Public trust is earned: Historical discrimination, carceral violence, and the COVID‐19 pandemic

Abstract Objective To assess whether knowledge of Tuskegee, the U.S. Immigration and Customs Enforcement (ICE) agency's detainment of children, and satisfaction with the George Floyd death investigation were associated with trust in actors involved in the development and distribution of coronavirus vaccines. Data Sources and Study Setting National survey with a convenience sample of Black (n = 1019) and Hispanic (n = 994) adults between July 1 and 26, 2021. Study Design Observational study using stratified adjusted logistic regression models to measure the association between ratings of the trustworthiness of actors involved in the development and distribution of coronavirus vaccines. Principal Findings Among Black respondents, lower satisfaction with the George Floyd death investigation was associated with lower trustworthiness ratings of pharmaceutical companies (ME: −0.09; CI: −0.15, 0.02), the FDA (ME: −0.07; CI: −0.14, −0.00), the Trump Administration (ME: −0.09; CI: −0.16, −0.02), the Biden Administration (ME: −0.07, CI: −0.10, 0.04), and elected officials (ME: −0.10, CI: −0.18, −0.03). Among Hispanic respondents, lower satisfaction was associated with lower trustworthiness ratings of the Trump Administration (ME: −0.14, CI: −0.22, −0.06) and elected officials (ME: −0.11; CI: −0.19, −0.02). Greater knowledge of ICE's detainment of children and families among Hispanic respondents was associated with lower trustworthiness ratings of state elected officials (ME: −0.09, CI: −0.16, 0.01). Greater knowledge of the US Public Health Service Study of Syphilis in Tuskegee was associated with higher trustworthiness ratings of their usual source of care (ME: 0.09; CI: 0.28, 0.15) among Black respondents (ME: 0.09; CI: 0.01, 0.16). Conclusions Among Black respondents, lower satisfaction with the George Floyd death investigation was associated with lowered levels of trust in pharmaceutical companies, some government officials, and administrators; it was not associated with the erosion of trust in direct sources of health care delivery, information, or regulation. Among Hispanic respondents, greater knowledge of the ICE detainments was associated with lower trustworthiness ratings of elected state officials. Paradoxically, higher knowledge of the Study of Syphilis in Tuskegee was associated with higher trustworthiness ratings in usual sources of care.

higher knowledge of the Study of Syphilis in Tuskegee was associated with higher trustworthiness ratings in usual sources of care.

K E Y W O R D S
health equity, questionnaire design, social determinants of health, survey research What is known on this topic • Scholars attributed low early uptake of the COVID-19 vaccines among Black and Hispanic Americans to medical mistrust.
• Medical mistrust is associated with experiences with racism and discrimination in the health care system and beyond.
• US media and the scientific community have named perceptions of historical and contemporary atrocities as contributors to low levels of trust of medical professionals among Black and Hispanic communities.

What this study adds
• Black respondents who were less satisfied with the George Floyd death investigation had lower trustworthiness ratings of pharmaceutical companies and government, but not sources of health care information, regulation, or delivery (i.e., usual care source, vaccine clinics, the FDA).
• Hispanic respondents with greater knowledge of the ICE's detainment of children and families had lower trustworthiness ratings of state elected officials.
• State-sanctioned structural violence in racially and ethnically minoritized communities is associated with racially and ethnically concordant lower levels of trust in COVID vaccinerelated actors.

| INTRODUCTION
The low early uptake of coronavirus vaccines among Black and Hispanic populations in the United States has been attributed to widespread medical mistrust rather than the misconduct of government and medical institutions. [1][2][3][4] Group-based knowledge of unequal treatment, neglect, and violence has led many to consider these institutions untrustworthy. Given the importance of laws and regulations in shaping health outcomes, 5,6 scholars have offered frameworks to expand analyses of institutional trust to distinguish between trust, distrust, mistrust, and trustworthiness and the important implications of each for public health and state actors. [7][8][9][10][11][12][13][14][15] Public perception of the trustworthiness (i.e., the perceived reliability of institutional actors to perform their duties or publicly stated commitments) of actors involved in the COVID-19 pandemic response predates the pandemic.
The reputations of these actors may have improved or worsened as a result of actions taken to address the pandemic. Yet, less is known about whether knowledge of historical and contemporary abuses, crimes, and discrimination perpetrated by these actors influences their trustworthiness among Black and Hispanic populations.
Discrimination experienced within the health care system is a known contributor to institutional mistrust. 16 For instance, despite mixed evidence, the US Public Health Service Study of Syphilis in Tuskegee has been repeatedly cited as a source of mistrust, particularly among Black Americans. [17][18][19][20][21][22][23][24] The United States has perpetrated or been complicit in health violence against Black, Hispanic, and Indigenous populations domestically and abroad (e.g., the sterilization of Native American women and girls through the Indian Health Service; the infection of prisoners, psychiatric patients, and sex workers to test the efficacy of penicillin in Guatemala), which led many to avoid health interventions proposed by actors perceived to act in partnership with those agencies and institutions. [25][26][27][28][29] Although knowledge of historical atrocities and common experiences of racial discrimination is known to have tangible health consequences, 30,31 this domain of knowledge has important behavioral consequences for public health actors that were underexplored during the COVID-19 pandemic.
Public health researchers and scholars across multiple disciplines have found police violence-disproportionately against Black Americans-has myriad health consequences in need of intervention. [32][33][34][35][36][37][38] Research shows that indirect police contact exacerbates health risks for people living in highly surveilled neighborhoods, [39][40][41][42] , and the effects of this police contact have varied by gender. 39,[43][44][45] Scholars have shown that routine exposure to police violence is linked to medical mistrust, 46 lower utilization of important health services, 39,47 and exacerbation of other known health disparities for youth and adults. 40,42,43,48,49 Sociolegal scholars have also begun examining how legal standards are being enforced to deter police misconduct, [50][51][52][53][54][55][56][57] and others are beginning to ask questions about how to use available 21st-century technological innovations to regulate law enforcement behavior more expansively. [58][59][60][61][62][63][64] However, research examining the impact of police violence-or (dis)satisfaction with governmental responses to police violence-on the political behaviors of American citizens and residents remains scant, despite evidence demonstrating that knowledge of this violence also has important political consequences (particularly among racially minoritized populations). [65][66][67] In this paper, we use a novel survey fielded during the summer of 2021 to empirically examine whether knowledge of historical atrocities and (dis)satisfaction with governmental responses to police violence influence trust in key actors involved with the development and distribution of the coronavirus vaccines. We tested three hypotheses to evaluate whether (1)

| Study design
We fielded a national online survey (COVID-19 Pandemic Trust, Racism, and State Violence Study) with a convenience sample of Black and Hispanic adults aged 18 and older between July 1 and July 26, 2021, using the Qualtrics survey platform. The data were scrubbed for duplicates, speeders (e.g., inappropriate completion time), flatliners (e.g., same response to all options), inattention (e.g., nonsensical responses), bad verbatim responses, and location outside of the U.S., ultimately yielding an achieved response rate of 46.6%. All respondents who completed the survey were provided a financial incentive of $6.25 on average by Qualtrics commensurate with the length of the survey (30 min). Our study protocol and survey instrument were approved by the Tulane University Social-Behavioral Institutional Review Board.

| Instrument and sample
Our instrument included adapted measures from the Uncovering COVID-19 Experiences and Realities (UnCOVER) Study fielded by the Society, Health, and Racial Equity Lab at Tulane University from May to July 2020, as well as the COVID Collaborative Study fielded by Langer Research Associates. 68,69 Since the COVID Collaborative study explicitly focused on Black and Hispanic respondents, we extend this work by restricting our final analytical sample to 1911 self-identifying Black (n = 1019) and Hispanic (n = 994) respondents (see Table 1) with complete responses for our primary outcome variables, predictor variables, and covariates.

| Measures
The outcome variables were the respondents' perceptions of the trustworthiness of actors involved in the development and distribution of the coronavirus vaccines, based on the survey items fielded during the COVID Collaborative Study. 69 Respondents were asked, Higher values indicate greater within-group perceptions of the trustworthiness of these actors.
We had three key predictor variables of interest (see Table 2). to delineate between respondents who knew a great deal/a lot/a moderate amount and those who knew little/none (reference group).
The literature is mixed on its assessment of the association between knowledge of the Tuskegee study and levels of trust and mistrust in medical institutions. 23,24 We sought to assess whether knowledge of the study correlates with perceptions of trustworthiness, given the resurgence of the study in popular media as a reported cause of low trust leading to low early uptake of the coronavirus vaccines.  Note: ME, average marginal effects; CI, 95% confidence interval; Stars indicate significance at <0.05*, <0.01**, <0.001***.
Note: The table reflects national from a convenience sample of people who self-identify as Black and Hispanic. All models were adjusted for respondent age, sex assigned a birth, education, income, whether they were born in or outside of the US, political affiliation, self-reported physical and mental health status, linked destiny, and where they consume their main source of news and information about the pandemic. Full regression results are included in a supplemental appendix.
Multiracial) to inform our analytical categorization of individual participants who may have selected one or more racial/ethnic identities.
Physical and mental self-reported health status was assessed by asking participants whether, in general, they would rate their physical and mental health as poor, fair, good, very good, or excellent. 71

| Analytic approach
We calculated summary statistics to assess the distribution of withingroup responses for our outcomes, exposures, and covariates (Tables 1-3). Rather than test these hypotheses by comparing relative differences in means between racial and ethnic groups (e.g., Black vs. White respondents), we explored these associations among Black and Hispanic respondents with stratified adjusted logistic regression models. Initially, we ran our analyses using the raw data from the fivepoint Likert scale as categorical variables; however, we ultimately analyzed these variables as dichotomous indicators to ease the analytical interpretation of our results (the results remained the same). We report the associations using marginal effects. 72  trustworthy (i.e., completely/mostly/somewhat trustworthy) (  (Table 4)

| Limitations
First, our data are cross-sectional and were collected using a nonprobabilistic online sample. Therefore, our estimates are not nationally representative and likely include bias due to coverage error associated with the sampling technique and mode of survey administration. We were also unable to determine the extent of survey non-response bias. Participants in our study, however, represent a diverse group of adults from all regions of the United States. To our knowledge, our study is the largest national sample of Black and Hispanic Americans who were asked about their knowledge of these historical and contemporary events. Second, we did not include community-based organizations and other trusted community settings where people were able to obtain vaccines. Finally, we also used single-item measures of trust that, while novel and potentially important, have not been validated or psychometrically tested.

| CONCLUSION
Trust in public and private institutions depends on their trustworthiness. Public health practitioners rely on a wellspring of trust in emergencies when the general population must act quickly (e.g., using masks and taking vaccines) to stop the spread of disease. But this trust must be earned by improving the trustworthiness of institutions.
The erosion of trust through untrustworthy actions of one institution or system (e.g., the criminal legal system) may spill over to other distinct institutions and systems (e.g., health care). 47  Paul Shafer https://orcid.org/0000-0003-0654-5821