screened but ineligible. Post-validation checks were performed including assessments of duration of time taken to complete survey (those who spent < 5 min on survey were excluded), assessment of concordance of items that follow logical flow (i.e. established DACA eligibility requirements met for those who have DACA (i.e., currently a DACA recipient [or in renewal pending status], expired lawful immigration as of June 15, 2012, living in the U.S. for at least 5 years prior to June 15, 2012, and not having a felony or a significant misdemeanor conviction). We excluded 24 participants who did not pass these validation checks. We have applied similar procedures to ensure sample validity in other studies [23, 24]. We excluded an additional 16 respondents who did not complete one or more questions included in these analyses. Missing variables are minimal and all less than 3%. The final analytic sample included 326 participants.
Respondents were asked the following questions about possible illness and testing: (a) Have you ever had, or thought you might have had, the Coronavirus, COVID-19? (referred to as had/suspected having COVID-19) (b) Did you ever receive a positive test result for COVID-19? (tested positive) (c) Did you delay or avoid getting tested and/or seeking treatment for COVID-related symptoms because of your immigration status? (avoided testing). Each question was a binary yes/no response.
Immigration enforcement encounters were the primary exposures of interest. Respondents were asked to indicate their encounters with the immigration system, immigration authorities, and law enforcement by reporting if: (a) there was ever a time they decided not to apply for one or more non-cash government benefits because of worries it would disqualify them, or a family member, from obtaining a green card or becoming a U.S. citizen; (b) they or someone they know experienced an immigration raid at work or at home; (c) someone they know had ever been detained or deported by immigration authorities; (d) they had ever faced deportation proceedings; (e) there was ever a time they decided not to leave their house or stayed away from certain areas to avoid the police or immigration authorities; (f) there was ever a time they decided to avoid travelling by car, bus, train, or plane to avoid internal checkpoints or TSA authorities; (g) they had ever been watched by a law enforcement officer on the street or a public place; (h) they had ever been stopped for no good reason by law enforcement; (i) they had ever been asked to show proof of their citizenship or legal status by a police officer or other law enforcement authority; (k) they had seen immigration authorities in their neighborhood, and if (k) they fear getting deported, reported as “all of the time,” “most of the time,” “some of the time,” and “no, I do not.” This question was dichotomized as 0 (no, I do not) and 1 (all/most/some of the time). All other immigration enforcement encounter questions were a binary yes/no response. The first question regarding applying for public benefits measures fears of the public charge rule. Under the Public Charge rule, those accessing public benefits and deemed likely or liable to become a public charge may be denied visas or inadmissible to the US.  immigration enforcement score was created using respondents’ total affirmative responses to each question (mean = 3.52, SD = 2.06) and ranged from 0–9, with higher scores indicating more immigration enforcement encounters. These questions were adapted from the Research on Immigrant Health and State Policy Study survey which aimed to develop measures of immigration enforcement experiences, including surveillance, policing, and deportatioon . These measures have been used in other studies with undocumented immigrants [20, 26].
Independent variables were comprised of demographic, socioeconomic, and health insurance measures. Demographic characteristics, included gender, which was reported as female and male. Race and ethnicity were reported as Latino or Asian, DACA status was coded as no DACA or DACA, and age was a categorical variable, 18–24, 25–30 and 31 + years. Socioeconomic variables included highest level of education, reported as high school or less than high school, some college/community college, and college or graduate school. Respondents indicated yes or no to questions about their employment status, school enrollment, and speaking English at home. Respondents reported their health insurance status and were asked to specify if it was a county health plan, Medi-Cal, school health plan, private/employee health plan or other health insurance. This categorical variable was dichotomized, so that “no” represented uninsured and “yes” included all other responses.
Descriptive and bivariate analyses were conducted to examine means, frequencies, and associations. We used multivariable logistic regression models to assess the association between immigration enforcement encounters and COVID-19 testing and treatment outcomes, adjusting for covariates. We included covariates that were theoretically associated with our main predictor and outcomes of interest. We then assessed bivariate associations using a p-value cut-off of 0.2. For a few covariates, such as gender, education level, and currently in school, we included these variables based on theoretical and conceptual importance even if they did not meet the 0.2 p-value cut-off. Finally, we assessed for collinearity using a variance inflation factor. We fit interaction models to assess differences between immigration enforcement encounters and the outcomes by race; however, we do not report those findings as there were no statistically significant results. We conducted sensitivity analyses where we used the continuous immigration enforcement score as a binary measure: (a) cut off at the mean (b) cut off at the median; and as (c) zero or one or more enforcement encounters. We observed similar overall patterns, net of covariates. Reporting one or more immigration enforcement encounters was significantly associated with higher odds of suspecting COVID-19 and reporting three or more encounters was significantly associated with higher odds of delaying or avoiding testing and/or seeking treatment for COVID-related symptoms because of immigration status [data not shown, available upon request]. All analyses were conducted using Stata version 15 and statistical significance was set at p < 0.05.