Influence of Trust in Information Sources on Self-Rated Health Among Latino Day Laborers During the COVID-19 Pandemic

Abstract

This study examined the relationship between trust in COVID-19 information sources and self-rated health (SRH) among Latino Day Laborers (LDLs) and whether mental health mediated this association. Participants (N = 300) recruited at 18 job-seeking locations were interviewed in Spanish during November and December 2021. Validated scales were used to measure trust in formal (e.g., broadcast news, newspapers, and radio) and informal sources (e.g., friends, family, and social media) and mental health (depression, anxiety, and stress), with SRH measured with a single item. Mediation analysis was conducted using Hayes’ SPSS PROCESS macro. Higher trust in formal sources of information was related to lower SRH, but this relationship was not mediated by mental health. However, depression and anxiety were associated with a decrease in SRH. There were no significant direct or indirect effects between trust in informal sources and SRH. Depression and anxiety remained significant predictors of lower SRH. Further research is warranted on the mechanisms underlying these associations and the differential impact of information sources on vulnerable populations, such as LDLs, during health crises.

1. Introduction

The COVID-19 pandemic has influenced general subjective well-being, particularly among individuals negatively affected by the accompanying economic downturn. In the United States (US), individuals who report poor health are more likely to be members of low-income, ethnic minority populations [1]. Contributing to this health disparity are environmental, social, and health-related exposures, social determinants, and cultural-related stressors [1], which may have an independent as well as a synergistic influence on health. For instance, individuals who experienced housing insecurity during the COVID-19 pandemic were shown to have increased psychological distress and decreased self-rated health (SRH) [2].

Latinos are considered a population vulnerable to COVID-19 due to structural and environmental determinants, social conditions, and personal risk factors that have an impact on their well-being [3]. According to the Centers for Disease Control and Prevention COVID-19 demographics and data on trends, 1,182,801 deaths occurred from 1 January 2020 to 30 April 2024, and the percentage of deaths among Hispanics was 14.71% (174,019 individuals) [4].

Although there is variation in Latino vulnerability to COVID-19, some groups, such as Latino Day Laborers (LDLs), were disproportionately affected by the pandemic. LDLs are a subgroup of mainly recently immigrated Latino men who engage in unregulated and informal work and are especially vulnerable to environmental stressors and occupational hazards due to the harsh, unfavorable conditions they confront at work. Due to these circumstances, Latino immigrant workers, including LDLs, had increased exposure to COVID-19 and were more susceptible to its economic and health-related repercussions than other ethnic groups [5].

The high COVID-19 infection and mortality rates experienced by LDLs are also partly due to barriers they encounter in seeking healthcare. As an example, Latinos across the US are targeted by policy rulings that may have an impact on their willingness to use medical care [6]. In the context of the COVID-19 pandemic, LDLs experienced limited access to preventive healthcare and federal programs [3]. In relation to preventive healthcare, mistrust has been reported to be one of the key barriers to COVID-19 diagnostic testing among all Latinos, including LDLs [5,7]. Contributing to mistrust is the inadequate, unreliable, and inconsistent COVID-19 information that has been circulated to and among marginalized communities, which has influenced their seeking healthcare services [7] and could have had an impact on their overall health. In addition, misinformation from various social sources, including family members, social networks, and Spanish-language media and news channels, has been a barrier to COVID-19 vaccine knowledge and uptake among Latino communities [8].

2. Literature Review

SRH is a reliable and valid measure of an individual’s general health [9,10,11,12]. Assessment of SRH involves an individual’s perception of health in the psychological, social, and physical realms [1] and the subjective assessment of various factors, including physical functioning and social engagement [13]. SRH is commonly measured as a single item that asks an individual to rate his or her general health, e.g., “In general, would you say your health is excellent, very good, good, fair, or poor?” [10,13].

SRH has been used extensively in public health-related, sociological, and physical health-related research studies [11,12,13] and found to be predictive of mortality, morbidity, individuals’ health behaviors, and health status [11,12,13,14]. SRH is also associated with mental health, as individuals who report decreased SRH have increased odds of also reporting depression [15]. Studies have reported a strong correlation between poor SRH and premature mortality and the risk of developing chronic health outcomes, relative to an individual who reports excellent SRH [1,16]. SRH is an independent predictor of survival, particularly for individuals who report having very good or excellent SRH [13].

Trust in people and institutions is a measure of social capital [17] and degree of connectedness [18] that may have an impact on an individual’s perceived health and may ultimately affect health outcomes. According to the American Psychological Association, trust is an individual’s confidence and reliance on a person’s or a group’s dependability and is a vital aspect of interpersonal relationships [19]. Institutional trust is individuals’ belief that governmental institutions will implement appropriate policies that are aligned with the expected behaviors of individuals and entities [20]. Interpersonal trust is individuals’ expectations that other individuals will engage in mutual enhancement of well-being without causing any harm [20]. By extension, informational trust may be understood as confidence in the dependability and credibility of institutional and interpersonal information sources.

Trust, measured as trust in people, feeling of reciprocity, and feeling safe in the neighborhood, is a dimension of social capital and has been positively associated with good SRH and psychological well-being, which indicates that individuals with higher levels of trust report better health outcomes compared to individuals with lower levels [21]. Measuring trust is critical to understanding the health of Latinos, specifically LDLs, as LDLs tend to report lower levels of trusted social relationships; one study found that 36% of LDLs reported a lack of trusted friends, and 33% reported a lack of trusted coworkers [9]. This limited connectedness is likely to influence Latino SRH and overall health.

Trust in health information and health advice [22] was critical during the COVID-19 pandemic. This type of trust includes formal and informal health information sources, such as traditional media, local and national organizations, social media, and interpersonal communication [23,24,25]. Lack of trust in the accuracy and reliability of COVID-19 diagnostic testing results contributed to underserved Latino communities’ reluctance to participate in COVID-19 mitigation practices, such as vaccination [7]. In addition, mistrust in the health system was a barrier to healthcare utilization, and misinformation regarding COVID-19 was associated with an increased risk of exposure to COVID-19 among Mexican immigrants [26]. Latino immigrants’ lack of documentation and identification was a significant barrier to their COVID-19 testing, and distrust in the handling of their personal data was a concern in their participation in COVID-19 diagnostic testing [7]. Distrust in public institutions, in general, complicates efforts to assist individuals who are experiencing distress due to disasters [27], including the COVID-19 pandemic.

In addition, individuals’ attitudes and knowledge concerning the COVID-19 pandemic were influenced by broadcast television news [28]. Latinos in the US depend on news broadcast on television channels, with an estimated 89% of Latino immigrants able to access Spanish news sources to some extent, and these sources were often preferred by Latinos as a means to obtain information relating to the COVID-19 pandemic [28]. Latinos had higher levels of trust in the information disseminated in Spanish by broadcasters who were of similar ethnicity, and Spanish television news sources had an average viewer count of 712, 827 for Telemundo and 1104, 233 for Univision [28]. A comparison of news disseminated by television networks shows that the COVID-19 pandemic had been covered more thoroughly by Cable News Network (CNN), as it reported more preventive response strategies, such as social distancing and potential lockdowns, than Fox News network [29]. One study found that individuals who trusted formal sources of information demonstrated lower preventive behaviors and higher engagement in risky behaviors, possibly due to the network’s political stance [30]. Additionally, in the US, individuals who had greater trust in television sources had lower engagement in protective behaviors [25], and increased viewing of traditional media was associated with lower COVID-19 vaccination uptake [31]. Another study found that mental distress (i.e., feeling nervous, worried, depressed, and loss of interest) was not significantly associated with trust in television sources concerning COVID-19 information [32].

Moreover, social media has been shown to be one of the main sources of information regarding the COVID-19 pandemic. False health information regarding COVID-19 proliferated on social media and created turmoil and panic among individual consumers [33]. Exposure to COVID-19 news and information on social media has been reported to be associated with adverse mental health outcomes and poor SRH, with individuals who were exposed to frequent use of social media reporting higher levels of anxiety and depression [15] in the context of the COVID-19 pandemic. Further, the more that individuals trusted COVID-19 information from social media sources, the higher their levels of anxiety as compared to individuals who had less trust in social media, as many social media sources provided inaccurate and unreliable information regarding the COVID-19 pandemic [33]. Moreover, individuals who trusted social media for COVID-19 information reported higher odds of severe mental distress compared to those who did not trust social media [32].

In the context of the COVID-19 pandemic, social, mental, and physical health outcomes, including anxiety, depression, emotional distress, and obesity, increased among adults in the US [34]. Moreover, due to the lack of clear and sufficient COVID-19 information, poorer mental health among immigrants in the US has intensified and worsened [35]. Research provides evidence of the association between mental health, SRH, and trust in the context of the COVID-19 pandemic. For instance, one study explored trust in institutions (e.g., healthcare system, news media) and found that older adults who had a higher score on trust were also more likely to report a higher score on perceived health [36].

Latino immigrants have been susceptible to poor mental health due to the disproportionate economic burden they have confronted, including financial setbacks, such as loss of employment and housing insecurity, and reported experiencing mental health problems during the COVID-19 pandemic, including symptoms consistent with anxiety disorders and depression [37]. In addition, Latino immigrants reported that fear of contracting the COVID-19 virus and social isolation contributed to their mental health concerns [37]. Undocumented Latino immigrants were at an increased risk for the adverse health outcomes of COVID-19, including psychological distress due to the nature of their job as essential workers, salary reductions, and workplace shutdowns, resulting in financial insecurity [38]. In addition to these economic burdens, misinformation regarding COVID-19 resulted in overall mistrust, and individuals demonstrated having anxiety and stress, manifested by a fear of going out, and these mental health symptoms led to undesirable health behaviors [39].

Demographic characteristics have played an important role in the health impact of the COVID-19 pandemic and the information access and resources to mitigate it [13,40,41]. Several studies conducted with Latinos indicate that members of this population were differentially affected over the course of the COVID-19 pandemic [8,40,41]. For example, individuals with less than a high school degree were more likely to be employed in essential jobs with a higher risk of COVID exposure [42]. Similarly, low-income Latinos with unstable employment and limited access to healthcare faced a similar risk of infection [43,44]. The influence of age appears to have been bifurcated, as Latinos 20–60 years of age had the highest age-related infection rate, while those over 60 confronted higher mortality rates [45].

Access to COVID-19-related information followed similar demographic trends: Individuals with less education and lower health literacy had less access to relevant health information [43,46,47,48]. While younger Latinos were more likely to use social media, older adults had lower digital literacy and faced limited access to COVID-19-related health information [47]. Disparities in access to relevant information also extended to income [49] and years living in the United States, with lower-income Latinos and those with undocumented status lacking reliable access to the Internet and digital services [50]. Consistent with these perspectives, we hypothesize that demographic characteristics such as age, education, income, marital status, and time in the United States will moderate the influence of trust in information on SRH.

Establishing trust between community members and testing administrators was a crucial facilitator in COVID-19 diagnostic testing among Latinos in the US [7]. In a study on the influence of trust on compliance with COVID-19 contact tracers among various racial and ethnic groups, Latinos exhibited the least amount of trust, having the lowest scores for trust in healthcare professionals, contact tracers, and governmental health authorities [51], as influenced by the significant socioeconomic adversities they faced during the COVID-19 pandemic. In addition, in a study among US Latinos and immigrants from Latin America, respondents who consumed Spanish-language news were more inclined to trust journalists as compared to those who consumed primarily English-language media [24]. Further, trust in Spanish-language journalists led to a better evaluation of state and local officials’ COVID-19 response and was associated with an increased likelihood of supporting collective health efforts [24].

Contextual and social factors have been linked to poorer mental health outcomes in LDLs, including unstable housing, experiences of discrimination and acculturative stress, and the absence of a partner [3]. These contextual and social factors can increase vulnerability to disease and contribute to adverse mental and physical health outcomes [3]. Structural stress has been positively associated with symptoms of anxiety and depression, and financial strain has likewise been linked with worse mental health outcomes in LDLs [3,52].

Higher odds of poor or fair health among Latinos were associated with age, while higher levels of education and income were associated with lower odds of reporting poor or fair health [40]. Further, lower levels of acculturation, as measured by years of living in the US, were associated with worse SRH among Mexican immigrants [13]. Among Mexicans living in the U.S., the foreign-born individuals had lower odds of reporting poor or fair SRH compared to those who were U.S. born [41]. In addition, among Mexicans who reside in the US, regardless of birthplace, married individuals had 14% lower odds of reporting poor or fair health relative to the unmarried [41].

COVID-19 information disseminated during the pandemic was sometimes distorted to create disinformation, leading to distrust in both its source and content [7,8,33]. Distrust was felt widely across many sectors of the U.S. population, but as evidence indicates, it was felt more acutely in groups with greater exposure and fewer resources to prevent COVID, such as Latino day laborers [3,5,7,8,26]. The objective of this study was to assess the association between trust in information sources and SRH. As trust in information is critical not only for adopting prescribed practices but also for overall well-being, we hypothesize that self-rated health (SRH) would be differentially influenced by the source of information, with greater trust in information stemming from social sources contributing to higher SRH and greater trust in broadcast sources contributing to lower SRH [21,25,30,31]. As lack of trust can adversely affect both mental health and SRH, we also hypothesized that depression, anxiety, and stress would mediate the relationship between trusted information sources and SRH [9,34,39]. As evidence indicates that these relationships may vary by individual factors, we also hypothesized that demographic characteristics would also modify them [13,40,41]. Exploring the association between mental health, SRH, and information trust is particularly important for LDL, as understanding these relationships can inform the factors that lead to their overall well-being and the adoption of COVID-19 mitigation practices.

3. Materials and Methods

3.1. Background

A rapid needs assessment survey was conducted to characterize LDLs’ experience of the COVID-19 pandemic and to assess how stressors and protective factors experienced during the pandemic influenced the mental health of LDLs. The study was funded by UTHealth Houston and the National Institute of Minority Health and Health Disparities (R01MD012928-05) and was approved by the Committee for the Protection of Human Subjects of the University of Texas Health Science Center at Houston (HSC-SPH-18-0337).

3.2. Recruitment

Data were collected during November and December 2021. Participants were recruited in the Houston metropolitan area from day labor “corners” or locations where LDLs gather to seek work (i.e., parking lots of home improvement stores, convenience stores, gas stations, apartment complexes, public parks, and street intersections).

A corner sampling strategy was developed prior to recruitment. A list of 30 previously observed corners was stratified according to the number of LDLs observed. Corners with fewer than 15 LDLs observed were classified as “small.” Corners at which 15 to 29 LDLs were observed were classified as “medium.” Corners with 30 or more LDLs were classified as “large.” Corners deemed to be adjacent (within approximately three blocks of each other) were considered as a single location and classified according to the aggregate number of LDLs observed across the individual corners. Of the 30 locations, 17 were small, 7 were medium, and 6 were large.

An overall goal of 300 participants was set for the study. Based on prior observations, a goal of 80 participants each from small and medium locations and 140 from large locations was specified. Within each size strata, locations were randomly ordered using a random number generator. Corners within each stratum were visited in the randomly determined order.

Participants had to be 18 years of age or older, self-identify as Latino, be at the corner, seeking work, and have previously sought work at a corner. Trained interviewers explained the purpose of the study, determined LDL eligibility, and consented eligible candidates. Surveys were administered on iPads in Spanish. A total of 416 LDLs were observed and 314 were approached at 18 corners. Of these, 304 consented, and 300 surveys were completed.

3.3. Measures

Demographic measures included participants’ age, years in the US, income, years of schooling, marital status, and the number of days in the last month that work was found at the corner. Those who had worked at least one day were asked how much they earned on a typical day. Thirty-day income was computed by multiplying days worked by typical daily earnings. Marital status was coded as single, never married; married or living with a partner; or formerly married (divorced, widowed, or separated). To reduce the skewness in the distribution of the income variable, seven values of monthly income greater than USD 3000 (ranging from USD 3150 to USD 9600) dollars were set to USD 3000. For the analyses reported below, monthly income for each participant was divided by 100. Dummy variables were created for each category of marital status, “single”, used as the reference categories, respectively.

Mental health was measured using three previously validated scales. Depression was measured with the 7-item Center for Epidemiologic Studies Depression Scale (CES-D) [53]. A sample item is, “In the last week, how often would you say you didn’t feel like eating; your appetite was poor?” Responses were recorded on a 4-point scale with 0 = not at all (less than 1 day); 1 = a little (1–2 days); 2 = frequently (3–4 days); 4 = a lot (5–7 days). Cronbach’s α for the scale was 0.87. State anxiety was measured using a six-item survey [54]. A sample item is, “In general today, do you feel nervous?” Three items that reflected a positive state (e.g., “calm”) were reverse scored. Responses were recorded on a four-point scale, with 0 = not at all; 1 = somewhat; 2 = moderately so; 3 = very much so. Cronbach’s α for the scale was 0.73.

Stress was measured using a six-item version of the Perceived Stress Scale [55]. A sample item is, “In the past 30 days, how often have you been upset because of something that happened unexpectedly?” Responses were recorded on a 5-point scale, with 0 = never; 1 = almost never; 2 = sometimes; 3 = fairly often; 4 = very often. Cronbach’s α for the scale was 0.88. For each scale, the scale score was computed as the mean value of non-missing responses.

Trust in sources of information was measured using 15 items, developed for this study. The scale was informed by a study conducted by the University of Chicago National Opinion Research Center. Using the item, “How much do you trust the Coronavirus (COVID-19) information disseminated by the following sources of information?” Sources included television networks, conversations with family or friends or coworkers, and social media. Responses were recorded using a 4-point scale, with 0 = no trust. 1 = some trust; 2 = a great deal of trust; 3 = a lot of trust. Participants could refuse to answer any item or indicate if they did not know how much they trusted a given source or that a particular source was not applicable to them.

To validate the content of the sources of information scale, we conducted an exploratory factor analysis of the trust items using principal axis factoring extraction and oblimin rotation. Three factors (“Spanish and other networks”, “other media”, and “other sources”) were extracted, which accounted for 44%, 3.9%, and 3.8%, respectively, of the variance in the items. After reviewing the content of the factors and the basic hypotheses, the research team made a decision to retain and combine items into two trust scales. Trust in formal sources of information (formal trust) consisted of trust in broadcast media, including Telemundo, Univision, CNN, Fox, newspapers, and radio stations. Trust in informal sources of information (informal trust) consisted of trust in social network members and social media, including conversations with friends or coworkers, conversations with family, websites or online news, social media, and WhatsApp. Variables excluded due to poor factor loadings were trust in PBS, official press releases or conferences, medical professionals, and opinion polls. For the analyses presented below, response options of “a great deal” and “a lot” for the trust items were combined. Responses of “refused”, “don’t know”, and “not applicable” were treated as missing.

Reliability of the two recoded trust in information scales was established using Cronbach’s α. The reliability coefficient for the recoded formal trust items was α 0.88, and for informal trust items, it was α 0.81. In the initial computation of the formal trust reliability coefficient, a total of 127 cases had to be excluded, as data analysis in SPSS only accepts cases with non-missing responses for each item in a given scale (most missing cases were due to the recomputation of the Univision and Telemundo as separate variables, or because the participant did not use the particular media source, thus being coded as “not applicable”). To validate and check the stability of the formal trust scale, we used matrix expectation maximization with the full sample (N = 300), a procedure recommended by Weaver and Maxwell (2014) to handle missing data in the SPSS reliability analysis [56]. Using this imputation allowed us to replicate the formal trust reliability coefficient (α 0.87) using all cases (N = 300).

Self-rated health was not subjected to reliability analysis, as it is measured by a single item, “In general, would you say your health is…?” Responses were measured on a 5-point scale with 0 = poor; 1 = fair; 2 = good; 3 = very good; 4 = excellent [10,13]. This single item measure has been used extensively in research and it is widely considered both reliable and valid [9,10,11,12].

4. Data Analysis

Descriptive statistics were computed for the study measures using SPSS version 29.0. Means and standard deviations were computed for continuous variables, such as age and number of days worked. Frequencies were computed for marital status.

As noted, we were interested in assessing the association between each informational trust measure and SRH and whether any associations were mediated by the individual mental health measures. Our model of interest is presented in Figure 1. Mediation Model, based on the traditional mediation approach presented by Kenny [57] but updated for SPSS using Hayes’ PROCESS Macro [58]. Three steps were carried out for each of the six possible combinations of trust (formal and informal) and mental health (depression, anxiety, and stress).

In the first step, SRH was regressed on the given trust variable in a linear regression. Each of the demographic variables (age, marital status, years in the US, years of school, and 30-day income) was entered as a covariate. No mental health measure was entered at this stage. The coefficient for trust represented the total effect of trust on SRH, as shown by path c in Figure 1. Mediation Model. In the second step, the given mental health mediator was regressed on the given trust measure, controlling for the demographic variables. The coefficient for trust represented the effect of trust on the mental health mediator, corresponding to path a in Figure 1. Mediation Model. In the third step, SRH was regressed on both the trust measure and the mental health mediator, controlling for demographic measures. The coefficient for the mental health variable gave the effect of the variable controlling for trust, represented by path b in Figure 1. Mediation Model. The coefficient for trust gave the direct effect of trust on SRH, represented by path c’ in Figure 1. Mediation Model.

The indirect effect of trust on SRH via mental health was equal to the product of the coefficients for paths a and b in Figure 1. Mediation Model. A significant indirect effect indicates a significant interaction effect. The total effect of trust on SRH (path c) described above is equal to the sum of the indirect and direct effects.

5. Results

The overall sampling goal was met, with a total of 300 LDLs surveyed. Strata-specific goals were also met, with 80 participants recruited from six small corners, 80 recruited from six medium corners, and 140 recruited from six large corners. A total of 271 of 300 (90.3%) participants had complete data for each of the study variables and formed the sample for this study. Characteristics of the sample are shown in

Table 1. Marital Status and Other Demographic Characteristics (N = 271).

Variable	N		%
Marital status
Single, never married	128		47.2%
Married, living with a partner	102		37.6%
Formerly married	41		15.1%
Variable	Possible Range	Observed Range	M	SD
Age		18.4–76.1	45.3	12.1
Years in the US		1 month–59.3 years	13.7	12.2
Years of school		0–22.0	7.7	4.4
30-day income		0–2500.00	906.79	691.10
Trust
Formal	0–2.0	0–2.0	1.2	0.6
Informal	0–2.0	0–2.0	1.1	0.5
Mental health
Depression	0–3.0	0–3.0	0.9	0.7
Anxiety	0–3.0	0–2.7	0.8	0.6
Stress	0–4.0	0–4.0	1.6	0.9
Self-rated health	0–4.0	0–4.0	1.9	1.3

. On average, participants were in their mid-40s, had been in the US for approximately 14 years, and had completed approximately eight years of schooling. On average, participants had earned approximately USD 900 in the 30 days prior to their survey. Nearly one-half had never been married, and nearly two-fifths were married or living with a partner.

Mean values of the trust measures indicated an average response of “some trust” in the items. The mean score for depression indicated an average frequency of “a little”, based on the response options reported above. The mean score for anxiety indicated an average response of “somewhat” to the items. The mean stress score indicated an average response between “almost never” and “sometimes.” The mean SRH score indicated an average level of SRH of “good.” There were no significant differences in these characteristics between those with complete data and those without.

Frequencies of the individual formal and informal trust items are shown in

Table 2. Trust in Sources of Information Items.

Trust Variable	No	Some	A Great Deal/A Lot of	Total Valid Cases	Not Applicable	Don’t Know	Refused	Total
CNN/CNN in Spanish	33	100	70	203	52	15	1	271
Valid %	16.3%	49.3%	34.5%	100.1%	-	-	-	-
Total %	12.2%	36.9%	25.8%	-	19.2%	5.5%	0.4%	100.0%
Fox News/Fox	42	85	65	192	61	18	0	271
News in Spanish
Valid %	21.9%	44.3%	33.9%	100.1%	-	-	-	-
Total %	15.5%	31.4%	24.0%	-	22.5%	6.6%	0.0%	100.0%
Telemundo	34	94	86	214	12	4	2	232
Valid %	15.9%	43.9%	40.2%	100.0%	-	-	-	-
Total %	14.7%	40.5%	37.1%	-	5.2%	1.7%	0.9%	100.1%
Univision	26	95	100	221	10	1	0	232
Valid %	11.8%	43.0%	45.2%	100.0%	-	-	-	-
Total %	11.2%	40.9%	43.1%	-	4.3%	0.4%	0.0%	99.9%
Daily or weekly newspapers	56	103	58	217	41	13	0	271
Valid %	25.8%	47.5%	26.7%	100.0%	-	-	-	-
Total %	20.7%	38.0%	21.4%	-	15.1%	4.8%	0.0%	100.0%
Radio stations	53	120	73	246	22	3	0	271
Valid %	21.5%	48.8%	29.7%	100.0%	-	-	-	-
Total %	19.6%	44.3%	26.9%	-	8.1%	1.1%	0.0%	100.0%
Conversations with family	33	97	138	268	2	1	0	271
Valid %	12.3%	36.2%	51.5%	100.0%	-	-	-	-
Total %	12.2%	35.8%	50.9%	-	0.7%	0.4%	0.0%	100.0%
Conversations with friends or coworkers	59	116	95	270	1	0	0	271
Valid %	21.9%	43.0%	35.2%	100.1%	-	-	-	-
Total %	21.8%	42.8%	35.1%	-	0.4%	0.0%	0.0%	100.1%
Websites or online news sites	56	109	61	226	34	10	1	271
Valid %	24.8%	48.2%	27.0%	100.0%	-	-	-	-
Total %	20.7%	40.2%	22.5%	-	12.5%	3.7%	0.4%	100.0%
Social media	81	105	57	243	21	7	0	271
Valid %	33.3%	43.2%	23.5%	100.0%	-	-	-	-
Total %	29.9%	38.7%	21.0%	-	7.7%	2.6%	0.0%	99.9%
WhatsApp groups	102	100	37	239	27	5	0	271
Valid %	42.7%	41.8%	15.5%	100.0%	-	-	-	-
Total %	37.6%	36.9%	13.7%	-	10.0%	1.8%	0.0%	100.0%

. For each item, the frequency of each possible response is presented in the first row of the table. The second row presents percentages for each valid response, i.e., excluding responses of “refused,” “don’t know,” and “not applicable.” The third row presents percentages for all possible responses.

Table 3. Mediation Analysis.

Model	Path (Figure 1)	B	95% Confidence Interval
Model 1
Total effect Formal trust → self-rated health	c (indirect + direct)	−0.327	(−0.587, −0.068)
Formal trust → depression	A	0.106	(−0.042, 0.255)
Depression → self-rated health	B	−0.383	(−0.591, −0.176)
Indirect effect	a * b	−0.041	(−0.113, 0.025)
Direct effect	c’	−0.287	(−0.541, −0.032)
Model 2
Total effect Formal trust → self-rated health	c (indirect+ direct)	−0.327	(−0.587, −0.068)
Formal trust → anxiety	A	−0.046	(−0.167, 0.074)
Anxiety → self-rated health	B	−0.487	(−0.744, −0.231)
Indirect effect	a * b	0.023	(−0.035, 0.093)
Direct effect	c’	−0.350	(−0.604, −0.096)
Model 3
Total effect Formal trust → self-rated health	c (indirect + direct)	−0.327	(−0.587, −0.068)
Formal trust → stress	A	0.040	(−0.152, 0.233)
Stress → self-rated health	B	−0.156	(−0.0319, 0.007)
Indirect effect	a * b	−0.006	(−0.051, 0.029)
Direct effect	c’	−0.321	(−0.580, −0.063)
Model 4
Total effect Informal trust → self-rated health	c (indirect + direct)	0.004	(−0.277, 0.284)
Informal trust → depression	A	0.068	(−0.091, 0.227)
Depression → self-rated health	B	−0.405	(−0.641, −0.196)
Indirect effect	a * b	−0.028	(−0.104, 0.041)
Direct effect	c’	0.031	(−0.242, 0.305)
Model 5
Total effect Informal trust → self-rated health	c (indirect + direct)	0.004	(−0.277, 0.284)
Informal trust → anxiety	A	−0.025	(−0.153, 0.104)
Anxiety → self-rated health	B	−0.471	(−0.730, −0.212)
Indirect effect	a * b	0.012	(−0.052, 0.083)
Direct effect	c’	−0.008	(−0.282, 0.267)
Model 6
Total effect Informal trust → self-rated health	c (indirect + direct)	0.004	(−0.277, 0.284)
Informal trust → stress	A	−0.026	(−0.232, 0.180)
Stress → self-rated health	B	−0.161	(−0.326, 0.003)
Indirect effect	a * b	0.004	(−0.036, 0.046)
Direct effect	c’	0.000	(−0.280. 0.279)

Note. B = Unstandardized regression coefficient.

presents the results of the mediation analyses with unstandardized regression coefficients and 95% confidence intervals. As shown in the table, for each of the models of formal trust and SRH (Models 1 through 3), there was a significant and inverse total effect of formal trust on SRH (corresponding to path c in Figure 1. Mediation Model). On average, the greater one’s trust in formal sources of information, the lower one’s SRH. In considering depression as a potential mediator of this association (Model 1), there was a significant and inverse association between depression and SRH (path b). A greater level of depression was associated with lower SRH. There was no significant association, however, between formal trust and depression (path a). Consequently, there was no significant indirect effect of formal trust on SRH through an association with depression (a * b).

A similar result was found when considering anxiety as a mediator of the association between formal trust and SRH (Model 2). Although there was an association between increased anxiety and lower SRH, there was no association between formal trust and anxiety, and no significant indirect effect was observed. A significant direct effect was observed. As with depression, the association of formal trust with lower SRH was not mediated by anxiety.

In Model 3, formal trust was not significantly associated with stress. Greater stress was associated with lower SRH, but, in contrast to depression and anxiety, this association was not significant at the p < 0.05 level. There was no significant indirect effect and thus no mediation effect of stress on the association between formal trust and lower SRH. There was a significant direct effect of formal trust on SRH.

In terms of the study objectives, a significant association was found between increased trust in formal sources of information and decreased SRH. Associations were found between increased depression and anxiety and lower SRH. Thus, while there were associations between formal trust and SRH and between anxiety and depression and SRH, these associations appear to be independent of each other, and there was no mediated effect of formal trust on SRH via mental health.

As shown in Models 4 through 6 in Table 3, there were no significant total or direct effects of informal trust on SRH. Depression and anxiety were inversely associated with SRH. Greater stress was associated with lower SRH, but this association was not significant at the p < 0.05 level. Informal trust was not associated with any of the measures of mental health, and there were no significant indirect effects.

It must be noted that the demographic factors used as covariates in mediational analyses did not have a consistent influence on the tested relationships, except in the case of mental health. In Model 1 (formal trust and depression), having been formerly married compared to having never been married was associated with greater depression (β = 0.306; 95% CI = 0.036, 0.576). In Model 2 (formal trust and anxiety), greater income was associated with decreased anxiety (β = −0.012; 95% CI = −0.022, −0.002). In Model 4 (informal trust and depression), having been formerly married compared to having never been married was associated with greater depression (β = 0.323; 95% CI = 0.053, 0.593). In Model 5 (informal trust and anxiety), greater income was associated with decreased anxiety (β = −0.012, 95% CI = −0.022, −0.002).

Again, with regard to the study objectives, there was no significant association between trust in informal sources of association and SRH, nor were there significant associations between informal trust and the mental health measures. As with formal trust, there were significant associations between depression and anxiety and decreased SRH.

6. Discussion

The research objective of this study was to assess the association between trust in information sources and SRH and whether any association would be mediated by mental health. We hypothesized that trust would influence SRH, with trust in informal sources of information linked to higher SRH and trust in formal sources linked to lower SRH; we also expected that anxiety, depression, and stress would mediate this relationship. However, our results showed that these mental health factors did not mediate the relationship between trust in information sources and SRH. Instead, we found two independent pathways. We found one pathway between greater trust in formal sources of information and lower SRH, and another pathway between greater anxiety and depression and lower SRH. As these pathways are independent, they may have different relationships with SRH and different explanations. As shown in the Results, the demographic factors used as covariates in mediational analyses did not have a consistent influence on the relationships tested, with the exception of mental health. Having been formerly married compared to having never been married was associated with greater depression (Model 1). Greater income was associated with decreased anxiety (Model 2). Having been formerly married compared to having never been married was associated with greater depression (Model 4). Greater income was associated with decreased anxiety (Model 5). These findings may contribute to understanding factors which may impact mental health among LDLs and thus contribute to SRH. As the focus of this study is on the influence of information sources on SRH, the important relationship between mental health and SRH among LDLs would have to be explored in a separate report. Below, we elaborate on the role of information sources and their relationship with SRH.

Our results suggest that the information received during the COVID-19 pandemic led to lower SRH. It is possible that the often-negative content of the information about COVID-19 portrayed in broadcast media channels, including Univision and Telemundo, led to a lower appraisal of personal well-being. Indeed, previous evidence indicates that individuals who seek general health information tend to report better SRH, while individuals who seek disease-specific information report worse SRH [59]. Much of the information disseminated through these media channels highlighted the ill effects of COVID-19 and the probability of disease and even death. Although the generalized negative tone of the COVID-19 information was not exclusive to these media sources, their assumed prestige and social influence among Spanish-speaking viewers may have played a stronger role in their personal appraisal of wellbeing, as Univision and Telemundo were the two media sources the study participants trusted the most. This pattern of findings also may contribute to the larger discussion of the ways that the nature, tone, and quality of information have the capacity to influence not only knowledge and awareness of health problems but also the perceived well-being of individuals.

The lack of influence of informal sources of information on either mental health or SRH must be noted. These sources include family, friends, coworkers, and various types of social media. This is a striking finding, considering that information from one’s social network is a valuable component of social capital. However, it is possible that the great variation in trust reported by the different social sources may have confounded the results. On the one hand, 51% of the respondents reported that the information provided by their family was trusted greatly, while the information provided by social media and WhatsApp groups was the least trusted. In order to test the potential beneficial effect of social sources of information on SRH, the role of the family would have to be explored separately, an approach consistent with the very prominent role of the family in Latino culture.

6.1. Strengths

Our study has several strengths, including our rigorous sampling and recruitment process, as well as our timely data collection, which was captured at a time when COVID-19 variants and subvariants were of utmost concern and precautions were still encouraged, providing timely insights into the information environment and health perceptions of LDLs.

6.2. Limitations

However, the study also had limitations. There were significant differences in demographic and mental health measures between cases used to compute the initial Cronbach’s reliability of the formal and informal trust measures. Bias may have existed to the extent reliability scores for those excluded would have differed from those that were used. But as noted, we were able to compute reliability using Weaver and Maxwell’s (2014) method to handle missing values [56]. In addition, the relatively small sample size may have limited the power of the statistical analyses. As noted, the purpose of the original study was to conduct a rapid needs assessment regarding COVID-19 among local LDLs. A sample size of 300 was determined sufficient for the analyses for which the study proposed. The mediation analyses presented herein were conducted as secondary analyses, with the fixed sample. In addition, some cases were lost due to incomplete data, resulting in a final sample size of 271. Given the observed effect sizes for some of the paths in our models, our analyses were likely underpowered. Future studies should be conducted with larger samples to assess these initial findings.

Another limitation was the influence of the media market to which LDLs are exposed. Locally, LDLs are exposed to national media channels like Univision and Telemundo, which are the most prominent Spanish-speaking channels in the US. As is the case with most broadcast media, these channels also cover state and local news, including locally relevant information about COVID-19. While most COVID-19-related health messages were similar across these traditional channels, the local news may have varied across locations, possibly having an impact on trust in formal sources across locations. We assume this may be a potential confounder that limits the generalizability of our study but do not know of any previous study that has tested this relationship among Latinos. The relative impact of local versus national news would need to be tested in a separate study.

In addition, since this study employed a cross-sectional design, it was conducted at a specific point in time and cannot establish causal inferences [60]. The survey represents attitudes, beliefs and practices at a single point in time which may have changed with the evolution of the pandemic. Although participants were randomly selected according to their corner location (randomization was conducted at the corner cluster level), it is possible that the results are more representative of the local population and of other corners with similar working patterns and conditions as Houston. Therefore, it is not clear that the results are representative of LDLs in the United States. We also acknowledge that some of our measures of trust may have been appropriate for local LDLs even when they were based on previously validated studies. Overall, the results are based on a carefully designed and implemented survey, and the obtained results are a valid representation of the influence of the pandemic on the trust of media sources and perceived health among these workers.

7. Conclusions

The results from our current study contribute to the literature and add to existing knowledge by demonstrating that formal trust, depression, and anxiety are associated with decreased SRH. However, these effects of trust and mental health on SRH are independent of each other. Individuals’ trust in formal and informal sources merits further exploration, as such sources can provide different types of information that lead to health behaviors and personal assessments that may ultimately influence perceived well-being. Building on these findings, communicators might tailor COVID-19 and future health messaging to be transparent by providing credible, evidence-based health information and delivering health messages with empathy and cultural relevance, fostering trust while minimizing potential negative influences on well-being. Although our focus was on the trust in information sources, future research should also examine how the frequency and specific content of news exposure affect trust in information and its effect on SRH. Our findings on formal sources of information, however, may also be applicable to other vulnerable populations with low levels of trust in the media channels that they are exposed to, and the impact on their well-being may extend to issues beyond the COVID-19 pandemic.