Abstract
Coronavirus disease 2019 (COVID-19) has disproportionately and negatively affected communities of color in the United States, especially Black, Latinx, and Indigenous populations. We report a cluster of COVID-19 cases among the Maya in Alameda County, California, most of whom were misclassified in public health data as nonindigenous Spanish-speaking people. We conducted a retrospective cohort study of all COVID-19 tests performed from April 1 through May 31, 2020, at Alameda Health System. A total of 1561 tests from 1533 patients were performed, with an overall test positivity rate of 17.0% (N = 265). We used the language field from the electronic health record to identify 29 patients as speaking an indigenous Mayan language; by medical record review, we identified 52 additional Maya patients. Maya patients had a test positivity rate of 72.8% as compared with 27.1% (P < .001) for nonindigenous Latinx patients and 8.2% (P < .001) for all other patients. In our sample, 39.6% of patients who had a positive test result for COVID-19 were hospitalized, 11.3% required admission to the intensive care unit (ICU), and 4.9% died of COVID-19. Maya patients had lower rates of hospitalization, ICU admission, and 30-day in-hospital mortality than non-Maya patients. We shared our data with the county health department to inform responses for education, testing, and isolation for Maya patients in Alameda County. Ongoing COVID-19 public health efforts should assess the community prevalence of COVID-19 in the Maya community and other indigenous communities and implement interventions that are linguistically and culturally appropriate.
Keywords: COVID-19, health disparities, indigenous communities, public health
Coronavirus disease 2019 (COVID-19) has disproportionately affected communities of color in the United States, especially Black, Latinx, and indigenous populations. 1,2 In California, the prevalence of disease has been disproportionate among Latinx patients; as of October 18, 2020, Latinx people comprised 39% of the state population but 61% of the state’s COVID-19 cases. 3 In response, local health departments have tailored COVID-19 testing and prevention strategies in Latinx neighborhoods, with a focus on messaging in Spanish. 4
In Alameda County, California, clinicians and community-based organizations (CBOs) that comprise the Alameda County COVID-19 Latinx Task Force noted in May 2020 a high prevalence of COVID-19 infection among people in the local Maya community who primarily speak Mam, an indigenous Mayan language, and who often speak little or no Spanish. Testing data submitted to the Alameda County Public Health Department (ACPHD), however, appeared to misclassify most Maya patients as Spanish speaking, thereby hindering an actionable, language-appropriate local public health response to the COVID-19 pandemic.
An estimated >500 000 Maya have immigrated to the United States during the past 20 years, and approximately 10 000 people from the Maya Mam ethnic group live in Oakland, California. 5,6 Based on data from Alameda County Care Connect using health insurance payor sources, it is estimated that Maya patients access care primarily through Alameda Health System (AHS), the county’s public health system; approximately half of the Maya patients in Alameda County access ambulatory care through AHS, and nearly all emergency department and inpatient encounters among Maya patients are at AHS hospitals (personal communication, Sheilani Alix, Alameda County Care Connect, August 2020).
In partnership with CBOs, ACPHD, Alameda County Care Connect, and Health Care for the Homeless, a team of health care providers at AHS joined the Alameda County COVID-19 Latinx Task Force in early May 2020 in an effort to assess the cluster of COVID-19 cases identified in the Maya community. We used data from the electronic health record (EHR) to determine the test positivity rate among Maya patients presenting to AHS for COVID-19 testing. The purpose of this effort was to direct public health resources toward linguistically and culturally appropriate testing, isolation, and quarantine procedures for Maya patients in Alameda County.
Methods
Design
We conducted a retrospective cohort study of all COVID-19 tests performed from April 1 through May 31, 2020, at AHS. We shared summary findings, including the absolute number and test positivity rate of Maya patients with COVID-19, with the public health department and CBOs. We de-identified all data before analysis. The AHS Institutional Review Board approved this project as exempt from review.
Setting and Population
A total of 444 COVID-19 cases were reported to ACPHD by April 1, 2020, and 3433 COVID-19 cases by May 31, 2020. 7 During the study period, COVID-19 testing was limited to patients with signs and symptoms of COVID-19. Surveillance testing for asymptomatic patients was not available because of limited testing supplies. AHS is a public health system with 5 acute care hospitals and 4 primary care clinics and serves as the safety-net health system for Alameda County. In 2019, 36% of patients at AHS self-identified as Hispanic/Latinx, 25% as Black/African American, 11% as Asian, 15% as White, 3.5% as Native Hawaiian/Other Pacific Islander, and 1% as American Indian/Alaska Native (unpublished data, AHS, 2019).
Outcome Measures
The primary outcome was the COVID-19 test positivity rate among indigenous Maya patients as compared with non-Maya Latinx patients and all patients tested. Secondary outcomes were the frequencies of comorbidities identified by the Centers for Disease Control and Prevention as risk factors for severe COVID-19 disease, 8 proportion of patients admitted to the hospital or the intensive care unit (ICU), and the 30-day in-hospital mortality rate for patients with COVID-19. We also conducted a sensitivity analysis of hospitalization rates for Maya and non-Maya patients without medical comorbidities to assess clinical outcomes and disparities between the populations.
Data Abstraction
We de-identified patients’ COVID-19 test results and data on self-identified race, ethnicity, and language from the EHR into REDCap, a secure web-based application. 9 We reviewed the EHR for mention of common Mayan languages as the patient’s primary language among patients who self-identified as Hispanic/Latinx or as primary Spanish-language speakers. Because more than 30 indigenous Mayan languages exist, we developed a list of the most common dialects spoken by the local Maya community (Mam, Maya, K’iche’, Q’eqchi’, Kaqchikel). We developed the list in collaboration with CBOs and the Mam language interpreter at AHS. We also collected data from the EHR on medical comorbidities, inpatient level of care, and 30-day in-hospital mortality. We added the information to the REDCap data set for patients diagnosed with COVID-19. We adhered to STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) guidelines for observational studies. 10
Data Analysis
We de-identified data from REDCap and exported data into Stata version 15.1 (StataCorp LLC). We reported continuous variables as medians and means and categorical variables as proportions or percentages. We made comparisons by using Fisher exact and Mann-Whitney tests between outcomes variables. We considered P < .05 to be significant for comparisons between data points.
Outcomes
From April 1 through May 31, 2020, AHS performed 1561 COVID-19 tests on 1533 patients at AHS, with an overall test positivity rate for patients of 17.0% (N = 265). Of the 1561 COVID-19 tests conducted, 561 (35.9%) were from patients who self-identified as Latinx (Maya and non-Maya), 552 (35.4%) as Black or African American, 230 (14.7%) as White, 140 (9.0%) as Asian or Pacific Islander, and 7 (0.4%) as American Indian/Alaska Native or Native Hawaiian/Other Pacific Islander. Of the 265 patients who tested positive for COVID-19, the mean age was 50.5, 108 (40.8%) were female, and 146 (55.1%) were enrolled in Medicaid or the county health insurance program (Table 1).
Table 1.
Demographic characteristics of patients tested for COVID-19, stratified by indigenous, Latinx, and non-Latinx ethnicity, Alameda Health System, California, April 1–May 31, 2020 a
| Characteristic | All COVID-19 tests (N = 1561) | COVID-19 tests among non-Maya Latinx patients (n = 480) | COVID-19 tests among Maya patients (n = 81) | COVID-19 tests among non-Latinx patients (n = 1000) | ||||
|---|---|---|---|---|---|---|---|---|
| Negative (n = 1296) | Positive (n = 265) | Negative (n = 350) | Positive (n = 130) | Negative (n = 22) | Positive (n = 59) | Negative (n = 924) | Positive (n = 76) | |
| Age, mean (median), y | 55.5 (55.3) | 50.5 (46.9) | 50.0 (45.7) | 47.7 (45.1) | 60.1 (38.1) | 48.8 (42.1) | 57.4 (58.1) | 56.5 (56.5) |
| Sex | ||||||||
| Female | 625 (48.2) | 108 (40.8) | 171 (48.9) | 59 (45.4) | 14 (63.6) | 27 (45.8) | 440 (47.6) | 22 (28.9) |
| Male | 671 (51.8) | 157 (59.2) | 179 (51.1) | 71 (54.6) | 8 (36.4) | 32 (54.2) | 484 (52.4) | 54 (71.1) |
| Health insurance | ||||||||
| Medicaid | 712 (54.9) | 111 (41.9) | 186 (53.1) | 50 (38.5) | 18 (81.8) | 32 (54.2) | 508 (55.0) | 29 (38.2) |
| Medicare | 310 (23.9) | 35 (13.2) | 33 (9.4) | 8 (6.2) | 1 (4.5) | 1 (1.7) | 276 (29.9) | 26 (34.2) |
| County b | 49 (3.8) | 35 (13.2) | 45 (12.9) | 26 (20.0) | 0 | 6 (10.2) | 4 (0.4) | 3 (3.9) |
| Commercial | 51 (3.9) | 5 (1.9) | 16 (4.6) | 3 (2.3) | 0 | 0 | 35 (3.8) | 2 (2.6) |
| Other or unknown c | 174 (13.4) | 79 (29.8) | 70 (20.0) | 43 (33.1) | 3 (13.6) | 20 (33.9) | 101 (10.9) | 16 (21.1) |
Abbreviation: COVID-19, coronavirus disease 2019.
aAll values are presented as number (percentage) unless otherwise indicated.
bRefers to the Health Program of Alameda County, a county program that provides affordable health care to uninsured people living in Alameda County. To be eligible, an individual must be an Alameda County resident and be living at 0%-200% of the federal poverty level, not be eligible for state Medicaid, and not be enrolled in private health insurance.
cIncludes other government-provided health insurance, worker’s compensation, and unknown health insurance type.
We identified from the language field in the EHR 29 patients as speaking an indigenous Mayan language; from a medical record review, we identified 52 additional patients as Mayan. All Maya patients spoke Mam, except 1 patient who spoke Maya. Maya patients had a test positivity rate of 72.8% (59 of 81), as compared with 27.1% (130 of 480; P < .001) for non-Maya Latinx patients and 7.6% (76 of 1000; P < .001) for all other patients (Table 1).
The mean age of Maya patients with a positive COVID-19 test result was 48.8, 27 (45.8%) self-identified as female, and 38 (64.4%) were enrolled in state Medicaid or the county health insurance program. The mean age of non-Maya Latinx patients with a positive COVID-19 test result was 47.7, 59 (45.4%) self-identified as female, and 76 (58.5%) were enrolled in state Medicaid or the county health insurance program.
Most Maya patients with a positive COVID-19 test result (66.1%; 39 of 59) had no medical comorbidities, whereas 35.0% (72 of 206) of non-Maya patients with a positive COVID-19 test result had no medical comorbidities (P < .001; Table 2). Overall, 105 of 265 (39.6%) patients with a positive COVID-19 test result in our sample were hospitalized: 30 (11.3%) were admitted to the ICU and 13 (4.9%) died in hospital within 30 days. Clinical outcomes were significantly more favorable among Maya patients than among non-Maya patients. One-quarter (27.1%; 16 of 59) of Maya patients compared with half (50.9%; 105 of 206) of non-Maya patients were hospitalized (P = .03), 1.7% (1 of 59) of Maya patients were admitted to the ICU as compared with 14.1% (29 of 206) of non-Maya patients (P = .01), and no Maya patients died, whereas 6.3% (13 of 206) of non-Maya patients died (P = .01). Hospitalization rates of Maya and non-Maya patients with no known medical comorbidities were not significantly different from each other (20.5% vs 15.3%; P = .60).
Table 2.
Comorbidities and clinical course of Maya and non-Maya patients with a positive COVID-19 test result, by indigenous Maya identity, Alameda Health System, California, April 1–May 31, 2020 a
| Characteristics and outcomes | All patients (N = 265) | Non-Maya patients (n = 206) |
Maya patients (n = 59) | P value b |
|---|---|---|---|---|
| Comorbidities | ||||
| Cardiac disease c | 24 (9.1) | 24 (11.7) | 0 | .01 |
| Renal disease d | 28 (10.6) | 26 (12.6) | 2 (3.4) | .04 |
| COPD or asthma | 21 (7.9) | 20 (9.7) | 1 (1.7) | .045 |
| Transplant recipient | 1 (0.4) | 1 (0.5) | 0 | .59 |
| Sickle cell disease | 1 (0.4) | 1 (0.5) | 0 | .59 |
| Diabetes | 61 (23.0) | 54 (26.2) | 7 (11.9) | .02 |
| Hypertension | 72 (27.2) | 69 (33.5) | 3 (5.1) | <.001 |
| Obesity | 57 (21.5) | 47 (22.8) | 10 (16.9) | .33 |
| Cirrhosis | 15 (5.7) | 13 (6.3) | 2 (3.4) | .40 |
| None | 111 (41.9) | 72 (35.0) | 39 (66.1) | <.001 |
| Clinical outcomes | ||||
| Hospitalized | 105 (39.6) | 89 (43.2) | 16 (27.1) | .03 |
| Intensive care unit | 30 (11.3) | 29 (14.1) | 1 (1.7) | .01 |
| 30-day in-hospital mortality | 13 (4.9) | 13 (6.3) | 0 | .01 |
Abbreviations: COPD, chronic obstructive pulmonary disease; COVID-19, coronavirus disease 2019.
aAll values are number (percentage). Percentages total to >100 across comorbidities because some patients had multiple comorbidities.
bSignificance between Maya and non-Maya patients was determined using a 2-sided Fisher exact test, with P < .05 considered significant.
cIncludes coronary artery disease and congestive heart failure.
dIncludes stage 3 chronic kidney disease and higher.
Lessons Learned
During April–May 2020, the COVID-19 test positivity rate among Maya patients seeking care at a safety-net health system was 3 times that of non-Maya Latinx patients and nearly 10-fold that of non-Latinx patients. The test positivity rate exceeded 70% among Maya patients and points to a substantial outbreak and undertesting in this community. Maya patients were significantly less likely than non-Maya patients to have medical comorbidities, be hospitalized, or be admitted to the ICU, and no Maya patients died of COVID-19 during the study period. Although studies on inequities in COVID-19 prevalence among Latinx communities have been published, 1,11,12 to our knowledge, no literature is available on the rates of COVID-19 infection in indigenous Maya communities.
Misclassification of Indigenous Maya as Nonindigenous Spanish Speakers
Indigenous communities may be undercounted or uncounted in COVID-19 registries. In our study, nearly two-thirds of Maya patients were misclassified as native nonindigenous Spanish speakers, an injustice common to indigenous communities in the United States. 13 When seeking health care, many Maya patients ask for a Spanish interpreter because they have had experiences in the United States where this is the only option available. This common practice is not only inadequate, because many patients speak little or no Spanish, but also leads to misclassification of language preference and indigenous identity in the EHR and in data collected by the public health department. Indigenous Maya may speak 1 of more than 30 Mayan dialects, which makes providing adequate interpretation services challenging. This phenomenon in the Maya community has received attention during the ongoing immigration crisis at the southern US border. Six children died at the southern US border between 2016 and 2020, 5 of whom were indigenous Maya from Guatemala; the children’s parents received medical screening forms in English with a Spanish translation, a language these families did not speak fluently. 14 Access to linguistically appropriate care for the Maya community is critical during the COVID-19 pandemic to educate and empower the community and to curb transmission of the virus.
Indigenous Maya Community at Increased Risk for COVID-19
Indigenous Maya communities are at an increased risk for COVID-19 transmission for several reasons. Many live in high-occupancy housing and have jobs as essential workers in high-exposure, low-wage jobs; one CBO estimated that up to 70% of the day laborers in Oakland, California, were Maya. 15 Given financial constraints, compliance with shelter-in-place regulations and the ability to adequately isolate or quarantine are difficult. Maya patients had high rates of enrollment in public health insurance programs, suggesting that strategies for prevention, testing, and contact tracing for this community could leverage these payor sources.
Although we did not collect data on immigration status, fears of immigration enforcement may have served as a barrier to seeking care or engaging with health systems. One study of Latinx patients from Alameda County demonstrated that undocumented immigrants may delay seeking emergency care for up to 2 or 3 days because of fear of deportation. 16 Although Maya patients in our study had less severe disease than other groups, fears of COVID-19 may outweigh concerns about immigration status.
We found that although the prevalence of COVID-19 among Maya patients was high, the disease severity in this community was lower than among non-Maya patients. This discrepancy was likely due to the younger age and lower rates of medical comorbidities among Maya patients than among non-Maya patients; two-thirds of Maya patients had no medical comorbidities compared with fewer than half of non-Maya patients. In a sensitivity analysis of patients without medical comorbidities, Maya and non-Maya patients had similar rates of hospitalization.
Public Health Collaboration and Response
At AHS, a patient’s primary language is assigned at the time of registration based on a brief registration questionnaire conducted by a registration clerk. Early in the COVID-19 pandemic, health care providers at AHS and at CBOs noted an increase in the number of patients with a positive COVID-19 test result among Maya Mam–speaking patients. However, the primary language for these patients was incorrectly assigned as Spanish in most cases. As a first step, we alerted ACPHD of the cluster of cases in the local Maya community and provided a list of Mam-speaking patients who received a positive test result for COVID-19, of whom approximately 80% were misclassified as Spanish-speaking in the county registry. We then worked with CBOs and the Alameda County COVID-19 Latinx Task Force and determined the need to conduct a cohort study using the EHR to gather accurate data on language for indigenous patients.
The data collected as part of our study informed county officials and CBOs about the extent of the COVID-19 cluster in the Maya community. We surmised that the test positivity rate exceeding 70% was evidence that COVID-19 was underdiagnosed and likely widespread in the Maya community of Alameda County. CBOs, Alameda County Health Care for the Homeless, and the ACPHD used these data to develop linguistically appropriate interventions, target testing strategies, and acquire funds to hire additional Mayan language interpreters and contact tracers.
Challenges
Although this case study represents a collaboration among a safety-net health system, CBOs, and a county public health department, substantial challenges remain in mitigating COVID-19 transmission in Alameda County. Among the Maya community, mistrust of the medical system exists because of historical trauma and mistreatment. For example, community leaders have reported that some Maya believe it is possible to contract COVID-19 at testing sites (personal communication, Carmelina Calmo, ACPHD, August 2020). At the institutional level, Alameda County resources are stretched thin, similar to many areas where community transmission of COVID-19 in marginalized communities is widespread. Moreover, the Maya community faces economic realities that put the community at increased risk for infection with COVID-19. In addition, public health data collection does not require granular reporting beyond threshold languages and ethnicities, making dissemination of accurate data for the Maya community challenging. Therefore, data available through partnerships with health systems that use ethnicity and language information from the medical record may allow for more precise identification of clusters among indigenous communities.
This case study had several limitations. First, this was a retrospective cohort study in a single safety-net health system. Although AHS provides a substantial proportion of care to immigrant communities in Alameda County, many patients are tested outside the health system. Testing for COVID-19 in Alameda County was inequitably distributed during the study period, and few testing sites were available for many lower-income communities. Second, the cross-section of patients in this study likely overrepresents patients with severe symptoms and immigrant and communities of color as compared with county data as a whole. Third, our methodology for medical record review likely underestimated the number of Maya patients, because many patients may not have indicated their indigenous ethnicity during the clinical encounter in a health care setting due to a history of systemic racism. Fourth, our cohort study presents the test positivity rate as a surrogate marker for disease transmission, and a seroprevalence study would provide more accurate data on the extent of the disease in this community.
These data support other well-documented inequities in the COVID-19 pandemic and highlight a previously unreported outbreak in the Maya community in Alameda County, most of whom were misclassified as Latinx Spanish speakers. Health systems and public health departments should work collaboratively to accurately identify patients who speak indigenous and other languages in the United States that are not commonly spoken. Ongoing COVID-19 public health strategies should tailor testing, prevention, and isolation strategies using linguistically and culturally appropriate techniques in collaboration with community partners and stakeholders.
Acknowledgments
The authors acknowledge the Alameda County COVID-19 Latinx Task Force, Alameda County Care Connect, Alameda County Health Care for the Homeless, Alameda County Public Health Department, Street Level Health, Gabriela Galicia, Carmelina Calmo, Laura Lopez, Damon Francis, MD, Kathleen Clanon, MD, Sheilani Alix, and Matt Beyers for their collaborative work with the Maya community in Alameda County. The authors also acknowledge the helpful comments by 2 anonymous reviewers.
Footnotes
Declaration of Conflicting Interests: The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding: The authors received no financial support for the research, authorship, and/or publication of this article.
ORCID iD
Erik S. Anderson, MD https://orcid.org/0000-0001-7759-4327
References
- 1. Webb Hooper M., Nápoles AM., Pérez-Stable EJ. COVID-19 and racial/ethnic disparities. JAMA. 2020;323(24):2466-2467. 10.1001/jama.2020.8598 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Price-Haywood EG., Burton J., Fort D., Seoane L. Hospitalization and mortality among Black patients and White patients with COVID-19. N Engl J Med. 2020;382(26):2534-2543. 10.1056/NEJMsa2011686 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. California Department of Public Health . COVID-19 race and ethnicity data. Accessed October 20, 2020. https://www.cdph.ca.gov/Programs/CID/DCDC/Pages/COVID-19/Race-Ethnicity.aspx
- 4. University of California San Francisco . Initial results of Mission District COVID-19 testing announced. Accessed August 1, 2020. https://www.ucsf.edu/news/2020/05/417356/initial-results-mission-district-covid-19-testing-announced
- 5. LeBaron A. Maya in the US: the Maya Heritage Community Project. Accessed June 10, 2020. https://lcfgeorgia.org/maya-in-the-us-by-alan-lebaron-phd-the-maya-heritage-community-project
- 6. Garofoli J. We are invisible: coronavirus just made census counting harder in the Bay Area. San Francisco Chronicle. April 16, 2020. Accessed June 10, 2020. https://www.sfchronicle.com/politics/article/We-are-invisible-Coronavirus-just-made-15203827.php
- 7. Alameda County Health Care Services Agency, Public Health Department . COVID-19 dashboard. Accessed August, 1 2020. https://ac-hcsa.maps.arcgis.com/apps/opsdashboard/index.html#/1e0ac4385cbe4cc1bffe2cf7f8e7f0d9
- 8. Centers for Disease Control and Prevention . Coronavirus disease 2019 (COVID-19): people with certain medical conditions. Accessed July 1, 2020. https://www.cdc.gov/coronavirus/2019-ncov/need-extra-precautions/people-with-medical-conditions.html
- 9. Harris PA., Taylor R., Thielke R., Payne J., Gonzalez N., Conde JG. Research electronic data capture (REDCap)—a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform. 2009;42(2):377-381. 10.1016/j.jbi.2008.08.010 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. von Elm E., Altman DG., Egger M. et al. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. Ann Intern Med. 2007;147(8):573-577. 10.7326/0003-4819-147-8-200710160-00010 [DOI] [PubMed] [Google Scholar]
- 11. Williams DR., Cooper LA. COVID-19 and health equity—a new kind of “herd immunity”. JAMA. 2020;323(24):2478-2480. 10.1001/jama.2020.8051 [DOI] [PubMed] [Google Scholar]
- 12. Misa N-Y., Perez B., Basham K. et al. Racial/ethnic disparities in COVID-19 disease burden & mortality among emergency department patients in a safety net health system [published online September 24, 2020]. Am J Emerg Med. 10.1016/j.ajem.2020.09.053 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Haozous EA., Strickland CJ., Palacios JF., Solomon TGA. Blood politics, ethnic identity, and racial misclassification among American Indians and Alaska Natives. J Environ Public Health. 2014;2014:1-9. 10.1155/2014/321604 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Nolan R. A translation crisis at the border. The New Yorker. December 30, 2019. Accessed July 1, 2020. https://www.newyorker.com/magazine/2020/01/06/a-translation-crisis-at-the-border
- 15. Rasilia A. One Fruitvale organization is providing a safety net for hundreds of Oakland’s struggling day laborers. Berkeleyside. May 14, 2020. Accessed August 1, 2020. https://www.berkeleyside.com/2020/05/14/one-fruitvale-organization-is-providing-a-safety-net-for-hundreds-of-oaklands-struggling-day-laborers
- 16. Rodriguez RM., Torres JR., Sun J. et al. Declared impact of the US president’s statements and campaign statements on Latino populations’ perceptions of safety and emergency care access. PLoS One. 2019;14(10):e0222837. 10.1371/journal.pone.0222837 [DOI] [PMC free article] [PubMed] [Google Scholar]