The Usage-Trust Gap: Information Sources, Trust, and COVID-19 Knowledge Among American Indian and Alaska Native Adults in Rural Michigan

Abstract

American Indian and Alaska Native (AI/AN) communities experienced disproportionate COVID-19 morbidity and mortality, particularly in rural areas with limited public health infrastructure. This study examined primary COVID-19 information sources among AI/AN adults in rural Michigan and evaluated how trust in these sources relates to health knowledge, attitudes, and vaccination behaviors. We conducted a prospective, randomized pre-post interventional study among 273 adults at a tribal health clinic in rural Isabella County, Michigan (2022–2024). Participants were assigned to receive a culturally tailored educational video or infographic, and surveys assessed COVID-19 knowledge, vaccine attitudes, information sources, and perceived reliability. Social media was the most frequently used information source but was rated as less reliable, whereas healthcare workers (HCWs) were considered the most trusted. Reliance on HCWs and personal relationships was associated with higher baseline vaccine knowledge and greater uptake of influenza vaccination. Both educational formats resulted in modest improvements in COVID-19 knowledge and vaccine attitudes. While no consistent differences were observed between formats overall, infographic-based education was associated with greater gains in select vaccine knowledge domains among participants who relied on trusted interpersonal or clinical information sources. These findings highlight a “usage-trust gap” in rural AI/AN health communication, where frequently used information channels are not necessarily the most trusted. Culturally tailored messaging delivered through trusted clinical and interpersonal networks may enhance the effectiveness of public health communication and support vaccine uptake in underserved communities.

1. Introduction

American Indian and Alaska Native (AI/AN) communities have long experienced profound health inequities rooted in historical trauma, systemic underfunding of tribal health services, and enduring barriers to care. These structural injustices have manifested in disproportionately high burdens of chronic disease and a persistent mistrust of mainstream healthcare institutions [1]. The COVID-19 pandemic intensified these disparities, with AI/AN populations experiencing some of the highest infection, hospitalization, and mortality rates in the United States. Contributing factors included socioeconomic vulnerability, crowded housing, comorbid conditions, and disruptions to food and healthcare systems [2].

In this context, culturally congruent and trusted health communication became essential. However, messaging in AI/AN communities is shaped by longstanding federal neglect and skepticism toward external authorities [3]. Identifying which information channels AI/AN adults rely upon—and how trust in those sources influences health decision-making—is critical to designing effective interventions. Prior research emphasizes the importance of trusted messengers, including local healthcare workers (HCWs) and tribal leaders, yet there remains limited empirical evidence linking specific information sources to measurable changes in health knowledge, attitudes, or behaviors in AI/AN populations [4,5,6,7,8].

This study addresses these gaps by examining both the primary COVID-19 information sources used by AI/AN adults and how trust in these sources relates to measurable outcomes, including health knowledge, vaccine attitudes, and vaccination behaviors. Despite growing literature on health communication in AI/AN populations, limited evidence directly links specific information sources to quantifiable changes in knowledge or behavior, particularly in rural AI/AN communities [9,10]. These questions are particularly relevant in rural Michigan, where tribal communities face provider shortages, transportation barriers, and limited broadband access [11]. Michigan represents a critical yet understudied setting in AI/AN health research. The majority of existing literature has focused on tribal populations in the Western and Southwestern United States, leaving Midwestern and Eastern AI/AN communities comparatively underrepresented despite potentially distinct sociocultural and healthcare contexts [9,12,13]. This study evaluates the effectiveness of culturally tailored educational tools and examines for whom—and through which information pathways—they function most effectively. In addition, we further hypothesize the presence of a “usage-trust gap,” in which frequently accessed information sources may not align with those perceived as most credible, and explore this dynamic through a proposed Relational-Cultural Trust Model.

2. Methods

2.1. Study Design and Setting

We conducted a pre/post-intervention, randomized, prospective study to assess the effectiveness of two educational modalities, video and infographic, in improving COVID-19 knowledge and vaccine-related attitudes among AI/AN adults in rural Michigan. In this study, “effectiveness” was defined as the extent to which educational interventions improve COVID-19-related knowledge in the short term. This study represents a pre-specified secondary exploratory analysis, focusing specifically on information sources and perceived reliability, of a previously published prospective, randomized educational intervention dataset [14]. Rural status was defined using RUCC classifications, with nonmetropolitan counties (RUCC 4–9) considered rural. The participating clinic is located in Isabella County, Michigan, classified as RUCC 6 (nonmetropolitan county with an urban population of 2500–19,999, adjacent to a metropolitan area).

Data were collected between February 2022 and January 2024 at a single outpatient Indian Health Service (IHS) clinic affiliated with the Saginaw Chippewa Indian Tribe and Central Michigan University (CMU). The clinic provides primary care services to approximately 4100 active patients annually.

After obtaining informed consent, participants completed a pre-survey, were randomly assigned to view either a short educational video or an infographic, and then completed a post-survey. The intervention and data collection process were completed in a single clinic visit. The CMU College of Medicine Institutional Review Board and the Saginaw Chippewa Indian Tribe approved all study procedures.

2.2. Participants

Eligible participants were English-speaking adults aged 18 years or older who received care at the participating IHS clinic. Research assistants recruited participants from the clinic’s waiting area using convenience sampling during routine operating hours. Exclusion criteria included individuals under 18 years old, non-patients, or those unable to provide informed consent. Participants completed the survey using a study-provided tablet or their personal device via QR code access. Upon completion, participants received a $10 gift card as compensation.

2.3. Survey Instruments and Educational Interventions

Participants were randomly assigned to a 1:1 ratio using a computer-generated randomization sequence embedded within the survey platform, ensuring allocation concealment at the point of enrollment. Both educational interventions, a video or static infographic, conveyed information about COVID-19 transmission, prevention, and vaccination safety. Both materials were culturally tailored through collaboration with clinicians, tribal leaders, and individuals familiar with AI/AN health communication practices. Tailoring included framing of health messages to align with community values and experiences. After viewing the assigned material, participants completed an identical post-intervention survey. Survey items measured seven COVID-19 knowledge domains and two vaccine attitude constructs. Knowledge items were scored from 0 (incorrect) to 2 (correct), while vaccine acceptance and concern were rated from 1 (unlikely) to 7 (most likely). The educational materials were based on information from the Centers for Disease Control and Prevention (CDC) and World Health Organization (WHO). Content validity was established through expert review by clinicians and public health researchers.

2.4. Statistical Analysis

We used paired-sample t-tests to compare pre- and post-intervention mean scores within each group. Mann–Whitney U tests compared changes between intervention types. Given the exploratory nature of this analysis, adjustments for multiple comparisons were not applied; findings should therefore be interpreted cautiously. All analyses were conducted usingIBM SPSS Statistics version 29 (IBM Corp., Armonk, NY, USA), with statistical significance set at p < 0.05. The Central Michigan University Institutional Review Board approved the study protocol.

3. Results

3.1. Participant Demographics and Survey Reliability

A total of 273 participants completed the study (Figure 1). The majority were female (n = 183, 67.0%) and non-Hispanic (n = 224, 91.1%). Age distribution ranged from 18–24 years (14.2%) to 65+ years (10.3%), with most participants residing in a nonmetropolitan, micropolitan county (RUCC 6) (n = 216, 79.4%). Political affiliation was diverse, including Democrats (29.7%), Independents (24.2%), Republicans (14.9%), and other affiliations (31.2%) (

Table 1. Demographic characteristics of participants. Descriptive statistics of participant characteristics, including gender, age, ethnicity, residence, political affiliation, COVID-19–related experiences, and adherence to Centers for Disease Control and Prevention (CDC)–recommended precautions. CDC-recommended precautions include measures such as mask-wearing, hand hygiene, physical distancing, and vaccination. Values are presented as n (%) or median (interquartile range).

Variable	Total n (%)
Gender	273
Female	183 (67.0%)
Male	85 (31.0%)
Other	5 (1.8%)
Age Group	273
18–24 years old	39 (14.2%)
25–34 years old	60 (22.0%)
35–44 years old	57 (20.9%)
45–54 years old	45 (16.5%)
55–64 years old	44 (16.1%)
65+ years old	28 (10.3%)
Ethnicity	246
Hispanic	22 (8.9%)
Non-Hispanic	224 (91.1%)
Residence	272
Metropolitan	35 (12.9%)
Micropolitan	216 (79.4%)
Non-Metropolitan	21 (7.7%)
Political Affiliation	269
Republican	40 (14.9%)
Democrat	80 (29.7%)
Independent	65 (24.2%)
Something else	84 (31.2%)
Was your employment status impacted by the COVID-19 pandemic?	234
Yes	128 (54.7%)
No	106 (45.3%)
Essential worker/household has essential worker	273
Self/Family	139 (50.9%)
No	134 (49.1%)
Since January 2020, have you been tested for the COVID-19 virus?	273
Yes	239 (87.5%)
No	31 (11.4%)
Not Sure	3 (1.1%)
Have you or anyone in your social circle ever tested positive for COVID-19?	273
Yes	246 (90.1%)
No	22 (8.1%)
Not Sure	5 (1.8%)
Do you personally know someone who has been hospitalized or died from COVID-19?	273
Yes	158 (57.9%)
No	106 (38.8%)
Not Sure	9 (3.3%)
Self-Risk	273
None/Don’t Know	56 (20.5%)
One risk	127 (46.5%)
Two or more risks	90 (33.0%)
Other-Risk	237
None/Don’t Know	65 (23.8%)
One risk	110 (40.3%)
Two or more risks	98 (35.9%)
Following CDC-recommended COVID-19 precautions	268
At least 5 precautions	137 (51.1%)
More than 5 precautions	131 (48.9%)
Median Number of CDC precautions followed from list	5.00 (3.00–8.00)

).

Regarding COVID-19 experiences, over half reported employment impacts due to the pandemic (n = 128, 54.7%), and 50.9% identified themselves or a household member as an essential worker. Most participants had been tested for COVID-19 (87.5%), and 90.1% reported that they or someone in their social circle had tested positive. Additionally, 57.9% personally knew someone who had been hospitalized or died from COVID-19. Self-reported risk factors for severe COVID-19 ranged from none (20.5%) to two or more (33.0%). Participants followed a median of five CDC-recommended precautions (IQR 3–8), with 48.9% following more than five precautions (Table 1).

The internal consistency of the survey items was acceptable, with Cronbach’s alpha coefficients of 0.663 for the 29-item questionnaire and 0.725 for the 7-item multiple-choice questions, indicating adequate reliability of the survey instrument. Missing data were handled conservatively; cases were excluded only when both pre- and post-survey responses were missing.

3.2. Primary Information Sources and Perceived Reliability

Participants reported multiple sources of COVID-19 information, with social media (n = 126), healthcare workers (HCWs, n = 104), and television (TV, n = 98) identified as the top three sources (

Table 2. Mean reliability scores for nine sources of COVID-19 information. Mean and standard deviation (SD) of participants’ perceived reliability for various information sources, including television, radio, print, social media, websites, healthcare workers, personal relationships, religious leaders, and government representatives. Higher scores indicate greater perceived reliability. All values are presented as n, mean, and SD.

Source	Total Count (n)	Reliability Count (n)	Mean Reliability Score (M)	SD
TV	98	85	3.45	1.01
Radio	36	30	2.97	1.29
Print	26	24	2.54	1.06
Social media	126	118	2.86	0.98
Web	61	57	3.28	1.19
Healthcare Worker	104	101	4.13	0.90
Personal Relationships	96	93	3.42	0.99
Religious Leader	16	10	2.10	1.79
Governor	27	24	3.71	0.95

). Although social media was the most frequently used source, it received moderate reliability ratings (M = 2.86, SD = 0.98). HCWs were rated as the most reliable source overall (M = 4.13, SD = 0.90), followed by governors (M = 3.71, SD = 0.95) and TV (M = 3.45, SD = 1.01). Personal relationships (M = 3.42, SD = 0.99) were also moderately reliable, whereas religious leaders were rated the lowest (M = 2.10, SD = 1.79).

3.3. Associations Between Information Sources and COVID-19 Knowledge and Attitudes

Analysis of pre-survey differences indicated that participants who primarily relied on healthcare workers (HCWs) demonstrated significantly higher knowledge scores on several vaccine-related items compared to non-users. Specifically, HCW users scored higher on vaccine effectiveness (user M = 1.63, non-user M = 1.28, p = 0.002), actions after vaccination (user M = 1.51, non-user M = 1.27, p = 0.034), reduce the risk of transmission of COVID-19 (user M = 1.24, non-user M = 1.03, p = 0.004), ways COVID-19 spreads (user M = 1.58, non-user M = 1.33, p = 0.012), vaccine facts and protection (user M = 1.60, non-user M = 1.29, p = 0.011), and vaccine development and safety (user M = 0.89, non-user M = 0.66, p = 0.038). Web-based sources were also associated with higher scores on vaccine facts and protection (user M = 1.67, non-user M = 1.33, p = 0.013) and vaccine development and safety (user M = 0.95, non-user M = 0.69, p = 0.048). Participants who relied on personal relationships demonstrated significantly higher scores on reduce the risk of transmission of COVID-19 (user M = 1.24, non-user M = 1.04, p = 0.008), ways COVID-19 spreads (user M = 1.61, non-user M = 1.32, p = 0.002), and vaccine facts and protection (user M = 1.60, non-user M = 1.30, p = 0.011) (

Table 3. Association between information source use and COVID-19–related knowledge, attitudes, and vaccine beliefs. Comparison of virus knowledge, vaccine knowledge, attitudes, and precautionary behaviors among users and non-users of nine information sources. Data are presented as mean (SD) or median (interquartile range) where appropriate. p-values are derived from non-parametric Mann-Whitney U tests.

Measure	Group	Television	Radio	Print	Social Media	Websites	Healthcare Workers	Personal Relationships	Religious Leaders	Government
Demographics
Median number of CDC precautions followed from list	Median (IQR)	6.00 (4.00–9.00)	7.00 (5.00–9.75)	6.00 (4.75–8.00)	5.00 (4.00–7.25)	6.00 (5.00–8.00)	6.00 (5.00–8.00)	6.00 (4.00–8.00)	6.50 (5.00–9.75)	7.00 (5.00–10.00)
Virus Knowledge
Reduce the risk of transmission of COVID-19	User	1.05 (0.52)	1.03 (0.60)	1.12 (0.51)	1.11 (0.54)	1.23 (0.52)	1.24 (0.51)	1.24 (0.49)	1.19 (0.40)	1.11 (0.50)
	Non-User	1.14 (0.60)	1.12 (0.57)	1.11 (0.58)	1.11 (0.61)	1.08 (0.58)	1.03 (0.60)	1.04 (0.60)	1.11 (0.59)	1.11 (0.58)
	p-value	0.181	0.374	0.996	0.966	0.073	0.004	0.008	0.637	0.967
Ways COVID-19 spreads	User	1.46 (0.70)	1.39 (0.73)	1.50 (0.64)	1.39 (0.73)	1.43 (0.71)	1.58 (0.62)	1.61 (0.58)	1.38 (0.71)	1.52 (0.70)
	Non-User	1.40 (0.73)	1.43 (0.68)	1.41 (0.73)	1.45 (0.72)	1.42 (0.72)	1.33 (0.77)	1.32 (0.77)	1.42 (0.72)	1.41 (0.72)
	p-value	0.539	0.620	0.666	0.435	0.988	0.012	0.002	0.714	0.445
Vaccine Attitudes and Beliefs
Concern about taking COVID-19 vaccine	User	2.28 (1.11)	2.26 (1.09)	2.50 (1.10)	2.40 (1.13)	2.13 (1.08)	2.22 (1.07)	2.35 (1.10)	2.38 (1.08)	2.22 (0.97)
	Non-User	2.43 (1.17)	2.39 (1.68)	2.36 (1.16)	2.36 (1.17)	2.45 (1.17)	2.47 (1.19)	2.39 (1.18)	2.38 (1.16)	2.39 (1.17)
	p-value	0.315	0.538	0.545	0.786	0.064	0.092	0.858	0.974	0.491
Vaccine Knowledge
Vaccine facts and protection	User	1.35 (0.89)	1.58 (0.77)	1.62 (0.75)	1.39 (0.90)	1.67 (0.67)	1.60 (0.74)	1.60 (0.74)	1.62 (0.71)	1.41 (0.93)
	Non-User	1.44 (0.87)	1.38 (0.89)	1.38 (0.89)	1.42 (0.86)	1.33 (0.92)	1.29 (0.94)	1.30 (0.93)	1.39 (0.89)	1.41 (0.87)
	p-value	0.345	0.23	0.228	0.859	0.013	0.011	0.011	0.376	0.887
Vaccine effectiveness	User	1.47 (0.86)	1.58 (0.80)	1.65 (0.74)	1.38 (0.92)	1.59 (0.78)	1.63 (0.75)	1.55 (0.80)	1.56 (0.81)	1.67 (0.73)
	Non-User	1.38 (0.92)	1.39 (0.91)	1.39 (0.91)	1.44 (0.88)	1.36 (0.92)	1.28 (0.95)	1.34 (0.94)	1.40 (0.90)	1.39 (0.91)
	p-value	0.514	0.235	0.167	0.607	0.106	0.002	0.091	0.548	0.136
Actions after vaccination	User	1.49 (0.81)	1.47 (0.81)	1.54 (0.81)	1.37 (0.89)	1.44 (0.84)	1.51 (0.82)	1.46 (0.84)	1.44 (0.81)	1.56 (0.80)
	Non-User	1.29 (0.94)	1.34 (0.91)	1.34 (0.90)	1.35 (0.91)	1.33 (0.91)	1.27 (0.93)	1.31 (0.93)	1.35 (0.90)	1.34 (0.91)
	p-value	0.112	0.538	0.306	0.854	0.488	0.034	0.219	0.870	0.248
Awareness of side effects	User	1.29 (0.85)	1.33 (0.83)	1.31 (0.84)	1.29 (0.86)	1.39 (0.80)	1.35 (0.81)	1.44 (0.76)	1.13 (0.96)	1.33 (0.83)
	Non-User	1.35 (0.85)	1.32 (0.85)	1.33 (0.85)	1.36 (0.84)	1.31 (0.86)	1.31 (0.87)	1.27 (0.88)	1.34 (0.84)	1.33 (0.85)
	p-value	0.496	0.967	0.837	0.437	0.563	0.942	0.178	0.378	0.972
Vaccine development and safety	User	0.80 (0.90)	0.92 (0.96)	0.85 (0.96)	0.68 (0.87)	0.95 (0.93)	0.89 (0.93)	0.84 (0.92)	0.50 (0.81)	1.07 (0.95)
	Non-User	0.73 (0.91)	0.73 (0.90)	0.74 (0.90)	0.81 (0.94)	0.69 (0.90)	0.66 (0.89)	0.70 (0.90)	0.77 (0.91)	0.72 (0.90)
	p-value	0.476	0.262	0.607	0.319	0.048	0.038	0.190	0.270	0.054

Statistically significant results are presented in bold.

). No significant differences were observed across groups for knowledge regarding vaccine side effects.

3.4. Impact of Educational Intervention Format on Knowledge and Attitudes

Post-intervention analyses compared video- versus infographic-based education across nine primary information sources (

Table 4. Comparison of COVID-19 knowledge and vaccine perceptions by message format (video vs. infographic) across nine information sources. Mean (SD) scores for virus knowledge, vaccine concerns, vaccine knowledge, and attitudes following exposure to video or infographic materials across different information channels. p-values represent differences between video and infographic groups.

Measure	Group	Television	Radio	Print	Social Media	Websites	Healthcare Workers	Personal Relationships	Religious Leaders	Government
Virus Knowledge
Reduce the risk of transmission of COVID-19	Video	1.22 (0.74)	1.32 (0.67)	1.46 (0.52)	1.40 (0.71)	1.41 (0.62)	1.48 (0.68)	1.48 (0.64)	0.89 (0.78)	1.31 (0.75)
	Infographic	1.33 (0.68)	1.41 (0.79)	1.38 (0.77)	1.44 (0.63)	1.59 (0.66)	1.50 (0.66)	1.55 (0.63)	1.29 (0.95)	1.57 (0.76)
	p-value	0.519	0.594	1.000	0.920	0.171	0.908	0.496	0.408	0.350
Ways COVID-19 spreads	Video	1.31 (0.79)	1.26 (0.80)	1.31 (0.86)	1.42 (0.73)	1.55 (0.63)	1.52 (0.71)	1.55 (0.68)	1.44 (0.73)	1.31 (0.95)
	Infographic	1.35 (0.75)	1.12 (0.85)	1.31 (0.75)	1.39 (0.72)	1.53 (0.67)	1.57 (0.56)	1.54 (0.69)	0.86 (0.69)	1.50 (0.65)
	p-value	0.852	0.639	0.920	0.775	0.966	0.982	0.930	0.142	0.793
Vaccine Concern
Concern about taking the COVID-19 vaccine	Video	2.31 (1.19)	2.42 (1.17)	2.85 (1.06)	2.62 (1.24)	2.21 (1.18)	2.15 (1.15)	2.67 (1.20)	2.0 (1.20)	2.0 (0.95)
	Infographic	2.13 (1.07)	2.40 (1.18)	1.92 (0.90)	2.13 (0.99)	2.23 (1.12)	2.20 (0.97)	2.11 (0.92)	2.33 (0.82)	2.15 (0.90)
	p-value	0.485	0.973	0.040	0.023	0.914	0.664	0.017	0.573	0.611
Vaccine Knowledge
Understanding of vaccine protection	Video	1.10 (0.96)	1.32 (0.95)	1.23 (1.01)	1.45 (0.88)	1.69 (0.66)	1.44 (0.85)	1.45 (0.85)	1.11 (1.05)	1.38 (0.96)
	Infographic	1.57 (0.79)	1.35 (0.93)	2.0 (0.0)	1.73 (0.65)	1.81 (0.54)	1.75 (0.64)	1.84 (0.50)	2.0 (0.0)	1.86 (0.54)
	p-value	0.010	0.925	0.101	0.056	0.394	0.025	0.007	0.142	0.302
Vaccine effectiveness	Video	1.51 (0.85)	1.74 (0.78)	1.77 (0.60)	1.48 (0.86)	1.79 (0.56)	1.67 (0.72)	1.53 (0.82)	1.67 (0.71)	1.38 (0.96)
	Infographic	1.65 (0.75)	1.76 (0.57)	2.0 (0.0)	1.67 (0.71)	1.75 (0.67)	1.68 (0.69)	1.73 (0.65)	2.0 (0.0)	1.86 (0.54)
	p-value	0.350	0.876	0.511	0.217	0.961	0.962	0.169	0.470	0.302
Knowledge about actions post-vaccination	Video	1.45 (0.84)	1.42 (0.84)	1.54 (0.78)	1.39 (0.89)	1.69 (0.66)	1.46 (0.85)	1.33 (0.88)	1.33 (0.87)	1.38 (0.96)
	Infographic	1.57 (0.82)	1.47 (0.87)	1.38 (0.96)	1.64 (0.76)	1.66 (0.75)	1.55 (0.81)	1.63 (0.78)	1.43 (0.98)	1.86 (0.54)
	p-value	0.349	0.802	0.840	0.068	0.95	0.514	0.049	0.758	0.302
Awareness of vaccine side effects	Video	1.33 (0.90)	1.58 (0.77)	1.77 (0.60)	1.42 (0.84)	1.59 (0.73)	1.48 (0.80)	1.63 (0.63)	1.22 (0.97)	1.15 (0.99)
	Infographic	1.45 (0.79)	1.59 (0.80)	1.54 (0.78)	1.52 (0.80)	1.66 (0.70)	1.57 (0.71)	1.54 (0.71)	1.43 (0.98)	1.64 (0.75)
	p-value	0.592	0.925	0.511	0.493	0.639	0.631	0.604	0.681	0.259
Vaccine development and safety	Video	0.84 (0.87)	0.84 (0.90)	0.92 (0.86)	0.94 (0.92)	0.93 (0.92)	1.10 (0.85)	0.88 (0.94)	0.44 (0.53)	1.0 (1.0)
	Infographic	0.86 (0.87)	0.88 (0.99)	1.0 (0.91)	0.89 (0.89)	1.34 (0.90)	1.05 (0.90)	0.98 (0.88)	1.00 (1.0)	1.14 (0.86)
	p-value	0.896	0.950	0.840	0.796	0.077	0.764	0.547	0.299	0.756

Statistically significant results are presented in bold.

). Mann–Whitney U tests were used due to the ordinal nature of the Likert-scale items. No consistent pattern of significant differences was observed across most knowledge and attitude domains. However, participants obtaining information from HCWs and personal relationships demonstrated significantly higher post-survey scores on select vaccine knowledge and attitude items (e.g., understanding of vaccine protection, knowledge about actions post-vaccination) following infographic interventions (p < 0.05). Additional pre- and post-intervention analyses of vaccine attitudes, beliefs, and behavioral intentions across information sources are provided in Tables S1 and S2.

Overall, both infographic and video interventions were associated with modest improvements in post-survey attitudes, suggesting that either format may enhance knowledge and perceptions of COVID-19. Further investigation is warranted to determine the relative effectiveness of the two formats across different information sources.

3.5. Relationship Between Information Source and Vaccination Behaviors

Significant associations were observed between primary information sources and both influenza vaccination in the past year and plans to receive the influenza vaccine during the current season (

Table 5. Relationship between information source and vaccination behaviors (influenza and COVID-19). Frequencies and percentages of participants reporting influenza vaccination in the past year, plans to receive the influenza vaccine during the current season, and COVID-19 vaccination status, stratified by primary information source. Associated p-values represent results of Pearson’s Chi-Square tests. For variables in which Pearson’s Chi-Square test assumptions are violated, Fishers Exact test/Likelihood Ratio test and associated p-values are reported.

Measure	Category	TV	Radio	Print	Social Media	Websites	Healthcare Workers	Personal Relationships	Religious Leaders	Government
Received Flu Shot in the Past Year	n	97	36	26	123	61	103	95	16	26
	Yes	34 (35.1%)	21 (58.3%)	12 (46.2%)	71 (57.7%)	37 (60.7%)	69 (67.0%)	63 (66.3%)	9 (56.3%)	18 (69.2%)
	No	63 (64.9%)	15 (41.7%)	14 (53.8%)	52 (42.3%)	24 (39.3%)	34 (33.0%)	32 (33.7%)	7 (43.8%)	8 (30.8%)
	p-value	0.076	0.954	0.202	0.958	0.619	0.017	0.038	0.891	0.218
Plans to Get Flu Shot This Season	n	98	36	26	126	61	104	96	16	27
	Yes	58 (59.2%)	22 (61.1%)	13 (50.0%)	65 (51.6%)	40 (65.6%)	66 (63.5%)	59 (61.5%)	9 (56.3%)	19 (70.4%)
	Don’t Know/Undecided	11 (11.2%)	3 (8.3%)	3 (11.5%)	23 (18.3%)	6 (9.8%)	15 (14.4%)	16 (16.7%)	0 (0.0%)	3 (11.1%)
	No	29 (29.6%)	11 (30.6%)	10 (38.5%)	38 (30.2%)	15 (24.6%)	23 (22.1%)	21 (21.9%)	7 (43.8%)	5 (18.5%)
	p-value	0.173	0.374	0.772	0.333	0.054	0.006	0.016	0.201	0.126
COVID-19 Vaccination Status	n	98	36	26	126	61	104	96	16	27
	Vaccinated	60 (61.2%)	21 (58.3%)	16 (61.5%)	67 (53.2%)	45 (73.8%)	69 (66.3%)	62 (64.6%)	9 (56.3%)	19 (70.4%)
	Unvaccinated, Hesitant	25 (25.5%)	11 (30.6%)	7 (26.9%)	44 (34.9%)	13 (21.3%)	25 (24.0%)	27 (28.1%)	6 (37.5%)	5 (18.5%)
	Unvaccinated, Not Hesitant	13 (13.3%)	4 (11.1%)	3 (11.5%)	15 (11.9%)	3 (4.9%)	10 (9.6%)	7 (7.3%)	1 (6.3%)	3 (11.1%)
	p-value	0.199	0.966	0.887	0.152	0.030	0.144	0.362	0.770	0.346

Statistically significant results are presented in bold.

). A significant relationship emerged between information source and receipt of the influenza vaccine in the prior year. Participants who relied on healthcare workers had one of the highest influenza vaccination rates (67.0%, p = 0.017). Similarly, those using personal relationships as their main source also demonstrated high uptake (66.3%, p = 0.038). These findings suggest that interpersonal or clinician-based information pathways were associated with higher real-world vaccination behavior.

Information source was also significantly associated with intent to receive a flu vaccine this season, with meaningful differences observed among participants who relied on healthcare workers (p = 0.006) and personal relationships (p = 0.016). Both groups demonstrated high intention to vaccinate: 65.6% among healthcare-worker users and 63.5% among those drawing primarily on personal relationships. These patterns further reinforce the influence of trusted interpersonal networks in shaping preventive health behaviors.

4. Discussion

4.1. Variability in COVID-19 Knowledge Across Information Sources

Findings from this study demonstrate that COVID-19 knowledge among AI/AN adults varies meaningfully based on their primary information source. While social media was the most utilized channel, it was simultaneously rated among the least reliable. This duality reflects its role as both a low-barrier communication tool and a potential source of misinformation [15]. The moderate reliability scores observed in this cohort parallel national trends: while social media facilitates rapid dissemination, it frequently propagates non-evidence-based content that may contribute to public confusion and undermine trust in government and healthcare institutions [16]. These patterns are particularly salient in AI/AN communities, where historical trauma and structural underfunding have already strained relationships with mainstream health systems.

Despite these challenges, social media remains a viable platform for culturally tailored outreach, provided that digital literacy disparities and information saturation are addressed [17,18]. However, mitigation efforts such as automated content filtering may encounter resistance in marginalized populations due to political polarization and a deep-seated mistrust of external authorities [19].

Conversely, participants who relied on HCWs or print media exhibited higher baseline knowledge across multiple vaccine domains. This finding is consistent with prior literature supporting the role of clinician-delivered health communication. Prior research suggests that information delivered by clinicians or reputable print sources correlates with higher vaccine acceptance [20]. The high reliability ratings attributed to HCWs in this study underscore their essential role in delivering accurate, evidence-based information to AI/AN communities. Website-based sources also yielded higher scores in vaccine effectiveness knowledge, suggesting that high-quality digital resources can enhance health literacy when the content is perceived as transparent and trustworthy.

These findings align with a growing body of literature emphasizing the importance of culturally congruent and trust-centered health communication in Indigenous populations. A systematic review identified key components of effective communication, including cultural congruence, Indigenous involvement in message design, and the use of trusted messengers [7,21]. However, prior work has also highlighted a persistent lack of empirically grounded, community-specific interventions in AI/AN populations, with many programs relying on adaptations of non-indigenous models [10,22]. The present findings contribute to this gap by illustrating measurable differences in knowledge associated with information source trust within a rural AI/AN population.

4.2. COVID-19-Related Behaviors Stratified by Information Sources

Rural AI/AN communities frequently face structural barriers including limited broadband access, transportation challenges, and persistent health workforce shortages. In such contexts, reliance on high-volume digital messaging alone could exacerbate disparities, particularly in settings with limited broadband access or digital infrastructure. These findings suggest that embedding culturally tailored materials within existing rural clinical sites may help mitigate these structural limitations.

Information sources not only shaped knowledge but also influenced health behaviors. Participants relying on web-based sources reported higher adherence to CDC-recommended precautions and greater booster vaccination intentions, consistent with research linking high-quality online information to public health compliance [23].

Furthermore, participants who identified HCWs as their primary source were more likely to have received the influenza vaccine and intended to receive future doses. This is consistent with prior international studies showing that clinician recommendations are among the strongest predictors of vaccine uptake [24]. Notably, interpersonal networks also played a crucial role; participants citing personal connections as their primary source exhibited higher knowledge scores, suggesting that peer-to-peer communication remains a vital, albeit informal, complementary pathway of health information exchange.

The observed influence of HCWs and interpersonal networks is consistent with prior studies in AI/AN communities, where trust-based relationships play a central role in shaping health behaviors. For example, research in AN and Southwest tribal communities has demonstrated that vaccine decision-making is frequently influenced by trusted individuals, including clinicians, elders, and community health representatives [25]. These findings support the hypothesis that interpersonal and relational pathways may function as critical conduits for translating health information into action.

4.3. Media Trust and the Role of Health Literacy

No consistent superiority of one educational format was observed overall. However, infographic delivery appeared particularly effective among participants who relied on clinician-based and interpersonal information pathways. Overall, the findings suggest a highly uneven trust landscape within AI/AN communities. HCWs received the highest reliability scores, highlighting their status as the most credible messengers. While government and web-based sources also scored relatively well, this contrasts with previous literature highlighting a legacy of federal mistrust in AI/AN healthcare contexts [3]. Meanwhile, social media usage continues to outpace trust, reflecting a persistent “usage-trust gap” consistent with national Pew Research findings [26].

This pattern may reflect broader historical and cultural factors shaping trust in health information. For many AI/AN individuals, trust is not merely a reflection of content accuracy but a measure of cultural congruence—the degree to which information respects tribal sovereignty and indigenous knowledge [7]. Historical legacies of unethical research and “faceless” institutional bureaucracies have often fostered a protective distrust of mainstream media and federal messaging. Consequently, for information to be deemed “culturally acceptable,” it must often bypass traditional top-down channels and be filtered through community-based participatory frameworks [4].

In this context, the high reliability of HCWs likely stems from their ability to provide interpersonal, relational accountability that national media outlets lack. Ultimately, the variability across sources emphasizes the need for communication strategies that elevate trusted community messengers while strengthening the cultural resonance of the digital channels frequently used by AI/AN adults.

More broadly, these findings are consistent with literature highlighting structural and informational barriers in rural populations. Rural residents often experience reduced access to high-quality health information and may have lower health literacy, particularly in settings with limited broadband infrastructure [27,28]. At the same time, clinicians remain among the most trusted sources of health information in rural communities, reinforcing the importance of embedding communication strategies within existing care relationships [22,29]. Importantly, the literature suggests that effective messaging in both rural and AI/AN populations should prioritize culturally grounded, non-fear-based approaches that emphasize protection of family and community rather than top-down or fear-driven messaging strategies [10]. This may help explain why interpersonal and clinician-based communication pathways were associated with stronger knowledge and behavioral outcomes in this study.

4.4. Proposed Model: The Relational-Cultural Trust Framework

Based on these findings, we propose the Relational-Cultural Trust Model (RCTM) as a framework for understanding health communication within AI/AN communities (Figure 2). In rural tribal settings, where public health infrastructure and specialty access are limited, clinician-patient relationships may serve as the primary conduit for evidence-based information. These findings highlight the potential importance of strengthening rural tribal health workforce capacity alongside digital communication strategies.

At the core of the RCTM are “Primary Trusted Messengers,” specifically healthcare workers and interpersonal networks, who provide the relational depth necessary to overcome historical institutional mistrust [5]. Surrounding this core are “Secondary Information Channels,” such as social media and web-based resources, which serve as high-volume access points but lack the inherent reliability required to drive behavior change on their own. The effectiveness of these secondary channels may depend on their alignment with the values and messaging of trusted interpersonal sources.

The RCTM is further supported by evidence from Community Health Representative (CHR) and community health worker models in Indigenous populations. CHRs, tribally embedded health workers, have demonstrated measurable improvements in chronic disease outcomes, screening uptake, and patient engagement through culturally grounded, relationship-based interventions [29]. These models emphasize that trust, cultural alignment, and relational continuity are central mechanisms through which health information leads to behavior change. Notably, the existing evidence base for CHR-led and culturally grounded interventions is heavily concentrated in Southwestern tribal communities, with relatively limited data from Midwestern and Eastern AI/AN populations [9]. This geographic gap further illustrates the importance of examining trust and communication pathways in settings such as rural Michigan.

By prioritizing this relational-cultural hierarchy, public health strategies may move beyond broad-based messaging toward approaches that are more responsive to community context and trust dynamics. While the RCTM is informed by observed patterns in the data, it should be interpreted as a conceptual and hypothesis-generating framework rather than a formally validated model. The model is intended to provide a structured lens through which to understand how trust, cultural congruence, and information pathways interact in AI/AN communities.

4.5. Limitations

While this study provides valuable insights into health communication within AI/AN communities, several limitations warrant consideration. First, the single-center design may restrict the generalizability of these findings to other geographically and socioculturally diverse tribal nations with differing healthcare infrastructures. Second, the use of convenience sampling among clinic patients introduces potential selection bias. Participants engaged in healthcare systems may be more likely to trust healthcare workers and possess higher baseline health literacy, potentially inflating the observed association between clinician-based information sources and positive health behaviors. This may limit the applicability of findings to individuals with lower healthcare access or engagement. Third, the reliance on self-reported data may be subject to recall or social desirability biases, potentially affecting the accuracy of reported vaccination histories and trust ratings. Fourth, the absence of measures assessing prior exposure to misinformation and baseline trust in information sources may introduce residual confounding, complicating interpretation of observed associations within this AI/AN sample. Finally, the study’s focus on immediate pre–post changes precludes conclusions regarding long-term knowledge retention or sustained behavioral shifts. Future multi-site, longitudinal research is needed to build upon these foundational findings and further explore the durability of information pathways in diverse AI/AN contexts.

5. Conclusions

This study highlights the critical role of trusted messengers and culturally congruent communication in addressing health disparities within AI/AN communities. Our findings reveal a distinct “usage-trust gap”: while social media serves as a primary information gateway, HCWs and personal networks remain the most credible drivers of health literacy and preventive behavior. The significant association between HCW-led communication and higher reported vaccination rates reinforces the necessity of relational accountability in public health strategy. Furthermore, while both video and infographic interventions proved effective for short-term knowledge gains, their impact was most pronounced when they aligned with the values of trusted interpersonal sources. These findings should be interpreted in light of the study’s single-center design and potential selection bias, which may limit generalizability. Broader multi-site and longitudinal studies are needed to confirm these patterns and evaluate their applicability across diverse Indigenous populations.

Ultimately, these results support the proposed RCTM, suggesting that effective health communication in AI/AN contexts must move beyond top-down messaging. For rural AI/AN communities, effective pandemic response strategies must extend beyond digital campaigns and invest in the stability of tribal health systems. Investment in rural clinician workforce development, community health representative programs, and culturally grounded messaging partnerships may yield more sustainable gains in vaccine uptake and public health preparedness than broad, centralized communication strategies alone.