Contexts of Drinking- and Alcohol-Related Problems Among Whites and Hispanics on and off the US/Mexico border in California

Abstract

Background

This paper examines the association between drinking context use by Whites and Hispanics on and off the US/Mexico border and alcohol problems.

Methods

Data come from a household sample of 1209 adults 18 to 39 years of age resident in Imperial County on the California/Mexico border; and Kern, Tulare, and Madera in California’s Central Valley. Data were collected on the phone or online and analyzed with an ordinal generalized linear model.

Results

The pattern of statistically significant associations between the frequency and the volume of drinking in different contexts varies across problem types. Furthermore, some contexts of drinking are associated with problems in more than one area. For instance, frequency of drinking at bars/pubs is associated with social problems, risky sex, and fights, but not with injuries. Injuries are associated with the frequency of drinking at home alone or with family and at restaurants. Volume of drinking at bars/pubs is also significantly associated with three different contexts: social problems, injury, and fights. But the volume of drinking at the home of friends or relatives is associated with fights only. Border location is an effect modifier, changing the effect of frequency of drinking at bars and pubs from protective to a factor of risk for social problems and fights.

Conclusion

These results provide support for the social ecology of drinking and micro environmental factors or risk. The effect of border location on frequency of drinking in bars/pubs underlines the importance of the macro environment in problem generation.

Introduction

Alcohol-related problems are associated with a variety of factors. They are closely related to the amount and patterns of drinking, to drinkers personal characteristics, and the environment in which drinking occurs. Studies of this latter factor are guided by socioecological theory, and focus on the contexts where drinking takes place, the characteristics of these contexts, and those of the drinkers in the context [1, 2]. Drinker selection of specific contexts to drink, such as parties, family gatherings, or restaurants, is influenced by the commercial availability of alcohol, and the presence of such places in the community. Drinkers select drinking context based on availability, characteristics, and on previous rewarding experiences. Once a specific drinking context is selected, drinkers influence one another’s alcohol consumption in the context [2]. For instance, bars have been associated with greater likelihood of aggression compared to other contexts, and this seems particularly so in crowded bars [3–5], and in bars with younger, risk taking, and impulsive patrons [2, 6] These bars may attract such a clientele because of lax policies about behavior, as well as characteristics that are attractive to a particular type of patron such as cheaper drinks, louder music, or larger crowds. In this way, a context may encourage or constrain drinking or encourage or constrain specific dinking-related problem behaviors [7].

The broader epidemiological literature on drinking and problems across ethnic groups has not paid attention to contexts of drinking and their association with problems. This literature has examined cross-ethnic differences in drinking, with recent pertinent results from national surveys indicating that American Indians/Alaskan Natives (AI/AN) commonly have heavier drinking patterns than other groups, that Asian Americans usually drink less than all others, while results from comparisons across Blacks, Whites, and Hispanics vary. For instance, data from the 2020 National Survey on Drug Use and Health (NSDUH) [8] show that binge drinking (4+ drinks for women/5+ for men on the same occasion) was slightly higher among Hispanics (26.6%) than Whites (24.6%) and Blacks (22.8%), and lowest among Asians (13.5%). Heavy alcohol use in the past month (4+ for women/5+ for men on one occasion in 5 or more days) was highest among AI/AN (9.1%), followed by Whites (7.8%), followed by Hispanics and Blacks (6.1% and 6%, respectively), and then Asians (2.5%). Previous studies with US Hispanics only have focused on, for example, drinking differences by Hispanic origin, with acculturation to the USA. [9–11], with birthplace [11, 12], and social disadvantage [13–15], but not drinking contexts.

A previous paper which analyzed the data set herein described the types and use of drinking contexts by White and Hispanic residents of California living on and away from the California/Mexico border [16]. The most frequent drinking context by both ethnic groups, independent of location on or off the border was “drinking at home alone or with family,” reported by a third to about a little more than half of the drinkers. In the Central Valley, the mean number of drinks per occasion for Hispanics was higher than that for Whites when drinking at home alone or with family, drinking at home with friends, and drinking at bars/pubs. On the border, the number of drinks was higher when drinking at home with friends for White drinkers and at bars and pubs for Hispanic drinkers. In multivariable analyses, location on or off the border was not associated with any drinking context, but Hispanics were 3 and almost 2 times more likely to drink more frequently than Whites at home with friends, and at friends’ and relatives’ homes, respectively.

This paper extends these previous analyses by examining the association between contexts and problems grouped in four areas: social problems, risky sex, injuries, and fights. The US/Mexico border area is of interest because of its well-known increased alcohol availability due to a lower legal drinking age (18 years of age in Mexico versus 21 years of age in the USA), lower alcohol prices, and later closing times of bars and clubs in Mexico than the USA [17–19]. Whites and Hispanics are two groups of interest because in 2018 together they comprised 76% of the California population (White, 37%; Hispanics, 39%) [20]. Hispanics, mostly of Mexican origin, are the majority in most population sites on the border [21].

Based on previous findings about differences between on and off border drinking [22–24] and drinking context use and preferences by drinkers [2, 16, 25], the magnitude and type of the association (positive or negative) between contexts and problems will vary. Following a socioecological framework, the goal of the analysis reported here is to measure how significant these variations in context use and drinking are between Hispanics and Whites on and off the border. This will indicate the extent to which prevention programs should pay attention to contexts of drinking and problems when developing prevention interventions for these two groups. Finally, because of the increased availability of alcohol on the border and lax drinking rules at bars/pubs there [17–19], border location is expected to modify the effect of frequency and volume of drinking at bars/pubs on problems related to injuries and fights.

Methods

Sample and Data Collection

Data are from a household survey of 1209 (final sample size) 18 to 39 year old adults, residing in three US counties of California’s Central Valley: Kern, Tulare, and Madera and one US county on the California/Mexico border, Imperial. These counties are agricultural and mostly with relatively small towns, with populations between 52,000 and 120,000. The separation between the southernmost areas where households were sampled in the Central Valley and the border was about 220 miles.

Households were selected using a United States Postal Service list assisted address-based sample purchased from a commercial vendor. Households were eligible if at least one of the residents was 18 to 39 years of age and self-identified as White or Hispanic. Respondents were sent a pre-announcement letter describing the study and offered the option to opt out of the survey by contacting the survey research firm. Surveys were conducted from June 2019 to May 2020 in English or Spanish. Respondents were given the option of being interviewed on the phone or respond to a questionnaire online. About 90% of the respondents opted to be interviewed or respond to the online questionnaire in English. About 86% responded to the survey online. Respondents gave verbal consent to participate in the survey and received $20.00 remuneration. The study was approved by the Institutional Review Board of the Pacific Institute for Research and Evaluation.

The survey overall cooperation rate and the response rate, using version #4 of the American Association for Public Opinion Research [26], in the Central Valley were 95% and 72.5%, respectively, while on the border they were 12% and 7%. The low rates on the border may have resulted from the time overlap between the survey and the COVID-19 pandemic, which disproportionately affected Imperial County [27], and by restrictive federal government immigration policies directed at asylum seekers at the US/Mexico border and undocumented Latin American immigrants already in the USA. This could have influenced potential respondents behavior, making it less likely that non-citizens and those born abroad would participate in the study, and also increasing under-reporting of problematic behavior. However, internal validity was secured by group matching respondents on the border and the Central Valley on age and ethnicity, maintaining uniformity of respondent selection and data collection, and conducting analyses with controls for potential confounders. Post stratification weights were used to match respondents on gender, age, and ethnicity to known proportions on the US Census Bureau 2018 American Community Survey (American Community Survey (ACS) (census.gov), accessed on July 7, 2020).

Measurements

Alcohol-Related Problems

Respondents were asked about nine different types of problems they might have experienced (Yes/No for each problem) after drinking alcohol in the past 12 months. These problems were grouped in four areas as follows: risky sex (having unplanned sex, having sex without protection); social problems (did something later regretted, criticized for drinking, hangover interfered with school, work, at home); injuries (got hurt or injured); and fights (got in a physical fight). Each problem present was scored as “1” and scores in each problem area were represented by a count of problems present.

Context Specific Drinking

A “drink” was defined as a 12-ounce can of beer, a 5-ounce glass of wine, or a 1.5-ounce shot of liquor. Respondents were asked for the number of days in which they had a drink in various drinking contexts: home alone or with family, home with friends, friends or relatives’ home, restaurant, and bar/pub. Respondents were also asked how many drinks they typically consumed in each context. Variables representing frequency of drinking (F) and volume (V) were estimated. Continued volumes of use (V–F), for each setting, were also calculated as [(F × typical number of drinks) minus F]. F represents independent effects related to greater numbers of drinking occasions (exposure) and V–F represents independent effects related to greater average drinks consumed per drinking occasion, two critical dimensions of dose-response [28]. The advantage of this approach is that it reduces collinearity in estimators of effects associated with frequency and volume of drinking providing a non-biased representation of dose–response effects [29]. All drinking measures were scaled to a 365-day metric.

Overall Drinking

Respondents were asked the number of days on which they had at least one drink of beer, wine, or liquor in the past 4 weeks, or among those who had not used alcohol in the previous 4 weeks, the past year. Those who had used alcohol were then asked on how many days they drank one, more than one, three or more, six or more, and nine or more drinks, and the maximum number of drinks over that time. These data, as those for drinking in specific contexts, were fit using a validated log-logistic model of these responses [30, 31] providing the same measures of drinking frequency (F), average drinking quantity (Q), volume (V=F×Q), and continued drinking volumes, V–F, a dose-response measure of heavier drinking [32].

Drinking in Mexico

Respondents were asked about how many times they had visited Mexico in the past 12 months. Those that answered at least once were then asked on how many of those days they had consumed at least one drink of any alcoholic beverage. Answers were coded as “0” for no days and as “1” for one or more days.

Border Location

This variable identified sample respondents interviewed away from the border in the US counties of Kern, Madera, and Tulare, coded as “0,” and those who were interviewed in Imperial County, also on the USA, but on the border, coded as “1.”

Impulsivity

This was measured assessing respondents’ agreement with the following three statements: I often act on the spur-of-the-moment without stopping to think; you might say I act impulsively; and many of my actions seem to be hasty [33, 34]. Response categories ranged from “not at all” to “quite a lot,” with scores ranging from one to four per item. Scores were divided into quarters, and the scale was dichotomized with the three bottom quarters coded as “none” and the top quarter coded as “one.” Cronbach’s alpha was .86.

Perceived Risk of Drinking and Driving and Other Consequences

This was assessed with five items asking respondents to evaluate the likelihood that “something bad would happen” if they drove over the speed limit, drove while drunk, drove without a seat belt, drank a lot, and had sex with someone they just met. Five response categories ranged from very likely to very unlikely. Scores were divided into quarters, and the scale was dichotomized with the three bottom quarters coded as “none” and the top quarter coded as “one.” Cronbach’s alpha was .88.

COVID Shelter-In-Place

This variable was coded as “0” for those interviewed before the shelter-in-place order and as “1” for those interviewed during the order, which closed bars, pubs, clubs, and restaurants in California. On March 19, 2020, California’s Department of Alcoholic Beverage Control (ABC) allowed liquor and “cocktail-to-go” home delivery together with meals, in addition to the already existing permission for wine and beer delivery. In addition, the US/Mexico border was closed to all non-essential (tourism and recreational) travel on March 21, 2020, and remained so until November 8, 2021. On this date, entry into the USA from Mexico became open to COVI-19 fully vaccinated individuals.

Sociodemographic Variables

Gender

A dichotomous variable coded as men and women.

Age

The age of respondents was analyzed as a categorical variable: 18–25, 26–29, and 30–39.

Ethnicity

This was based on self-identification. Only Whites and Hispanics were eligible to participate in the study. This variable was coded as “0” if respondents were Hispanic and as “1” if they were non-Hispanic White. Following American Psychological Association guidelines, this variable is seen as representing “shared cultural characteristics such as language, ancestry, practices, and beliefs” (https://apastyle.apa.org/style-grammar-guidelines/bias-free-language/racial-ethnic-minorities accessed on 11/10/2022).

Income

This variable had three categories: $20,000 or less, $20,001 to $60,000, and $60,001 and more.

Employment

This was a dichotomy coded as “0” for currently working and as “1” for not currently working.

Education

Respondents were categorized into four education categories: (a) less than high school; (b) completed high school or graduate equivalency exam (GED); (c) some college or technical or vocational school; (d) completed 4-year college or technical school or higher.

Marital Status

This 3-category variable was coded as follows: (a) married; (b) separated or divorced, and (c) single.

Statistical Analyses

Analyses were conducted on weighted data. About 90% of survey weights fell between 0.0 and 2.0. The statistical software used was Stata’s 17 with “svy” prefix [35]. The overall number of different types of problems connected to context-specific drinking (Table 1) was estimated by running a simple linear regression model for each of the problem domains adjusting for each drinking context one at a time [36]. The resulting coefficients were multiplied by the mean drinking frequency in each context by drinkers in our sample and scaled to a population size of 10,000. The resulting numbers are yearly frequencies related to any use in the context.

Table 1.

Prevalence of specific types of problems among drinkers who drank at a context in the past 12 months by location and ethnicity. Weighted percentages (unweighted N)

	Social problems		Risky sex		Injuries		Fights
Context	White	Hispanic	White	Hispanic	White	Hispanic	White	Hispanic
2a. Central Valley
Home alone or with family	30.9%ns	34.00%	19.0%ns	29.00%	12%ns	7.90%	6.7%ns	5.00%
	(195)	(297)	(196)	(297)	(196)	(297)	(196)	(297)
Friends or relatives’ home	37.3%ns	43.20%	27.4%ns	30.80%	16.2%ns	7.90%	12.5%ns	5.30%
	(112)	(224)	(112)	(224)	(112)	(224)	(112)	(224)
Home with friends	37.5%ns	38.50%	32.5%ns	30.80%	14.9%ns	9.30%	12.0%ns	7.00%
	(105)	(214)	(106)	(214)	(106)	(214)	(106)	(214)
Restaurant	32.9%ns	37.80%	27.8%ns	28.00%	14.7%ns	11.90%	11.4%ns	8.10%
	(78)	(148)	(79)	(148)	(79)	(148)	(79)	(148)
Bar/pub	43.1%ns	50.50%	49.3%ns	38.40%	20.9%ns	15.30%	12.3%ns	11.20%
	(41)	(89)	(41)	(89)	(41)	(89)	(41)	(89)
2b. Border
Home alone or with family	25.0%ns	28.40%	12.7%ns	18.50%	1.0%ns	4.50%	9.8%ns	3.00%
	(52)	(289)	(52)	(288)	(52)	(289)	(52)	(290)
Friends or relatives’ home	20.0%ns	295%	10.8%ns	22.90%	2.8%ns	7.00%	7.7%ns	3.20%
	(37)	(236)	(37)	(234)	(37)	(236)	(37)	(237)
Home with friends	36.1%ns	31.70%	19.6%ns	21.10%	4.1%ns	5.80%	16.6%ns	3.50%
	(36)	(200)	(36)	(199)	(36)	(200)	(36)	(201)
Restaurant	29.1%ns	34.10%	16.3%ns	18.90%	5.2%ns	6.00%	6.8%ns	3.30%
	(41)	(214)	(41)	(214)	(41)	(214)	(41)	(214)
Bar/pub	45.4%ns	37.80%	16.9%ns	24.80%	4.6%ns	6.90%	6.70%	2.50%
	(27)	(143)	(27)	(143)	(27)	(143)	(27)	(143)

Bonferroni corrected chi square independently for “a” Central Valley and “b” Border (.05 /20) ^ns: not significant

Multivariable analysis of the association between problems and contexts (Table 2) was conducted using Stata’s “oglm” procedure, which fits an ordinal generalized linear model for count responses [37, 38]. The outcomes were the 4 areas of alcohol-related problems described above, coded as follows: risky sex, 0 to 2; social problems, 0 to 3; fights 0 to 1; and injury, 0 to 1. As error variance increases with drinking frequencies and volumes of drinking and is a source of bias in model estimation, the analysis also controls for these sources of heteroskedasticity models.[32, 39]. Continuous variables (e.g., impulsivity, perceived risk, days drinking in Mexico) were categorized to address skewness and uneven distribution of respondents across answer categories. Independent variables were selected based on previous evidence of associations with drinking [40–42]. A correlational analysis as well as a variance inflation analysis using Stata’s “vif” command (vif range 1.2 to 8.4; mean vif=3.35) did not indicate multicollinearity in the variables in the paper.

Table 2.

Expected number of drinking-related problems specific to drinking contexts for a population of 10,000 drinkers

	Types of problems
Context	Social	Risky sex	Injury	Fight
Home alone or with family	1337	386	286	189
Home with friends	896	405	181	195
Friends or relatives’ home	931	481	148	163
Restaurant	579	302	101	144
Bar/pub	613	376	91	149

Results

Sample Characteristics

Women represented 60% of the sample, which also was the proportion of participants 30-39 years of age (data not shown). About 71% of the sample self-identified as Hispanic, 41% had some college or technical education, 53% were married, 85% were US born, 36% were unemployed, 29% had an annual family income of $20,000 or less, and 40% were Catholic. Bonferroni corrected chi square (.05/9=.005) testing differences in sample characteristics between the border and the Central Valley showed statistically significant differences for ethnicity (Central Valley Hispanic 61%; border Hispanic 85%), birthplace (Central valley US born 90%; border 79%), and religion (Central Valley Catholic 34%; border 49%).

Problem Type Percentage by Location and Ethnicity

There were no statistically significant differences in problem type percentage between Whites and Hispanics (Table 1) in any location or context.

Expected Number of Problem Type by Drinking Context

The number of different problem types per 10,000 population varied across contexts (Table 2). Drinking home alone or with family was the context connected with the largest number of social problems, injuries, and fights. However, the number of social problems associated with drinking home alone or with family was 3.5 times larger than the number of problems related to risky sex, 4.7 times larger than injuries, and 7 times larger than fights. Risky sex was mostly connected with drinking at home with friends. In general, on-premises contexts such as restaurants and bars/pubs were not associated with as many problems as off-premises contexts such as drinking at home or at friends’ and relatives’ homes.

Multivariable Analysis

Context specific frequency of drinking had positive and statistically significant associations with three types of problems depending on context (Table 3). These included: the associations of frequency of drinking at bars/pubs with social problems, risky sex, and fights; frequency of drinking at home alone or with family with risky sex, and injuries; frequency of drinking at home with friends with risky sex;, and frequency of drinking at friends and relatives homes with injury. Context specific continued volume of drinking also had associations with a variety of problems, three of which were statistically significant: volume of drinking at bars and pubs was significantly associated with social problems; continued volume of drinking at home with friends and at restaurants were associated with injuries; and continued drinking at home with friends, continued drinking at friends and relatives homes, and continued drinking at bars/pubs were associated with fights.

Table 3.

Ordinal generalized linear regression model of alcohol-related problems on frequency and volumes of drinking, location, and sociodemographic and psychological variables

	Social problems	Risky sex	Injury	Fight
Independent variables	AOR (95% CI)	AOR (95% CI)	AOR (95% CI)	AOR (95% CI)
Context frequency
Home alone or with family	1.002 (0.994, 1.011)	1.004 (1.002, 1.005)**	1.013 (1.006, 1.021)**	0.992 (0.975, 1.009)
Home with friends	0.996 (0.982, 1.011)	0.994 (0.989, 0.999)*	0.985 (0.966, 1.004)	1.021 (0.995, 1.047)
Friends/relatives homes	1.009 (0.998, 1.021)	1.003 (0.998, 1.008)	1.017 (1.004, 1.031)*	0.969 (0.934, 1.006)
Restaurant	1.007 (0.996, 1.019)	0.995 (0.986, 1.005)	1.042 (1.019, 1.066)**	1.009 (0.983, 1.036)
Bars and pubs	0.961 (0.928, 0.994)*	1.028 (1.000, 1.057)*	0.976 (0.937, 1.016)	0.928 (0.880, 0.979)**
Context continued volume
Home alone or with family	1.000 (0.999, 1.001)	1.000 (1.000, 1.000)	1.000 (0.999, 1.001)	1.001 (1.000, 1.002)
Home with friends	0.999 (0.998, 1.000)	1.000 (1.000, 1.001)	1.003 (1.000, 1.005)*	0.998 (0.996, 1.000)*
Friends/relatives homes	1.000 (0.999, 1.001)	1.000 (0.999, 1.001)	0.999 (0.997, 1.002)	1.003 (1.000, 1.005)*
Restaurant	0.999 (0.997, 1.001)	0.999 (0.995, 1.002)	0.977 (0.958, 0.996)*	0.998 (0.995, 1.001)
Bars and pubs	1.005 (1.000, 1.010)*	0.998 (0.993, 1.002)	1.007 (1.001, 1.014)*	1.016 (1.006, 1.026)**
Interaction terms
Bars/pubs drinking frequency × border	1.044 (1.002, 1.087)*	0.983 (0.949, 1.018)	1.067 (1.007, 1.131)*	1.084 (1.031, 1.140)**
Bar/pub continued volume × border	0.996 (0.991, 1.001)	1.004 (0.997, 1.011)	0.985 (0.974, 0.996)**	0.985 (0.974, 0.995)**
Sociodemographic
Hispanic (ref: White)	1.177 (0.69, 2.008)	1.231 (0.753, 2.014)	0.432 (0.133, 1.400)	0.172 (0.053, 0.562)**
Border location (ref: Central Valley)	0.352 (0.192, 0.644)**	0.67 (0.383, 1.172)	0.354 (0.055, 2.290)	0.390 (0.115, 1.322)
Heteroskedasticity terms	Coefficient (95%CI)	Coefficient (95%CI)	Coefficient (95%CI)	Coefficient (95%CI)
Overall drinking frequency	0.080 (0.025, 0.136)**	0.048 (−0.009, 0.106)	0.136 (−0.025, 0.298)	−0.002 (−0.092, 0.089)
Overall continued volume	0.007 (−0.002, 0.015)	0.022 (0.010, 0.034)**	0.010 (−0.010, 0.030)	−0.005 (−0.017, 0.006)
Overall drinking frequency squared	−0.003 (−0.007, 0.000)*	−0.006 (−0.009, −0.002)**	−0.009 (−0.016, −0.002)*	0.002 (−0.001, 0.005)

Data are weighted. Heteroskedastic coefficients are not transformed to odds ratios. **p<0.01, *p<0.05. Analyses also controlled for gender, age, education level, employment status, income, birthplace, religion, impulsivity, perceived risk, COVID-19 shelter in place, ever used marijuana, and drinking in Mexico

Border and Hispanic Ethnicity Effects

The interaction between frequency of drinking and location had statistically significant associations with social problems, injury, and fights (Table 3). These associations showed that on the border, a higher frequency of drinking at bars/pubs was associated with increased odds of social problems, of injuries and of fights. On the other hand, increased continued volume of drinking at bars /pubs on the border was associated with decreased odds of injuries and of fights. The main effect for Hispanic ethnicity was protective against fights, while that for border location was protective against social problems.

Discussion

One of the variables of interest in this paper is ethnicity, which had no statistically significant associations with problems in any context or location in Table 1. Furthermore, ethnicity had only one statistically significant effect as a protective factor for drinking related fights in the multivariable analysis in Table 3. This latter effect may be associated with going out for a drink with friends or family, a common occurrence among Mexican Americans [40], which might minimize aggressive interactions. It is also possible that the lack of differences between Whites and Hispanics is sample homogeneity due to age restriction, 18 to 39 years of age, for eligibility to participate in the study. While this group has a higher prevalence rate of drinking, heavier drinking, and problems, all of which have a positive effect on power, it may also make ethnic groups more like one another in drinking and its consequences.

More broadly, previous analyses in the alcohol epidemiology literature show that differences in drinking and problem rates between Whites and Hispanics are dependent on the type of outcome being compared, date of data collection, and study methods. For instance, 2014 NSDUH data reviewed by Vaeth et al. [43] basically show no differences in the rate of binge drinking (5 or more standard drinks on occasion) (24.7% vs. 23.5%) and in the rate of heavy drinking (5 or more binge episodes in the previous month) between Whites and Hispanics. In the 2020 NSDUH [8], results show that binge drinking was slightly higher among Hispanics (26.6%) than Whites (24.6%), and heavy drinking in the past month was higher among Whites (7.8%) than Hispanics (6.1%), albeit these differences are not large. In data from the 2012 National Epidemiologic Survey on Alcohol and Related Conditions III (NESARC III), heavy episodic drinking (one or more episodes of binge drinking per month) was higher for Hispanics (31.6%) than Whites (24.8%) [44]. Regarding drinking problems, and using alcohol use disorders (AUD) as an example, analyses of NESARC III data showed no differences between Hispanic and Whites for past 12-month DSM-5 AUD (White 14%, Hispanic 13.6%), but Whites had a higher rate of lifetime AUD than Hispanics (32.6% vs. 22.9%) [45]. Previous analyses of AUD based on the data set in this paper also showed no differences in AUD between Whites and Hispanics [46]. Other evidence also indicates that time trends in problems across ethnic groups may differ, increasing or minimizing disparities across groups [47].

Rates of problems per 10,000 population in Table 2 are higher for off-premises contexts such as one’s own home and friends’ and relative’s homes and especially for social problems and risky sex. These contexts have been associated with higher frequency of use [16, 48] and are also usually associated with concurrent drinking in other contexts. Even though these contexts may not be associated with a high problem generation per drinker, the fact that they are used by a higher number of drinkers leads to higher rates of problems at the population level. Drinkers start drinking in one of these contexts and then move on to drink in on-premises contexts such as restaurants and bars/pubs [1].

The multivariable results in Table 3, which control for the effect of other factors, do not show the same picture. Frequency of drinking at one’s own home and friends and relative’s homes are associated with risky sex and injuries only. This was somewhat surprising because these problems tend to be associated with drinking in on-premises contexts, which sometimes can be unruly, such as bars/pubs. In fact, the results show an association between frequency of drinking at bars/pubs and social problems, risky sex, and fights. Huckle et al. [29] reported a positive association between increased frequency of drinking at bars/nightclubs and “alcohol disorderly behavior,” while Sumetsky et al. [36] reported null results for frequency of drinking at bar/club and risky sex, social problems, physiological problems, and driving after drinking too much.

The associations found between continued volume of drinking and problems did not show the same pattern as that for frequency of drinking, confirming that exposure effects can be at times independent of the effect associated with increased volume of drinking in the same context. That is, drinkers do not need to consume alcohol beyond a first drink to suffer consequences from their exposure to a particular drinking environment. In this regard, bars have been identified as environments especially associated with higher risk of certain types of problems such as injuries and fights [2, 49, 50].

Some of this special effect of bars on certain types of problems is reflected in the results of the interaction between border location and frequency and volume of drinking at bars/pubs in Table 3. Here there is also a confluence of community level and drinking context specific effects that reflects well the framework of the social ecological model of drinking proposed by Gruenewald et al. [2]. Understanding the social ecology of drinking, i.e., the interactions between drinking contexts characteristics, drinking companions, patterns of drinking in various contexts, and their association with problems is important to develop and implement prevention interventions. For instance, if the objective is to prevent drinking related aggression and associated injuries, paying attention to on-premises drinking outlets such as bars, their environment, their opening and closing times, shaping patron interactions to avoid aggression, and providing server training on how to prevent intoxication are necessary factors to consider when planning interventions.

The border, as discussed in the Introduction, has been characterized as a place where the increased availability of alcohol plus lax policies about intoxication and aggressive behavior in bars/pubs and nightclubs commonly lead to aggression both sexual, mostly directed towards women, and non-sexual [18, 51]. As in Huckle et al. [29], the risk effect on injuries/fights herein is associated with exposure to bars/pubs but not with volume of drinking. In fact, continued volume of drinking has a protective effect against injuries and fights. Given alcohol’s complex effects on behavior, a threshold blood alcohol concentration has not been identified yet for alcohol and aggression [52]. It is possible that more intoxicated drinkers are inhibited from involvement on physical aggression because of loss of motor ability due to alcohol intake.

Strengths of this study include data collection in purposively selected counties in a large state, with respondent matching for age and ethnicity to maximize internal validity, and the focus on two large population groups, one of which, Hispanics, represent a large ethnic minority group in California and the USA. Regarding limitations, the sample response rate on the border (Imperial County) was low. Drinking amounts and other drinking related information were self-reported, which could lead to under-reporting. The study design was cross-sectional, which does not allow for assessments of temporal associations. Some of the reported relationships can therefore be bidirectional. Not all White and Hispanic population subgroups are identified in the study. Sexual minorities, the LGBTQ community, have higher rates of heavier drinking (e.g., binge drinking) and alcohol-related problems than other groups. Future research should focus on such groups, which face increased stress due to diminished access to resources such as housing, education, employment, and health care.

In conclusion, results show that drinking problems are independently connected with two important dimensions of drinking, frequency, and continued volume of drinking in a context. The association between exposure to a particular context and problems is an important finding supporting the social ecology of drinking and micro environmental factors or risk. The risk linked to the effect of border on frequency of drinking in bars/pubs underlines the importance of the macro environment, the border, in problem generation. The identification of different uses of contexts and different rates of problems related to contexts use between Hispanics versus Whites on- and off-the-border tells us what to prevent and what groups to target to reduce problems most efficiently. If Hispanics and Whites had the same pattern of context use and problem risks related to contexts, then there would be no need to differentiate prevention efforts by population group or locale. If there are differences, then there are reasons to differentiate prevention efforts and the study tells us some of the differences that matter.

Author Contributions

RC conceptualized the paper, executed part of data analyses, and took main responsibility for writing. PV helped conceptualize the paper and in the interpretation of results and editing and literature review. PG guided data analysis, also providing support for result interpretation and guidance for review of past findings. BP supported data assembling and analyses. ZK contributed to data analysis execution, also coding variables, and provided comments on drafts of the manuscript.

Funding

Research for and preparation of this manuscript was supported by National Institute on Alcohol Abuse and Alcoholism National Research Center grant P60-AA06282 to the third author.

Declarations

Ethics Approval

The study was approved by the Institutional Review Board of the Pacific Institute for Research and Evaluation.

Consent to Participate

Informed consent was obtained from all individual participants included in the study.

Competing Interests

The authors declare no competing interests.