Data were drawn from the public use file of the 1999 National Household Survey on Drug Abuse
+(14). This survey is designed to provide national estimates of the use of licit and illegal substances by Americans 12 years old or older. The target population included residents of households, shelters, rooming houses, and dormitories, as well as civilians living on military bases. To increase the accuracy of drug use estimates, young people 12–25 years of age were oversampled. Survey respondents were interviewed about their use of a variety of substances, problems related to substance use and dependence, use of substance abuse services, and demographic characteristics.
The 1999 National Household Survey on Drug Abuse data collection method involved in-person interviews at the respondent’s place of residence and incorporated procedures developed to increase their cooperation and willingness to report drug use behavior honestly. These procedures combined computer-assisted personal interviews with audio computer-assisted self-interviews, which were designed to provide a private, confidential setting for answering sensitive questions.
This study of secondary data analysis of existing files was declared exempt from review by the Research Triangle Institute Institutional Review Board.
Of the 32,628 adults 18–64 years of age who were interviewed, 17,379 (53%) were women and 15,249 (47%) were men. Alcohol use in the past year was reported by 12,146 (70%) of the women and 11,754 (77%) of the men (χ2=89.89, df=1, p<0.001). Of the alcohol users, 40% were 18–34 years old, 9% were African American, 10% were Hispanic, 4% were other minority groups, 55% had at least some college education, 27% had never been married, and 42% had a family income of less than $40,000 (weighted percents). There were slight gender variations in education (10% of female alcohol users did not complete high school, compared with 14% of male users) and marital status (24% of female users had never been married, compared with 29% of male users).
+
Definitions of Variables
The 1999 National Household Survey on Drug Abuse
+(14) included a series of questions to assess alcohol dependence in the past year based on seven DSM-IV criteria: 1) tolerance, 2) withdrawal, 3) inability to cut down or stop alcohol use, 4) spending a great deal of time obtaining or drinking alcohol or recovering from its effects, 5) reducing or giving up important activities because of alcohol use, 6) using alcohol more often than intended, and 7) having physical or psychological problems because of alcohol use. Respondents who met three or more of these criteria in the previous year were classified as alcohol dependent.
Dependence on psychotherapeutic medication referred to dependence on any nonmedical use of prescription-type stimulants, sedatives, pain relievers, and tranquilizers, excluding over-the-counter drugs. Other illicit drug dependence referred to dependence on marijuana, cocaine, hallucinogens, heroin, or inhalants in the past year.
Use of alcohol treatment services was defined broadly as use in the past year of any service for alcohol-related problems at any location. We defined two types of treatment locations—specialty and nonspecialty—on the basis of respondents’ answers to a question about the main place where they received their most recent treatment services. A specialty setting refers to residential alcohol/drug rehabilitation facilities, nonresidential alcohol/drug rehabilitation facilities or programs, and mental health facilities or centers. A nonspecialty setting includes private doctors’ offices, hospitals, jails or prisons, court-sponsored programs, self-help groups, church-run programs, schools, workplaces, or community/city programs.
+
Perceived need for services
The subgroup of respondents who reported that they did not receive treatment or counseling for their alcohol use in the previous year were asked if they needed alcohol-related treatment or counseling during this period.
We examined the following respondent characteristics: age, race/ethnicity, education, marital status, total family income, and population density of the area where the respondent lived
+(14). For the analysis of perceived need for alcohol treatment services and use of treatment services, we also determined how many children younger than 18 years of age were living with respondents because of the potential association with use of treatment services among women
+(15).
The National Household Survey on Drug Abuse used a multistage probability sampling method to select participants. To ensure the representativeness of the sample, analysis weights were developed to adjust for variation in household selection, nonresponse, and poststratification of the selected sample to census data
+(16). Analyses were conducted with SUDAAN software to adjust for the National Household Survey on Drug Abuse’s complex multistage survey design
+(17). All percentages reported in this paper are weighted estimates; sample sizes are unweighted.
We began by calculating the rates and correlates of alcohol dependence among alcohol users. Among the subset of respondents who were alcohol dependent, we determined the extent and correlates of their use of alcohol treatment services and, in those who did not use these services, their perceived need for treatment. We then conducted logistic regression procedures to identify 1) the characteristics of alcohol dependence, 2) correlates of use of alcohol treatment services, and 3) correlates of perceived need for such services. We used a stepwise method to select variables to be included in the final logistic regression model and determined their respective odds ratios. An adjusted odds ratio denotes the strength and direction of the association between a response variable (e.g., use of treatment services) and a given correlate, while controlling for other potentially confounding characteristics. An odds ratio greater than one indicates a positive association between the two variables; an odds ratio smaller than one denotes an inverse association.
+
Descriptive Prevalence Estimates
Of all adults 18–64 years old (unweighted N=32,628), significantly fewer women than men met criteria for alcohol dependence in the previous year (
+Table 1). Among the subgroup of respondents who had used alcohol in the past year, women were significantly less likely than men to be classified as alcohol dependent (
+Table 1).
There was no difference between men and women in rate of comorbid drug dependence among alcohol-dependent adults (17% for women versus 16% for men). Alcohol-dependent women and men were equally likely to meet criteria for dependence on psychotherapeutic medications (5% of women and 2% of men) and other drugs (14% of women and 16% of men). Few alcohol-dependent adults used alcohol treatment services (13% of women and 12% of men). Of those who did not receive services, few perceived a need for treatment (8% of women and 10% of men). None of these differences between men and women was statistically significant.
Many users of alcohol treatment services also received treatment for drug problems in the previous year. Of those using alcohol treatment services (N=181), 42% of women and 51% of men reported using services for alcohol problems only. Alcohol-dependent women and men were equally likely to receive services at a specialty addiction/mental health treatment setting (6% of women and 4% of men) or nonspecialty treatment location (7% of women and 8% of men). We found similar results even when we restricted the specialty setting to substance abuse treatment facilities (4% of women and 3% of men) and compared it with all other non-substance-abuse treatment locations (8% of women and 9% of men).
+
Correlates of Past-Year Alcohol Dependence
We generated gender-specific logistic regression models to identify correlates of alcohol dependence among respondents who had used alcohol in the past year (
+Table 2). In our unadjusted analyses, age, education, marital status, family income, and drug dependence were each independently associated with alcohol dependence among female alcohol users. Even after we controlled for comorbid drug dependence, all variables remained significant. Female alcohol users 18–25 years old were twice as likely as female users 35–64 years old to be dependent on alcohol. Female alcohol users without a high school diploma (relative to college graduates), those who were white (relative to Hispanics), those with a family income of $75,000 or more (relative to those with incomes of $20,000–$74,999), and those who never had been married were more likely to report alcohol dependence. In addition, dependence on psychotherapeutic medications was associated with a 16-fold increase in odds of alcohol dependence, and other illicit drug dependence was associated with a nine-fold increase.
Among male alcohol users, being white (relative to being African American), never having been married, having a family income of less than $20,000, and dependence on psychotherapeutic medications and other drugs were associated with greater odds of alcohol dependence.
+
Correlates of Using Alcohol Treatment Services
The odds ratios for use of alcohol treatment services among alcohol-dependent adults are presented in
+Table 3. In the adjusted logistic regression, use of treatment services was associated with age, race/ethnicity, population density, and psychotherapeutic dependence. Women who were 18–25 years of age were less likely than women who were 35–64 years of age to receive services. African Americans (relative to whites), those living in metropolitan areas (relative to nonmetropolitan areas), and those with comorbid psychotherapeutic dependence were more likely to receive services.
Except for age (like younger women, younger men were much less likely than those 35–64 years old to use services), correlates of use of treatment services among alcohol-dependent men were different from those of alcohol-dependent women. The men who had one child living with them were four times as likely as those without children to use services. Alcohol-dependent men who also were dependent on illicit drugs were three times as likely as those without comorbid substance abuse to use services when their demographic characteristics were held constant.
To understand whether the use of alcohol treatment services varied by alcohol dependence symptoms
+(18), we used logistic regression analyses to examine their relationship. As shown in
+Table 4, after we held age, race/ethnicity, and education constant, tolerance, alcohol-related emotional or physical problems, and being unable to cut down on alcohol use increased the odds of using treatment services among women. For men, greater use of services was highly associated with alcohol-related emotional or physical problems but not with other symptoms.
+
Correlates of Perceived Need for Alcohol Treatment Services
Finally, we examined the characteristics associated with perceived need for alcohol treatment services among alcohol-dependent adults who did not use any alcohol treatment services in the previous year (
+Table 5). After holding comorbid drug dependence constant, we found that young women (those 18–25 years of age), women with a family income of $75,000 or more, and women who had two or more children living with them were less likely to report needing services. Alcohol-dependent women with comorbid psychotherapeutic dependence were seven times as likely as those without comorbid substance use to report needing services.
Among alcohol-dependent men who did not use any alcohol abuse service in the previous year, the perceived need for services was related to age but not to other characteristics examined. Even after holding comorbid drug dependence constant, we found that young men (those who were 18–25 years of age) were less likely than those who were 35–64 years old to report needing alcohol treatment.