Prevalence of Substance Use Among Racial & Ethnic Subgroups in the U.S.
The National Household Survey on Drug Abuse (NHSDA) is the primary source of data on the prevalence of substance use in the U.S. [ The National Commission on Marihuana and Drug Abuse sponsored the first two surveys, conducted in 1971 and 1972. The National Institute on Drug Abuse (NIDA) sponsored the NHSDA from 1974 to 1991. In October 1992, responsibility for conducting the NHSDA and preparing reports was moved to the Office of Applied Studies (OAS) within the newly created Substance Abuse and Mental Health Services Administration (SAMHSA).] This continuing cross-sectional survey of persons aged 12 and over has been conducted periodically since 1971. NHSDAs conducted prior to 1991 had sample sizes of less than 10,000 completed interviews. The sample size increased substantially in 1991 with oversampling in six large metropolitan areas and an expansion of the national sample. These two changes increased the overall sample size to 32,594 respondents. The 1991 research design was largely retained in the 1992 and 1993 surveys, with only modest changes in the questionnaire, sample stratification, and weighting procedures. There were 28,832 respondents in the 1992 NHSDA and 26,489 respondents in the 1993 NHSDA. Substantial revisions were made in both the questionnaire and research design beginning with the 1994 NHSDA.
The NHSDA provides estimates of the prevalence of substance use, consequences of substance use (e.g., problem substance use), and trends and demographic patterns of substance use and abuse. The "surveyed population," that is, the population for which estimates can be made using NHSDA data, is the civilian, noninstitutionalized population aged 12 and older in the United States. The surveyed population is believed to comprise more than 98% of the total population aged 12 and older in the United States. The subpopulations excluded are those residing in institutional group quarters (e.g., prisons, nursing homes, residential treatment centers), those with no permanent address (e.g., homeless people not in shelters), and active military personnel. NHSDA estimates of the prevalence, consequences, and patterns of substance use and abuse are published by the Substance Abuse and Mental Health Services Administration (SAMHSA) in its periodic Main Findings series (SAMHSA, 1993, 1995b, c).
The large total sample size of more than 87,000 respondents is the main reason why this report presents data from the combined 1991-93 NHSDAs rather than data from earlier or more recent NHSDAs. A large sample size is needed to obtain satisfactorily precise estimates of drug use prevalence in racial/ethnic subgroups that are small relative to the total population. A second reason for our choice of the combined 1991-93 NHSDAs is that each of these surveys obtained large oversamples from six major metropolitan areasChicago, Denver, Los Angeles, Miami, New York, and Washington, D.C.. More than 50% of all respondents in the combined 1991-93 surveysabout 47,000 of the 87,000 total respondentsresided in one of these six metropolitan areas. Because the six oversampled metropolitan areas contain large populations of many of the small racial/ethnic subgroups that are analyzed in this report, the oversampling had the effect of increasing the statistical precision of estimates for these subgroups. Both the large total sample size of the combined 1991-93 NHSDAs and the oversampling of large metropolitan areas helped to make the estimates presented in this report more accurate than similar estimates based on other data sources. As discussed in 1.3, previous analyses of racial/ethnic patterns of substance use in the U.S. have been greatly limited by small sample sizes. Starting in 1999, the sample size of the NHSDA will be increased to approximately 70,000 completed interviews per year. The sample will be designed to ensure that every state is represented. This expansion and redesign of the sample will facilitate analyses such as the ones included in this report, and make it possible to do more. In particular, it is expected that it will be possible to produce reliable estimates of substance use prevalence for some Asian subgroups.
As racial/ethnic minorities continue to increase their percentage of the U.S. population, understanding differences among racial and ethnic subgroups in substance use, risk factors, and odds of receiving treatment becomes increasingly important to policy. While most U.S. residents continue to be non-Hispanic white, the percentage who were Asian/Pacific Islander, black, Native American, or Hispanic increased from about 17% in 1980 to 20% in 1990 and, according to Census Bureau projections, will further grow to 28% in 2000, 32% in 2010, and 40% in 2030 (U.S. Bureau of the Census, 1992). According to the same projections, the fastest growing racial/ethnic subgroups are Hispanics and Asian/Pacific Islanders. Between 1980 and 1990, Hispanics increased their share of the U.S. population by almost 50%, from 6.4% to 9.0%, and Asian/Pacific Islanders nearly doubled their share, from 1.5% to 2.9%. Given that most individuals begin using drugs during adolescence (SAMHSA, 1996a), it is significant that racial/ethnic minorities are expected to account for more than one-third of U.S. students in kindergarten through twelfth grade by the year 2000 (U.S. Department of Education, 1994).
Data from the 1990 U.S. Census show that race/ethnicity continues to be important in American society (Harrison and Bennett, 1995). Among males at every educational level, ethnic minorities black, Native American, Asian, and Hispanic menearn substantially less than their white counterparts, usually 10 to 20 cents on each dollar of earnings (there are no comparable differences between the earnings of minority and white women.) While the percentage of blacks with high school degrees has increased rapidly in the post-World War II period, from less than 50% to more than 75%, blacks and other minorities, except Asians, still remain much less likely to receive bachelor's degrees than whites. [ The percentages of males 25 years and older in 1990 who completed college equal about 11% for blacks, 10% for Native Americans, 42% for Asian/Pacific Islanders, 10% for Hispanics, and 26% for whites. The corresponding figures for females are 12%, 9%, 32%, 8%, and 19% (Harrison and Bennett, 1995).] Poverty rates are more than three times as high among blacks, Hispanics, and Native Americans as among whites. [ Based on the 1990 Census data, the percentage of families in poverty in 1989 equaled about 24% among blacks, 27% among Native Americans, 12% among Asian/Pacific Islanders, 22% among Hispanics, and 7% among whites. While poverty rates declined by about 20% between 1969 and 1989, the differentials among racial/ethnic subgroups remained roughly the same (Harrison and Bennett, 1995).]
Given the socioeconomic disparities among racial/ethnic subgroups, it is not surprising that many minority subgroups have a greater prevalence of substance abuse than individuals in the total U.S. population (NIDA, 1994). Living in relatively unhealthful situations, as evidenced by high levels of poverty and illiteracy, substantially increases vulnerability to substance abuse. The prevalence of substance abuse is also generally higher in urban areas than in suburban or rural areas (SAMHSA, 1993, 1995b, 1995c). Minorities, particularly blacks and Hispanics, may have a higher prevalence of substance abuse than other Americans because the minorities are disproportionately concentrated in central cities.
The adverse consequences to racial/ethnic minorities of substance abuse include substantial health risks. For example, the death rate from alcohol-related causes is three times higher among blacks than among whites (Gary, 1986). In seven major U.S. cities, liver morbidity rates (cirrhosis and other liver diseases associated with alcohol use) were found to be about three times higher among black males aged 25 to 34 than among same-age white males (Herd, 1985). Similarly, cirrhosis mortality among Native Americans has been estimated to be at least four times the rate in the general population (Beauvais and LaBoueff, 1985). Substance abuse also increases likelihood of violence: Harvey (1985) estimated that black men were five times more likely than white men to be victims of homicide, and that at least half of both offenders and victims (regardless of race) had been drinking alcohol at the time of the homicide.
The purpose of this report is to analyze racial and ethnic patterns of substance use in the United States, using a more detailed classification of race/ethnicity than has been possible in previous reports that used the National Household Survey on Drug Abuse. The NHSDA has long been used to monitor trends in substance use among the largest racial and ethnic categories of the United States, including non-Hispanic blacks, non-Hispanic whites, Hispanics, and others. But the relatively small sample sizes of some racial/ethnic subgroups, such as Asian/Pacific Islanders, particular Hispanic national origin groups, and Native Americans, have made it difficult to precisely estimate the prevalence of substance use in these subgroups using data from a single survey year. Small sample sizes have also made it difficult to compare substance use patterns in racial and ethnic subgroups while statistically controlling for such important factors as gender, education, employment, and family income.
This report uses the pooled 1991, 1992, and 1993 NHSDAs to distinguish eleven subgroups. These include Native Americans, Asian/Pacific Islanders, non-Hispanic blacks, non-Hispanic whites, and seven subgroups of Hispanics defined based on national or geographical self-identification: Caribbean, Central American, Cuban, Mexican, Puerto Rican, South American, and Other Hispanic. The large sample size of the pooled 1991-93 NHSDAs, more than 87,000 respondents, also allows us to compare substance use prevalence across racial/ethnic subgroups while statistically controlling for sociodemographic characteristics, including age, gender, educational attainment, employment, and family income.
The particular questions that we explore in this report include:
·How do the prevalences of alcohol, cigarette, and illicit drug use vary across racial/ethnic subgroups?
·Are there important racial/ethnic differences in the age and gender patterns of substance use? For example, do some racial/ethnic groups tend to use drugs at earlier ages than others? Is substance use more heavily concentrated among males, or among females, in some racial/ethnic groups than in the total U.S. population?
·To what extent do the different social, demographic, and economic profiles of racial/ethnic subgroups account for their different patterns of substance use? For example, is residence in large metropolitan areas always associated with an increased prevalence of illicit drug use, regardless of the racial/ethnic category?
·Are racial/ethnic patterns of dependence on alcohol different from corresponding patterns of substance use?
·Which racial/ethnic subgroups appear to be most in need of illicit drug abuse treatment?
Previous research based on major national surveys has focused primarily on differences in substance use prevalence among three major racial/ethnic groupsHispanics, non-Hispanic Blacks, and non-Hispanic whites. More is known about racial/ethnic differences among adolescents than among adults in the U.S. because several major national surveys focus on substance use among adolescent respondents. [ This is an appropriate emphasis given that most individuals begin using drugs during adolescence and that adolescent drug use predicts drug use in later life (e.g., SAMHSA, 1996a).] The University of Michigan's Monitoring the Future (MTF) Survey (Johnston et al., 1995), a school-based survey excluding adolescents who were not enrolled, shows that Hispanics have the highest rates of past-month use for most substances in eighth grade, whites have the highest rates in twelfth grade, and blacks in grades eight through twelve are consistently lowest in the use of most substances. For example, based on the combined 1993-94 MTFs, past-month marijuana use was 6% among white eighth-graders and 18% among white twelfth-graders. The corresponding prevalences among Hispanics and blacks were 12% and 15%, and 5% and 13%, respectively. Black secondary school students were also low in binge drinking, with only 14% of black twelfth-graders reporting five or more drinks in the past two weeks, as compared with 32% of whites and 24% of Hispanics. Bachman et al. (1991) combined data from MTF surveys conducted between 1976 and 1989 to show that, during 1976-89, Asian American high school seniors had rates of alcohol, cigarette, and illicit drug use that were even lower than the rates of black seniors, while Native American high school seniors had rates of alcohol, cigarette, and illicit drug use that were higher than those of Hispanic or white seniors.
Another school-based survey, the 1993 Youth Risk Behavior Surveillance System (YRBSS, Centers for Disease Control and Prevention, 1995), suggests that access to illegal drugs may be an important factor affecting racial/ethnic differences in adolescent drug use: About 42% of Hispanic male high school students reported that they had been offered or sold an illegal drug on school property during the 12 months preceding the survey, compared with only 29% of white males and 20% of black males.
In contrast to MTF and YRBSS, NHSDA data for adolescent respondents are not restricted to adolescents who are enrolled in school. The differences between black adolescents and the other major racial/ethnic subgroups in illicit drug use are generally smaller in the NHSDA than in the school-based surveys. One possible explanation is that drug use is generally higher among high-school dropouts than among high-school students of the same age (Oetting and Beauvais, 1990), and blacks are relatively likely to drop out (Harris and Bennett, 1995).
In each of the three broad subgroups, male adolescents are more likely than female adolescents to be past or present substance users, and, at least for cigarettes and marijuana, the gender difference is more pronounced among blacks than among Hispanics or whites. For example, the percentages of males and females aged 12 to 17 reporting past-year marijuana use equaled 15% and 8% among blacks, compared with 14% and 13% among Hispanics, and 16% and 14% among whites (SAMHSA, 1997b, Table 3.5).
Data from the Hispanic Health and Nutrition Examination Survey (HHANES), a national probability sample of Hispanics aged 12 to 74 conducted during 1982 through 1984, suggested that patterns of substance use may differ significantly among major Hispanic subgroups. Mexican Americans and Puerto Ricans were more likely to be past or present drug users than Cuban Americansfor example, 42% of Mexican Americans and 43% of Puerto Ricans, but only 20% of Cuban Americans, reported having ever used marijuana in their lifetimes (NIDA, 1987). Moreover, HHANES suggested that substance use may be higher among Hispanics who are relatively assimilated into American society: Hispanics who preferred to be interviewed in English were two to three times as likely ever to have used drugs as Hispanics who preferred to be interviewed in Spanish. Among Hispanics who had ever used marijuana, Mexican Americans were inferred to have begun use at earlier ages than Puerto Ricans. Given the positive association between beginning drug use and dropping out of school (Chavez and Swaim, 1992), a possible explanation is a higher rate of school dropout among Mexican American adolescents than among Puerto Rican adolescents (see Chapter 3).
Reliable estimates of substance use among Native Americans in the total U.S. continue to be unavailable. (See Appendix B for a discussion of sampling strategies for improving the precision of estimates for Native Americans and other small racial/ethnic subgroups.) Data from a survey of Native American seventh through twelfth graders living on reservations suggest that substance use is generally high among Native American adolescents living on reservations (NIDA, 1995). For example, 83% of Native American high school seniors reported using marijuana in their lifetime, compared with 35% of non-Native American high school seniors in the 1993 MTF. Yet these statistics may not be representative of Native American secondary students in the U.S. because only about 30% of Native Americans live on reservations (Judkins et al., 1992).
Most research on substance use among Asian/Pacific Islanders is based on small surveys conducted by individual researchers or state agencies, with the most consistent finding being that, while the prevalences of substance use and abuse are not insignificant, Asian/Pacific Islanders use alcohol, cigarettes, and illicit substances less frequently than other Americans (Trimble et al., 1987; Murakami, 1989). Consistent with this finding, data from the pooled 1991-93 NHSDAs reported in Chapter 4 below indicate that, across all ages and all drug categories, Asian/Pacific Islanders have lower prevalence of substance use than Hispanics, non-Hispanic blacks, and non-Hispanic whites. Given the extensive ethnic diversity of the Asian/Pacific Islander category used in this report, these and other findings should be interpreted with caution; averages for the overall group may mask significant variations in the prevalence of substance use among ethnic subgroups that are not identified here.
The same data reported in Chapter 4 suggest that Native Americans generally have higher prevalence of substance use, for all drug categories except alcohol, than Hispanics, non-Hispanic blacks, and non-Hispanic whites (see also NIDA, 1995). However, even given the large sample size of the pooled 1991-93 NHSDA data, the number of Native American respondents is too small (n = 416) to support statistically significant findings.
Knowledge of racial/ethnic differences in substance use among adults in the U.S. depends especially heavily upon the NHSDA, because NHSDA is the only major federal survey that regularly collects detailed, confidential data on substance use prevalence among adults aged 18 and older in the U.S. household population. This report provides an unusually detailed presentation of racial/ethnic differences in substance use, both among adolescents and among adults, by pooling data from the 1991-93 NHSDAs.
Chapter 2 discusses the measurement of adolescent substance use, dependence on substances, and need for illicit drug abuse treatment in the 1991-93 NHSDAs. The chapter also discusses the measurement of racial/ethnic identifications in the 1991-93 NHSDAs, which used questionnaire items on race and Hispanic origin that closely resembled those of the 1990 U.S. Census. Chapter 2 shows that the estimated population percentages in each of eleven racial/ethnic subgroups based on the 1991-93 NHSDA are close to the corresponding percentage in the 1990 U.S. Census. Finally, Chapter 2 discusses the sociodemographic control variables and statistical methods used to analyze the effects of family structure on adolescent substance use.
Chapter 3 presents data on the demographic and socioeconomic characteristics of the eleven racial/ethnic subgroups: Native American, Asian/Pacific Islander, Hispanic-Caribbean, Hispanic-Central American, Hispanic-Cuban, Hispanic-Mexican, Hispanic-Puerto Rican, Hispanic-South American, Hispanic-other, non-Hispanic black, and non-Hispanic white. For all respondents aged 12 and older, the sociodemographic control variables include gender, region, population density, language of the NHSDA interview (Spanish vs. English), and family income. For respondents aged 12 to 17, the sociodemographic control variables also include family structure (residing in a household with two biological parents versus residing with fewer than two biological parents) and school dropout status (enrolled in school vs. not enrolled). For respondents aged 18 and older, the sociodemographic control variables also include educational attainment, marital status, employment status, and number of own children (none versus one or more). For each age group, the results show that racial/ethnic subgroups differ substantially in their sociodemographic profiles, which might account for some racial/ethnic differences in substance use.
Chapter 4 presents the estimated percentages of individuals in each of the eleven racial/ethnic subgroups who used alcohol, cigarettes, and marijuana and other illicit substances; who drank alcohol and smoked cigarettes heavily; who were dependent on alcohol; and who needed illicit drug abuse treatment. We examine the associations between race/ethnicity and substance use both in the total population aged 12 and older and within subclasses defined by age and gender.
Chapter 5 extends the analysis by investigating the associations between race/ethnicity and substance use, after controlling statistically for sociodemographic variables that were introduced in Chapters 2 and 3. Because of limitations of sample size in this detailed analysis, results for Native Americans are not presented and we are only able to distinguish four subclasses of Hispanics: Hispanic-Cuba, Hispanic-Mexico, Hispanic-Puerto Rico, and all other Hispanics.
The final chapter (Chapter 6) summarizes the findings and presents directions for future research.
Appendix A discusses sample weighting, statistical estimates and tests, and suppression of estimates due to low precision.
Appendix B discusses possible changes in the NHSDA sample design that would result in improved estimates of drug use prevalence in small racial/ethnic groups.
There are four important limitations of this report's estimates of substance use prevalence within racial/ethnic subgroups and comparisons of substance use prevalence across racial/ethnic subgroups. These are a) limitations in the measurement of substance use, b) limitations in the measurement of race/ethnicity, c) limitations in the sociodemographic control variables that are used to analyze racial/ethnic differences in substance use, and d) limitations in the precision of statistical estimates due to the small sample sizes of some racial/ethnic subgroups.
Measurements of substance use in the NHSDA are based on respondents self-report, and their value depends on respondents truthfulness and memory. Several studies have established the validity of self-report data (see Harrison, Haaga, and Richards, 1993; NIDA, 1992). The NHSDA procedures encourage honesty and recall; nevertheless, some under- and over reporting was very likely to occur.
Measurements of race/ethnicity in this report are based on respondents' self-identifications with a small number of closed-end response categories that were designed to be consistent with federal statistical policy (see Section 2.2). This measurement approach has four possible problems: First, self-reported racial/ethnic identifications may imperfectly gauge a complex social reality. For example, Isajiw (1974) discusses twelve different "dimensions" of ethnicity including common national or geographical origin, ancestry, culture, religion, race or physical characteristics, and languagesome of which may be imperfectly reflected in respondents self-reports. Research at the U.S. Bureau of the Census has shown that small changes in the questionnaire instruments that are used to collect data on racial/ethnic identifications can sometimes produce large shifts in the distribution of reported racial/ethnic identifications (Pinal and Lapham, 1992). Second, the questionnaire instruments that were used in the NHSDA to measure race/ethnicity did not allow a respondent to identify with more than one racial subgroup or with more than one Hispanic subgroup. Persons of mixed racial or ethnic backgrounds may have found these questions difficult to answer meaningfully. Third, consistent with federal statistical policy (U.S. Department of Commerce, 1978), this report classifies self-identified Hispanics as members of Hispanic-origin groups regardless of whether these individuals also identified themselves as Native American, Asian/Pacific Islander, black, or white. Some Hispanic persons who came to the U.S. from Latin American countries with high concentrations of indigenous peoples might be better classified as Native Americans than as Hispanics (Forbes, 1991). Fourth, with the exception of two questions on Hispanic origin (see 2.2), the NHSDA did not collect data on ancestry, which makes it impossible to distinguish such large national-origin subgroups as Filipino-Americans, Vietnamese-Americans, and Chinese-Americans (Barringer et al., 1993). In this report, non-Hispanic members of these and other Asian national origin subgroups are classified simply as "Asian/Pacific Islanders [ In conjunction with the expanded sample size beginning in 1999, data on country of ancestry will be collected for Asian and Pacific Islander racial/ethnic groups to provide the ability to estimate drug use prevalence for the largest subgroups.] ."
The control variables that are used to analyze racial/ethnic differences in substance use in this report include basic demographic variables such as sex and age; geographic characteristics such as region and population density; measures of socioeconomic status such as educational attainment, school enrollment status, employment status, health insurance coverage, receipt of welfare, and family income; measures of family structure and family background such as living arrangements, marital status, and number of own children; and characteristics of the interview itself (whether the interview was conducted in Spanish or in English). These control variables gauge social dimensions along which racial/ethnic subgroups vary and suggest possible explanations of racial/ethnic differences in substance use. For example, adolescents living in households with two biological parents are relatively unlikely to use illicit drugs (SAMHSA, 1996c), and the percentage of adolescents living in such households varies markedly across racial/ethnic subgroups (see Table 3.2 below). Yet many potential control variables that might be important in accounting for racial/ethnic differences in behavior could not be employed in the analyses of this report because these variables were not measured in the 1991-1993 NHSDAs. For example, foreign birth, foreign parentage, and citizenship/naturalization status might be important in interpreting patterns of substance use in racial/ethnic subgroups that experienced large in-migration flows in recent decades (Lieberson and Waters, 1988), but these variables were not measured in the 1991-93 NHSDAs.
A final limitation pertains to the sample sizes that are available in the combined 1991-93 NHSDAs. The large combined sample size of more than 87,000 respondents makes it possible to present more detailed descriptions of racial/ethnic patterns of substance use than was previously possible. However, the racial/ethnic subgroups that are analyzed in this report are still extremely heterogeneous. In part this heterogeneity results from the problems of measuring race/ethnicity, especially the absence of data on ancestry, which precludes analyzing subgroups of Asian/Pacific Islanders. Previous research makes it clear that in addition to sociodemographic differences, patterns of substance use and problems related to substance use vary quite a bit between the ethnic subgroups that make up this category (for an overview see Zane and Sasao, 1992). But because questions that would identify these subgroups were not asked in the 1991-93 NHSDAs, the findings presented here cannot distinguish these variations. It is possible that in some instances, they may cancel each other out to present a very different picture for the group as a heterogeneous whole. The heterogeneity of the racial/ethnic groups used in this report also reflects the relatively small sizes of many racial/ethnic subgroups in the U.S. population and the difficulties of designing the NHSDA sample to provide sufficient sample representation of small racial/ethnic subgroups. For example, "Native American" is also a heterogeneous ethnic category, encompassing such subgroups as Eskimos and Aleuts as well as many tribes of American Indians within the continental United States (Snipp, 1986). Yet the small sample size of Native Americans in the combined 1991-93 NHSDAs (n = 416) makes it impossible to include even the broad category "Native American" in the detailed analyses of Chapter 5.
Chapter 6 summarizes some of the important conclusions and policy implications of this report. Appendix B focuses on sample redesign options that might be used to increase the precision of drug use prevalence estimates for small racial/ethnic subgroups.
Return to Table of Contents (TOC)
This page was last updated on May 19, 2008.