At-Risk Drinking and Workplace Conditions among Latino Day Laborers

Abstract Background : Latino Day Laborers (LDL) face a variety of factors which have been associated with at-risk drinking. The objective of this study was to assess the association of at-risk drinking with measures of work site conditions. Methods : Data from surveys conducted with 307 LDL in Houston, TX in 2015 were analyzed. Sociodemographic information and measures of exposures to hazardous products at the worksite, adverse working conditions, and work stressors were collected. Measures of positive working climate at the jobsite and a climate that promoted jobsite safety were also included. Participants were administered the Hazardous Use items from the Alcohol Use Disorders Identification Test of Consumption (AUDIT-C). Participants were classified as low-risk or at-risk drinkers based on AUDIT-C score. Logistic regression models were run to assess the associations of the sociodemographic and worksite related variables with drinking status. Results : One-hundred-five (34.2%) participants were classified as at-risk drinkers. At-risk drinking was associated with past-month income, being formerly married (compared to having never married), and lack of housing. At-risk drinking was also associated with measures of a positive working climate and a climate that promoted jobsite safety. Conclusions : One-third of our participants were classified as at-risk drinkers. At-risk drinking was associated with stressors in the form of lack of housing and no longer having a spouse but was also associated with increased income and with positive workplace factors. At-risk drinking was thus a function of both stressors and positive factors, including a positive work site.


Introduction
Day laborers are individuals who gather at street corners, retail parking lots (often home improvement stores), and other locations where they approach or are sought by potential employers.Valenzuela et al., (2006) estimated that there were approximately 117,600 persons in the United States who were seeking or performing day labor on a typical day.Of the 2660 day laborers interviewed in their nationwide sample, the majority were male immigrants.Most had come from Mexico or Central America, were undocumented, spoke primarily Spanish, and had limited formal education, and their earnings from day labor were modest.Subsequent studies on Latino day laborers (LDL) have shown similar results (Bacio et al., 2014;Duke et al., 2010;Nelson et al., 2012;Organista et al., 2019).
Many LDLs come to the United States alone, with the goal of earning money to remit to family members in their country of origin.(Nelson et al., 2012;Worby et al., 2014).Once in the United States, LDL often shares living quarters with other workers, and many experience periods of homelessness.The condition of LDL in the United States has been characterized in terms of "structural vulnerability" (Worby et al., 2014).This vulnerability encompasses distal factors, such as immigration and work authorization policies, and more immediate influences, such as lack of income or housing.

Drinking patterns and background factors
Overall, Hispanics are more likely to be abstinent than are non-Hispanic Whites.Those who do drink, however, are more likely to consume higher amounts of alcohol on the occasions that they do drink.Specifically, they are more likely to engage in binge drinking, defined as having five or more drinks per occasion for males and four or more drinks per occasion for females (National Institute on Alcohol Abuse & Alcoholism, 2021).Unhealthy use of alcohol among Latinos has been associated with younger age, being single or separated, being of Mexican descent, and time in the United States (Caetano et al., 2008;Sanchez, 2015;Vaeth et al., 2009).
Some Latino males may be socialized in a culture that encourages them to conceal emotions and, when under duress, to turn to more avoidant coping strategies, such as drinking (Abreu et al., 2000).Cultural norms also may be more accepting of drinking to intoxication, beginning in youth (Kulis et al., 2008).In addition to dealing with cultural factors, Latino immigrants face challenges related to their immigration and racial/ethnic status, such as language barriers, lack of documentation, and isolation from family, which may increase the risk for unhealthy use (Negi, 2011).
In their qualitative interviews with LDL with unhealthy levels of drinking, as measured by the three-item Alcohol Use Disorders Identification Test of Consumption (AUDIT-C), Ornelas et al. (2015) found that episodes of binge drinking were usually seen on the weekends.Social services providers interviewed as part of the study referred to such use as "fiesta drinking." Such binge drinking was seen as acceptable by many of the participants.The authors added, however, that drinking was seen as a means of coping with feelings of vulnerability and isolation.Worby and Organista (2013) noted a "money to burn" phenomenon among the day laborers whom they interviewed.Workers were often paid on Saturdays in their home countries, and having cash facilitated drinking.They also found that some laborers saw being offered one or two beers by an employer as an appropriate reward for physical labor.In addition, laborers reported instances of employers promoting after-work drinking with them.Ornelas et al. (2016) found that 18.3% of 104 LDL surveyed had never drunk alcohol.Nevertheless, based on scores on the 10-item version of the AUDIT (which includes measures of the consequences of alcohol consumption), 65.4% of participants had scores indicative of hazardous drinking, and 49.0% reported consuming five or more drinks on a single occasion on a monthly basis during the previous year.In keeping with earlier studies, the mean AUDIT score was associated with a lack of stable housing and higher levels of depression.In another quantitative study of LDL, 126 of 181 (69.6%) drank at unhealthy levels, as measured by the AUDIT (Ornelas et al., 2019).
As noted above, cultural, sociodemographic, economic, and emotional factors may influence drinking behavior.An additional set of immediate factors relates to LDL's work settings.LDL are engaged primarily in occupations such as construction, landscaping, and moving, with construction work as one of the most dangerous in terms of non-fatal and fatal injuries (Walter et al., 2002).Using data from the Medical Expenditure Panel Survey, Dong et al. (2010) found that Latino construction workers were more likely than white non-Latino workers to suffer non-fatal work-related injuries.In a study of occupational injuries in an urban trauma center, Latino workers suffered a disproportionate number of injuries and were more likely to be injured by machinery and hand tools (Forst et al., 2010).Notably, LDL often engages in a succession of different jobs, and many lack adequate training.Further, when the work is dangerous, protective equipment may not always be provided, and many LDLs may not be given rest breaks (Burgel et al., 2015).In this context of inconsistent and often dangerous work, alcohol use may further endanger the safety and health of these workers.

Work conditions
Hazardous work exposures represent another factor potentially related to heavy drinking.Admittedly, the relationship between work hazards and drinking may be equivocal, as the stress created by exposure to dangerous conditions may precede or follow from drinking behavior.Grzywacz et al. (2012) study of immigrant Latino residential construction workers assessed exposures to perceived hazards, such as working with potentially dangerous machinery or equipment.The mean exposures score was at the midpoint of the possible range of 6 to 24.In a related study of work exposures, Ochsner et al. (2008) surveyed 117 LDL in New Jersey, where the majority of those surveyed expressed "great concern" regarding getting sick from breathing dust and getting sick from exposure to chemicals.Pransky et al. (2002) studied 427 nonagricultural immigrant Latino workers.Over one-quarter of participants reported exposures to 10 or more potentially harmful substances, e.g., pesticide, bleach, or physical risks, e.g., slippery floors.
Although these studies are valuable in documenting workplace conditions faced by LDL, they focus primarily on the association of these factors with workplace injuries.Fewer studies have assessed the association of workplace conditions with drinking.Martin et al. (1996) developed a conceptual model of employee problem drinking.According to their model, drinking is a reaction to stressful or nonrewarding work and a social context that includes work-based drinking networks and norms that support drinking.Conversely, drinking may be reduced by the presence of supportive relationships with others at work, including supervisors.Their analysis of data from the National Employee Survey found that increased job pressures increased the odds of experiencing a measure of job-based escapist drinking.Drinking with coworkers was also associated with escapist drinking, while support from a supervisor was associated with decreased odds.Bacio et al. (2014) studied 89 LDL in Los Angeles and found that participants had an average score on the AUDIT indicative of non-hazardous drinking, and based on a structural equation model, there was no direct link between day labor stress and alcohol use.In their study of Latino migrant day laborers in the San Francisco Bay area, Organista et al. (2019) found that a measure of working conditions based on time worked and income in the previous week was inversely associated with measures of depression and desesperación (a feeling of frustration and disappointment that arises from not progressing economically).Lower levels of these measures were, in turn, associated with lower scores on the AUDIT.Negi et al. (2020) conducted qualitative interviews with 25 LDL in Baltimore.They assessed workplace victimization in terms of wage theft, abandonment at the worksite, poor working conditions (e.g., working long hours without breaks), and verbal abuse.Street-level victimization was defined as physical abuse, robbery, or assault.Approximately one-third (32.0%) of their participants had experienced victimization, primarily workplace-related.The mean score on the AUDIT was 1.6 times higher among those who reported victimization.
In summary, LDL are workers who are vulnerable to hazardous and adverse working conditions as well as problematic drinking.Research has documented the association between workplace-related factors and problem drinking as well as the protective effects of supportive relationships.In this study, we seek to contribute to this literature by presenting a systematic assessment of the association of problem drinking with measures of exposures to hazardous workplace materials and items, adverse workplace conditions, and workplace stressors.We also examine the association of at-risk drinking with measures of a positively perceived working climate and a positively perceived workplace safety climate.

Methods
Data were collected as part of a project to develop a pilot community-based program to reduce the risk for workplace injuries among LDL, funded by the National Institute for Minority Health and Health Disparities.The study was approved by our Institutional Review Board.There were no conflicts of interest with respect to this research.

Study location and sample
The study was conducted at locations within the Houston metropolitan area at which Latino men sought work.These locations included the parking lots of home improvement and convenience stores, areas adjacent to apartment complexes, and street corners.Locations were referred to collectively as "corners."The identification of potential study locations was based on repeated observations conducted by a former day laborer familiar with such locations and other members of the study field team.Observations were conducted between November 2013 and July 2014, during which a total of 44 day-labor corners were identified.Corners were classified according to the number of LDL observed.Small corners were defined as those with 20 or fewer LDLs.Medium corners were defined as those with between 21 and 44 LDL.Large corners were defined as those with 45 or more LDL.A sample size of 300 participants was deemed adequate to achieve the study goal of conducting a needs assessment to determine the risk and protective factors related to workplace injuries among LDL.To ensure a representative sample of hiring locations in the study area, we set quotas of 60 participants from small corners, 150 from medium corners, and 90 from large corners.A stratified random sampling procedure was used to determine the order in which corners would be visited, with corner as the selection unit and the worker as the unit of recruitment.

Participant selection
Study participants were LDL who were recruited from Houston corners.Participant eligibility criteria included (a) being 21 years of age or older, (b) self-identifying as Hispanic or Latino, and (c) presently looking for work on the corners.Trained Latino lay health workers approached potential participants, explained the purpose of the study, and asked whether they wanted to participate.Individuals were told they could pause the survey if they had an offer of employment and resume the survey later.Interested LDL were then presented with the study's informed consent and asked to give verbal consent.All eligible individuals who agreed to participate were included in the study until the sample quota for each corner was met.

Interview procedures
Surveys were conducted at the corners by trained bilingual interviewers.A Spanish version of the survey was programmed in Qualtrics and administered using iPads.The survey lasted between 45 and 60 min, and the project coordinator was on site to oversee procedures and address any issues.All participants were given the option to discontinue the interview or complete it at a later time if they needed to participate in daily hiring/labor activities.Interviews were conducted away from other groupings of laborers to ensure privacy, and participants could refuse to answer any question.Upon survey completion, participants were compensated with a $20 supermarket gift card and received a copy of their informed consent and a list of community resources that included our program's contact information.
The corner survey was completed between mid-November 2014 and the end of February 2015.A total of 890 Latino day laborers were approached by the field team and informed of the study.Of these, 386 (43.4%) expressed interest in the study.A total of 331 of the 386 LDL completed the survey.The remaining 55 LDLs were not surveyed, as sampling quotas had been exceeded.A total of 76 surveys were conducted at small corners, given a quota of 60; 152 surveys were conducted at medium corners, given a quota of 150; and 103 surveys were conducted at large corners, given a quota of 90.
Of the 504 who declined to participate, a reason for refusal was available from 283 (56.2%).The most common reasons were currently "looking for work" or "waiting for a boss" (53.4%), "not having time at the moment" (19.4%), and "not being interested" (15.5%).

Measures
Sociodemographic and background characteristics included date of birth, with age computed as of the time of the interview; time in the United States; time seeking work at the corners; and years of school.Frequency of work was measured by the question, "In the past month, how many days did you work in a typical week?"Earnings were measured by asking, "In the past month, how much did you (more or less) get paid per workday (in dollars)?"Monthly earnings were computed by multiplying the number of days worked per week by daily earnings by four (weeks).
Country of origin was coded as United States (including Puerto Rico); Mexico; Central America (Honduras, El Salvador, Nicaragua, or Guatemala); Cuba; or Other.Marital status was coded as single (never married), married or living with a partner, or formerly married.Language generally spoken was coded as no English or some English.Current living arrangements were originally coded as a house, apartment, shelter, or another place.Those who were living in a house or apartment were classified as housed and those who were living in a shelter, as unhoused.Those who reported that they were living in another place were asked to specify their living arrangements.These respondents were classified as housed or unhoused based on their responses as described in the Results sections.Self-rated health was coded as "bad/just okay or so-so, "good, " or "very good/excellent." Participants also were asked whether they had suffered a work-related injury or illness in the last year severe enough to cause them to miss work or feel that they should not work or serious enough to seek medical attention.
Refusal to answer an item was set to missing.Age, time in the United States, time on the corners, and earnings were set to missing for participants who could not recall the requested information.Five workplace measures were developed, as described below.

Exposures to hazardous products at the work site
Exposures to hazardous products were assessed with 10 researcher-developed items that used the stem, "In the last 12 months, how often have you worked with …?" A sample item was "asbestos." Responses were recorded using a 4-point scale (1 = never, 2 = sometimes, 3 = many times, 4 = all the time).In addition to "refused" (which was an option for all items), other response options were "do not know" and "does not apply." A higher score on this measure indicated greater exposures to hazardous products or equipment.For the final study sample described below, Cronbach's α for the scale was .86.

Adverse work conditions
Adverse work conditions were assessed with 15 researcher-developed items that used the stem, "How often do your work conditions include …?" A sample item was "slippery or wet floors." Responses were recorded using a 4-point scale (1 = never, 2 = sometimes, 3 = many times, 4 = all the time).Higher scores indicated greater experience with adverse work conditions (α = 0.87).

Work stressors
Work stressors consisted of seven researcher-developed work-related items chosen from a measure of critical events, and items used the stem, "Please tell me how often in the last 12 months …" A sample item was, "After finishing your work, you were not returned to the place where you were picked up." Responses were recorded with a 3-point scale (1 = never; 2 = one or a few times (1-3), 3 = Very often (4+).Other response options included "does not apply" and "do not know/remember." Higher scores indicated greater experience with these events (α = 0.72).These measures are analogous to the workplace victimization items used by Negi et al. (2020), and we retained their label of "work stressors" for this study.

Positive working climate
Positive working climate was measured by 11 items developed from focus group sessions with workers and others familiar with the day labor community.Items used the stem, "In your last full day at work as a day laborer, would you say …?" A sample item was, "You were paid what was promised/agreed upon." Responses were recorded using a 4-point scale (1 = strongly disagree, 2 = disagree, 3 = agree, 4 = strongly agree).Options included "do not know." A higher score on this measure indicated a greater positive working climate (α = 0.82).

Positive safety climate
Positive safety climate was adapted from the 6-item scale used in Fugas et al. (2012).The current study used the stem, "Thinking about the last day labor job you worked in for a full day, would you say that your boss …?" A sample item was, "Considered whether you knew how to do the job safely?"Responses were recorded as 0 = no and 1 = yes.Other response options included 5 = do not know/remember and 6 = not applicable.Higher scores on this scale reflected a greater positive safety climate (α = 0.85).
Responses of "refused, " "do not know, " and "not applicable" were set to missing.For each scale, scores were computed as the mean response to the non-missing items in the scale.A complete listing of items contained in the five scales is presented in the Supplemental Tables.
Alcohol use was measured by the three items in the AUDIT-C (Bush et al., 1998).The three items constitute the Hazardous Use portion of the 10-item AUDIT, which contains additional items to measure the consequences of use (Babor et al., 2001).The interviewer prefaced the items by stating, "Now I am going to ask you some questions about your use of alcoholic beverages during this past year." Participants were first asked, "How often do you have a drink that contains alcohol?" Response options were "never, "monthly or less, " "2 to 4 times a month, " "2 to 3 times a week, " and "4 or more times a week." Those who reported drinking were then asked, "How many drinks that contain alcohol do you have on a typical day when you are drinking?"Response options were 1 or 2; 3 or 4; 5 or 6; 7, 8, or 9; and 10 or more.These participants were next asked, "How often do you have five or more drinks on one occasion?"Response options were "never, " "less than monthly, " "monthly, " "weekly, " and "daily or almost daily." The use of five or more drinks for the third item was a modification to the original item, which specifies six or more drinks for male respondents.National Institute on Alcohol Abuse and Alcoholism (2021) defines binge drinking as consumption sufficient to bring an individual's blood alcohol concentration to .08.For males, this corresponds to consuming five or more drinks in a two-hour period.
Responses for each item were scored from 0 to 4, and a total score was computed as the sum of the three items, with potential scores as ranging from 0 to 12. AUDIT-C Hazardous Use scores under 4 are considered indicative of low-risk drinking.For the present study, the score was dichotomized as low-risk (0-3) and at-risk (4-12; Bush et al., 1998).
Participants could refuse to answer any of the three alcohol use items or indicate that they did not know the correct response.These responses were set to missing.

Planned missing values
To reduce the survey burden on participants, we created three versions of the survey.Different subsets of the items for exposures to hazardous products, adverse work conditions, work stressors, positive working climate, and positive safety climate were included.Participants were randomly assigned one of the three versions.For each scale with planned missing items, the SPSS imputation procedure was run, using the Markov Chain Monte Carlo (MCMC) method.Five imputations were run for each scale, and the resulting files were saved.The value for each item was computed as the mean in the five imputations.Non-integer imputed values were rounded to an integer value.These files were then merged into a single database.Responses of "refused, " "do not know, " or "not applicable" were left as missing.The number of planned missing cases for each scale item is also presented in the Supplemental Tables.

Data analysis
Frequencies were computed for categorical variables, such as marital status and country of origin.Ranges, means, standard deviations, and medians were computed for continuous items, such as age and time in the United States.We assessed the associations of the continuous sample characteristic measures with drinking risk using independent-sample t-tests.We assessed the associations of the categorical measures with risk using Pearson chi-square tests.We then assessed the association of the mean scale scores with drinking risk using independent-samples t-tests.To assess the simultaneous association of the study variables with drinking risk, we first conducted a multivariable logistic regression analysis, with drinking risk coded as low or at-risk as the dichotomous dependent variable and the sociodemographic and workplace variables entered together as the independent variables (for these analyses, past-month income was divided by 100 so that the results would represent the effect of a $100 increase in income on the adjusted odds of at-risk drinking).Given the non-theoretical and exploratory nature of the current study, we then conducted analyses using conditional forward entry.Analyses were conducted with SPSS, v. 29; and a significance level of p < 0.05, two-tailed, was used for all analyses.

Results
Of the 331 surveys, non-missing values for the AUDIT-C score and independent variables were available for 307 LDL (92.7%), and analyses were conducted with these surveys.Participants had a mean age of 43.3 years (SD = 10.3,Mdn = 43.7).They had been in the United States for a mean of 13.1 years (SD = 10.0,Mdn = 11.0) and had been looking for work on the corners for a mean of 3.8 years (SD = 4.6, Mdn = 2.0).On average, participants had 7.6 years of school (SD = 3.7, Mdn = 8.0).Mean past-month income was $1203.68 (SD = $667.15,Mdn = $1080).
As shown in Table 1, nearly 90% of participants were born in Central America (45.6%) or Mexico (43.0%).Just over one-half (53.1%) were currently married or living with a partner.Nearly three-fifths (59.0%) did not speak English.A total of 290 participants reported living in a house or apartment.Four reported living in a shelter, and 13 reported living in another place.Of these 13, nine reported living in a trailer, living with a friend in a room, or living in some other room or quarter.These nine were classified as housed.The remaining four reported living in abandoned homes, living on the street, or being homeless.These participants and the four who reported living in a shelter were classified as unhoused.This provided a total of 299 (97.4%) participants who were housed and 8 (2.6%) who were unhoused.Just over one-half (50.2%) of participants rated their health as "bad" or "just okay or so-so." A total of 63 participants (20.5%) reported experiencing a work-related severe injury or illness in the past year.
The AUDIT-C Hazardous Use items are summarized in Table 2.For the frequency of alcohol use, 43.3% of participants reported they had not drunk alcohol in the previous year.Ten percent (10.4%) drank two or more times a week.Of those who drank, 36.2% reported having one or two drinks on a typical day when they drank, while 19.0% reported consuming seven or more drinks.One-quarter (25.3%) of past-year drinkers stated that they had never had five or more drinks on one occasion, while 28.7% reported that they had done so weekly or daily, or almost daily.Overall, 28.7% of all participants had consumed five or more drinks on one occasion on a monthly or greater basis in the last year.
Of the 307 participants, 202 (65.8%) were classified as low-risk drinkers and 105 (34.2%), as at-risk, based on the AUDIT-C score.In the bivariable independent samples t-tests, at-risk status and low-risk status did not differ by mean age (42.7 vs. 44.4,p = 0.159), years in the United States (14.5 vs. 12.4,p = 0.074), or years on the corners (3.7 vs 3.9, p = 0.735).They also did not differ by years of school (7.7 vs. 7.5, p = 0.663) or past-month income ($1305.30vs. $1150.86,p = 0.054).As shown in Table 3, drinking status was significantly associated with country of origin, marital status, and living arrangements in the bivariable chi-square analyses.
As shown in As shown by the adjusted odds ratios (aOR) in Table 6, being formerly married compared to being single (never married) and being unhoused compared to being housed were associated with at-risk drinking in the initial regression model.Although Positive Working Climate and Positive Safety Climate scores were significantly associated with at-risk drinking in the bivariable analyses, they were not in the multivariable model.Further, Positive Working Climate and Positive Safety Climate scores were strongly correlated (ρ = 0.499, p < 0.001).
To account for potential collinearity, in the subsequent forward entry analysis, we first ran the regression, omitting Positive Safety Climate scores.As shown in Table 7, in this model, at-risk drinking was associated with past-month income, living arrangements, and Positive Working Climate score.The regression was then rerun, omitting Positive Working Climate scores.In this model, at-risk drinking was associated with past-month income, being formerly married, and not being housed.At-risk drinking was also associated with a Positive Safety Climate score.

Discussion
Our overall sample was disadvantaged in terms of education, income, ability to speak English, and self-rated health.With regard to alcohol use, 43.3% of the LDL surveyed had not drunk alcohol in the previous year, which is higher than the 18.3% reported by Ornelas et al. (2016).Over one-quarter (28.7%) of our sample had consumed five or more drinks on one occasion on a monthly or greater basis in the past year.This contrasts with 49.0% reported by Ornelas et al. (2016).
Participants in their study appear to have been more likely to have reported being single or divorced and to have reported being homeless or living in temporary housing.
Although not directly assessed in this study, during field observations, we received testimonies of workers who indicated that they were provided a six-pack by an employer as a reward for a job well done.This pattern corroborates the findings of Worby and Organista (2013).Nevertheless, it appears as though occasions of drinking and hazardous alcohol consumption are less frequent in the current sample than in those of prior studies.Findings related to sociodemographic characteristics and at-risk drinking are consistent with previously conducted studies.Although not significant in the multivariable analyses, an association at the bivariable level between country of origin and at-risk drinking was found.The rate of at-risk drinking appeared higher among those born in the United States.Among those born outside the United States, the rate appeared higher among those of Mexican descent as compared to other countries (Caetano et al., 2008;Sanchez, 2015;Vaeth et al., 2009).We observed an association between drinking risk and unstable housing, as reported by Ornelas et al. (2019).This finding should be interpreted with caution, given the small number of unhoused participants.It does suggest, however, that, although a lack of housing may be rare among LDL, it may be a factor in at-risk drinking when it does occur.Compared to those who had never married, those who were formerly married had greater odds of at-risk drinking.As noted by Ornelas et al. (2015), drinking may be a means for some to cope with feelings of isolation.We also observed an association between past-month income and at-risk drinking.This is in keeping with Worby and Organista (2013) belief that having "money to burn" may lead to drinking.
The measures related to the physical work environment were not related to drinking risk.Thus, although workers may express concern regarding exposures (Ochsner et al., 2008) or report encountering physical risks (Pransky et al., 2002), these did not lead to greater at-risk drinking in this study.Unlike Negi et al. (2020), we did not find an association between the Work Stressors score and drinking risk.The results for the Positive Working Climate and Positive Safety Climate scales showed that positive actions undertaken by an employer at a job can be associated with more hazardous drinking.It may be that these positive climates encourage bonding and a celebratory mood that facilitates drinking.
The findings also underscore the personal conditions that may contribute to at-risk drinking.Workers who no longer had a partner (formerly married) were more likely to be at-risk drinkers as compared to those who had never married.As noted, although few participants were unhoused, those who were, were more likely to be at-risk drinkers.The drinking that occurs under these circumstances may be the result of a worker's trying to mitigate the effect of stressful life events or living conditions by drinking more heavily.
We considered a variety of sociodemographic characteristics associated in the literature with vulnerability.Measures of workplace conditions included exposures to hazardous products, adverse work conditions, and workplace stressors.Measures also included positive working and safety climates.These measures have been validated in our previous studies.The association of each item with at-risk drinking, as measured the AUDIT-C, was assessed.From the analyses, a parsimonious multivariable model of at-risk drinking among our sample of LDL was derived.This allowed us to address the research question, namely the association of both adverse and positive workplace factors with at-risk drinking in LDL.
Although this study makes a meaningful contribution to the literature with regard to the contributing factors of at-risk drinking among LDL, it is subject to certain limitations.Our analyses were not a formal test of any theoretical model.Future studies should consider an a priori model of at-risk drinking, such as the one developed by Martin et al. (1996), which would include measures of drinking networks and coworker and employer support.Future studies also should consider measures of mental health, such as depression and desesperación, and whether these act as mediators between workplace-based measures and at-risk drinking.Organista et al. (2019) found paths between working and living conditions and depression and desesperación and between living conditions and alcohol use in LDL.Addressing structural vulnerability to psychological distress can be done at multiple social-ecological levels, such as community-based support and governmental efforts, to expand work authorizations.
The timeframes specified for study items were not consistent across items.The AUDIT-C score was based on past-year alcohol use.In addition, the AUDIT-C is intended to serve as a screening instrument; it does not establish a diagnosis.The study did not consider the consequences of at-risk drinking.Exposures to hazardous products and work stressors also were measured with regard to the past year.In addition, the Adverse Work Conditions items did not specify a timeframe.Further, the Positive Working Climate and Positive Safety Climate Items were framed in terms of the participant's last day of work as a day laborer.
Functionally, there was likely no difference between responses of "never" and "does not apply" for the Exposures to Hazardous Products and Work Stressors items.Notably, there were no responses of "does not apply" for any of these items.
As noted, the majority of LDL approached declined to participate in the study.The characteristics of these individuals were not collected, and, thus, the results may be biased to the extent that non-participants and participants differed.
As with all cross-sectional studies, these results cannot establish causality.For example, we cannot establish whether a positive workplace climate leads to at-risk drinking or at-risk drinkers are more likely to find positive climates.Likewise, whereas lack of housing may lead to at-risk drinking, it may be that such drinking leads to loss of housing.Similarly, at-risk drinking may result from separation or divorce, or it may contribute to these.The study reported here was conducted in 2015, and results are limited by the extent to which circumstances have changed after that time period, specifically the potential impact of COVID.This study was conducted in a large metropolitan area, and results are limited by the extent to which this setting differs from others.
Data were collected in the form self-reported responses.Although previous studies have found high rates of abstinence among LDL, our rate of past year abstinence was, as noted, higher than in these studies.In addition, our observed rate of binge drinking was lower.Results will be biased to the extent that reported use was a result of respondents' providing "socially desirable" responses.Negi (2011) reported that efforts to reduce this bias in her study of substance use in LDL included efforts to build trust and rapport with workers and conducting interviews away from other workers, efforts followed in this study.Finally, values for many of the scale items were imputed.As noted, the results that use these values did not differ from those that used the original data.
Future studies could be strengthened by longitudinal studies with consistently framed items that take into account the factors that emerged from the COVID outbreak.Corner size was not considered in the current analyses but could be used in future research (along with other corner characteristics) to develop multilevel models.Research also could be strengthened by considering the social bonds between laborers and employers and their association with alcohol use.

Table 2 .
auDit-c Hazardous use items.
Table 4, drinking risk was not associated with either the Exposures to Hazardous Products score or the Adverse Work Conditions score.Drinking risk also was not associated with the Work Stressors score.For both the Positive Working Climate and Positive Safety Climate scales, the mean score was higher among the at-risk drinkers.As shown in Table 5, Exposures to Hazardous Products, Adverse Work Conditions, and Work Stressors scales were positively correlated with each other and inversely correlated with Positive Working Climate and Positive Safety Climate.The latter two scales were positively correlated.

Table 3 .
Drinking risk by sample characteristics.

Table 4 .
Working condition by drinking risk.

Table 5 .
Spearman correlations between working conditions.

Table 6 .
Logistic Regression analyses: all variables entered simultaneously.