Influence of Medical Mistrust on Prevention Behavior and Decision-Making among Minoritized Youth and Young Adults During the COVID-19 Pandemic

Abstract

Background.

Medical mistrust (MM) is seen as a barrier to assessing healthcare needs and addressing health disparities; however, limited literature has focused on assessing MM for vulnerable populations, especially racial/ethnic minority and sexual/gender minority YYA.

Methods.

Between February 2021 and March 2022, we conducted the Youth and Young Adults COVID-19 Study, a prospective cohort of minoritized YYA aged 14 to 24 years (n = 1,027), within the United States and its territories. Participants were recruited through a combination of paid social media ads, outreach with organizations serving marginalized youth, and an existing registry, targeting racial and ethnic minority and LGBTQ+ youth for a study on COVID-19 health behaviors. Multiple multinomial logistic regression models were developed to examine associations between demographics and three dimensions of MM including healthcare experience, government information, and scientific information.

Results.

Most participants were between the ages of 18 and 21 years (48.3%), identified as Hispanic (33.3%) or white (22.5%), and bisexual or pansexual (34.3%). Queer YYA had higher odds of reporting worse personal healthcare experiences than their straight peers. The odds of gay/lesbian YYA that reported somewhat or extreme trust in doctor’s sources were two times higher than their straight peers. Except for those who identified as Asian, racial/ethnic minority YYA were less likely to report somewhat or extreme trust in the CDC’s general information or its COVID-19 data than white YYA. Transgender and gender diverse YYA were more than twice as likely to report being very or extremely influenced by statistics of the dangers of COVID-19 than cisgender YYA.

Conclusions.

Our study indicated the importance of incorporating marginalized identities into the assessment of medical mistrust to better understand YYA’s health prevention and treatment behaviors and to develop public health prevention and treatment strategies, especially for minoritized communities.

INTRODUCTION

Medical mistrust (MM) is an umbrella term encompassing both general feelings of mistrust in the medical institution as a whole and mistrust specific to one disease or context (e.g., COVID-19). MM is seen as a barrier to health promotion and addressing health disparities, especially among highly marginalized populations like sexual and gender minorities (SGM), racial/ethnic minorities (REM) identities, and those at the intersections of these identities.(1) The origins of MM are often attributed to negative historical interactions with health systems due to experiences of discrimination. For example, historical events, such as the Tuskegee syphilis study, which eroded trust in medical systems in the Black community, and the Eugenics movement of the 1970s, which included the forced sterilization of many Latinx communities, continue to have a profound effect on Black and Latinx communities’ health decision-making.(2) Similarly, history of (mis) categorization of minoritized sexual and gender identities as mental disorders and sexual deviance by health authorities like the American Psychological Association (APA) in the 1950s and in the earlier versions of Diagnostic and Statistical Manual of Mental Disorders (DSM) by APA has significantly contributed to MM among SGM populations.(3) Furthermore, continued stigmatization, discrimination, and blame assigned to one’s sexual orientation and/or gender identity during every public health crisis (e.g., HIV pandemic,(4) mpox epidemic(5)) has dramatically increased MM among SGM populations.(3) The influence of these historical experiences of trauma are reinforced by continued institutional racism and exposure to medical microaggressions.(6)

The onset of the COVID-19 pandemic revealed and exacerbated pre-existing disparities between racial/ethnic minorities and white individuals,(7, 8) especially those with SGM identities.(9–11) Frequent changes in messaging, ongoing political tensions, inconsistent state-level policies throughout the US, and a lack of resources dedicated to minoritized communities (e.g., SGM and REM populations) synergistically interacted with extant MM to undermine COVID-19 mitigation programming, thus rendering many prevention strategies less effective.(12, 13) Prior studies have demonstrated that high levels of MM are associated with poorer management of health conditions,(14) lower adherence to prescriptions and medical recommendations,(15) lower utilization of healthcare services,(16) and decreased participation in preventive care.(17)

As devasting as the COVID-19 pandemic has been, COVID-19 is largely a preventable disease. The risk of COVID-19 infection can be significantly reduced through preventive behaviors such as mask-wearing, social distancing, and vaccination.(18) Despite the effectiveness of these strategies, a large proportion of the overall US population has continued to be unconvinced about the utility of these approaches, especially in communities where MM is relatively high like SGM,(19) REM,(20) and those at the intersection of these identities.(21) These challenges have resulted in the emergence of narratives that position certain communities as non-compliant and reservoirs of disease that perpetuate the ongoing COVID-19 pandemic.(22–25) These narratives ignore the impact of pre-existing MM among minoritized communities due to negative healthcare experiences and lack of trust in official government channels and scientific information. The lack of a robust understanding regarding the onset of MM, especially during public health crises and among minoritized populations, limits the reach, acceptability, and effectiveness of any current and future public health programming.

Youth and young adults (YYA), or those aged 14–24 years old, may be vulnerable to MM due to a lack of experience and knowledge in navigating the healthcare system, limited access to reliable information, past negative experiences with healthcare providers, and a general skepticism towards authority figures.(26–28) YYA also are more likely to be influenced by factors such as peer-pressure and other societal influences,(29, 30) and a desire for autonomy and independence,(31) which can lead to questioning or doubting medical advice and treatment options. Despite this vulnerability, especially among youth with minoritized identities, YYA are frequently underrepresented in MM literature due to the long-standing notion that their guardians are the primary decision makers regarding healthcare. However, emerging research indicates that many YYA develop health-related attitudes and beliefs that are not aligned with those of their past or current guardians. As such, there is a need to understand and assess MM among YYA.(32–36) Focusing on the response of YYA to the COVID-19 pandemic as a case study, we aim to fill this gap by examining sociodemographic predictors of MM among YYA. More specifically, we examine three indicators of medical mistrust: 1) negative experiences with the healthcare system, 2) distrust in government sources of information, and 3) distrust of scientific sources of information. We expect YYA with minoritized backgrounds would report higher MM compared to their majority counterparts.

METHODS

1. Study Design

Data were collected as a part of the Youth and Young Adults COVID-19 Study—a prospective cohort of minoritized YYA—conducted between February 2021 and March 2022. Eligibility criteria were (1) being 14 to 24 years of age, (2) residing in the United States/US territories, (3) having access to the internet, (4) being willing to complete a follow-up survey in 6 months, and (5) providing informed consent. We first sent out a screener to participants who were interested in the study, and then eligible participants who provided informed consent were invited to complete the survey. For participants under 18 years of age, the ability to understand study procedures and decisional capacity was first assessed based on the UCSD Task Force on Decisional Capacity’s procedures for the determination of decisional capacity in persons participating in research,(37) using a modified version of the Evaluation to Consent Form (38–40). Informed consent for all participants was obtained electronically. Participants were recruited through paid social media advertisements, outreach with organizations that served LGBTQ+, Indigenous, and Latinx youth, and an existing participant registry. Participants who completed the baseline survey received a digital $30 VISA card. Study procedures were approved by Northwestern University Institutional Review Board through expedited review.

2. Measures

2.1. Outcome variables

The historical legacies of the Tuskegee Syphilis Study, the HIV/AIDS crisis, and the Forced Sterilization Programs have all contributed to a pervasive sense of mistrust towards the healthcare system, government information, and scientific information among marginalized communities.(14, 41–44) The Tuskegee experimenťs deliberate experimentation on African American men without their informed consent,(41) the governmenťs slow response to the HIV/AIDS epidemic,(45) and the forced sterilization of thousands of people,(45, 46) often without their knowledge or consent, have all eroded trust in these institutions. As a result, it is not surprising that many individuals in these communities approach public health information with skepticism and mistrust. In recognition of this complex issue, we have chosen to focus on three key dimensions that play a significant role in shaping medical mistrust with a focus on participants’ decision-making on COVID-19 prevention and treatment behaviors. These dimensions include (1) healthcare experiences, (2) government sources of information, and (3) scientific sources of information. Each of the domains included two factors:

Healthcare Experiences.

1. Personal healthcare experience. Participants were asked, “Within the past 12 months when seeking health care, how do you feel your experiences compared to other persons living in the United States?” Response options were grouped into three categories: worse than others, the same, and better than others.

2. Trust in doctor’s sources. Participants were asked, “How much would you trust your doctor’s sources of information for making decisions about COVID-19 prevention and treatment behaviors?” Response options were grouped into three categories: neutral/not trustworthy, somewhat trustworthy, and extremely trustworthy.

Government Sources of Information.

1. Trust in CDC's COVID-19 information. Participants were asked, “How much would you trust information from CDC or other government bodies for making decisions about COVID-19 prevention and treatment behaviors?” Response options were grouped into three categories: neutral/not trustworthy, somewhat trustworthy, and extremely trustworthy.

2. Influence of CDC’s COVID-19 data. Participants were asked, “If you were to receive information about COVID-19, how influential would data about COVID-19 trends from the CDC be on your decision-making about COVID-19 prevention and treatment behaviors?” Response options were grouped into three categories: slightly/somewhat influential, very influential, and extremely influential.

Scientific Sources of Information.

1. Influence of COVID-19 statistics. Participants were asked, “If you were to receive information about COVID-19, how influential would the statistics that show the danger of COVID-19 be on your decision-making about COVID-19 prevention and treatment behaviors?” Response options were grouped as slightly/somewhat influential, very influential, and extremely influential.

2. Influence of up-to-date scientific information. Participants were asked, “If you were to receive information about COVID-19, how influential would the most up-to-date scientific information be on your decision-making about COVID-19 prevention and treatment behaviors?” Response options were grouped as slightly/somewhat influential, very influential, and extremely influential.

2.2. Demographics

A detailed description of demographic measures is presented in Supplement A. Demographic measures included age in three groups: 1) 14–17 years, 2) 18–21 years, and 3) 22–24 years; race/ethnicity in seven groups: 1) American Indian/Alaska Native (AI/AN), 2) Asian, 3) Black or African American, 4) Hispanic, 5) Multiracial, and 7) white; sexual identity in six groups: 1) Asexual or asexual spectrum, 2) Bisexual or pansexual, 3) Gay or lesbian, 4) Straight, 5) Queer, and 6) Questioning; gender identity in six groups: 1) Agender, 2) Gender Queer, 3) Man/Boy, 4) Non-binary, 5) Questioning, and 6) Woman/Girl; and gender modality in three groups: 1) Cisgender, 2) Trans and Gender Diverse, and 3) Not Sure.

3. Data Cleaning and Analytic Sample

In total, 2,395 eligible individuals participated in the YYA COVID-19 Study by providing informed consent. After thorough data cleaning procedures were implemented: participants who missed attention check questions, provided inconsistent demographic identities between the screening and main survey, or duplicate or invalid email addresses (n = 1,191), and those who did not complete the survey (n = 149) were excluded, resulting a final sample of 1,055. For this study, we excluded those with missing values (n=28); response options of “Not listed” (n = 9), “Prefer not to answer” or “Don’t know” (n=1); “Native Hawaiian or other Pacific Islander” for race/ethnicity (n = 5); and “Two-spirit” for gender identity (n=6; not mutually exclusive) due to small sample size, resulting in a final analytic sample of 1,027.

4. Statistical Analyses

Data cleaning and recoding, and statistical analyses were conducted in RStudio version 4.2.1 (RStudio, Boston, MA). Frequencies and percentages were calculated for demographics (Table 1). Multinomial logistic regression models were developed to estimate the associations between demographic characteristics and each of six categorical outcomes (Tables 2 – 4). Demographics including age, race/ethnicity, sexual identity, gender identity, and gender modality were adjusted for each of the models. Adjusted odds ratios (aORs) and 95% confidence intervals (CIs) were calculated for all regression models.

Table 1.

Demographic Characteristics and Main Outcomes (N=1027).

Demographics	N (%)
Age
22–24	359 (35.0)
14–17	172 (16.7)
18–21	496 (48.3)
Race/Ethnicity
White	231 (22.5)
American Indian or Alaska Native	97 (9.4)
Asian	75 (7.3)
Black	182 (17.7)
Hispanic	342 (33.3)
Multiracial	100 (9.7)
Sexual Identity
Straight	226 (22.0)
Asexual/Ace Spectrum	53 (5.2)
Bisexual/Pansexual	352 (34.3)
Gay/Lesbian	227 (22.1)
Queer	147 (14.3)
Questioning	22 (2.1)
Gender
Woman/Girl	513 (50.0)
Agender	10 (1.0)
Gender Queer	29 (2.8)
Man/Boy	277 (27.0)
Non-binary	168 (16.4)
Questioning	30 (2.9)
Gender Modality
Cisgender	659 (64.2)
Trans and Gender Diverse	331 (32.2)
Not Sure	37 (3.6)
Outcomes
Healthcare Experiences
Personal Healthcare Experience
Worse than others	597 (58.1)
The same	337 (32.8)
Better than others	93 (9.1)
Trust Level in Doctors' Sources
Neutral/Not trustworthy	83 (8.1)
Somewhat trustworthy	324 (31.5)
Extremely trustworthy	620 (60.4)
Government Information
Trust level in CDC's general COVID-19 Information
Neutral/Not trustworthy	135 (13.1)
Somewhat trustworthy	318 (31.0)
Extremely trustworthy	574 (55.9)
Influence level in CDC’s COVID-19 Data
Slightly/Somewhat Influential	200 (19.5)
Very influential	492 (47.9)
Extremely influential	335 (32.6)
Scientific Information
Influence level in COVID-19 statistics
Slightly/Somewhat Influential	190 (18.5)
Very influential	474 (46.2)
Extremely influential	363 (35.3)
Influence level in up-to-date scientific information
Slightly/Somewhat Influential	126 (12.3)
Very influential	585 (57.0)
Extremely influential	316 (30.8)
Total	1027

Table 2:

Multinomial Logistic Regression: Associations between Healthcare Experiences and Demographic Characteristics.

	Healthcare Experience
	Personal Healthcare Experience						Trust Level in Doctors’ Sources
	The Same VS. Worse than others			The Same VS. Better than others			Neutral/Not trustworthy VS. Somewhat trustworthy			Neutral/Not trustworthy VS. Extremely trustworthy
Predictors	aOR	95% CI		aOR	95% CI		aOR	95% CI		aOR	95% CI
(Intercept)	0.12	0.05	0.31	2.64	1.63	4.28	6.20	2.28	16.85	12.04	4.59	31.56
Demographics
Age
22–24	Ref			Ref			Ref			Ref
14–17	0.43	0.19	0.98	0.73	0.48	1.11	1.88	0.72	4.92	3.03	1.20	7.62
18–21	0.86	0.52	1.42	0.80	0.58	1.09	1.19	0.71	2.01	1.31	0.80	2.16
Race/Ethnicity
White	Ref			Ref			Ref			Ref
American Indian or Alaska Native	2.06	0.76	5.58	0.64	0.37	1.12	0.94	0.27	3.33	0.53	0.15	1.84
Asian	2.16	0.76	6.18	0.74	0.40	1.35	3.74	0.43	32.42	2.01	0.24	17.20
Black	1.91	0.87	4.20	0.37	0.24	0.59	0.23	0.09	0.57	0.14	0.06	0.35
Hispanic	1.68	0.79	3.56	0.77	0.52	1.15	0.28	0.12	0.69	0.25	0.10	0.58
Multiracial	0.82	0.24	2.82	1.06	0.61	1.84	0.71	0.21	2.44	0.55	0.17	1.84
Sexual Identity
Straight	Ref			Ref			Ref			Ref
Asexual/Ace Spectrum	1.76	0.48	6.50	1.28	0.61	2.68	1.62	0.31	8.42	2.65	0.55	12.79
Bisexual/Pansexual	1.44	0.71	2.91	0.92	0.62	1.36	1.08	0.55	2.14	1.26	0.66	2.42
Gay/Lesbian	1.40	0.65	3.03	1.05	0.68	1.61	2.44	1.07	5.59	2.37	1.06	5.26
Queer	3.12	1.26	7.76	1.53	0.86	2.69	0.64	0.27	1.54	0.63	0.28	1.44
Questioning	2.81	0.74	10.69	0.61	0.23	1.64	2.68	0.31	23.04	1.61	0.19	13.72
Gender Identity
Woman/Girl	Ref			Ref			Ref			Ref
Agender	--	--	--	1.37	0.26	7.31	1.28	0.11	14.58	0.59	0.05	6.55
Gender Queer	0.59	0.10	3.60	0.94	0.33	2.73	0.52	0.11	2.50	0.55	0.13	2.29
Man/Boy	0.83	0.45	1.53	0.65	0.45	0.93	1.11	0.57	2.17	1.34	0.70	2.54
Non-binary	0.80	0.29	2.20	0.71	0.37	1.34	1.76	0.56	5.55	1.40	0.47	4.17
Questioning	1.15	0.09	15.18	1.14	0.18	7.11	6.84	0.41	112.77	15.83	0.87	289.39
Gender Modality
Cisgender	Ref			Ref			Ref			Ref
Trans and Gender Diverse	2.03	0.88	4.65	2.20	1.32	3.67	0.98	0.39	2.51	1.25	0.52	3.02
Not Sure	1.55	0.15	15.62	0.68	0.13	3.54	0.48	0.07	3.19	0.15	0.02	1.22

“ -- ” Unstable estimates due to small sample cells.

RESULTS

Most participants were between the ages of 18 and 21 years (48.3%) and identified as Hispanic (33.3%) or white (22.5%; Table 1). Approximately one-third of the sample identified as bisexual or pansexual (34.3%), 22.1% identified as gay or lesbian, and 22.0% identified as straight. Substantially fewer individuals identified as asexual/ace spectrum, queer, or questioning. The majority of the sample identified as cisgender (64.2%) and nearly one-third (32.2%) identified as transgender and gender diverse.

Over half of the participants (58.1%) reported their personal healthcare experiences were worse than others during the COVID-19 pandemic. Most participants (60.4%) found their doctor’s sources in making decisions on COVID-19 prevention and treatment behaviors to be extremely trustworthy. Over half of the participants (55.9%) were extremely trusting of the CDC’s information about COVID-19, but only about one-third (32.6%) thought the CDC’s COVID-19 data were extremely influential.

Nearly one-half of participants (46.2%) thought statistics about the dangers of COVID-19 were very influential in decision-making. Up-to-date scientific information was more influential, with 57.0% indicating it was very influential in their decision-making.

Multinomial Logistic Regression Models

Healthcare Experiences

Healthcare experiences included two main factors: personal healthcare experience and trust level in the provider (Table 2). Personal healthcare experiences within the past 12 months were explored using three levels: the same (reference group), worse than others, and better than others. Overall, the data fit the model for this factor well (X² = 94.624, p < 0.01).

Queer YYA had higher odds of reporting worse personal healthcare experiences than their straight peers (aOR = 3.12; 95% CI: 1.26, 7.76). Black YYA (vs. white) and man/boy (vs. woman/girl) were significantly less likely to report better personal healthcare experiences (aOR = 0.37; 95% CI: 0.24, 0.59 and aOR = 0.65; 95% CI: 0.45, 0.93, respectively). Trans and gender diverse YYA were significantly more likely to report better personal healthcare experiences than cisgender YYA (aOR = 2.20; 95% CI: 1.32, 3.67).

Trust in doctor’s sources of information was explored using three levels: neutral/not trustworthy (reference group), somewhat trustworthy, and extremely trustworthy. Overall, the data fit the model for personal healthcare experiences well (X² = 86.727, p < 0.01).

Results showed that both Black and Hispanic YYA were significantly less likely to report somewhat or extreme trust in doctor’s sources than their white peers. Conversely, the odds of gay/lesbian YYA that reported somewhat (aOR = 2.44; 95% CI: 1.07, 5.59) or extreme trust (aOR = 2.37; 95% CI: 1.06, 5.26) were two times higher than their straight peers.

Government Sources of Information

Government sources of information were assessed through two factors: trust in the CDC’s COVID-19 information and the influence of the CDC’s COVID-19 data (Table 3). The data fit two separate models well (trust: X² = 86.343, p < 0.01; influence: X² = 90.81, p < 0.01).

Table 3:

Multinomial Logistic Regression: Associations between Trustworthiness in Government Information and Demographic Characteristics.

	Government Sources of Information
	Trust Level in CDC's General COVID-19 Information						Influence Level in CDC’s COVID-19 Data
	Neutral/Not trustworthy VS. Somewhat trustworthy			Neutral/Not trustworthy VS. Extremely trustworthy			Slightly/Somewhat Influential VS. Very influential			Slightly/Somewhat Influential VS. Extremely influential
Predictors	aOR	95% CI		aOR	95% CI		aOR	95% CI		aOR	95% CI
(Intercept)	4.22	1.90	9.38	8.16	3.83	17.37	3.34	1.68	6.62	4.08	2.11	7.89
Demographics
Age
22–24	Ref			Ref			Ref			Ref
14–17	0.90	0.48	1.69	1.43	0.80	2.56	1.51	0.88	2.59	1.35	0.80	2.26
18–21	1.71	1.09	2.70	1.99	1.29	3.06	1.60	1.08	2.39	1.84	1.26	2.68
Race/Ethnicity
White	Ref			Ref			Ref			Ref
American Indian or Alaska Native	0.30	0.13	0.69	0.18	0.08	0.40	0.24	0.11	0.51	0.19	0.09	0.40
Asian	2.77	0.58	13.27	2.00	0.43	9.27	0.73	0.29	1.84	0.60	0.25	1.49
Black	0.25	0.12	0.52	0.16	0.08	0.32	0.25	0.13	0.48	0.18	0.10	0.33
Hispanic	0.54	0.27	1.08	0.43	0.22	0.83	0.32	0.17	0.58	0.36	0.20	0.63
Multiracial	0.60	0.24	1.49	0.48	0.20	1.12	0.42	0.19	0.91	0.36	0.17	0.75
Sexual Identity
Straight	Ref			Ref			Ref			Ref
Asexual/Ace Spectrum	0.55	0.19	1.64	0.80	0.30	2.14	0.39	0.15	1.00	0.81	0.36	1.82
Bisexual/Pansexual	0.76	0.42	1.35	0.84	0.49	1.46	1.02	0.61	1.69	1.39	0.86	2.26
Gay/Lesbian	1.09	0.57	2.07	0.98	0.53	1.82	1.18	0.68	2.06	1.41	0.83	2.42
Queer	0.71	0.31	1.61	0.88	0.41	1.91	1.53	0.74	3.17	1.74	0.86	3.49
Questioning	1.99	0.39	10.10	1.33	0.27	6.56	0.81	0.27	2.40	0.47	0.15	1.54
Gender Identity
Woman/Girl	Ref			Ref			Ref			Ref
Agender	0.56	0.03	10.13	1.82	0.19	17.25	1.66	0.24	11.61	1.10	0.18	6.55
Gender Queer	0.88	0.21	3.79	0.92	0.25	3.41	1.22	0.34	4.37	0.95	0.30	3.01
Man/Boy	0.95	0.57	1.61	0.84	0.51	1.37	1.03	0.65	1.63	0.92	0.59	1.42
Non-binary	1.48	0.58	3.77	1.01	0.42	2.44	1.65	0.73	3.73	1.13	0.54	2.38
Questioning	7.29	0.65	81.12	12.04	1.12	129.98	4.56	0.56	37.29	10.99	0.77	156.12
Gender Modality
Cisgender	Ref			Ref			Ref			Ref
Trans and Gender Diverse	1.05	0.50	2.23	1.29	0.64	2.59	0.60	0.31	1.15	0.94	0.52	1.70
Not Sure	0.27	0.04	1.70	0.16	0.02	1.02	0.56	0.10	3.01	0.11	0.01	1.18

Except for those who identified as Asian, racial/ethnic minority YYA were less likely to report somewhat or extreme trust in the CDC’s general information or its COVID-19 data than white YYA. Specifically, American Indian/Alaska Native YYA (somewhat neutral: aOR = 0.30; 95% CI: 0.13, 0.69; extremely neutral: aOR = 0.18; 95% CI: 0.08, 0.40) and Black YYA (somewhat neutral: aOR = 0.25; 95% CI: 0.12, 0.52; extremely neutral: aOR = 0.16; 95% CI: 0.08, 0.32) were significantly less likely to report somewhat or extreme trust in CDC’s information than their white peers. Similarly, except for those who identify as Asian, racial/ethnic minority YYA were less likely to report being very or extremely influenced by CDC’s data than white participants.

In comparison to those who identified as white, American Indian/Alaskan Native and Black identifying individuals had lower odds of reporting that the CDC’s general information was somewhat or extremely trustworthy. Individuals who identified as Hispanic had lower odds of reporting that the CDC’s general information was extremely trustworthy when compared to their white counterparts.

Scientific Sources of Information

Scientific sources of information were assessed through two factors: influence of COVID-19 statistics and influence of up-to-date scientific information (Table 4). The data fit two separate models well (statistics: X² = 66.58, p < 0.01; scientific information: X² = 89.91, p < 0.01).

Table 4:

Multinomial Logistic Regression: Associations between Trustworthiness in Scientific Information and Demographic Characteristics.

	Scientific Sources of Information
	Influence Level in the COVID-19 Statistics						Influence Level in Up-to-Date Scientific Information
	Slightly/Somewhat Influential VS. Very influential			Slightly/Somewhat Influential VS. Extremely influential			Slightly/Somewhat Influential VS. Very influential			Slightly/Somewhat Influential VS. Extremely influential
Predictors	aOR	95% CI		aOR	95% CI		aOR	95% CI		aOR	95% CI
(Intercept)	2.21	1.18	4.14	2.11	1.15	3.90	9.36	3.60	24.36	18.33	7.29	46.09
Demographics
Age
22–24	Ref			Ref			Ref			Ref
14–17	1.30	0.74	2.29	1.18	0.68	2.05	1.55	0.81	2.96	0.85	0.45	1.60
18–21	0.96	0.65	1.43	1.05	0.72	1.53	1.01	0.63	1.61	1.07	0.69	1.65
Race/Ethnicity
White	Ref			Ref			Ref			Ref
American Indian or Alaska Native	0.56	0.28	1.12	0.52	0.26	1.04	0.34	0.12	0.94	0.16	0.06	0.44
Asian	1.05	0.45	2.44	1.19	0.52	2.72	0.53	0.16	1.77	0.38	0.12	1.22
Black	0.35	0.20	0.63	0.40	0.23	0.71	0.21	0.08	0.51	0.12	0.05	0.27
Hispanic	0.74	0.43	1.28	0.93	0.55	1.59	0.21	0.09	0.51	0.20	0.09	0.47
Multiracial	0.86	0.41	1.77	0.65	0.32	1.34	0.26	0.09	0.73	0.20	0.07	0.52
Sexual Identity
Straight	Ref			Ref			Ref			Ref
Asexual/Ace Spectrum	1.00	0.38	2.60	1.47	0.59	3.65	0.65	0.20	2.10	1.02	0.34	3.03
Bisexual/Pansexual	1.11	0.68	1.82	1.51	0.94	2.44	0.83	0.46	1.48	0.99	0.57	1.73
Gay/Lesbian	1.48	0.85	2.56	1.60	0.94	2.73	0.90	0.48	1.68	0.96	0.53	1.75
Queer	1.13	0.55	2.32	1.75	0.88	3.47	0.58	0.24	1.40	1.13	0.51	2.51
Questioning	1.13	0.36	3.57	0.77	0.23	2.60	0.60	0.16	2.22	0.62	0.18	2.21
Gender Identity
Woman/Girl	Ref			Ref			Ref			Ref
Agender	--	--	--	--	--	--	0.56	0.04	7.26	1.07	0.11	10.30
Gender Queer	0.40	0.10	1.55	0.62	0.18	2.12	0.49	0.11	2.11	0.54	0.14	2.07
Man/Boy	0.76	0.48	1.20	0.78	0.50	1.21	0.68	0.41	1.14	0.71	0.44	1.16
Non-binary	0.79	0.33	1.89	0.67	0.29	1.56	1.00	0.36	2.79	1.04	0.39	2.73
Questioning	2.11	0.22	19.96	3.60	0.33	38.94	3.53	0.14	88.11	4.09	0.20	85.31
Gender Modality
Cisgender	Ref			Ref			Ref			Ref
Trans and Gender Diverse	2.11	1.04	4.29	2.27	1.15	4.49	1.63	0.75	3.58	1.91	0.91	3.97
Not Sure	0.94	0.15	6.03	0.44	0.06	3.31	1.04	0.09	12.12	1.01	0.10	9.71

“ -- ” Unstable estimates due to small sample cells.

Significant differences in two factors were found by race/ethnicity and gender modality. Black YYA were less likely to report being very or extremely influenced by scientific information than their white peers (very influential: aOR = 0.35; 95% CI: 0.20, 0.63; extremely influential: aOR = 0.40; 95% CI: 0.23, 0.71). Additionally, all other racial/ethnic minority groups except Asians were less likely to be influenced by up-to-date scientific information than white YYA. Transgender and gender diverse YYA were more than twice as likely to report being very or extremely influenced by statistics of the dangers of COVID-19 than cisgender YYA (very influential: aOR = 2.11, 95% CI: 1.04, 4.29; extremely influential: aOR = 2.27, 95% CI: 1.15, 4.49).

DISCUSSION

MM among YYA remains an understudied topic that plays a role in the utilization of healthcare services and receptiveness to guidance from healthcare professionals. Scholarly consensus underscores the significant impact of MM, especially among YYA, on exacerbating disparities in both healthcare experiences and outcomes.(26, 47) When left unaddressed, MM can lead to a cascade of negative consequences, including increased risk behavior, delays in seeking essential care, poor adherence to treatment plans, and avoidance of preventive services. This is particularly concerning for minoritized populations like SGM and REM, and those at the intersection of both identities, who already face systemic barriers to accessing quality healthcare. Therefore, the intersection of race/ethnicity, gender identity, and sexual orientation should be incorporated into health interventions and outreach to better understand the unique barriers that populations at the intersection of multiple minoritized identities face.

Although past research consistently highlights worse outcomes for those with minoritized backgrounds, this study did not always reinforce this assumption, at least not among YYA. In this study, adolescents, in general, were less likely to report worse healthcare experiences while transgender and gender diverse participants reported better health experiences compared to their cisgender peers. Perhaps, the involvement of unobserved protective factors can explain the better healthcare experiences reported by those who identify as transgender and gender diverse. At the same time, our results demonstrate worse reported healthcare experiences among those identifying as queer, Black, or men.

Participants from REM backgrounds like those from Black, Native American and Alaskan Native, and Hispanic communities reported that they did not consider the CDC to be a trustworthy source of general COVID-19 information and data. Perhaps the CDC's responses to the pandemic, including mixed messaging and a lack of clear guidance, contributed to this mistrust. For instance, during the early stages of the pandemic, the CDC and other health organizations provided conflicting information about mask-wearing, social distancing, and other mitigation strategies.(48, 49) This lack of coordination and clarity left many communities with unanswered questions and concerns, which only added to the sense of uncertainty and mistrust. This significant mistrust in the CDC suggests more work must be done by the CDC and its related health institutions to restore the trust of minoritized communities who were historically harmed and stigmatized by the US health system. However, fixing such strained relationships cannot be done simply with a change in messaging tone. This requires an approach that is historically informed and fosters collaboration with the community when implementing projects that seek to improve outreach and service delivery. US health institutions must demonstrate their willingness to address concerns and express a commitment to these communities.

Healthcare providers, as the first point of contact between the public and the US healthcare system, are uniquely positioned to mitigate MM. However, they face a steeper challenge to reduce MM among some minoritized communities over others. In this study, those from Black and Hispanic communities reported less trust in healthcare providers and physicians while participants who were younger and identified as gay/lesbian reported higher trust in healthcare providers and physicians. This suggests that more should be done to increase trust between healthcare providers and those from REM backgrounds. Research has demonstrated that race concordance between physicians and patients is associated with better patient-physician communication.(50, 51) Healthcare providers should incorporate structural factors into their practice.

Public health’s success is, among other things, reliant on the translation of data through inclusive messaging and information sharing. However, only those aged 18–21 years reported that data influenced their health-related decision-making more than older adults aged 22–24 years. At the same time, those from Black, Hispanic, and Alaska Native/American Indian communities, as well as those who identified as multiracial, reported that data did not influence their health-related decision-making. This suggests that communication strategies solely based on health data might not particularly be able to improve engagement with prevention strategies. Perhaps this lack of engagement with data-driven messaging is due to a history of exclusion from data or presentation of data that is not relevant or does not recognize the experiences of these minoritized communities. One sign of progress is that national-level and large data collection tools such as CDC’s COVID Data Tracker(52) have been tracking COVID-related information such as deaths and vaccination coverage by race and ethnicity including Black, Indigenous, and people of color (BIPOC). Governments, research institutions, and healthcare providers should consider the effort to educate these minoritized communities about available data, providing guidance and support on how to access and use this information for informed decision-making.

Trans and gender diverse communities were more likely to report COVID-19 statistics of danger as being influential in their decision-making when compared to their cisgender peers. Black communities were less likely to report that statistics of danger were influential compared to their white peers. Although these communities experience discrimination, they are unique and have differing responses to communication strategies that rely on presenting danger.

The diversity in responding to the statistics of danger included in formal health messaging should be further explored by researchers. Our results suggest that different communities do not perceive and internalize information the same way, and some might not find it helpful in addressing their concerns, especially during public health crises. For example, with the exception of Asian YYA, other YYA on average did not find up-to-date information of use or influential in their decision-making. Factors like information overload and message fatigue may also play a significant role in this decision-making,(53) which should be taken into consideration when evaluating the role of messaging interventions and decision making. Lastly, research around SGM populations tends to see this community as a monolith which might lead to development of messaging approaches that do not adequately address unique needs of subpopulations. Recognizing the diversity present in SGM populations can help researchers to better understand the nuance in responses to health information that exists within the SGM community.

LIMITATIONS

This study is not without limitations. Our specific focus on YYA prevents our findings from being generalized to the broader US population. Additionally, our data did not account for the beliefs and perspectives of parents, guardians, and friends within the social networks of YYA, whose perspectives and healthcare experiences could strongly shape MM in YYA. Past research highlights regional variations in healthcare experiences and attitudes towards the government as a reliable source of information. Unfortunately, our analyses did not control for these regional differences. Examining the influence of state-level factors on MM while accounting for varying policies and public health practices could be a valuable area for future research. Furthermore, variables such as education, socioeconomic status, and other social determinants of health were not included since they are correlated with age. Lastly, the study seeks to address the underrepresentation of YYA in research and treat them as individuals in order to avoid conflating guardian’s wealth with their own. Therefore, guardian socioeconomic status was not included to avoid conflating the guardian’s status with participant’s status. Additionally, our participant recruitment method via paid social media ads and organizational outreach may have introduced a selection bias, potentially skewing the sample towards certain subgroups. Paid social media ads may have attracted tech-savvy individuals with higher socioeconomic status, while outreach with organizations serving marginalized youth may have recruited participants who are more comfortable with their identities, potentially inflating representation of certain subgroups. Conversely, this strategy may have excluded participants who are not connected to these organizations or do not use social media, potentially leading to underrepresentation of certain groups, such as those living in rural areas, and limiting the study's generalizability and ability to make inferences about the broader population of minoritized youth. Identifying and addressing health disparities is critical for those at the intersections of minoritized identities; however, this study did not conduct further intersectional analyses due to limited sample sizes of individuals at the intersections of race/ethnicity, sexual identity, gender identity, and gender modality. Finally, due to its recency, we used COVID-19 as an example of a public health crisis. However, unique characteristics and the nature of each public health crisis must be taken into account when developing a public health response, especially for minoritized populations.

CONCLUSION

Medical mistrust is seen as a barrier to assessing healthcare needs and addressing health disparities; however, limited literature has focused on assessing medical mistrust for vulnerable populations, especially among minoritized YYA. Our study addressed this gap by examining multiple dimensions of medical mistrust among YYA, including not only personal healthcare experiences but also trust in doctors’ sources, government, and scientific information during the COVID-19 pandemic. Our findings indicated the importance of incorporating minoritized identities into the assessment of medical mistrust to understand prevention and treatment behaviors of minoritized populations and to better identify health disparities and develop public health prevention and treatment strategies targeting YYA who may experience MM.