problems with the police. Other consequences included on the survey related to health problems (e.g., interference with regular activities caused by drinking); involvement in accidents causing injury or property damage; and financial problems. Each item was rated by degree of severity. For example, having a spouse or significant other feel that a respondent should act differently when drinking was rated as less severe than a situation in which the respondent's drinking was threatening to break up the relationship.

Table 6 (Hilton 1991a) presents data from the 1984 National Alcohol Survey on the prevalence among current drinkers of social and other consequences of alcohol during the past year. Problem scores were obtained by summing severity ratings for all items. Individuals having scores between one and three were considered to have at least minimal problems; those having scores of four or more were considered to have at least moderate problems; and those having scores of

eight or more were considered to have high levels of problems. Hilton (1991a) noted that the most frequently reported consequences were problems with spouse and problems with friends, whereas accidents were quite rare. At least minimal levels of drinking-related belligerence were quite common due to the frequent occurrence of heated arguments, but the difference between minimal and moderate problem levels for belligerence was large, suggesting that actual fights occurred much less often than arguments.

A similar gap was observed between minimal and moderate health problems. Many people believed that alcohol consumption was harmful to their health, but relatively few had had an alcohol-related illness or had been told by a doctor to cut down on their drinking (Hilton 1991a). Overall, 14 percent of the men and 6 percent of the women who drank experienced moderate levels of consequences (Hilton 1991a).

The distribution of moderate and high levels of social and other alcohol-related consequences

* Less than 0.5 percent.

among current drinkers in the population was also reported (Hilton 1991a). Both moderate and high levels of consequences occurred most frequently among black men (19.2 percent and 13.2 percent, respectively), compared with 13.0 and 6.6 percent for white men; consequences for Hispanic men were lowest at 10.6 and 5.6 percent. Percentages of those reporting moderate and high levels of consequences were lower for women, and the racial or ethnic differences were not as pronounced: Moderate levels of consequences occurred among 6.2 percent of black women, 5.8 percent of white women, and 3.8 percent of Hispanic women; high levels occurred among 2.3 percent of white women, 2.1 percent of black women, and 0.9 percent of Hispanic women.

The proportions of respondents having scores of two or more on the measure of social or other drinking consequences in the 1979 and 1984 National Alcohol Surveys were compared (Hilton 1991c). No change in the prevalence of drinkingrelated consequences was observed in either sex. Essentially all groups of respondents, especially younger men and women, indicated less tolerant attitudes toward the acceptability of alcohol. One theory has suggested that drinking consequences should be sensitive to changes in popular attitudes toward alcohol; however, this finding does not support this view.

Among respondents who reported current drinking problems at the moderate level (Hilton 1991a), the percentage reporting at least two problems ranged from 75 percent (for those experiencing problems with friends or health problems) to 93 percent (those having problems with their spouse). Despite this tendency for respondents to have more than one problem, only weak relationships between problem items or problem categories were observed. Hilton (1991a) concluded that the failure to observe clustering suggests that the experience of alcohol-related problems is not confined to a small number of alcohol abusers or dependents.

These findings have significant implications for prevention strategies. As will be discussed in Chapter 12, Prevention of Alcohol-Related Problems, a narrow focus on individuals meeting the criteria for a diagnosis of alcohol abuse or dependence would fail to prevent many of the alcohol-related problems that occur among the much larger group of individuals whose drinking puts them at risk for problems even though they would not be diagnosed as having an alcohol use disorder (Kreitman 1986).

Alcohol Abuse and Alcohol Dependence

The ability to estimate the prevalence of alcohol use disorders in U.S. population samples is a relatively recent development. Two important studies in this area are the Epidemiologic Catchment Area (ECA) Study conducted in the early 1980s (Robins and Regier 1991) and the 1988 Alcohol Supplement to the National Health Interview Survey (NHIS) (Grant, Harford et al. 1991). Table 7 outlines differences between the ECA and the NHIS, including population differences and differences in the diagnostic criteria used.

The ECA study employed a structured interview (Robins et al. 1981) using criteria enabling the diagnosis of alcohol abuse and alcohol dependence based on the American Psychiatric Association's Diagnostic and Statistical Manual of Mental Disorders, Third Edition (1980), known as DSM-III. The 1988 National Health Interview Survey (Grant, Harford et al. 1991) employed criteria from a revised version of this diagnostic manual, known as DSM-III-R (American Psychiatric Association 1987). One effect of the manual change is that some individuals who would be defined as alcohol abusers according to DSM-III criteria are categorized as dependent under DSMIII-R criteria, but the overall rates of alcohol use disorders should not be affected (Cottler et al. 1991). For more information on the diagnosis of alcohol disorders, see Chapter 14, Diagnosis and Treatment of Alcoholism.

Age-, sex-, and ethnicity-specific prevalence rates for alcohol abuse and dependence (table 7) are drawn from the ECA (Helzer et al. 1991) and the 1988 NHIS (Grant, Harford et al. 1991). General trends in the prevalence of alcohol abuse and dependence according to age and sex in the total population are comparable in the two studies. Rates are higher among males than females, and they are highest among the youngest age cohorts (ages 18-29), decreasing with increasing age. The 1988 NHIS estimates of alcoholism in the total population are significantly higher than ECA estimates among persons under age 30 and women aged 30 through 44; however, estimates for the population older than this are very close, despite the differences in the two surveys noted above. It was estimated that in 1988 approximately 15.3 million individuals in the United States over age 18 met criteria for a DSM-III-R diagnosis of alcohol abuse or dependence or both (Grant, Harford et al. 1991).

Population Subgroups

Women

Past and present studies of alcohol use, in this and other cultures, have consistently found that men drink more than women and experience more adverse consequences of drinking (Wilsnack and Wilsnack 1991). For some time, there has been widespread concern that the gap between men's and women's drinking is narrowing. In a review of U.S. and Canadian survey research, Wilsnack and Wilsnack (1991) noted that drinking among women rose precipitously after World War II but appeared to have leveled off during the 1970s. Research in both countries over the past 20 years has revealed little evidence of major changes in consumption levels or alcohol-related problems among women in general, but some subgroup changes, based on such factors as age, ethnicity, employment, or marital status, may have occurred.

Past and present studies of alcohol use, in this and other cultures, have consistently found that men drink more than women and experience more adverse consequences of drinking.

In a related study, Hilton (1991d) used data from National Alcohol Surveys conducted between 1964 and 1984 to analyze trends in drinking over the past 20 years and found an increase in heavier drinking among individuals ages 21 through 34 in both genders. The proportion of these younger respondents who consumed 60 or more drinks per month increased among both men (15 to 23 percent) and women (4 to 7 percent), as did the proportions who had 5 or more drinks per sitting at least once a week. However, as indicated in table 4, heavier drinking among younger drinkers (ages 18 to 29) decreased from 21 to 18 percent among men and from 8 to 6 percent among women between 1983 and 1988 (Williams and DeBakey 1992). Thus, it may be that heavier drinking in both sexes peaked in the early 1980s and is now declining.

The male-to-female ratio for prevalence of alcohol abuse or dependence changed as a function of age in both the ECA and 1988 NHIS surveys (table 7). Grant, Harford et al. (1991) reported a striking effect in the 1988 NHIS data

set, where the rates for females more closely approximated the rates for males in the younger (18-29) age groups, with gender differences becoming more pronounced with increasing age; as previously noted, rates in the younger cohorts in general exceed rates for older cohorts. Thus, there has been speculation that alcohol abuse or dependence is increasing and that rates among women are increasing more rapidly than among men (Grant, Harford et al. 1991; Robins et al. 1988). It is notable, however, that in the 1988 NHIS, the higher rates for alcohol use disorders among younger groups were not accompanied by markedly higher rates of heavier drinking in these cohorts for either gender (tables 3A and 3B). Thus, there may be other, more complex explanations for the reported prevalence rates of alcohol abuse or dependence.

Data from the Behavioral Risk Factor Surveillance System documented a decline in alcohol use in the month prior to the survey by pregnant women from 32 percent in 1985 to 20 percent in 1988; however, the median number of drinks consumed each month remained constant among pregnant women who drank (Serdula et al. 1991). Alcohol use also did not change among less educated or younger (under age 25) women. Data from 1988 revealed that alcohol use among pregnant women was highest among those who smoked (37 percent) or were unmarried (28 percent). Thus, although results of this ongoing survey suggest an overall trend of reduced alcohol consumption among pregnant women in the United States, it is clear that further efforts should focus on reducing alcohol use among pregnant women who may already be at high risk for a poor pregnancy outcome (Serdula et al. 1991).

Gender differences associated with the nature of alcoholism have been investigated by examining male-to-female ratios for items used in structured alcoholism diagnostic interviews (Dawson and Grant in press), confirming earlier findings that gender differences exist in reporting alcoholrelated problems (Robbins 1989). Dawson and Grant (in press) found that men responded more frequently than women to all 41 items being assessed. The smallest gender differences were reported for questions dealing with physical or psychological effects of drinking and with powerlessness; these results are consistent with reports indicating that women tend to be vulnerable to the physical effects of alcohol (Dawson and Grant in press). The greatest gender differences were for items that addressed hazardous use,

legal problems, and having a spouse threaten to leave. Because these items are associated with higher alcohol consumption levels, the greater prevalence of high consumption among men may account for the male excess of these problems.

The national survey of women's drinking conducted in 1981 and its followup in 1986 represent a major contribution to our understanding of the etiology of problem drinking among women (Wilsnack et al. 1991). Among women identified in 1981 as nonproblem drinkers, young age, cohabitation, and lifetime use of drugs other than alcohol appeared to be the most consistent predictors of onset of problem drinking in the 1986 followup. Sexual dysfunction was the most consistent predictor of persistent (chronic) problem drinking. In another study, childhood sexual abuse was found to predict problem drinking among women (Wilsnack 1991). Other factors, such as being employed part-time, never marrying, or experiencing recent depression, were also predictive of chronic problem drinking (Wilsnack et al. 1991). These findings suggest that the personal and social factors predictive of the onset of problem drinking differ from those related to its continuation. Interestingly, divorce and separation were associated with lower levels of subsequent alcohol dependence among problem drinkers. The authors (Wilsnack et al. 1991) concluded that the effect of relationships on drinking behavior may have to do with the stress they engender: For example, divorce or separation might represent relief of marital stress, and cohabitation might create stress because of the lack of institutional support for the relationship.

Adolescents

Trends in drug and alcohol use among adolescents have been monitored since 1975 via annual national surveys of approximately 17,000 high school seniors. These surveys make up the Monitoring the Future study conducted by the Institute for Social Research at the University of Michigan. Beginning in 1976, the surveys were supplemented by followup studies; therefore, the most recent reports include data on young people between the ages of 18 and 30 (Johnston et al. 1991a, 1991b).

By far, alcohol was the drug used most often by 1990 high school seniors. Although most high school seniors cannot legally buy alcoholic beverages, 90 percent of them had tried alcohol,

compared with 64 percent who had tried cigarettes and fewer still who had tried other drugs; 32 percent of high school seniors reported occasions of heavy drinking (five or more drinks in a row) during the 2 weeks prior to the interview (Johnston et al. 1991a). Despite the fact that alcohol use is still high, drinking by high school students has been declining, as it has among U.S. adults. The monthly prevalence of alcohol use (i.e., the percentage of respondents using alcohol in the past 30 days) among seniors has gradually declined, from 72 percent in 1980 to 57 percent in 1990. Daily use declined from a peak of 6.9 percent in 1979 to 3.7 percent in 1990; and the 32-percent prevalence of heavy drinking in the prior 2 weeks mentioned above represents a decline from 41 percent in 1983.

According to Johnston et al. (1991a), high school students who are bound for college are less likely to report having had five or more drinks in a row during the prior 2 weeks than those who are not college bound, but college students are more likely than their peers of the same age to report occasions of heavy drinking (41 percent versus 33 percent, respectively). In contrast, daily drinking was more prevalent among those not in college than among college students (4.9 percent versus 3.8 percent, respectively).

Among women identified in 1981 as nonproblem drinkers, young age, cohabitation, and lifetime use of drugs other than alcohol appeared to be the most consistent predictors of onset of problem drinking in the 1986 followup.

By far, alcohol was the drug used most often by 1990 high school seniors.

Male high school seniors had a markedly higher prevalence of occasions of heavy drinking than female high school seniors (24 percent for females versus 39 percent for males in 1990), but this difference has been diminishing gradually during the past decade. Male college students and young adults also tend to drink substantially more than females in these groups. For example, in the 4 weeks prior to the study, 50 percent of college males had five or more