Skip to main content
NCBI home page
Alcohol and Alcoholism (Oxford, Oxfordshire) logoLink to Alcohol and Alcoholism (Oxford, Oxfordshire)
. 2020 Aug 13;56(3):360–367. doi: 10.1093/alcalc/agaa073

State Alcohol Policies, Binge Drinking Prevalence, Socioeconomic Environments and Alcohol’s Harms to Others: A Mediation Analysis

Won Kim Cook 1,✉, Libo Li 1, Thomas K Greenfield 1, Deidre Patterson 1, Timothy Naimi 2, Ziming Xuan 3, Katherine J Karriker-Jaffe 1
PMCID: PMC8085366  PMID: 32790857

Abstract

Aims

Alcohol policy effects on alcohol’s harms due to others’ drinking (AHTO) and contextual factors that may mediate such policy effects have been understudied. This study examines state binge drinking prevalence as a mediator of the relationship between state alcohol policy and socioeconomic environments and individual-level AHTO.

Methods

A nationally representative sample of US adults (N = 32,401; 13,873 males, 18,528 females) from the 2000, 2005, 2010 and 2015 National Alcohol Surveys and the 2015 National Alcohol’s Harm to Others Survey, administered in telephone interviews and based on random digit dialed sampling, were linked with state-level Alcohol Policy Scale (APS) scores, binge drinking prevalence and socioeconomic status (SES) data. Three 12-month AHTO measures were family/marriage difficulties, assault or vandalism and riding with drunk driver or having traffic accident. Three-level mediation analyses were conducted, controlling for gender, race, education, marital status, family problem-drinking history and state policing rate.

Results

The effects of the APS on reduced risks for assault/vandalism and drinking-driving harms were significantly mediated by reduced state binge drinking prevalence. The APS had no direct or indirect effect on family/marital trouble. State SES had significant indirect effects on increased risks for assault/vandalism and driving-related harm through increased state binge drinking prevalence and a direct effect on reduced family/marital problems.

Conclusions

A more stringent alcohol policy environment could reduce assault/vandalism and driving-related harm due to another drinker by lowering state binge drinking rates. Alcohol policies may not be effective in reducing family problems caused by another drinker more prevalent in low-SES states.

INTRODUCTION

Health and social effects of alcohol on the drinker are relatively well researched, but alcohol’s harms due to others’ drinking (AHTO) are less so (Casswell et al., 2011; Navarro et al., 2011). There is a growing body of literature on AHTO in the USA and internationally, but aside from two recent US studies (Greenfield et al., 2019; Trangenstein et al., 2020), little research has been reported on the effects of alcohol policies on AHTO. This is a notable gap. Alcohol policies, in essence, are public health responses intended to serve the public good by reducing the widespread health and social burden related to alcohol use for the greatest number of people (Babor et al., 2010). As alcohol’s harms to others are inherently farther-reaching than harms to the drinkers themselves, AHTO research has great potential to advance the alcohol policy agenda by focusing on the health and well-being of a broader population (Giesbrecht et al., 2010).

Also notably, with much of alcohol policy research to date having taken an ecological approach to examine associations between aggregate exposures and aggregate outcomes or focused on direct effects on individual-level drinking outcomes, limited attention has been paid to contextual factors that may mediate, undermine or reinforce policy effects. Alcohol policies alone may neither cause nor explain changes in alcohol consumption and alcohol-related harm (Matrai et al., 2014). Rather, since policy effects are likely to result from complex interplays with other contextual factors such as social, cultural, economic, religious and demographic factors, they can be better explained when appropriate contextual factors are considered simultaneously (Allamani et al., 2011). With some exceptions like the influences of alcohol industry (Babor et al., 2015) and urbanization (Matrai et al., 2014), contextual factors have been understudied in alcohol policy research.

In examining the relationship between alcohol policies and AHTO in this study, we consider two contextual factors, US state-level binge drinking prevalence and state-level living standards. Binge drinking prevalence varies a great deal across US states, with each state having unique characteristics that may shape drinking patterns and alcohol-related problems, including demographic compositions by age, gender, race/ethnicity, socioeconomic status (SES) and religious affiliations, and historical and cultural practices related to drinking (Kerr, 2010). Prior AHTO studies have found victims’ heavy drinking, higher annual alcohol intake, more frequent episodes of intoxication and more frequent visits to public drinking places to be among the most consistent correlates of AHTO (Rossow and Hauge, 2004). A permissive environment that tolerates drinking, an ‘anything goes’ attitude and rowdiness have been associated with both higher levels of alcohol consumption and alcohol-related harm (Hughes et al., 2011), which may potentially include harms to others. States where binge drinking is more prevalent may create environments more prone to AHTO, where heavy drinkers are more likely to be present and perhaps act badly when drinking. State alcohol policies intended to reduce binge drinking may lead to environments where binge drinking is less prevalent, which, in turn, may reduce AHTO risk.

State-level living standards or socioeconomic status (SES) is another contextual factor that may affect AHTO risk, independently or through its effects on binge drinking. Prior research has linked area-level socioeconomic deprivation to negative consequences of others’ drinking such as victimization from fighting or vandalism and family problems (Karriker-Jaffe and Greenfield, 2014). This may be because residents in disadvantaged neighborhoods tend to experience a higher level of chronic stress due to social and physical disorganization and depleted psychosocial resources to cope with this stress (Fitzpatrick and LaGory, 2000). Both stress and lack of positive coping resources in disadvantaged areas may increase hazardous drinking or harms due to such drinking in others (Mezuk et al., 2013). Similarly, economic hardship in low-SES states may engender more stressful social environments where binge drinking is more prevalent, which, in turn, may increase risk for AHTO.

Building on a recent study reporting that stringent state alcohol policy environments were inversely associated with aggression-related harms and drunk driving-related harm from someone else’s drinking for US adults younger than 40 years (Greenfield et al., 2019), the current study addresses two aims: (1) to examine state binge drinking prevalence as a mediator of the association between alcohol policies and individual-level AHTO and (2) to examine the relationship between state-level SES and AHTO risk, including mediation of this relationship by state binge drinking prevalence. We hypothesize that alcohol policies will reduce AHTO risk in part by reducing state binge drinking prevalence. We also hypothesize that lower state-level SES will be associated with higher risk for AHTO, partly through higher state binge drinking rates.

METHODS

Data

Data from the last four National Alcohol Surveys (NAS) conducted in 2000, 2005, 2010 and 2015 and the 2015 National Alcohol’s Harm to Others Survey (NAHTOS) were pooled to draw a nationally representative sample of US adults ages 18 or older (N = 32,401). All surveys were computer-assisted telephone interviews, based on list-assisted random digit dialed sampling. The 2015 NAS and NAHTOS used dual-frame (landline/mobile) sampling to include respondents who completed the interview via mobile phone. Survey cooperation rates for the surveys were estimated to be 58% (2000 NAS), 56% (2005 NAS), 52%, (2010 NAS) and 60% (2015 NAS and 2015 NAHTOS). These survey data were linked with prior-year state policy and state SES measures described below. The institutional review board of the Public Health Institute, Oakland, CA, approved all surveys, respondent consent procedures, geocoding protocols and data security measures.

Measures

AHTO measures

Five AHTO items in these surveys, all due to someone else’s drinking, were initially developed by the World Health Organization’s Study of Community Response to Alcohol-Related Problems (Rootman and Moser, 1984) and then incorporated into Fillmore’s seminal ‘social victims of drinking’ survey of the city of Berkeley, CA (Fillmore, 1985). Subsequently, they were used in the 1989 Canadian Alcohol and Other Drug Survey (Eliany et al., 1992). These items were family problems or marital difficulties; being pushed, hit or assaulted by someone; having property vandalized; passenger with drunk driver; and vehicular crash, all due to another drinker. The three AHTO measures we used, two of which combine two items each, indicated experiencing: (1) aggression-related harms (physical assault or property vandalism), (2) harms related to drinking driving and (3) family or marital problems.

State alcohol policy measures

The Alcohol Policy Scale (APS) was measured for each of the survey years. The APS assessed the strength of state alcohol policy environments for reducing binge drinking, defined as the cumulative effect of multiple concurrent policies (Xuan et al., 2015b). The APS encompasses 29 policies in five policy domains: physical availability of alcohol (outlet density, alcohol beverage control agency, hours of sale restriction, local option, days of sale restriction, state monopoly and keg registration); underage drinking (minimum legal drinking age, prohibition of providing alcohol to minors and prohibition of false identification for purchasing alcohol); alcohol-impaired driving policies (administrative license revocation, graduated driver’s license, roadside sobriety checkpoints, zero tolerance, open container prohibited in motor vehicle and ignition interlock laws); social host or dram shop laws (prohibition of sales to intoxicated persons, dram shop, social host laws, minimum age of server, house party laws and responsible beverage service training); and alcohol pricing (taxes, wholesale price restriction and retail price restriction). The Alcohol Policy Information System (National Institute on Alcohol Abuse and Alcoholism) was the primary source for 13 of the 29 policies, and 18 additional data sources were used to collect further information about policies and their key provisions in the 50 US states and Washington, DC from 2000 to 2014 (Naimi et al., 2014).

The APS is based on ratings of efficacy and implementation of each policy (typically involving each policy’s statutory design, such as provisions making the policy broadly applicable, effective and enforceable), developed with assistance from a panel of policy experts (Naimi et al., 2014; Xuan et al., 2015a, c). Further details on the APS are provided elsewhere (Naimi et al., 2014). The APS scores demonstrated good construct validity to predict lower odds of binge drinking and alcohol-impaired driving among adults (Naimi et al., 2014; Xuan et al., 2015a, c) and less youth drinking (Xuan et al., 2015c). The APS scores have a theoretical range from 0 to 100%, with higher scores denoting a more stringent alcohol policy environment (Naimi et al., 2014; Xuan et al., 2015a). For interpretability of coefficients, we divided the state APS scores by 10.

State-level contextual variables

State binge drinking prevalence for each year was indicated by the proportion of adults engaging in any heavy episodic drinking (5+ drinks men/4+ drinks women per drinking occasion) in the past 30 days. These estimates came from the 2000–2014 Behavioral Risk Factor Surveillance System (Centers for Disease Control and Prevention, 2017). In past research, binge drinking rates that vary across states have been a reasonable proxy for drinking culture found to be associated with drinking behaviors (e.g. binge drinking among college students)(Nelson et al., 2005). A state-year-level covariate, the number of police officers per capita also was controlled for as a proxy for alcohol policy enforcement. Not captured by the APS, this policy dimension may confound the relationship between APS and individual-level AHTO.

As a proxy for state-level SES, we used state median family income from the 2000 US Decennial Census and the 2005–2015 American Community Surveys (United States Census Bureau, 2015). For interpretability of coefficients we divided state median family income by 10,000.

Individual-level covariates

Personal demographic characteristics included as controls were respondent’s gender, age, race/ethnicity (black, Hispanic, and others, vs. white), marital status (married/cohabitating vs. separated/divorced/never married), family income (below the federal poverty level vs. higher income), education (college/advanced degree vs. lower education) and living with a family member or having a blood relative with a history of alcohol problems.

Statistical analysis

Our preliminary analyses included descriptive statistics for sample characteristics and polychoric correlations among continuous and dichotomous study variables. Given the nested data structure (individuals clustered within each state across survey years), we used three-level random intercepts models implemented in Mplus 7.4 to examine whether associations of APS and state median income with the AHTO measures were mediated by state binge drinking rates. The mediation models for each of the three outcomes took the same general form, with the specifications described below.

In each multilevel model, Inline graphic denotes the ith person’s observed AHTO outcome in the jth survey year and kth state, with estimation of Inline graphic based on a probit model. Our level-1 model (individual level) is specified as Inline graphic, where Inline graphic is the estimated mean score of the outcome in the jth year and kth state, and Inline graphic represents a vector of the ith person’s individual characteristics such as gender, age, race, education, marital status and family problem-drinking history. The state-year-specific score (Inline graphic accounts for clustering of outcomes among individuals surveyed in the same year from the same state. Our level-2 model (state level) was specified as Inline graphic, where Inline graphic is the mean value for the outcome measure in the kth state, Inline graphic represents the state binge drinking rate (the mediator), Inline graphic represents the APS or state median income (tested simultaneously), and Inline graphic represents the level-2 covariate (state policing rate).

Our level-2 mediator model was specified as Inline graphic. In this model, the state-specific value Inline graphic accounts for clustering of mediator variables among different survey years in a given state, which is predicted by the level-3 model, specified as Inline graphic, where Inline graphic is the overall mean of the outcome measure and Inline graphic is the state-specific deviation from that overall mean. We assume all error terms Inline graphic, Inline graphic, Inline graphic and Inline graphic are normally distributed. For the mediation analysis, the product Inline graphic is the indirect effect of interest; this product of coefficients represents the indirect effect of each state-level predictor (APS or state median income) on the outcome through the association with the state-level mediator (binge drinking rate), with each estimated indirect effect adjusting for the simultaneous effect through the other hypothesized mediation pathway.

We used Bayesian estimation with the default priors and estimation settings (Asparouhov and Muthén, 2010) to estimate the three-level model and simulate the posterior distribution of parameters for testing the significance of the mediated effects. The convergence of the estimation was evaluated using the potential scale reduction (PSR) convergence criterion, with a PSR value not much larger than 1 considered evidence of convergence (Gelman et al., 2004). The posterior predictive P value (PPP) provided by Mplus was used to evaluate the overall model fit, with a PPP value less than .05 used as indication of poor fit (Asparouhov and Muthén, 2010).

RESULTS

Sample characteristics

Table 1 shows the demographic characteristics of the weighted sample (with unweighted sample sizes). Females were somewhat overrepresented. About 40% of the sample were younger than 40. Due to oversampling, black (19.2% unweighted; 11.6% weighted) and Latino/Hispanic (19% unweighted, 13% weighted) subsamples were relatively well represented. Whites were over half of the sample (57.1% unweighted; 68.8% weighted), with small numbers of other racial/ethnic minorities like American Indians and Asian/Pacific Islanders. About 29.3% had college/advanced degrees. About 18.8% had family incomes below the federal poverty level for the number of dependents in the household. A slight majority lived with a family member who was a problem drinker when growing up or had a blood relative with alcohol problems. Table 1 also provides the 12-month prevalence of each of the three types of harms attributed to others’ drinking. Prevalence of assault/vandalism (4.9%) and driving-related harm (5.3%) were slightly higher than family-related harm (4.0%). Overall, over 1 in 10 (10.9%) of US adults reported experiencing any of these harms in the prior year.

Table 1.

Sample characteristics (n; weighted percentage)

Characteristics n (%)
All 32,401 (100.0)
Gender
 Female 18,528 (51.8)
 Male 13,873 (48.2)
Age, years
 18–29 4977 (21.1)
 30–39 5247 (18.8)
 40–49 5854 (19.1)
 50–59 6049 (18.2)
 60 or older 9659 (22.8)
Race
 Asian Pacific Islander 652 (3.0)
 Black 6234 (11.6)
 White 18,505 (68.8)
 Hispanic 6153 (13.0)
 American Indian 472 (2.2)
 Others 385 (1.4)
Educational level
 Less than high school 4375 (13.0)
 High school graduation 8978 (28.6)
 Some college education 8325 (29.2)
 College or advanced degree 10,574 (29.3)
Family income
 Below federal poverty level 5814 (18.8)
 Federal poverty level or higher 26,587 (81.2)
Marital status
 Married/living with partner 16,998 (60.7)
 Divorced/separated/widowed 9292 (20.1)
 Never married 6111 (19.3)
Family history of alcohol problems 15,860 (52.2)
AHTO outcomes
Assault/vandalism 1152 (4.9)
Driving-related harms 1182 (5.3)
Family-related harms 912 (4.0)
Any of the three types of harm 2566 (10.9)
Open in a new tab

Table 2 shows low to moderate correlations among study variables (ranging from 0 to 0.56). Significant correlations between AHTO outcomes and individual- and state-level variables were in the directions we hypothesized, suggesting the potential for testing our path models. The PPP values for the three final mediation models were 0.485 (95% confidence interval (CI): −29.070, 23.570) for the model with assault/vandalism as the outcome, 0.461 (95% CI: −27.286, 32.897) for driving-related harm and 0.455 (95% CI: −27.258, 28.940) for family problem, all of which were very close to .5 and indicated excellent model fit (Muthen and Asparouhov, 2012).

Table 2.

Correlations among study variables

1 2 3 4 5 6 7 8 9 10 11 12 13
1. Malea 1.00
2. Age −0.11 1.00
3. College degree+b 0.08 0.02 1.00
4. Income <FPLc −0.12 −0.06 −0.38 1.00
5. Marrried/cohabitatingd 0.18 −0.05 0.16 −0.27 1.00
6. Family alcohol problemse −0.04 −0.10 −0.01 0.00 0.01 1.00
7. APS −0.02 0.06 0.03 0.00 0.00 0.03 1.00
8. State median income −0.05 0.11 0.11 0.02 −0.03 0.00 −0.03 1.00
9. State binge drinking 0.03 −0.03 0.02 −0.03 0.00 0.00 −0.38 0.31 1.00
10. State policing −0.01 −0.02 0.00 −0.01 0.00 0.00 −0.06 0.02 0.01 1.00
11. Assault/vandalismf 0.14 −0.33 −0.13 0.18 −0.11 0.18 −0.03 −0.02 0.07 0.00 1.00
12. Driving-related harmg 0.13 −0.33 −0.02 0.10 −0.16 0.20 −0.03 −0.02 0.07 0.00 0.56 1.00
13. Family-related harmh −0.12 −0.21 −0.02 0.10 0.06 0.30 −0.01 −0.05 0.01 0.00 0.53 0.48 1.00
Open in a new tab

* P < .05

aWith female as reference category

bWith no college degree as reference category

cWith income ≥ federal poverty level (FPL)

dWith divorced/separated/widowed/never married as reference category

eWith no family history of alcohol problems as reference category

fWith no experience of assault/vandalism as reference category

gWith no experience of driving-related harm as reference category

hWith no experience of family-related harm as reference category

Figures 1–3 show the unstandardized coefficients and 95% CIs for the paths of interest. (Full results including estimates for individual-level covariates that are not the focus of the current study are shown in Supplementary Tables A–C.) We found support for an indirect effect of the APS on assault/vandalism (indirect effectAPS = −0.042, P < .0001) and drinking-driving harms (indirect effectAPS = −0.037, P < .001), through reduced state binge drinking rates. The APS was negatively associated with state binge drinking rates in the model for assault/vandalism (b = −1.614, P < .0001) and in the model for driving-related harm (b = −1.605, P < .0001), which, then, was positively associated with these two outcomes (b = 0.026, P < .0001 for assault/vandalism and b = 0.023, P < .001 for driving-related harm). The APS had no significant effect, direct or indirect, on family/marital trouble due to another’s drinking.

Fig. 1.

Fig. 1.

Open in a new tab

Results of multilevel path analysis testing direct and indirect effects of state alcohol policy environment and state SES on assault/vandalism.

Fig. 3.

Fig. 3.

Open in a new tab

Results of multilevel path analysis testing direct and indirect effects of state alcohol policy environment and state SES on family-related harm.

Fig. 2.

Fig. 2.

Open in a new tab

Results of multilevel path analysis testing direct and indirect effects of state alcohol policy environment and state SES on driving-related harm.

State median income had significant indirect effects, through increased state binge drinking rates, on assault/vandalism (b = 1.001, P < .0001) and driving-related harm (b = 0.998, P < .001). It was negatively associated with family/marital problems (b = −0.048, P < .001), but this was not mediated by state binge drinking rates.

DISCUSSION

Findings of this study partially support our hypotheses regarding the relationships between the APS and state binge drinking rate with the three AHTO outcomes. We found support for indirect effects of APS on assault/vandalism and driving-related harm through reduced state binge drinking rates, as well as indirect effects of higher state median income on these two outcomes through increased state binge drinking rates. The APS had neither direct nor indirect effects on family/marital problems, but we found support for a direct effect of lower state SES on family/marital problems.

These findings suggest the ways in which two state-level contextual factors—state binge drinking prevalence and socioeconomic environment—could affect AHTO. We found evidence that state binge drinking prevalence mediated the effect of the APS, suggesting that a more stringent state alcohol policy environment could lower state binge drinking rates, which, in turn, could reduce risk for assault/vandalism and driving-related harms caused by other drinkers. We found that state SES may influence risk for different types of AHTO differently: higher state SES increased risk for assault/vandalism and driving-related harm through increased state binge drinking rate, while lower state SES elevated risk for family/marital problems due to another’s drinking.

Our findings that alcohol policies could be more effective in reducing externalizing behaviors of the perpetrator (i.e. assault/vandalism and driving-related harms) that are likely to take place in public places and may be easier to prevent by reducing binge drinking are well aligned with past research showing that individuals who reported more frequent visits to public drinking places were more likely to experience harms caused by another drinker (harassment, physical assaults and vandalism) (Rossow and Hauge, 2004). Encounters or confrontations with aggressive strangers under the influence of alcohol are more likely to happen in an environment where binge drinking is more prevalent, increasing risk for related harms such as aggression, sexual assault, unintentional injury and traffic crashes (Rossow and Hauge, 2004; Hughes et al., 2011). Our findings show how a more stringent alcohol policy environment could reduce risk for types of AHTO likely to occur in public places by reducing binge drinking.

These indirect effects of the APS on AHTO are consistent with prior studies that used the APS to demonstrate the effects of the alcohol policy environment on drinking in adults and youth (Xuan et al., 2015a, b, c). As the risk for AHTO varies depending upon environment and circumstances, which the perpetrator or the victim of AHTO has little control, the APS that captures the policy ‘environment,’ where multiple alcohol policies operate in a combined or interactive way (Xuan et al., 2015c), is a useful and appropriate tool for assessing policy effects on AHTO.

In light of literature that characterizes the family as largely impervious to outside interventions (Room, 1994), the absence of a significant association between the APS and family/marital problems is not surprising. The relationship between alcohol consumption and family problems is complex, with negative impacts occurring over an extensive period of time in several interlocking aspects such as stress and threat to the family (Berends et al., 2012). There is a solid body of literature documenting harms caused by heavy/problematic drinkers in the family (Casswell et al., 2011; Berends et al., 2012; Greenfield et al., 2015). Building on this work, further research to disentangle the intra- and extra-household components of family lives on which policies can intervene for prevention and remediation appears to be critical for addressing family problems due to others’ drinking (Dussaillant and Fernandez, 2015).

In identifying low state-level SES as a risk factor for family problems due to others’ drinking, our findings point to the importance of the macro-level socioeconomic environment. As households with fewer financial resources are more prone to family/marital problems due to a higher level of financial strain (Paat, 2011), it is highly plausible that harms due to another family member’s drinking are more likely to occur in low-SES states. These findings are consistent with prior research that found higher odds of family problems in disadvantaged neighborhoods (Karriker-Jaffe and Greenfield, 2014), suggesting a similar mechanism by which socioeconomic disadvantage increases AHTO on a more macro level.

The indirect effects of state SES on assault/vandalism and driving-related harm through increased state binge drinking, though unexpected, are worth noting. Binge drinking is predicated in part upon the ability to afford alcohol in large quantities (Jiang and Livingston, 2015), and thus may be more prevalent in high-SES states. To our knowledge, this study is the first that reports high prevalence of binge drinking in high-SES states and suggests the need for more vigorous alcohol policy efforts in such states.

Study strengths and limitations

The current study has important strengths. Much of alcohol policy research to date has examined alcohol policies without addressing the possible role of contextual factors in the relationship between policies and outcomes. This is an important gap addressed here. Alcohol policies do not independently bring about changes in alcohol consumption and related harms, but rather they operate in complex relationships with other contextual factors. This may particularly be the case for AHTO, as the risks arguably lie in the sociocultural and physical environments and circumstances that may govern drinkers’ behaviors or influence their consequences. Building on the extant literature, we examine relationships of state-level exposures and mediators with individual-level indicators of second-hand harms due to someone else’s drinking. In improving understanding of the roles two contextual factors simultaneously play, this study meaningfully contributes to the literature, helping to lay the groundwork for future policy and community efforts to address AHTO.

We acknowledge several study limitations. While we found indirect paths by which APS is associated with AHTO through reduced state binge drinking prevalence, these findings do not establish causal relationships. The relationship between state binge drinking prevalence and AHTO may be due to drinking cultures (i.e. attitudes, behaviors and common practices related to drinking) in high-binge states that are more tolerant of the drunken comportment or ‘bad behaviors’ resulting from drinking (Room, 2001). Alternatively, this relationship may be due to the greater prevalence in high-binge states of physical environments where a larger number of heavy drinkers are present. Either or both of these cultural and physical environments might be mitigated by more stringent alcohol policies, but due to the lack of information in our data, we were unable to explore these mechanisms.

Small effect sizes for the significant direct and indirect effects we estimated are also a limitation. To a certain degree, the small effect sizes are not surprising given the long and complex pathways between alcohol policies (or state living standards) and AHTO. Other sociocultural, economic and religious contextual factors, as well as different ways in which alcohol is made available or consumed, may be involved in these pathways, but because of the lack of information on these factors, they were not accounted for in our models. Further, due to the lack of statistical techniques to test sensitivity to unmeasured confounding for models like ours, we were unable to do so. Currently available methods for testing the sensitivity of mediation analyses to unmeasured confounding are based upon single-level logistic regression models to estimate the effect of a single exposure (VanderWeele, 2016) or multilevel mediation models for continuous outcome variables with all variables measured at level 1 (the individual level) (Tofighi and Kelley, 2016), and thus they are not applicable to our multilevel mediation analysis using probit models (that include both individual-level and state-level random error terms to account for clustering) and simultaneously testing two exposures (the APS and state median income).

Given the small effect sizes and potential confounders unaccounted for, the current study is akin to prior studies whose findings are more valuable for identifying patterns of association between societal-level predictors and alcohol consumption or related outcomes than for generating precise estimates (Wilsnack et al., 2009; Cook et al., 2014). The data limitations of the current study suggest that future research consider other potential mediators or moderators of the effects of alcohol policies on AHTO such as permissive drinking norms and practices, especially in high-SES states where binge drinking and AHTO due to externalizing behaviors of drinkers are more prevalent. Such additional studies would help inform and meaningfully guide intervention efforts to address AHTO.

CONCLUSIONS

In rigorous analyses using a nationally representative sample of US adults, we identified two contextual factors associated with alcohol’s harm to others. One contextual factor (state binge drinking prevalence) mediates policy effects on harms due to externalizing behaviors (assault/vandalism and drinking driving) of other drinkers. The other (state SES) influences the mediator and undermines policy efforts, while independently affecting the other outcome, family/marital problems, which take place a more private sphere. These findings uniquely contribute to the alcohol policy literature, helping to improve understanding of the roles contextual factors play in influencing policy effects and highlighting the need to better account for them in development and implementation of alcohol policies.

Supplementary Material

Supplemental_Tables_submitted_agaa073
Click here for additional data file. (20.1KB, docx)

FUNDING

This work was supported by the US National Institute on Alcohol Abuse and Alcoholism (NIAAA) [P50AA005595 to Kerr, R01AA022791 to T.K.G. and K.J.K.-J. (Multiple-PIs), RO1AA018377 to T.N., R01AA023376 to T.N. and Z.X. (Multiple PIs) and RO1AA026268 to T.N.]. Content is the responsibility of the authors alone and does not reflect official positions of NIAAA, the National Institutes of Health or sponsoring institutions.

CONFLICT OF INTEREST STATEMENT

There is no commercial or any other conflict of interest for the authors to declare with regard to the manuscript or the subject matter.

References

  1. Allamani  A, Voller  F, Decarli  A, et al. (2011) Contextual determinants of alcohol consumption changes and preventive alcohol policies: A 12-country European study in progress. Subst Use Misuse  46:1288–303. [DOI] [PubMed] [Google Scholar]
  2. Asparouhov  T, Muthén  B. (2010) Bayesian analysis using Mplus: Technical implementation. https://www.statmodel.com/download/Bayes3.pdf (15 July 2019, date last accessed).
  3. Babor  T, Caetano  R, Casswell  S, et al. (2010) Alcohol: No Ordinary Commodity Research and Public Policy. New York, NY: Oxford University Press. [Google Scholar]
  4. Babor  TF, Robaina  K, Jernigan  D. (2015) The influence of industry actions on the availability of alcoholic beverages in the African region. Addiction  110:561–71. [DOI] [PubMed] [Google Scholar]
  5. Berends  L, Ferris  J, Laslett  A-M. (2012) A problematic drinker in the family: Variations in the level of negative impact experienced by sex, relationship and living status. Addict Res Theory  20:300–6. [Google Scholar]
  6. Casswell  S, You  RQ, Huckle  T. (2011) Alcohol's harm to others: Reduced wellbeing and health status for those with heavy drinkers in their lives. Addiction  106:1087–94. [DOI] [PubMed] [Google Scholar]
  7. Centers for Disease Control and Prevention . (2017) Behavioral Risk Factor Surveillance System Annual Survey Data. Atlanta, Georgia: U.S. Department of Health and Human Services, Centers for Disease Control and Prevention.
  8. Cook  WK, Bond  J, Greenfield  TK. (2014) Are alcohol policies associated with alcohol consumption in low- and middle-income countries?  Addiction  109:1081–90. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Dussaillant  F, Fernandez  M. (2015) Alcohol's harm to others' well-being and health: A comparison between Chile and Australia. Alcohol Alcohol  50:346–51. [DOI] [PubMed] [Google Scholar]
  10. Eliany  M, Giesbrecht  N, Nelson  M, et al. (1992) Alcohol and Other Drug Use by Canadians: A National Alcohol and Other Drugs Survey (1989) Technical Report. Ottawa: The Unit. [Google Scholar]
  11. Fillmore  KM. (1985) The social victims of drinking. Br J Addict  80:307–14. [DOI] [PubMed] [Google Scholar]
  12. Fitzpatrick  K, LaGory  M. (2000) Unhealthy Places: The Ecology of Risk in the Urban Landscape. New York, NY: Routledge. [Google Scholar]
  13. Gelman  A, Carlin  JB, Stern  HS, et al. (2004) Bayesian Data Analysis. London: Chapman & Hall. [Google Scholar]
  14. Giesbrecht  N, Cukier  S, Steeves  D. (2010) Collateral damage from alcohol: Implications of 'second-hand effects of drinking' for populations and health priorities. Addiction  105:1323–5. [DOI] [PubMed] [Google Scholar]
  15. Greenfield  TK, Cook  WK, Karriker-Jaffe  KJ, et al. (2019) The relationship between the U.S. state alcohol policy environment and individuals' experience of secondhand effects: alcohol harms due to others' drinking. Alcohol Clin Exp Res  43:1234–43. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Greenfield  TK, Karriker-Jaffe  KJ, Kaplan  LM, et al. (2015) Trends in alcohol's harms to others (AHTO) and co-occurrence of family-related AHTO: The four US national alcohol surveys, 2000-2015. Subst Abus  9:23–31. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Hughes  K, Quigg  Z, Eckley  L, et al. (2011) Environmental factors in drinking venues and alcohol-related harm: The evidence base for European intervention. Addiction  106:37–46. [DOI] [PubMed] [Google Scholar]
  18. Jiang  H, Livingston  M. (2015) The dynamic effects of changes in prices and affordability on alcohol consumption: An impulse response analysis. Alcohol Alcohol  50:631–8. [DOI] [PubMed] [Google Scholar]
  19. Karriker-Jaffe  KJ, Greenfield  TK. (2014) Gender differences in associations of neighbourhood disadvantage with alcohol's harms to others: A cross-sectional study from the USA. Drug Alcohol Rev  33:296–303. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Kerr  WC. (2010) Categorizing US state drinking practices and consumption trends. Int J Environ Res Public Health  7:269–83. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Matrai  S, Casajuana  C, Allamani  A, et al. (2014) The relationships between the impact of alcoholic beverage control policies, selected contextual determinants, and alcohol drinking in Spain. Subst Use Misuse  49:1665–83. [DOI] [PubMed] [Google Scholar]
  22. Mezuk  B, Abdou  CM, Hudson  D, et al. (2013) “White Box” epidemiology and the social neuroscience of health behaviors: The environmental affordances model. Soc Mental Health  3. doi: 10.1177/2156869313480892. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Muthen  B, Asparouhov  T. (2012) Bayesian structural equation modeling: A more flexible representation of substantive theory. Psychol Methods  17:313–35. [DOI] [PubMed] [Google Scholar]
  24. Naimi  TS, Blanchette  J, Nelson  TF, et al. (2014) A new scale of the U.S. alcohol policy environment and its relationship to binge drinking. Am J Prev Med  46:10–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. National Institute on Alcohol Abuse and Alcoholism . Alcohol Policy Information System (APIS). https://alcoholpolicy.niaaa.nih.gov (10 October 2019, date last accessed). [PMC free article] [PubMed]
  26. Navarro  HJ, Doran  CM, Shakeshaft  AP. (2011) Measuring costs of alcohol harm to others: A review of the literature. Drug Alcohol Depend  114:87–99. [DOI] [PubMed] [Google Scholar]
  27. Nelson  TF, Naimi  TS, Brewer  RD, et al. (2005) The state sets the rate: The relationship among state-specific college binge drinking, state binge drinking rates, and selected state alcohol control policies. Am J Public Health  95:441–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Paat  Y-F. (2011) The link between financial strain, interparental discord and children's antisocial behaviors. J Fam Violence  26:195–210. [Google Scholar]
  29. Room  R. (1994) Alcohol, Drugs and the Family: An Overview. Toronto, Canada: Addiction Research Foundation. [Google Scholar]
  30. Room  R. (2001) Intoxication and bad behaviour: Understanding cultural differences in the link. Soc Sci Med  53:189–98. [DOI] [PubMed] [Google Scholar]
  31. Rootman  I, Moser  J. (1984) Community Response to Alcohol-Related Problems. Geneva, Switzerland: World Health Organization. [Google Scholar]
  32. Rossow  I, Hauge  R. (2004) Who pays for the drinking? Characteristics of the extent and distribution of social harms from others' drinking. Addiction  99:1094–102. [DOI] [PubMed] [Google Scholar]
  33. Tofighi  D, Kelley  K. (2016) Assessing omitted confounder bias in multilevel mediation models. Multivar Behav Res  51:86–105. [DOI] [PubMed] [Google Scholar]
  34. Trangenstein  PJ, Subbaraman  MS, Greenfield  TK, et al. (2020) Association between state-level alcohol availability and taxation policies on the prevalence of alcohol-related harms to persons other than the drinker in the USA, 2000-2015. Drug Alcohol Rev  39:255–66. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. United States Census Bureau . (2015) American Community Survey: American Fact Finder. Suitland-Silver Hill, MD: United States Census Bureau.
  36. VanderWeele  TJ. (2016) Mediation analysis: A practitioner's guide. Annu Rev Public Health  37:17–32. [DOI] [PubMed] [Google Scholar]
  37. Wilsnack  RW, Wilsnack  SC, Kristjanson  AF, et al. (2009) Gender and alcohol consumption: Patterns from the multinational GENACIS project. Addiction  104:1487–500. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Xuan  Z, Blanchette  J, Nelson  TF, et al. (2015a) The alcohol policy environment and policy subgroups as predictors of binge drinking measures among US adults. Am J Public Health  105:816–22. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Xuan  Z, Blanchette  JG, Nelson  TF, et al. (2015b) Alcohol policies and impaired driving in the United States: Effects of driving- vs. drinking-oriented policies. Int J Alcohol Drug Res  4:119–30. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Xuan  Z, Blanchette  JG, Nelson  TF, et al. (2015c) Youth drinking in the United States: Relationships with alcohol policies and adult drinking. Pediatrics  136:18–27. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplemental_Tables_submitted_agaa073
Click here for additional data file. (20.1KB, docx)

Articles from Alcohol and Alcoholism (Oxford, Oxfordshire) are provided here courtesy of Oxford University Press

RESOURCES