Abstract
This cross-sectional study examines variation in SARS-CoV-2 infection risk and socioeconomic disadvantage among a Mayan-Latinx population in the Fruitvale, California, community.
Introduction
US Latinx populations are disproportionally affected by the SARS-CoV-2 pandemic, with higher rates of infection and associated morbidity and mortality.1 Although often treated as homogeneous, members of Latinx communities vary by national origin, immigration status, and language.2 Oakland, California, is home to many Latinx individuals and an estimated 10 000 Mayan individuals, many of whom speak Indigenous languages.3 Early in the pandemic, community-based organizations (CBOs) in Oakland, California, observed a high frequency of infections among Latinx individuals in general and even higher frequency among Mayan individuals.4 Local CBOs, the University of California, San Francisco (UCSF), and public health authorities formed a collaborative to offer diagnostic testing in Fruitvale, a diverse neighborhood that has among the highest cumulative infection rates in Alameda County.5 Using data from the resulting SARS-CoV-2 testing event, we examined variation in infection risk and socioeconomic disadvantage within the Fruitvale community.
Methods
This cross-sectional study was approved by the UCSF institutional review board and followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline. Free SARS-CoV-2 testing was provided for individuals of all ages on September 26 to 27, 2020. Adults gave verbal consent for themselves and for participating children. Anterior nasal swab samples were obtained for polymerase chain reaction (PCR) testing to detect the virus, and venous blood was collected to detect immunoglobin G antinucleocapsid antibodies. Adults completed a survey on sociodemographic characteristics at the testing event. Interviewers fluent in Spanish, 2 Mayan languages, and more than 4 other languages were available. We analyzed cross-sectional associations between demographic and socioeconomic indicators and SARS CoV-2 infection using χ2 tests and logistic regression analyses adjusted for age and sex and accounting for household clustering. Statistical significance was set at P < .05. Data analyses were conducted in Stata version 16 (StataCorp). Additional information regarding the methods appear in the eAppendix in the Supplement.
Results
We tested 1186 individuals (1034 [87.2%] adults; 152 [12.8%] children; 610 [51.4%] female participants; mean [SD] age, 40.0 [18.3] years); 108 (9.1%) were Mayan individuals, 661 (55.7%) non-Mayan Latinx individuals, and 417 (35.2%) non-Latinx individuals. Compared with other Latinx individuals, Mayan individuals were more likely to live in households with 5 or more people (49 [53.3%] vs 152 [32.6%]; P < .005), report food insecurity (53 [62.4%] vs 172 [41.8%]; P = .001), have difficulty finding work due to the pandemic (12 [13.0%] vs 31 [5.4%]; P = .01), lack a regular medical practitioner (55 [64.7%] vs 340 [76.2%]; P = .03), and have no health insurance (35 [38.0%] vs 118 [20.7%]; P < .001) (Table 1). Mayan individuals were also more likely to have limited English proficiency compared with other Latinx participants (49 [58%] vs 187 [46%]; P = .04); 41 (44.6%) spoke a Mayan language at home. Mayan and non-Mayan Latinx participants had significantly greater odds of having a positive PCR test compared with non-Latinx participants (Mayan: adjusted odds ratio [aOR], 16.66; 95% CI, 3.54-78.41; P < .001; non-Mayan Latinx: aOR, 8.48; 95% CI, 1.91-37.67; P = .004). Mayan individuals were significantly more likely to have positive serology results compared with non-Latinx participants (aOR, 5.58; 95% CI, 2.13-14.65; P < .001) (Table 2).
Table 1. Demographic and Socioeconomic Characteristics of Study Population by Ethnicity.
| Characteristic | No. (%) | P value for comparisona | ||||
|---|---|---|---|---|---|---|
| Non-Latinxb | Non-Mayan Latinx | Mayan | Non-Latinx vs Non-Mayan Latinx | Non-Mayan Latinx vs Mayan | Non-Latinx vs Mayan | |
| All participants | ||||||
| No. | 417 | 661 | 108 | NA | NA | NA |
| Age, y | ||||||
| 0-17 | 45 (10.8) | 91 (13.8) | 16 (14.8) | .19 | .16 | .07 |
| 18-49 | 223 (53.7) | 358 (54.2) | 68 (63.0) | |||
| 50-64 | 111 (26.8) | 145 (21.9) | 19 (17.6) | |||
| ≥65 | 36 (8.7) | 67 (10.1) | 5 (4.6) | |||
| Sexc | ||||||
| Female | 203 (52.5) | 356 (56.3) | 51 (49.5) | .23 | .20 | .60 |
| Male | 184 (47.6) | 276 (43.7) | 52 (50.5) | |||
| Participants 18 y or older | ||||||
| No. | 372 | 570 | 92 | NA | NA | NA |
| Live in Fruitvale | 143 (38.7) | 275 (48.3) | 66 (71.7) | .004 | <.001 | <.001 |
| Work in Fruitvale | 51 (13.7) | 74 (13.0) | 15 (16.3) | .72 | .39 | .54 |
| Individuals in household, No.d | <.001 | |||||
| 1-4 | 247 (85.2) | 314 (67.4) | 43 (46.7) | <.001 | <.001 | <.001 |
| ≥5 | 43 (14.8) | 152 (32.6) | 49 (53.3) | |||
| Income reduced due to pandemic | 59 (15.9) | 140 (24.6) | 29 (31.5) | .002 | .16 | .001 |
| Lost job due to pandemic | 34 (9.1) | 84 (14.7) | 17 (18.5) | .01 | .35 | .01 |
| Difficulty finding work due to pandemic | 25 (6.7) | 31 (5.4) | 12 (13.0) | .40 | .01 | .05 |
| No health insurance | 29 (7.8) | 118 (20.7) | 35 (38.0) | <.001 | <.001 | <.001 |
| Have regular health care practitionere | 219 (80.2) | 340 (76.2) | 55 (64.7) | .21 | .03 | .003 |
| Household income <$50 000/yf | 130 (58.8) | 293 (77.1) | 61 (87.1) | <.001 | .06 | <.001 |
| Worried about ability to buy enough foodg | 99 (37.1) | 207 (49.4) | 52 (61.2) | .002 | .05 | <.001 |
| Not enough food to last until paycheckh | 91 (35.1) | 172 (41.8) | 53 (62.4) | .09 | .001 | <.001 |
| Low English proficiencyi | 12 (18.8) | 187 (46.2) | 49 (58.3) | <.001 | .04 | <.001 |
Abbreviation: NA, not applicable.
P values by Mantel-Haenszel χ2 test.
Non-Latinx included participants who indicated the following racial identity categories: White (114 [27.5%]), Black (125 [30.1%]), Asian (65 [15.7%]), multiracial (22 [5.3%]), other (25 [6.0%]), NA (26 [6.3%]), and unknown (38 [9.2%]).
Data missing for 30 non-Latinx, 29 non-Mayan Latinx, and 5 Mayan participants.
Data missing for 82 non-Latinx and 104 non-Mayan Latinx participants.
Data missing for 99 non-Latinx, 124 non-Mayan Latinx, and 7 Mayan participants.
Data missing for 151 non-Latinx, 190 non-Mayan Latinx, and 22 Mayan participants.
Data missing for 105 non-Latinx, 151 non-Mayan Latinx, and 7 Mayan participants.
Data missing for 113 non-Latinx, 158 non-Mayan Latinx, and 7 Mayan participants.
Low proficiency corresponded to reporting speaking a language other than English at home and having answered not well or not at all to the question, “How well do you speak English?” Data on English proficiency collected only for participants who reported speaking a language other than English at home (66 non-Latinx, 423 non-Mayan Latinx, and 90 Mayan participants). Of those who reported speaking a language other than English at home, data on English language proficiency was missing for 18 non-Mayan Latinx and 6 Mayan individuals. An additional 97 Latinx and 78 non-Latinx individuals did not report the language spoken at home so were not asked about English proficiency.
Table 2. Association of Sex, Age, and Ethnicity With Positive SARS-CoV-2 PCR and Serology Test Results Among Participants in the Fruitvale Testing Event, September 2020.
| Characteristic | PCR test | Serology test | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Tested, No. | Positive, No. (%) | OR (95% CI) | P value | aOR (95% CI)a | P value | Tested, No. | Positive, No. (%) | OR (95% CI) | P value | aOR (95% CI)a | P value | |
| Sexb | ||||||||||||
| Female | 570 | 14 (2.5) | 1 [Reference] | NA | 1 [Reference] | NA | 454 | 48 (10.6) | 1 [Reference] | NA | 1 [Reference] | NA |
| Male | 485 | 25 (5.2) | 2.16 (1.12-4.17) | .02 | 2.25 (1.16-4.39) | .02 | 364 | 29 (8.0) | 0.73 (0.48-1.11) | .15 | 0.70 (0.45-1.09) | .11 |
| Age, y | ||||||||||||
| 0-17 | 146 | 10 (6.9) | 1 [Reference] | NA | 1 [Reference] | NA | 57 | 7 (12.3) | 1 [Reference] | NA | 1 [Reference] | NA |
| 18-49 | 605 | 25 (4.1) | 0.59 (0.28-1.25) | .17 | 0.57 (0.27-1.21) | .14 | 516 | 57 (11.1) | 0.89 (0.36-2.19) | .80 | 1.06 (0.39-2.90) | .90 |
| 50-64 | 262 | 3 (1.2) | 0.16 (0.03-0.83) | .03 | 0.16 (0.03-0.80) | .03 | 203 | 14 (6.9) | 0.53 (0.19-1.46) | .22 | 0.61 (0.19-2.04) | .43 |
| ≥65 | 98 | 1 (1.0) | 0.14 (0.02-1.15) | .07 | 0.15 (0.02-1.29) | .08 | 90 | 8 (8.9) | 0.70 (0.20-2.41) | .57 | 0.77 (0.18-3.18) | .71 |
| Ethnicity | ||||||||||||
| Non-Latinx | 381c | 2 (0.5) | 1 [Reference] | NA | 1 [Reference] | NA | 304d | 18 (5.9) | 1 [Reference] | NA | 1 [Reference] | NA |
| Non-Mayan Latinx | 626 | 28 (4.5) | 8.87 (2.00-39.43) | .004 | 8.48 (1.91-37.67) | .005 | 485 | 49 (10.1) | 1.79 (0.86-3.69) | .12 | 1.76 (0.78-3.95) | .17 |
| Mayan | 102 | 9 (8.8) | 18.34 (3.85-87.25) | <.001 | 16.66 (3.54-78.41) | <.001 | 76 | 19 (25.0) | 5.30 (2.30-12.19) | <.001 | 5.58 (2.13-14.65) | <.001 |
Abbreviations: aOR, adjusted odds ratio; NA, not applicable; OR, odds ratio; PCR, polymerase chain reaction.
Models adjusted for sex and age. All models clustered on the household level to account for within-household correlation and correct standard errors.
Nonbinary or sex data missing from 5 participants.
Non-Latinx participants who received PCR testing included participants who indicated the following racial identity categories: White (107 [28.1%]), Black (114 [29.9%]), Asian (61 [16.0%]), multiracial (21 [5.5%]), other (23 [6.0%]), NA (21 [5.5%]), and unknown (34 [8.9%]).
Non-Latinx participants who received serology testing included participants who indicated the following racial identity categories: White (89 [29.3%]), Black (80 [26.3%]), Asian (54 [17.8%]), multiracial (18 [5.9%]), other (19 [6.3%]), NA (25 [8.2%]), and unknown (19 [6.3%]).
Discussion
We found that Latinx participants were more likely to have current SARS-CoV-2 infection than non-Latinx participants, reflecting state and national trends.1 Our data highlight heterogeneity within the Latinx community, with Mayan individuals having even higher risk than other Latinx individuals. Findings related to socioeconomic disadvantage, including large household size, low income, and food insecurity, likely reflect the heightened susceptibility of Mayan individuals to the pandemic.6 In addition, limited English proficiency and access to health care pose challenges for effective public health messaging.
Limitations that reduce generalizability include that the study analyzed a convenience sample of those seeking testing, and that testing and medical referrals were prioritized over questionnaire completion, resulting in missing data. Unmeasured confounding factors could attenuate results.
This study underscores the need to consider heterogeneity within Latinx communities and to prioritize subgroups with higher risks, such as Mayan individuals, in health policies and outreach. Limited Spanish and English proficiency reduce this population’s access to health information and care. Few CBOs or public health departments have Mayan language speakers to provide information and perform contact tracing. Those that do are underresourced. Understanding differential risks within the heterogenous Latinx population can guide more efficient targeting of services. Failure to engage with communities with higher risks increases the likelihood of ongoing transmission and may hinder SARS CoV-2 vaccine uptake.
eAppendix. Supplementary Methods
References
- 1.Chen Y-H, Glymour MM, Catalano R, et al. Excess mortality in California during the coronavirus disease 2019 pandemic, March to August 2020. JAMA Intern Med. 2020. doi: 10.1001/jamainternmed.2020.7578 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Rodriguez-Diaz CE, Guilamo-Ramos V, Mena L, et al. Risk for COVID-19 infection and death among Latinos in the United States: examining heterogeneity in transmission dynamics. Ann Epidemiol. 2020;52:46-53.e2. doi: 10.1016/j.annepidem.2020.07.007 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Garofoli J. “We are invisible”: coronavirus just made census counting harder in Bay Area. San Francisco Chronicle. Published April 16, 2020. Accessed January 28, 2021. https://www.sfchronicle.com/politics/article/We-are-invisible-Coronavirus-just-made-15203827.php
- 4.Foo PK, Perez B, Gupta N, et al. High rates of COVID-19 infection among Indigenous Maya at a US safety-net health system in California. Public Health Rep. Published online February 16, 2021. doi: 10.1177/0033354921990370 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Alameda County Public Health Department. COVID-19 data. Accessed January 5, 2021. https://covid-19.acgov.org/data.page
- 6.Cervantes L, Martin M, Frank MG, et al. Experiences of Latinx individuals hospitalized for COVID-19: a qualitative study. JAMA Netw Open. 2021;4(3):e210684. doi: 10.1001/jamanetworkopen.2021.0684 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
eAppendix. Supplementary Methods