Key PointsQuestion Was the 2017 “Muslim ban” executive order associated with changes in health care utilization by people born in Muslim-majority countries living in Minneapolis-St. Paul, Minnesota?
Findings This cohort study of 252 594 patients found that after the executive order was issued, there was an increase in missed primary care appointments and increased emergency department visits among people from Muslim-majority countries living in Minneapolis-St. Paul.
Meaning Changes in health care utilization among people from Muslim-majority countries after the Muslim ban may reflect changes in population health influenced by federal immigration policy.
Importance The health effects of restrictive immigration and refugee policies targeting individuals from Muslim-majority countries are largely unknown.
Objective To analyze whether President Trump’s 2017 executive order 13769, “Protecting the Nation from Foreign Terrorist Entry into the United States” (known as the “Muslim ban” executive order) was associated with changes in health care utilization by people born in targeted nations living in the US.
Design, Setting, and Participants This retrospective cohort study included adult patients treated at Minneapolis-St. Paul HealthPartners primary care clinics or emergency departments (EDs) between January 1, 2016, and December 31, 2017. Patients were categorized as (1) born in Muslim ban–targeted nations, (2) born in Muslim-majority nations not listed in the executive order, or (3) non–Latinx and born in the US. Data were analyzed from October 1, 2019, to May 12, 2021.
Exposures Executive order 13769, “Protecting the Nation from Foreign Terrorist Entry into the United States.”
Main Outcomes and Measures Primary outcomes included the number of (1) primary care clinic visits, (2) missed primary care appointments, (3) primary care stress-responsive diagnoses, (4) ED visits, and (5) ED stress-responsive diagnoses. Visit trends were evaluated before and after the Muslim ban issuance using linear regression, and differences between the study groups after the executive order issuance were evaluated using difference-in-difference analyses.
Results A total of 252 594 patients were included in the analysis: 5667 in group 1 (3367 women [59.4%]; 5233 Black individuals [92.3%]), 1254 in group 2 (627 women [50%]; 391 White individuals [31.2%]), and 245 673 in group 3 (133 882 women [54.5%]; 203 342 White individuals [82.8%]). Group 1 was predominantly born in Somalia (5231 of 5667 [92.3%]) and insured by Medicare or Medicaid (4428 [78.1%]). Before the Muslim ban, primary care visits and stress-responsive diagnoses were increasing for individuals from Muslim-majority nations (groups 1 and 2). In the year after the ban, there were approximately 101 additional missed primary care appointments among people from Muslim-majority countries not named in the ban (point estimate [SE], 6.73 [2.90]; P = .02) and approximately 232 additional ED visits by individuals from Muslim ban–targeted nations (point estimate [SE], 3.41 [1.53]; P = .03).
Conclusions and Relevance Results of this cohort study suggest that after issuance of the Muslim ban executive order, missed primary care appointments and ED visits increased among people from Muslim-majority countries living in Minneapolis-St. Paul.
The 2016 US presidential election was marked by anti-Muslim and anti-immigrant rhetoric, and the subsequent Trump administration introduced multiple restrictive immigration policies targeting individuals from Muslim-majority and Latin American countries.1 On January 27, 2017, President Trump issued executive order 13769, “Protecting the Nation from Foreign Terrorist Entry into the United States,”2 commonly referred to as the “Muslim ban.” The first iteration of the Muslim ban suspended the US Refugee Resettlement Program and prevented citizens from 7 Muslim-majority countries (Iraq, Syria, Iran, Libya, Somalia, Sudan, and Yemen) from traveling or immigrating to the US. The ban underwent multiple legal challenges but was upheld by the US Supreme Court in 2018.3 On January 20, 2021, President Biden repealed the Muslim ban by executive order.4
Policies like the Muslim ban exacerbate heightened levels of discrimination, hostility, and “othering” that US Muslims experience.5 Over the past 2 decades, there has been an increase in hate crimes6 and social hostility7,8 directed toward US Muslims—experiences that negatively affect health. Following the September 11 attacks, Arab Americans, including Muslim Arab Americans, experienced increased rates of anxiety, depression, and low birth weights.5,9-12 Islamophobia and restrictive entry policies have been associated with worse health outcomes among migrants, placing Muslim American immigrants and refugees at increased vulnerability as they are additionally targeted by immigration and refugee policy changes, above and beyond discrimination faced by US Muslims more broadly.13-15 Recent research has shown increased incidence of preterm birth among people from Muslim-majority countries after the Muslim ban was issued.16 However, it is largely unknown and can be difficult to measure how health and health care utilization in Muslim American immigrant and refugee communities changes in response to shifting sociopolitical climates due to several methodological challenges. National health and health care surveys, as well as administrative data sets, do not routinely capture religious affiliation, and naming and country-of-origin algorithms are not precise enough to distinguish people who are Muslim.17,18
To describe changes in health care utilization of people from Muslim-majority countries after enactment of the Muslim ban, we examined changes in primary care and emergency department (ED) utilization by people from Muslim ban–targeted nations living in the Minneapolis-St. Paul, Minnesota, metropolitan area, which is home to the largest Somali Muslim community in the US.19
We conducted a retrospective, cohort study documenting trends in primary care and ED utilization, missed scheduled appointments, and stress-responsive conditions among individuals from Muslim ban–targeted nations living in Minneapolis-St. Paul 1 year before (January 1, 2016) to 1 year after (December 31, 2017) issuance of the ban. For outcomes with similar visit and diagnosis trends before the ban was issued, we used a difference-in-difference analysis to compare differences in utilization trends between people born in Muslim ban–targeted nations and non–Latinx US-born citizens. Supplementary analyses compared trends among people born in a Muslim-majority nation not listed in the Muslim ban to non–Latinx US-born citizens. Patient demographic, visit, and diagnosis data were extracted from the HealthPartners electronic health record (EHR) by a HealthPartners data analyst. HealthPartners and the institutional review board of Yale University approved this study and waived the need for patient informed consent. All records were deidentified and assigned a unique study identification number before secure filing from HealthPartners to the Yale analytic team. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.20
Study Setting and Population
The Minneapolis-St. Paul metropolitan area has 3.63 million residents and the largest Somali Muslim population in the US. In 2017, approximately 252 000 area residents (59.3%) were White non-Latinx individuals, 81 900 (19.3%) were Black non-Latinx individuals, 26 800 (6.3%) were Asian non-Latinx individuals, and 40 900 (9.6%) were Latinx individuals.21 In 2016, there were approximately 17 889 people born in Muslim ban–targeted nations living in Minneapolis-St. Paul, of whom 15 808 (88.4%) were born in Somalia.21
We analyzed EHR data from HealthPartners, one of Minneapolis-St. Paul’s largest health care and insurance organizations, serving over 1.2 million patients at 55 primary care centers, 22 acute care centers, and 8 hospitals. Although religion is not recorded in the EHR, the HealthPartners EHR is unique in that it includes nation of origin information. This allowed us to characterize patients receiving care from January 1, 2016, to December 31, 2017, into 3 groups: (1) adults born in a country listed in the executive order (group 1) (Table 1; eTable 1 in the Supplement); (2) adults born in Muslim-majority nations not listed in the executive order (group 2) (Table 1, eTable 2 in the Supplement); and (3) US-born non–Latinx adults (Group 3) (Table 1). We excluded US-born Latinx patients (persons who identify as Latino, Latina, or Hispanic, or having lineage from Mexico or any nation in Central or South America) as they have been subject to distinct anti-immigrant rhetoric and policies that have important effects on their health and health care utilization.22,23
We hypothesized that after issuance of the Muslim ban, we would observe increased health care visits for stress-responsive diagnoses, increased primary care missed appointments, and increased ED visits. Primary outcomes included the number of (1) primary care visits, (2) missed primary care appointments, (3) primary care stress-responsive diagnoses, (4) ED visits, and (5) ED stress-responsive diagnoses.
Primary care visits, missed appointments, and stress-responsive diagnoses were analyzed as counts per 1000 people. Kept and missed visit trends were examined overall, regardless of diagnosis, and stress-responsive diagnoses were analyzed separately. We identified stress-responsive diagnoses through literature review and opinion by study team members with expertise in Muslim, immigrant, and refugee health and who provide primary care to people from Muslim-majority countries living in Minneapolis-St.Paul.9,23-30 Diagnoses considered to be stress responsive were agreed upon by consensus (eTable 3 in the Supplement) and included 138 International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10), codes in 6 categories: mental health, sleep disorders, gastrointestinal symptoms, neurologic symptoms, food-related disorders, and pain syndromes.
Emergency Department Utilization
Overall ED visits and ED stress-responsive diagnoses were also analyzed as counts per 1000 people. As previously described, ED stress-responsive diagnoses were identified through literature review and expert opinion, were agreed upon by consensus, and included 27 ICD-10 codes for acute coronary syndrome, assault, suicide attempt, and syncope (eTable 4 in the Supplement)9,23-32 as well as ambulatory sensitive conditions. Ambulatory sensitive conditions are conditions for which an ED visit or hospitalization is considered preventable through outpatient interventions and can be exacerbated by social stressors and inequalities.31,33 Ambulatory sensitive diagnoses included 21 ICD-10 codes for angina, asthma, congestive heart failure, chronic obstructive pulmonary disease, diabetes complications, and hypertension (eTable 4 in the Supplement).31,33
We used local linear regression to characterize visit and diagnosis trends in 2016 and 2017 and fit separate models for all outcomes: primary care visits, missed primary care appointments, primary care stress-responsive diagnoses, ED visits, and ED stress-responsive diagnoses. After documenting changes over time, we identified outcomes for which trends appeared to be similar among non–Latinx US-born individuals and individuals from Muslim ban–targeted nations in the preintervention period by examining the interaction between study group and time (30-day periods) in linear regression models. For outcomes that followed approximately parallel trends before the ban was issued, we conducted difference-in-difference analyses.
The primary difference-in-difference analyses estimated the change in outcomes between pre- and post-ban periods among individuals from Muslim ban–targeted nations above and beyond the change observed among non–Latinx US-born individuals. For the difference-in-difference analyses, we fit the linear regression model described in equation 1:
Y = β0 + (β1Muslim ban targeted) + (β2Muslim ban targeted × post-Muslim ban) + (βt period) + ε.
We compared the 360 days before and after the ban was issued, divided into 24 distinct, 30-day periods. Each outcome Y is a count per person per 30-day time period. The designations Muslim ban targeted and Muslim ban targeted × post-Muslim ban represent being from a Muslim ban–targeted nation and being from a Muslim ban–targeted nation after the ban was issued, respectively. For each model, we estimated the average difference in differences (β2) over increasing time intervals centered on the ban issuance date, beginning with 30 days before and after the ban was issued and increasing by 30-day increments to 360 days before and after issuance.
Demographic details were not included in the primary analysis because they do not vary over time. However, it is important to account for demographic details, because there are important differences between study groups that could influence health care utilization. In order to account for different characteristics between study groups and assess robustness, we compared changes in health care utilization by individuals from Muslim ban–targeted nations to 2 subgroups of non–Latinx US-born individuals demographically similar to individuals in group 1 (Muslim ban-targeted): a matched comparison group and a synthetic control. First, we matched on age, sex, race/ethnicity, and insurance type (Medicaid or Medicare, or commercial) to reweight members of group 3 (non–Latinx, US born) and identify a subset of group 3 with similar characteristics as group 1 (Muslim ban–targeted). We used the R package MatchIt, version 3.6.1, (R Core Team) to identify this reference group, then fit a weighted version of the model described in equation 1 (eTable 10 in the Supplement).34 Second, we used a generalized synthetic control method to reweight members of group 3 to produce a reference group with demographic characteristics and pre–Muslim ban outcomes more similar to those observed in group 1. Models were fit using the R package gsynth with parametric bootstrap standard errors.35 These robustness checks are reported in eTables 8-10 and in the eFigure in the Supplement. All P values were 2-sided, and P < .05 was considered significant. Data were analyzed from October 1, 2019, to May 12, 2021.
READ ON
Health Care Utilization Before and After the “Muslim Ban” Executive Order Among People Born in Muslim-Majority Countries and Living in the US | Emergency Medicine | JAMA Network Open | JAMA Network