key: cord-0896732-fu8cgs0c
authors: Karriker-Jaffe, Katherine J.; Chartier, Karen G.; Bares, Cristina B.; Kendler, Kenneth S.; Greenfield, Thomas K.
title: Intersection of familial risk and environmental social control on high-risk drinking and alcohol dependence in a US national sample of adults
date: 2020-09-29
journal: Addict Behav
DOI: 10.1016/j.addbeh.2020.106668
sha: fd1ccd1c941aae22851678d0660b4f39f053a35e
doc_id: 896732
cord_uid: fu8cgs0c

BACKGROUND: Effects of a family history of alcoholism may be moderated by area-level social control factors. We examine whether increased neighborhood alcohol availability (low social control environment) or increased presence of religious adherents in the county (high social control environment) interact with family history in relation to alcohol outcomes. METHODS: Weighted data from 12,686 adult drinkers (51% male; mean age 44; 80% White, 9% Black, 11% Hispanic) in three US National Alcohol Surveys were linked with data on area-level off-premise alcohol availability and adherence to religions with strong prohibitions against drinking. Family history density had four levels (family history negative, extended family only, first-degree relative(s) only, high family density). Dichotomous outcomes were past-year high-risk drinking and alcohol dependence. Logistic regression models with interaction terms assessed whether associations of family history with alcohol outcomes differed significantly by area-level social control. Stratified models assessed differences by sex and by race/ethnicity. RESULTS: In the full sample, effects of first-degree relatives and high family density on high-risk drinking strengthened as alcohol availability increased. This was replicated in the subsample of women and suggested in relation to dependence among men and Black drinkers. For White drinkers, higher religious social control reduced effects of first-degree relatives on high-risk drinking. CONCLUSIONS: Low social control—in particular, greater density of off-premise alcohol outlets—appears to exacerbate effects of a family history of alcoholism on high-risk drinking. Policy makers should consider differential benefits of decreasing alcohol availability for people from high-risk families to reduce high-risk drinking and alcohol problems.

Having biological relatives with alcohol problems increases an individual's risk for alcohol problems. 1, 2 The intergenerational transmission of alcohol use is attributed to multiple mechanisms, including genetic vulnerability 3 and learned attitudes about alcohol and drinking behaviors. 4, 5 Many people report a family history (FH) of alcohol problems: A recent national survey documented 21.0% of adults had a biological parent with a history of alcohol problems, and 37.6% had another biological relative with a history of alcohol problems. 6 Adults with familial alcohol problems are more likely to transition from at-risk drinking to alcohol use disorder (AUD). 7, 8 Secondhand effects of family members' alcohol problems also are significant, particularly for women 9 and children. 10 FH captures potentially heritable genetic and early family environment effects 11, 12 and is commonly measured by retrospective indicators of alcohol problems in parents or in first-degree relatives. 13 Some studies use a degree measure of FH density in both first-and second-degree relatives, 1,14 while other studies consider both the proportion of affected relatives and level of relatedness, [15] [16] [17] or presence of affected relatives across multiple generations. 1, 18, 19 Studies using these composite FH measures consistently find individuals from more densely-affected families at greater risk for hazardous alcohol use and AUD. In a US population sample, Dawson et al. 1 showed odds of AUD among drinkers were highest in individuals from high-density families with alcohol problems in both first-and second-or third-degree relatives (OR=2.79).

Odds of dependence also were higher in those with alcohol problems only in first-degree relatives (OR=1.91) compared to negative FH. 1 Studies on sex differences in the effect of FH in adult samples are relatively limited.

There is prior evidence that the effects of FH are stronger in males than females, particularly when the effects of childhood exposure to family drinking are controlled. 11 We know less about the influence of FH on adult drinking behaviors in non-White populations, as demonstrated by a 2012 meta-analysis in college students which found only four of 53 study samples were fully non-White, as were fewer than one in four participants across studies. 13 There also is a gap in current scientific knowledge about FH effects in later life, as most analyses of age-related genetic effects to date focus on adolescence and young adulthood, [20] [21] [22] [23] with few examples of studies that include adults at midlife or older ages. 24

In addition to liabilities transmitted within families, social environments contribute to alcohol use and AUD. Differential environmental exposures lead to sex and racial/ethnic differences in alcohol outcomes. [25] [26] [27] Prior models of joint genetic and environmental factors 28, 29 further suggest genetic effects on alcohol use are more pronounced in adverse environments (with low social control) and become muted in protective environments (with high social control).

Because there are few FH studies in adults that examine intersections with environmental context, we review evidence supporting social control mechanisms from twin and measured gene studies. Most of these studies have focused on proximal environmental exposures. 30 Alternatively, the current study focuses on two area-level indicators of social control: neighborhood alcohol availability and county-level adherence to religions with strong prohibitions against drinking.

Environmental social control processes decrease excessive alcohol use by regulating access to alcohol and helping to maintain social order through more restrictive social norms and involvement with prosocial institutions (including places of worship) that discourage alcohol use. 28, 29 In areas with greater alcohol availability, alcohol consumption is higher, 31, 32 and restricting the number of alcohol outlets in an area reduces alcohol problems. 33 Kendler et al. 34 found more pronounced genetic effects on alcohol consumption when access to alcohol was high, and this finding replicates at the area level. Specifically, one study of young adults found heritability of frequent drinking was stronger in areas with higher levels of alcohol availability than in areas without alcohol outlets, 35 suggesting a lack of social control strengthens expression of genetic predispositions to heavy alcohol use.

Other studies have shown religious participation and affiliated practices act as social controls on alcohol use. Twin studies show genetic effects on alcohol use are attenuated at increasing levels of religiosity. 36, 37 Chartier et al. 38 found genetic markers associated with alcohol metabolism had weaker effects on high-risk drinking as a function of increasing religious participation. This is the first study, to our knowledge, to examine the moderating effect of arealevel religiosity. Because religious denominations play a role in both promoting abstinence and reducing heavy drinking among those who choose to drink, 39 we extend prior studies by examining area-level religious adherence and FH in relation to both high-risk drinking and dependence among current drinkers, while controlling for individual-level religious affiliation.

Some studies have shown area-level effects on alcohol outcomes vary by sex. [40] [41] [42] [43] Other work suggests neighborhood areas may have stronger effects for Blacks/African Americans a and Hispanics/Latinos/Latinas b than Whites. 25, [44] [45] [46] Further, religious social controls on drinking may be stronger for women than for men, 39 and they may be stronger for Blacks, who are more likely to belong to religions that strongly discourage drinking alcohol, and for Hispanics, who show high levels of religiosity. 39 Thus, we examine interactions between FH and area-level social control factors separately by sex and race/ethnicity. As individuals age into early and midadulthood, alcohol-specific genetic factors increase in influence, 34, 47 with effects moderated by interpersonal social controls. 48 As such, it is important to examine joint effects of FH and arealevel social control in adult respondents across the lifespan. a Hereafter, Blacks. b Hereafter, Hispanics.

The current study, like Dawson et al., 1 examines FH across levels of relatedness in a sample of US adults. We examine associations of FH density with two alcohol outcomes, expecting drinkers with alcohol problems in close relatives c or from more densely-affected families d will be more likely to be high-risk drinkers and to have alcohol dependence. We also expect individuals residing in areas with greater alcohol availability to be more likely to be highrisk drinkers and to have dependence, while those residing in areas with greater adherence to religions that discourage or prohibit drinking will be less likely to be high-risk drinkers or alcohol dependent. To extend prior work, we examine whether relationships of FH with high-risk drinking and dependence vary by environmental context. We assess interactions of FH with our two social control measures. We expect environments characterized by low social control (greater alcohol availability) will be associated with a stronger relationship of FH with alcohol outcomes and environments associated with high social control (more religious adherents) will conversely be associated with weaker relationships. We explore differences in these effects by sex and by race/ethnicity as well. oversamples of Black and Hispanic respondents, and interviews were conducted in either English or Spanish. An introductory consent script was used, with the requirement for written proof of consent waived by the relevant Institutional Review Boards. Data were weighted to be population-representative at the time of data collection. More details on NAS methodology can be found elsewhere. 49 To ensure respondents had recent exposure to alcohol, the current sample was limited to 13,793 past-year drinkers (64.7% of weighted total sample), of whom 12,833 (91.8%) had data on their neighborhood of residence at the time of the survey. To allow subgroup analyses by race/ethnicity, we excluded 147 people who identified with any group other than White, Black, or Hispanic, as the other groups were too small for sub-analysis and too heterogeneous for combined analysis. The final, weighted analysis sample was 51.3% male and 80% White, 9% Black and 11% Hispanic (Table 1) . A majority (61.6%) of Hispanics were born in the US.

Approximately 33% of the sample came from each survey, with no differences in distribution across surveys by sex or race/ethnicity.

2.2.1 High-risk drinking e was defined based on daily and weekly drinking guidelines. 50,51 Drinkers who exceeded daily and/or weekly guidelines in the past year were classified as at-risk drinkers, with low-risk drinkers as the referent. 52 validated in prior NAS data. 53 A dichotomous variable indicated whether respondents reported at least one symptom in three or more of the seven criteria domains in the past year. f e These guidelines recommend no more than 4 drinks in any single day and no more than 14 drinks per week on average for men and no more than 3 drinks in a day and no more than 7 drinks per week on average for women. f The DSM-IV domains assessed were withdrawal; tolerance; drinking despite consequences; unsuccessful efforts to reduce drinking; drinking more than intended; time spent drinking/recovering; and giving up activities because of drinking. n, number of observations. a Sex differences, b Racial/ethnic differences. a,b p<0.05, aa,bb p<0.01, aaa,bbb p<0.001 for each test 1 Past-year drinking more than 3 or 4 drinks/day or more than 4 or 14 drinks/week (for women and men, respectively); 2 Past-year alcohol dependence symptoms in 3 or more of 7 domains; M, mean; SD, standard deviation

any of their blood relatives have ever been a "problem drinker or alcoholic", with affirmative responses followed by a checklist of relatives. For this study, FH was implemented as a four-level variable: negative history (referent); extended family only (grandparents, aunts/uncles, cousins);

first-degree relatives only (parents, siblings); and both first-degree and extended family (high FH density 

Analyses were conducted using Stata 64 to accommodate sampling and non-response weights adjusting for survey design. Preliminary analyses found FH was not significantly associated with the area-level social control variables. Associations of alcohol outcomes with FH and area-level social control were assessed using adjusted logistic regression models that included all demographic controls and indicators for survey year (using 2000 as referent).

Because the national samples were selected through RDD methods, multilevel modeling was not necessary. The multiplicative interactions were tested simultaneously, as correlations between the two area-level social control variables were modest. h Interactive effects were assessed using multiple-degrees-of-freedom contrasts of the joint effects of the interaction (adjusted Wald tests), as well as tests of marginal effects of FH across different levels of the area-level variables. Due to generally limited power for tests of interactions, 65 those interactions that did not reach p<0.10 were removed to improve interpretability of the main effects. 66 Preliminary analyses (see Supplemental Tables S1-S2) showed interactions of neighborhood alcohol availability with sex in relation to high-risk drinking, as well as an interaction of FH with race/ethnicity in relation to dependence, so we used stratified models for subgroup analyses, even though three-way interactions (such as sex X FH X outlet density) were not statistically significant. All demographic covariates and an indicator for survey year were included in the stratified models, although only selected coefficients are reported (full models available upon request).

High-risk Drinking. In the full sample (Table 2) , there was an interaction of FH with alcohol availability in relation to high-risk drinking. Contrasts showed the interaction was present for the combined group of respondents with either an affected first-degree relative or with both first-degree and extended family members affected (F(2, 18951)=3.06, p=.047). This group had a significant increase in high-risk drinking compared to negative FH, and these FH effects h The correlation between neighborhood alcohol outlet density and county-level density of adherents to religions with strong prohibitions on drinking was -0.13. became stronger as alcohol availability increased (Figure 1, left panel) . The average marginal effect of FH in only extended family members (compared to negative FH) was associated with increases in high-risk drinking at low levels of alcohol availability (Figure 1, right panel) , but this FH effect became non-significant as alcohol availability increased to 10+ outlets per square mile. However, the interaction of alcohol availability with only extended family was not significant Dependence. The interaction of FH of alcohol problems with alcohol availability was not present for alcohol dependence. In the reduced main effect model (Table 2) , FH density was increasingly associated with higher odds of dependence, as expected, but none of the arealevel variables were associated with alcohol dependence.

For female drinkers (see Table 3 ), there was an interaction of FH with alcohol availability in relation to high-risk drinking. Contrasts showed the interaction was significant for the group of women with either an affected first-degree relative or a high-density FH (adjusted For women there also was an interaction of FH with alcohol availability in relation to dependence. Contrasts showed this was significant for women with only extended FH of alcohol problems, but the pattern of effects suggested lower probability of dependence for these women if they lived in areas with high alcohol availability (Figure 4, left panel) .

For male drinkers, there were no interactions for high-risk drinking. In reduced main effects models (Table 3) , FH density was increasingly associated with higher odds of high-risk drinking, but none of the area-level variables were significant. For dependence, there was a marginal interaction of FH with alcohol availability. Contrasts showed this was just for men with only first-degree relatives with alcohol problems (Figure 4 , right panel).

For White drinkers (see Table 4 ), there was an interaction of FH with area-level religious adherence in relation to high-risk drinking. Contrasts showed the interaction was present for those with an affected first-degree relative (adjusted Wald test: F(1,20449)=5.30, p=.021). The average marginal effect of first-degree FH (compared to negative FH) on high-risk drinking was stronger at low levels of religious adherence in the county, and this effect became weaker as county-level religious adherence increased ( Figure 5 ). Stated another way, the risk of having one (or more) first-degree relatives with alcohol problems was exacerbated in areas with low religiosity and buffered in areas with high religiosity (Figure 6 ). Average marginal effects for respondents with only extended family or with high FH density (compared to negative FH), were not as strongly attenuated as county-level religious adherence increased. Note. Models also adjusted for age, sex, level of education, marital status, employment status, income and survey year. *** p<0.001, ** p<0.01, * p<0.05, + p<0.10. a Adjusted Wald tests were not statistically significant for the interaction of FH with religious density for dependence among White drinkers (F(3,20828)=0.36, p=.779); neither of the interactions of FH with liquor store density were statistically significant for White drinkers (both p>.10; detailed results available upon request), and they also were removed from the models to improve interpretability. b The interaction of FH with liquor store density for high-risk drinking among Black drinkers was not statistically significant (F(3,22005)=1.00, p=.391), and neither of the interactions of FH with religious density were statistically significant for Black drinkers (both p>.10; detailed results available upon request); these interaction terms were removed from the models. c None of the interactions of FH with liquor store density or FH with religious density among Hispanic drinkers were statistically significant (all p>.10; detailed results available upon request), so all interaction terms were removed. The interaction of FH and county-level religious adherence was not present in relation to dependence among White drinkers. In the reduced main effect model (Table 4) , FH was associated with higher odds of dependence, particularly for people with high FH density, but none of the area-level variables were associated with dependence.

For Black drinkers, there were no interactions of FH with the area-level variables in relation to high-risk drinking. In the reduced main effect models (Table 4 ), increasing FH density was associated with increased odds of high-risk drinking by Black drinkers. County-level religious adherence also was associated with higher odds of high-risk drinking. For Hispanic drinkers, there were no interactions for either high-risk drinking or dependence. In reduced main effect models (Table 4) , FH density was not significantly associated with high-risk drinking, and only the highest FH density level was associated with dependence. None of the area-level variables were significantly associated with high-risk drinking or dependence in Hispanic drinkers.

Familial and area-level factors were associated with high-risk drinking and alcohol dependence in this national sample of adults, with findings similar to prior research on social control of heavy alcohol use and associated risk behaviors. 28 For White drinkers, higher religious social control reduced associations of first-degree FH with high-risk drinking, similar to other work using individual-level religious practices to measure social control. [36] [37] [38] It was unexpected that these interactive effects did not emerge for women or for Blacks and Hispanics, given prior studies suggesting that religious social controls on drinking may be stronger for these demographic groups. 39 However, area-level religious effects were assessed when accounting for religious rules against drinking at the individual level, which was consistently associated with a lower likelihood of high-risk drinking across all sex and race/ethnicity subgroups. In post-hoc analyses, we tested whether individual-level religious prohibition on drinking moderated the associations of FH with the drinking outcomes, but there were no significant interactions in any of the three racial/ethnic subgroups for either outcome (all p>.10; detailed results available upon request). Further work to assess how individual-and area-level religion interact with each other, and with FH, in relation to adult drinkers' alcohol consumption patterns and problems would be informative.

As noted above, this study was conducted in a national sample of adults. Research has shown that, while genetic factors account for 28% of the variance in alcohol dependence in adolescence, in young adulthood (age 30 to 32) effects of genetic factors nearly double (58%) and remain above 50% into adulthood. 47 As such, in post-hoc analyses we checked for 3-way interactions (such as age X FH X outlet density) but these were not statistically significant when assessed several ways, including using the categorical age variable included in the adjusted models, using a continuous age variable, and using a dichotomous indicator separating respondents under age 40 from older respondents. It would be valuable for future studies to fully examine differences in FH effects and area-level social control on alcohol outcomes across the lifespan, including data from adolescents to older adults.

Our study used data from a large, nationally-representative sample of adult drinkers to examine associations of FH of alcohol problems with two different alcohol outcomes along the risk continuum. The large sample also permitted us to examine differences by sex and by race/ethnicity. These strengths are somewhat offset by a few limitations. First, despite the large sample, we may have had limited power to test some interactions in the stratified models.

Further, based on available survey measures, we used DSM-IV criteria for alcohol dependence, so our findings bear replication using DSM-5 criteria. With a large enough sample, this would allow distinction of risk factors for mild, moderate and severe AUD. Additionally, our measure of FH considered the presence of affected relatives across multiple generations; 1,18,19 however, we were unable to assess the proportion of affected relatives [15] [16] [17] or to parse out genetic influences from family environment effects. 11, 12 Most family method studies for AUD [71] [72] [73] show high specificity (correct indication of negative diagnosis) and moderate sensitivity (correct indication of positive diagnosis), but also acknowledge the potential for reporting bias based on the sex and age of both the informant and the subject, as well as the closeness of the relationship between the subject and informant. Future studies with more detailed FH measures would be informative in this regard. Finally, although we tested whether FH was associated with the social control indicators and did not find evidence for gene-environment correlation, i twin and adoption studies and longitudinal studies are better-suited to testing neighborhood selection effects.

Low social control-in particular, greater density of off-premise alcohol outlets-appears to exacerbate associations between family history of alcohol problems and high-risk drinking by adults. Sex-and racial/ethnic-specific differences did appear, suggesting that risk factors for AUD do not increase liability for alcohol use for all groups equally. Clinicians should consider not only family history but also area-level factors as they gather information and prepare treatment plans for their clients. Policymakers should consider benefits of decreasing alcohol availability, especially for people from high-risk families, to reduce high-risk drinking and associated problems. Retirement of outlet licenses for businesses lost during the recession due to the coronavirus 2019 (COVID-19) pandemic may provide a legislative opportunity to decrease alcohol availability.

Family history as a predictor of alcohol dependence

Development and vulnerability factors in adolescent alcohol use

Adjusted Wald tests from regression models testing associations of FH with logged values for alcohol outlet density and religious adherent density were not statistically significant

The heritability of alcohol use disorders: a metaanalysis of twin and adoption studies

Strategies for characterizing complex phenotypes and environments: general and specific family environmental predictors of young adult tobacco dependence, alcohol use disorder, and co-occurring problems

Disentangling the effects of parental drinking, family management, and parental alcohol norms on current drinking by Black and White adolescents

Distress and alcohol-related harms from intimates, friends, and strangers

Adult transition from at-risk drinking to alcohol dependence: the relationship of family history and drinking motives

Predictive value of family history on severity of illness: the case for depression, anxiety, alcohol dependence, and drug dependence

Alcohol's Secondhand Harms in the United States: New Data on Prevalence and Risk Factors

Alcohol's harm to children: findings from the 2015 United States National Alcohol's Harm to Others Survey

Estimating genetic and environmental effects of alcohol use and dependence from a national survey: a "quasi-adoption" study

Developmental evidence for at least two alcoholisms. I. Biopsychosocial variation among pathways into symptomatic difficulty

Does family history of alcohol problems influence college and university drinking or substance use? A meta-analytical review

Family history, alcohol use and dependence symptoms among young adults in the United States

Family history models of alcoholism: age of onset, consequences and dependence

Evaluating measures of family history of alcoholism: density versus dichotomy

How should we construct psychiatric family history scores? A comparison of alternative approaches from the Dunedin Family Health History Study

Family history of alcoholism and childhood adversity: joint effects on alcohol consumption and dependence

Transmission of alcohol use disorder across three generations: a Swedish National Study

The predictive power of family history measures of alcohol and drug problems and internalizing disorders in a college population

Alcohol Metabolizing Polygenic Risk for Alcohol Consumption in European American College Students

Family history density predicts long term substance use outcomes in an adolescent treatment sample

Developmental trajectory and environmental moderation of the effect of ALDH2 polymorphism on alcohol use

The influence of sociodemographic characteristics on familial alcohol problems: data from a community sample

Neighborhood disadvantage and adult alcohol outcomes: differential risk by race and gender

Less drinking, yet more problems: understanding African American drinking and related problems

Racial prejudice and unfair treatment: interactive effects with poverty and foreign nativity on problem drinking

The impact of gene-environment interaction on alcohol use disorders

Social context in gene-environment interactions: retrospect and prospect

Environmental influences on alcohol use: Informing research on the joint effects of genes and the environment in diverse U.S. populations

Hours and days of sale and density of alcohol outlets: impacts on alcohol consumption and damage: a systematic review

Evidence of a structural effect for alcohol outlet density: a multilevel analysis

Changing the density of alcohol outlets to reduce alcohol-related problems

Predicting alcohol consumption in adolescence from alcohol-specific and general externalizing genetic risk factors, key environmental exposures and their interaction

Neighborhood alcohol outlet density and genetic influences on alcohol use: evidence for gene-environment interaction

The moderating effect of religiosity on the genetic variance of problem alcohol use

The influence of religion on alcohol use initiation: evidence for genotype X environment interaction

Interactions Between Alcohol Metabolism Genes and Religious Involvement in Association With Maximum Drinks and Alcohol Dependence Symptoms

Religion and alcohol in the U.S. National Alcohol Survey: how important is religion for abstention and drinking? Drug Alcohol Depend

Area-level disadvantage and alcohol use disorder in northern Mexico

Chains of risk for alcohol use disorder: Mediators of exposure to neighborhood deprivation in early and middle childhood

Neighborhood disadvantage and adult alcohol outcomes: differential risk by race and gender

The Moderating Role of Gender in the Relationship Between Tobacco Outlet Exposure and Tobacco Use Among African American Young Adults

Alcohol-related problems among black, Hispanic and white men: the contribution of neighborhood poverty

Intersections of poverty, race/ethnicity, and sex: alcohol consumption and adverse outcomes in the United States

Poor, persecuted, young, and alone: Toward explaining the elevated risk of alcohol problems among Black and Latino men who drink

Social Relationships Moderate Genetic Influences on Heavy Drinking in Young Adulthood

National Institute on Alcoholism and Alcohol Abuse. Rethinking Drinking: Alcohol and your health

MD: National Institute on Alcoholism and Alcohol Abuse

Dietary Guidelines for Americans

American Psychiatric Association. DSM-IV: Diagnostic & Statistical Manual of Mental Disorders

Prevalence and correlates of DSM-IV and ICD-10 alcohol dependence

The roles of outlet density and norms in alcohol use disorder

Predicting the Population Health Impacts of Community Interventions: The Case of Alcohol Outlets and Binge Drinking

Alcohol outlets and binge drinking in urban neighborhoods: the implications of nonlinearity for intervention and policy

The Beverage Information Group. Liquor Handbook. Norwalk, CT: The Beverage Information Group Publications

The Beverage Information Group. Wine Handbook. Norwalk, CT: The Beverage Information Group Publications

In: Norwalk, CT: The Beverage Information Group Publications

Using ZIP code business patterns data to measure alcohol outlet density

Religious Congregations and Membership in the United States

Religious Census: Religious Congregations & Membership Study. Kansas City, MO: Association of Statisticians of American Religious Bodies

Stata Statistical Software: Release 15

Testing moderator and mediator effects in counseling psychology research

Multiple regression : testing and interpreting interactions

Neurobiology of addiction: a neurocircuitry analysis

Neighborhood density of alcohol outlets moderates genetic and environmental influences on alcohol problems

Interrelationship between family history of alcoholism and generational status in the prediction of alcohol dependence in US Hispanics

Alcohol availability and onset and recurrence of alcohol use disorder: examination in a longitudinal cohort with cosibling analysis

Biases in the diagnosis of alcoholism by the family history method

Accuracies and inaccuracies of the family history method: a multivariate approach

Comparison of direct interview and family history diagnoses of alcohol dependence