Abstract
Objectives:
American Indian adolescents who reside on or near reservations report higher levels of substance use than adolescents in other racial/ethnic groups. Little research has addressed patterns of use, which have important implications for prevention and treatment planning. The objective of our study was to describe substance use among a large, population-based sample of American Indian and white students who lived on or near reservations.
Methods:
We obtained data from 4964 students in grades 7-12 attending 46 schools on or near reservations throughout the United States during 4 academic years (2009-2013). Measures assessed current substance use for alcohol, heavy drinking, marijuana, cigarettes, inhalants, and other drugs. We used latent class analysis to identify patterns of substance use by grade group (grades 7-8 and grades 9-12) and race (American Indian or white).
Results:
For American Indians in both grade groups, we found 4 classes of substance use (in order of size): (1) nonusers; (2) marijuana and cigarette users; (3) alcohol, marijuana, and cigarette users; and (4) polysubstance users. For white students, we found 2 classes (nonusers and polysubstance users) among younger students and 4 classes (nonusers; alcohol, marijuana, and cigarette users; alcohol and cigarette users; and polysubstance users) among older students.
Conclusion:
We found significant differences in substance use patterns, especially at younger ages, between reservation American Indian students and white students attending the same schools. Combinations of substances used by American Indian adolescents were most likely to include marijuana, as compared with alcohol for white adolescents. Identifying subpopulations of users allows the design of interventions that will more efficiently and effectively address prevention and treatment needs of groups of individuals than would a one-size-fits-all approach.
Keywords: American Indian, substance use, reservation, latent class analysis
American Indian adolescents residing on or near reservations (hereinafter, reservations) have higher levels of substance use and initiate use earlier than other adolescents in the United States.1,2 Little research has addressed patterns of substance use—for example, whether a single drug or multiple drugs are used and in what combination—among American Indian adolescents. Variations in patterns have important implications for prevention, treatment, and public health outcomes. Compared with single drug use, use of more than 1 drug results in especially adverse outcomes, including psychological distress, violent behavior, sexual risk taking, noncompletion of school, and other problem behaviors.3–6 Understanding the various ways in which adolescents use and combine substances is essential for appropriately targeting prevention and treatment efforts.
Most studies that examine patterns of substance use indicate 1 class of nonusers and various numbers of classes consisting of use of 1 or multiple substances.7–11 Furthermore, patterns can differ by sex and race/ethnicity.8,9 Several studies examined patterns of substance use among American Indian adolescents, including one that investigated patterns among urban American Indian adolescents living in Arizona.12 That study identified 4 classes of substance use among students in grades 8, 10, and 12 at metropolitan schools: nonusers (69%); alcohol, tobacco, and/or marijuana users (17%); polysubstance users (6%); and users of tobacco, marijuana, and prescription drugs but rarely alcohol (4%). Another study that compared 2 American Indian reservation populations and a national sample, all aged 15-49, found 3 classes of past-year substance users for all 3 populations: abstainers, primarily alcohol users, and alcohol and drug users.13 A third study modeled lifetime substance use among adolescents in 5 American Indian communities in the West and found 4 classes of substance users: abstaining, predominantly alcohol users, predominantly alcohol and marijuana users, and plural substance users.14
Unlike studies of specific subpopulations of American Indian adolescents, our study assessed patterns of current substance use among a large, population-based sample of adolescents on reservations representing a broad range of tribal groups and geographic regions. To compare patterns of substance use by race, we also assessed patterns of substance use among white students who attended the same schools as the schools attended by the American Indian students in our study. White adolescents are the largest group of non–American Indian adolescents residing on or near reservations. Another objective of our study was to describe how combinations of substance use differed by age and sex. Such data could better inform the development and implementation of effective prevention and treatment programs.
Methods
Participants
Our study used cross-sectional data from a survey of students in grades 7-12 in 46 schools from 14 reservations across 4 academic years (2009-2013). Each academic year, we randomly sampled schools from a sampling frame of schools located within 25 miles of reservations or tribal lands and in which at least 20% of enrolled students were American Indian. Because the population of American Indians varies by region, we stratified the schools within the sampling frame by region. We randomly drew schools within each region so that the overall sample would reflect the regional distribution of American Indians living on reservations or trust lands based on US 2000 Census data.
School size averaged 125 students in grades 7 and higher, and on average, 80% of enrolled students (range, 66%-100%) completed the survey. The percentage of American Indian students enrolled in the sampled schools ranged from 21% to 100% (mean, 74%). The regional distribution of students was Northwest, 3.3%; Northern Plains, 51.0%; Southeast, 3.6%; Southwest, 30.7%; and Upper Great Lakes, 11.4%. No schools in the Northeast participated. Overall, 28 schools were located on reservation or tribal land; 84% of American Indian students and 38% of white students in the sample attended schools on the reservation.
We defined “white” as any student marking the “white” response and not marking the “American Indian” response for the single mark-all-that-apply question that asked about race/ethnicity; 2.2% of students also checked “Hispanic.” Thus, we used “white” in this study to include both non-Hispanic white and Hispanic white students. Identities of tribes and reservations were kept confidential. The Colorado State University Institutional Review Board approved all procedures, and for each participating school, we obtained the appropriate tribal and school board approvals.
Students completed the American Drug and Alcohol Survey, an instrument validated for reliability and validity among racial/ethnic minority young people.15,16 In total, 786 male and 714 female American Indian students in grades 7-8 (average [SD] age = 13.3 [1.2]) and 977 male and 1017 female American Indian students in grades 9-12 (average [SD] age = 15.9 [0.8]) completed the survey; 88 male and 93 female white students in grades 7-8 (average [SD] age = 13.3 [0.8]) and 659 male and 630 female white students in grades 9-12 completed the survey (average [SD] age = 15.9 [1.3]).
Procedures
All students in grades 7-12 in participating schools were eligible to complete the survey. Schools notified parents of survey administration via a media release and a letter mailed to all parents by the school that explained how to opt their child out of the survey. Fewer than 1% of students did not complete the survey because of lack of parental consent. A teacher or school staff member trained in human subjects procedures administered the survey during regular classroom hours. The survey took 20 to 45 minutes to complete. Students gave no identifying information, and they could leave any question blank that they did not wish to answer. Each school received a comprehensive report of their survey findings and $500 compensation for resources used to complete the survey process.
Measures
Six variables measured current substance use. Two variables measured alcohol use: frequency of alcohol use in the last month and frequency of consuming ≥5 drinks in a 2-hour period in the past 2 weeks (hereinafter, heavy drinking); for frequency, we used the following categories: 0 times = 0, 1 or 2 times = 1, and ≥3 times = 2. We used these same categories for a third variable, which measured frequency of marijuana use, and a fourth variable, which measured frequency of inhalant use in the last month. A fifth variable measured frequency of smoking cigarettes (0 = not at all; 1 = once in a while to 5 times daily; 2 = half-pack or more per day). The sixth variable measured other drug use: 1 = use in the last month of any other drugs, including tranquilizers without a prescription, cocaine, crack, other amphetamines, oxycontin, other narcotics, LSD, ecstasy, heroin, or methamphetamines, or 0 = no use of any of these drugs in the last month. We defined polysubstance use as multiple-substance use that included one other substance in addition to alcohol, marijuana, and cigarettes.
Analysis
We used latent class analysis (LCA) to identify groups of students with qualitatively distinct patterns of substance use. LCA is a technique for identifying subgroups of individuals who share similar characteristics within a population. These subgroups are called “latent classes” because they are otherwise unobservable. Latent classes may differ on predictive characteristics (eg, risk factors, protective factors) and on their response to treatment and prevention. LCA is a person-centered analysis (in contrast to a variable-centered analysis, such as factor analysis) that classifies individuals into unobserved (latent) clusters or classes according to their pattern of responses on indicator variables (here, substance use measures). Individuals in a given latent class are similar to each other with respect to item responses, whereas individuals across 2 classes are dissimilar with respect to item responses. Estimated parameters from LCA include the likelihood of any random person being in a latent class and the likelihood that an individual in a given latent class will indicate use of a given substance.
We conducted analyses using Mplus version 7.4; we accounted for the nested structure of the data (students within schools) by using a sandwich estimator to adjust standard errors.17,18 We first estimated unconditional models (ie, no covariate included) for number of classes from 1 to 5, estimating models separately by race (American Indian and white) and by grade group (7-8 and 9-12). We chose the number of classes based on the sample size–adjusted Bayesian information criterion (SABIC), with a lower SABIC indicating better fit, and on the interpretability of the classes. The bootstrap likelihood ratio test outperforms other measures of fit for determining the correct number of classes, but this test is not available for complex samples.19 However, 1 study found that SABIC performed similarly to the bootstrap likelihood ratio test in large samples.20
After we determined the number of classes for each grade group, we added sex, with boys as the reference class, as an auxiliary variable to assess its relationship to latent class. We examined odds ratios (ORs) of being in a given class to determine differences between boys and girls, with the nonuser class used as the reference group.
Results
Students in grades 9-12 had higher levels of substance use than students in grades 7-8 with 1 exception: American Indian girls and white girls in grades 7-8 had higher levels of inhalant use (12.4% and 5.6%, respectively) than their older counterparts (5.0% and 2.2%, respectively) (Table 1). Compared with white students, American Indian students had higher levels of current use of all substances, by grade and by sex, with 2 exceptions: alcohol use was higher among white boys (32.2%) and white girls (33.8%) compared with American Indian boys (28.0%) and American Indian girls (31.6%).
Table 1.
Prevalence of current substance use among American Indian and white students in grades 7-8 and 9-12 (N = 4964) in the United States, by sex, for academic years 2009-2013a
| Measure | American Indian, % (95% CI) | White, % (95% CI) | ||||||
|---|---|---|---|---|---|---|---|---|
| Grades 7-8 | Grades 9-12 | Grades 7-8 | Grades 9-12 | |||||
| Boys (n = 786) | Girls (n = 714) | Boys (n = 977) | Girls (n = 1017) | Boys (n = 88) | Girls (n = 93) | Boys (n = 659) | Girls (n = 630) | |
| Alcohol, no. of times in last monthb | ||||||||
| 0 | 84.0 (81.4 to 86.6) | 72.3 (69.0 to 75.6) | 72.0 (69.2 to 74.8) | 68.4 (65.5 to 71.3) | 86.0 (78.8 to 93.2) | 85.0 (77.7 to 92.3) | 67.8 (64.2 to 71.4) | 66.2 (62.5 to 69.9) |
| 1 or 2 | 11.3 (9.1 to 13.5) | 19.6 (16.7 to 22.7) | 16.4 (14.1 to 18.7) | 20.0 (17.5 to 22.5) | 8.4 (2.6 to 14.2) | 10.3 (4.1 to 16.5) | 20.2 (17.1 to 23.3) | 25.0 (21.6 to 28.4) |
| ≥3 | 4.7 (3.2 to 6.2) | 8.1 (6.1 to 10.1) | 11.6 (9.6 to 13.6) | 11.6 (9.6 to 13.6) | 3.6 (−0.3 to 7.5) | 4.7 (0.4 to 9.0) | 12.0 (9.5 to 14.5) | 8.8 (6.6 to 11.0) |
| Heavy drinking, no. of times in last 2 weeksc | ||||||||
| 0 | 88.9 (86.7 to 91.1) | 85.4 (82.8 to 88.0) | 82.1 (79.7 to 84.5) | 83.6 (81.3 to 85.9) | 96.2 (92.2 to 100.2) | 95.3 (31.0 to 99.6) | 84.4 (81.6 to 87.2) | 91.0 (88.8 to 93.2) |
| 1 or 2 | 6.5 (4.8 to 8.2) | 7.2 (5.3 to 9.1) | 9.9 (8.0 to 11.8) | 8.5 (6.8 to 10.2) | 0 | 0.9 (−1.0 to 2.8) | 9.3 (7.1 to 11.5) | 6.1 (4.2 to 8.0) |
| 3 | 4.6 (3.1 to 6.1) | 7.4 (5.5 to 9.3) | 8.0 (6.3 to 9.7) | 7.9 (6.2 to 9.6) | 2.8 (−0.6 to 6.2) | 3.8 (−0.1 to 7.7) | 6.3 (4.4 to 8.2) | 2.9 (1.6 to 4.2) |
| Marijuana, no. of times in last monthd | ||||||||
| 0 | 67.3 (64.0 to 70.6) | 66.8 (63.3 to 70.3) | 61.8 (58.8 to 64.8) | 61.8 (60.1 to 63.5) | 91.6 (85.5 to 97.7) | 90.6 (82.7 to 98.5) | 81.7 (79.2 to 84.2) | 88.0 (85.3 to 90.7) |
| 1 or 2 | 13.9 (11.5 to 16.3) | 13.6 (11.1 to 16.1) | 9.3 (7.5 to 11.1) | 11.2 (9.3 to 13.1) | 3.7 (−0.2 to 7.6) | 6.5 (1.5 to 11.5) | 6.5 (4.6 to 8.4) | 5.3 (3.6 to 7.0) |
| ≥3 | 18.8 (16.1 to 21.5) | 19.6 (16.7 to 22.5) | 28.9 (26.1 to 31.7) | 27.0 (24.3 to 29.7) | 4.7 (0.3 to 9.1) | 2.9 (−0.5 to 6.3) | 11.8 (9.3 to 14.3) | 6.7 (4.7 to 8.7) |
| Inhalants, no. of times in last monthe | ||||||||
| 0 | 94.7 (93.1 to 96.3) | 87.6 (85.2 to 90.0) | 95.5 (94.2 to 96.8) | 95.0 (93.7 to 96.3) | 98.2 (95.4 to 101.0) | 94.4 (89.7 to 99.1) | 97.4 (96.2 to 98.6) | 97.8 (96.7 to 98.9) |
| 1 or 2 | 4.1 (2.7 to 5.5) | 10.1 (7.9 to 12.3) | 2.5 (1.5 to 3.5) | 3.3 (2.2 to 4.4) | 0.9 (−1.1 to 2.9) | 5.6 (0.9 to 10.3) | 1.4 (0.5 to 2.3) | 1.4 (0.5 to 2.3) |
| ≥3 | 1.2 (0.4 to 2.0) | 2.3 (1.2 to 3.4) | 2.0 (1.1 to 2.9) | 1.7 (0.9 to 2.5) | 0.9 (−1.1 to 2.9) | 0 | 1.2 (0.4 to 2.0) | 0.8 (0.1 to 1.5) |
| Cigarettes, how oftenf | ||||||||
| Not at all | 65.0 (61.7 to 68.3) | 55.3 (51.7 to 58.9) | 55.5 (52.4 to 58.6) | 50.5 (47.4 to 53.6) | 88.8 (82.2 to 95.4) | 81.3 (73.4 to 89.2) | 76.3 (73.1 to 79.5) | 73.9 (70.5 to 77.3) |
| Once in a while to 5 per day | 26.9 (23.8 to 30.0) | 33.9 (30.4 to 37.4) | 27.9 (25.1 to 30.7) | 32.2 (29.3 to 35.1) | 8.4 (2.6 to 14.2) | 15.0 (7.7 to 22.3) | 16.8 (13.9 to 19.7) | 15.9 (13.0 to 18.8) |
| ≥Half-pack per day | 8.1 (6.2 to 10.0) | 10.8 (8.5 to 13.1) | 17.1 (14.7 to 19.5) | 17.3 (15.0 to 19.6) | 2.8 (−0.6 to 6.2) | 3.7 (−0.1 to 7.5) | 6.9 (5.0 to 8.8) | 10.2 (7.8 to 12.6) |
| Other drugsg | ||||||||
| No | 93.2 (91.4 to 95.0) | 90.0 (87.8 to 92.2) | 89.3 (87.4 to 91.2) | 90.0 (88.2 to 91.8) | 96.3 (92.4 to 100.2) | 97.2 (93.8 to 100.6) | 92.8 (90.8 to 94.8) | 95.5 (93.9 to 97.1) |
| Yes | 6.8 (5.0 to 8.6) | 10.0 (7.8 to 12.2) | 10.7 (8.8 to 12.6) | 10.0 (8.2 to 11.8) | 3.7 (−0.2 to 7.6) | 2.8 (−0.6 to 6.2) | 7.2 (5.2 to 9.2) | 4.5 (2.9 to 6.1) |
a Data source: The American Drug and Alcohol Survey.16
b In answer to the question, “How often in the last month have you had alcohol to drink?”
c In answer to the question, “During the last 2 weeks, how many times did you have 5 or more drinks in a 2-hour period?”
d In answer to the question, “How often in the last month have you used marijuana?”
e In answer to the question, “How often in the last month have you ‘sniffed’ or ‘huffed’ glue, gas, sprays, or anything like that to get high (do NOT include cocaine)?”
f In answer to the question, “How often do you smoke cigarettes?”
g In answer to the question, “Have you used any of these drugs to get high during the last month (tranquilizers, cocaine, crack, oxycontin, heroin, other narcotics, methamphetamine, other amphetamines, LSD, ecstasy)?” A positive response to any drug was coded as yes.
LCA Models
For both grade groups of American Indian students, 4-class models performed best, whereas for white students, a 2-class model for students in grades 7-8 and a 4-class model for students in grades 9-12 performed best (Tables 2 and 3).
Table 2.
SABICa goodness-of-fit values for a given number of classes and number of students in each class for American Indian students (n = 1500) and white students (n = 181) in grades 7-8 in a latent class analysisb of substance use among students living on or near reservations, United States, 2009-2013c
| By Race | 1-Class Model | 2-Class Modeld | 3-Class Model | 4-Class Modele | 5-Class Model |
|---|---|---|---|---|---|
| American Indian students | |||||
| SABIC | 10 319 | 9056 | 8884 | 8853 | 8865 |
| No. of students in class 1 | 1500 | 1134 | 929 | 786 | 754 |
| No. of students in class 2 | — | 366 | 126 | 213 | 111 |
| No. of students in class 3 | — | — | 445 | 432 | 488 |
| No. of students in class 4 | — | — | — | 69 | 77 |
| No. of students in class 5 | — | — | — | — | 70 |
| White students | |||||
| SABIC | 633 | 540 | 546 | — | — |
| No. of students in class 1 | 181 | 162 | 162 | — | — |
| No. of students in class 2 | — | 19 | 16 | — | — |
| No. of students in class 3 | — | — | 3 | — | — |
Abbreviations: —, does not apply; SABIC, sample size–adjusted Bayesian information criterion.
a The SABIC is a measure used for model selection among a set of models. A lower number represents a better fit.
b Latent class analysis is a technique for identifying subgroups of individuals within a population who share similar characteristics. These subgroups are called “latent classes” because they are otherwise unobservable. The technique was used to identify groups of students with qualitatively distinct patterns of substance use.
c Cross-sectional data were obtained from a survey of students in grades 7-12 in 46 schools from 14 reservations across 4 academic years (2009-2013). Students completed the American Drug and Alcohol Survey, Form AP-1.16
d The 2-class model performed best for white students.
e The 4-class model performed best for American Indian students.
Table 3.
SABICa goodness-of-fit values for a given number of classes and number of students in each class for American Indian students (n = 1994) and white students (n = 1289) in grades 9-12 in a latent class analysisb of substance use among students living on or near reservations, United States, 2009-2013c
| By Race | 1-Class Model | 2-Class Model | 3-Class Model | 4-Class Modeld | 5-Class Model |
|---|---|---|---|---|---|
| American Indian students | |||||
| SABIC | 15 019 | 13 059 | 12 811 | 12 690 | 12 696 |
| No. of students in class 1 | 1994 | 1383 | 1214 | 1221 | 1113 |
| No. of students in class 2 | — | 611 | 328 | 220 | 185 |
| No. of students in class 3 | — | — | 452 | 429 | 345 |
| No. of students in class 4 | — | — | — | 124 | 119 |
| No. of students in class 5 | — | — | — | — | 232 |
| White students | |||||
| SABIC | 7427 | 6460 | 6379 | 6336 | 6359 |
| No. of students in class 1 | 1289 | 1052 | 934 | 925 | 885 |
| No. of students in class 2 | — | 237 | 280 | 188 | 173 |
| No. of students in class 3 | — | — | 75 | 111 | 119 |
| No. of students in class 4 | — | — | — | 65 | 65 |
| No. of students in class 5 | — | — | — | — | 47 |
Abbreviations: —, does not apply; SABIC, sample size–adjusted Bayesian information criterion.
a The SABIC is a measure used for model selection among a set of models. A lower number represents a better fit.
b Latent class analysis is a technique for identifying subgroups of individuals within a population who share similar characteristics. These subgroups are called “latent classes” because they are otherwise unobservable. The technique was used to identify groups of students with qualitatively distinct patterns of substance use.
c Cross-sectional data were obtained from a survey of students in grades 7-12 in 46 schools from 14 reservations across 4 academic years (2009-2013). Students completed the American Drug and Alcohol Survey, form AP-1.16
d The 4-class model performed best for American Indian students and white students.
American Indian students
The 4 classes of substance use for students in American Indian grade groups were nonusers (786 of 1500 [52.4%] students in grades 7-8; 1221 of 1994 [61.2%] students in grades 9-12); alcohol, marijuana, and cigarette users (213 [14.2%] students in grades 7-8; 220 [11.0%] students in grades 9-12); marijuana and cigarette users (432 [28.8%] students in grades 7-8; 429 [21.5%] students in grades 9-12); and polysubstance users (69 [4.6%] students in grades 7-8; 124 [6.2%] students in grades 9-12) (Tables 2 and 3).
Among American Indian students, the class of marijuana and cigarette users had a similar profile to that of the class of alcohol, marijuana, and cigarette users, except for the alcohol items. For example, for students in grades 9-12 in the class of marijuana and cigarette users, the likelihood of current alcohol use was 0.35, whereas for heavy drinking, it was 0.02 (Figure 1). In contrast, these probabilities were 1.00 and 0.83, respectively, for the class of alcohol, marijuana, and cigarette users. Students in the class of alcohol, marijuana, and cigarette users in grades 7-8 were nearly twice as likely to use marijuana ≥3 times in the last month (probability = 0.50) than were students in the class of marijuana and cigarette users (probability = 0.28) and more than twice as likely to smoke a half-pack of cigarettes or more per day (probability = 0.28 and 0.12, respectively). However, for students in grades 9-12, the likelihood of more frequent marijuana and cigarette use was greater for students in the class of marijuana and cigarette users (probability = 0.60 for marijuana use and 0.36 for cigarette use) than for students in the class of alcohol, marijuana, and cigarette users (probability = 0.48 for marijuana use and 0.25 for cigarette use). Finally, the class of polysubstance users had probabilities greater than 0.90 for alcohol, heavy drinking, marijuana, and cigarette use; this class was most likely to indicate using substances ≥3 times in the last month. In addition, the probabilities of using other drugs among polysubstance users in both grade groups were greater than 0.60, and the probabilities of using inhalants were 0.53 for students in grades 7-8 and 0.36 for students in grades 9-12.
Figure 1.
Conditional probabilities of substance use for each estimated latent class for American Indian students in grades 7-8 and 9-12 living on or near reservations, United States, 2009-2013. Conditional probabilities of substance use among American Indian students in the last month are shown. A conditional probability is the probability that a randomly selected person in a class will have a nonzero response; a nonzero response indicates substance use in the last month. Each latent class is represented by a bar. Six variables measure substance use: (1) alcohol (probability of using alcohol ≥1 time in the last month), (2) heavy drinking (probability of having ≥5 drinks in a 2-hour period ≥1 time in the last 2 weeks), (3) marijuana (probability of using marijuana ≥1 time in the last month), (4) inhalants (probability of using inhalants ≥1 time in the last month), (5) cigarettes (probability of smoking cigarettes at least once in a while in the last month), and (6) other drugs (probability of using any other drug [eg, tranquilizers, cocaine, crack, oxycontin, heroin, other narcotics, methamphetamine, other amphetamines, LSD, or ecstasy] ≥1 time in the last month).
White students
The 2-class model for the 181 white students in grades 7-8 consisted of a class of nonusers (162 [89.5%] students) and a class of polysubstance users (19 [10.5%] students) (Table 1 and Figure 2). The probability of white students in grades 7-8 using substances 1 or 2 times in the last month was higher than the probability of using substances ≥3 times in the last month among students in the class of polysubstance users, with the exception of heavy drinking.
Figure 2.
Conditional probabilities of substance use for each estimated latent class for white students in grades 7-8 and 9-12 living on or near reservations, United States, 2009-2013. Conditional probabilities of substance use among white students in the last month. A conditional probability is the probability that a randomly selected person in a class will have a nonzero response; a nonzero response indicates substance use in the last month. Each class is represented by a bar. Six variables measure substance use: (1) alcohol (probability of using alcohol ≥1 time in the last month), (2) heavy drinking (probability of having ≥5 drinks in a 2-hour period ≥1 time in the last 2 weeks), (3) marijuana (probability of using marijuana ≥1 time in the last month), (4) inhalants (probability of using inhalants ≥1 time in the last month), (5) cigarettes (probability of smoking cigarettes at least once in a while in the last month), and (6) other drugs (probability of using any other drug [eg, tranquilizers, cocaine, crack, oxycontin, heroin, other narcotics, methamphetamine, other amphetamines, LSD, or ecstasy] ≥1 time in the last month).
The 4-class model for white students in grades 9-12 was (1) nonusers (925 of 1289 [71.8%] students); (2) alcohol, marijuana, and cigarette users (111 [8.6%] students); (3) alcohol and cigarette users (188 [14.6%] students); and (4) polysubstance users (65 [5.0%] students) (Table 3 and Figure 2). Students in the class of alcohol, marijuana, and cigarette users largely differed from students in the class of alcohol and cigarette users in that the probability of heavy drinking was 0 for those in the former class and 0.53 for those in the latter class, and the probability of marijuana use was 0.53 for those in the former class and 0.27 for those in the latter class. The probability of using alcohol ≥3 times in the last month was 0.44 among students in the class of alcohol and cigarette users but 0 for students in the class of alcohol, marijuana, and cigarette users. Finally, the class of polysubstance users had probabilities greater than 0.70 for alcohol use, heavy drinking, marijuana use, cigarette use, and other drug use, and the probability was ≥0.50 for using these substances ≥3 times in the last month.
Sex Differences by Class
Among American Indian students in grades 7-8, girls were more likely than boys to be in the class of polysubstance users (OR = 2.33; 95% confidence interval [CI], 1.34-4.06) and in the class of alcohol, marijuana, and cigarette users (OR = 2.34; 95% CI, 1.64-3.33). Among American Indian students in grades 9-12 and white students in grades 7-8, we found no differences in classification across sexes. Among white students in grades 9-12, girls were less likely than boys to be in the class of polysubstance users (OR = 0.52; 95% CI, 0.38-0.70).
Discussion
Using a population-based sample of American Indian and white adolescents who lived on or near reservations, our study estimated latent classes of substance use by race and grade group. We identified 4 classes of substance use for American Indian students in grades 7-12 and white students in grades 9-12 and 2 classes of substance use for white students in grades 7-8. The largest class for each group was nonusers, consistent with other LCA studies of adolescents, and the smallest class was polysubstance users.11,21,22 In addition, we found distinct patterns of substance use among American Indian adolescents compared with white adolescents, with relatively large classes of marijuana and cigarette users (>20%) among American Indian students in both grade groups but not among white students.
Our findings showed significant differences in patterns of substance use, especially at younger ages, between reservation American Indian students and white students attending the same schools. American Indian students in grades 7-8 had similar patterns to those of American Indian students in grades 9-12, with 4 classes found for each group. In contrast, patterns differed between white middle school students and white high school students. Among white students in grades 7-8, we found 2 classes at the extremes: a large group of nonusers (89.5%) and a small group of polysubstance users (10.5%). For white students in grades 9-12, four classes emerged: 2 extreme classes and 2 in-between classes (one of alcohol, marijuana, and cigarette users and one of alcohol and cigarette users). Our findings suggest that the substance use patterns of American Indian students are established earlier—by middle school—than they are for white students and that these patterns continue through high school. In contrast, a smaller percentage of white students than American Indian students in grades 7-8 used substances, but they appeared to be experimenting with multiple substances. This pattern changed for white students in grades 9-12: in high school, a greater percentage of students was using substances, and the 3 classes of users had relatively high probabilities of some type of alcohol use. The class of alcohol and cigarette users in our study was similar to classes of alcohol-only users and/or alcohol and tobacco users described in other studies.11 Interestingly, we found no class among white students that was similar to the marijuana and cigarette user classes among American Indian students in grades 7-8 and 9-12.
The differences between the white and American Indian classes in both grade groups could be due, in part, to more white students attending schools near a reservation rather than on a reservation. In general, substance use norms, substance availability, and other factors on a reservation may differ from those near a reservation. For example, 1 study found that reservation American Indian adolescents perceived less disapproval for marijuana use among peers and adults in the community than did white adolescents attending the same schools.23 On the other hand, these white students were exposed to many of the same sociocultural factors that were present on reservations and generally lived in areas where the socioeconomic status was below that of the general population of the United States; as such, the white students in our study were distinct from the general US population of white students. Whether differences between American Indian students and white students were due to differences in norms, availability, or other factors is a topic for future research.
Another study of American Indian adolescents, by Kulis et al,12 also found 4 classes of substance users. That study, which sampled American Indian students in grades 8, 10, and 12 in urban areas, found a large class of nonusers (69%) and a small class of polysubstance users (6%). The class of polysubstance users in our study of reservation adolescents was more likely than the class of polysubstance users in the study of urban adolescents to use all substances except other drugs (for which the likelihood was similar). The urban class of alcohol, tobacco, and marijuana users and the reservation class of alcohol, marijuana, and cigarette users were also similar, although the urban class was somewhat larger (17%) than the reservation classes (14.2% in grades 7-8 and 11.0% in grades 9-12). The class of alcohol, tobacco, and marijuana use in the urban study and the class of alcohol, marijuana, and cigarette use in our study reported high conditional probabilities of heavy drinking (>0.70) and, to a lesser extent, marijuana and cigarette use (urban >0.40; reservation >0.60). Finally, the reservation class of marijuana and cigarette use was similar to the urban “not alcohol” class, but the probabilities for marijuana and cigarette use were significantly higher for reservation adolescents than for urban adolescents, and the reservation classes were larger (28.8% in grades 7-8 and 21.5% in grades 9-12) than the urban class (4%). In summary, we found similarities in patterns of substance use between urban and reservation American Indian students, but reservation classes had higher probabilities of use within a class, had smaller numbers of nonusers and heavy alcohol users, and had larger numbers of nonalcohol users (ie, the class of marijuana and cigarette users).
Differences between urban and reservation classes may reflect differences between the environment of reservation adolescents and the environment of urban American Indian adolescents, especially as they reflect differences in availability of and norms surrounding use of alcohol and marijuana. Kulis et al12 noted that the small urban class of nonalcohol users may be related to favorable attitudes in American Indian communities toward abstinence, attitudes that may be due to high levels of awareness of the negative effects of alcohol on American Indians, an awareness also noted by May.24 For adolescents living on reservations, where abstinence among adults is often higher than abstinence among the general population and where restrictive alcohol policies may be in place, abstinence may lead, in part, to large classes of marijuana and cigarette users.25
Kulis et al12 found no differences by sex in latent classes among urban American Indian adolescents. In contrast, we found that among students in grades 7-8, American Indian girls were more likely than their male counterparts to be in the class of polysubstance users. This finding is consistent with the findings of another study, which showed that in early adolescence, Indigenous girls had higher rates of substance use than Indigenous boys, with differences dissipating by later adolescence.26 Our results suggest that reservation American Indian girls in middle school are at increased risk, compared with their male counterparts, for a pattern of using multiple substances beyond alcohol and marijuana.
Our results indicate that adolescents living on or near reservations, similar to other nonreservation adolescents, use substances in combination, and classes of users can be identified by using LCA. We found that the combinations of substances used by American Indian adolescents were most likely to include marijuana; in contrast, alcohol was more likely to be used among white adolescents. In addition, when compared by grade level, our American Indian sample had a smaller percentage of nonusers than the white sample, and for most substances, the probabilities of substance use were higher among American Indians. American Indian reservation adolescents were at particular risk for marijuana use, compared with white reservation adolescents, and white reservation adolescents had a greater risk for alcohol use. American Indian girls in grades 7-8 appeared to be initiating patterns of polysubstance use at a greater rate than their male counterparts.
Limitations
This study had several limitations. Although we used a large, geographically diverse sample of reservation schools, the sample was not a random sample of all reservation schools because participation was voluntary. Thus, our results may have been different had we obtained data from an entire population of reservation adolescents. However, to our knowledge, our sample is the largest such sample of reservation adolescents. The distribution of students across geographic regions was similar to the overall distribution of American Indians nationally, except that our sample consisted of a greater percentage of students located in the Southwest and Northern Plains. Our results may have been different had American Indian schools in the Northeast chosen to participate, but this region accounts for a small proportion of the complete sampling frame (2.0%). Because we administered surveys in schools, our data did not include adolescents who dropped out of school. Because of high rates of dropout among American Indian adolescents (67% graduation rate for the 2014-2015 school year27), our findings would likely have been different had we accounted for adolescents no longer in school.28 Caution is also warranted in interpreting findings for white students in grades 7-8 because of the small sample size, although all LCAs for this group converged successfully, and we encountered no statistical anomalies. Further study with larger sample sizes from this subgroup should be conducted.
Conclusion
Most substance use prevention programs designed for adolescents implement a primary prevention approach that targets the entire population, but emerging evidence indicates the effectiveness of secondary prevention approaches for this age group, in which adolescents are matched to a condition according to individual characteristics.29 Identification of subpopulations of users—for example, through LCA—allows the design of interventions that will efficiently and effectively address prevention and treatment needs of groups of individuals who share certain characteristics.30 Our study found that for our sample of adolescents living on or near reservations, substance use was not homogenous; thus, applying a one-size-fits-all approach may waste limited funds and resources for prevention and intervention.31 In addition, our study underscores that American Indian adolescents residing on reservations are at high risk for substance use, especially at an early age, and that likelihood is high for marijuana use in combination with cigarettes and, in some cases, other substances.
Our results have important public health implications for how best to target already limited resources for prevention and intervention. Particular attention should be directed to American Indian girls, who appear to be initiating patterns of polysubstance use at greater levels than their male counterparts. Further research is needed to determine what factors may lead to early establishment of substance use among these groups of reservation adolescents.
Footnotes
Declaration of Conflicting Interests: The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding: The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Research reported in this article was supported by the National Institutes of Health, National Institute on Drug Abuse (R01 DA003371-27A1).
References
- 1. Stanley LR, Harness SD, Swaim RC, Beauvais F. Rates of substance use of American Indian students in 8th, 10th, and 12th grades living on or near reservations: update, 2009-2012. Public Health Rep. 2014;129(2):156–163. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Stanley LR, Swaim RC. Initiation of alcohol, marijuana, and inhalant use by American-Indian and white youth living on or near reservations. Drug Alcohol Depend. 2015;155:90–96. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Kelly AB, Chan GC, Mason WA, Williams JW. The relationship between psychological distress and adolescent polydrug use. Psychol Addict Behav. 2015;29(3):787–793. [DOI] [PubMed] [Google Scholar]
- 4. Morley KI, Lynskey MT. Polysubstance use, mental health and high-risk behaviors: results from the 2012 Global Drug Survey. Drug Alcohol Rev. 2015;34(4):427–437. [DOI] [PubMed] [Google Scholar]
- 5. Kelly AB, Evans-Whipp TJ, Smith R, et al. A longitudinal study of the association of adolescent polydrug use, alcohol use and high school non-completion. Addiction. 2015;110(4):627–635. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Kokkevi A, Kanavou E, Richardson C, et al. Polydrug use by European adolescents in the context of other problem behaviours. Nordic Stud Alcohol Drugs. 2014;31(4):323–342. [Google Scholar]
- 7. Bohnert KM, Walton MA, Resko S, et al. Latent class analysis of substance use among adolescents presenting to urban primary care clinics. Am J Drug Alcohol Abuse. 2014;40(1):44–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Cranford JA, McCabe SE, Boyd CJ. Adolescents’ nonmedical use and excessive medical use of prescription medications and the identification of substance use subgroups. Addict Behav. 2013;38(11):2768–2771. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Khan MR, Cleland CM, Scheidell JD, Berger AT. Gender and racial/ethnic differences in patterns of adolescent alcohol use and associations with adolescent and adult illicit drug use. Am J Drug Alcohol Abuse. 2014;40(3):213–224. [DOI] [PubMed] [Google Scholar]
- 10. Marti CN, Stice E, Springer DW. Substance use and abuse trajectories across adolescence: a latent trajectory analysis of a community-recruited sample of girls. J Adolesc. 2010;33(3):449–461. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Tomczyk S, Isensee B, Hanewinkel R. Latent classes of polysubstance use among adolescents—a systematic review. Drug Alcohol Depend. 2016;160:12–29. [DOI] [PubMed] [Google Scholar]
- 12. Kulis SS, Jager J, Ayers SL, Lateef H, Kiehne E. Substance use profiles of urban American Indian adolescents: a latent class analysis. Subst Use Misuse. 2016;51(9):1159–1173. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Whitesell NR, Beals J, Mitchell CM, Novins DK, Spicer P, Manson SM. Latent class analysis of substance use: comparison of two American Indian reservation populations and a national sample. J Stud Alcohol. 2006;67(1):32–43. [DOI] [PubMed] [Google Scholar]
- 14. Mitchell CM, Plunkett M. The latent structure of substance use among American Indian adolescents: an example using categorical variables. Am J Community Psychol. 2000;28(1):105–125. [DOI] [PubMed] [Google Scholar]
- 15. Oetting ER, Beauvais F. Adolescent drug use: findings of national and local surveys. J Consult Clin Psychol. 1990;58(4):385–394. [DOI] [PubMed] [Google Scholar]
- 16. Tri-Ethnic Center for Prevention Research. The American Drug and Alcohol Survey, Form AP-1. Fort Collins, CO: Colorado State University; 2009. [Google Scholar]
- 17. Muthén LK, Muthén BO. Mplus User’s Guide. 7th ed Los Angeles, CA: Muthén & Muthén; 2015. [Google Scholar]
- 18. White HR. A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity. Econometrica. 1980;48(4):817–838. [Google Scholar]
- 19. Nylund KL, Asparouhov T, Muthén BO. Deciding on the number of classes in latent class analysis and growth mixture modeling: a Monte Carlo simulation study. Struct Equ Modeling. 2007;14(4):535–569. [Google Scholar]
- 20. Tein JY, Coxe S, Cham H. Statistical power to detect the correct number of classes in latent profile analysis. Struct Equ Modeling. 2013;20(4):640–657. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Snyder SM, Smith RE. Do physical abuse, depression, and parental substance use influence patterns of substance use among child welfare involved youth? Subst Use Misuse. 2015;50(2):226–235. [DOI] [PubMed] [Google Scholar]
- 22. Conway KP, Vullo GC, Nichter B, et al. Prevalence and patterns of polysubstance use in a nationally representative sample of 10th graders in the United States. J Adolesc Health. 2013;52(6):716–723. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Swaim RC, Stanley LR, Beauvais F. The normative environment for substance use among American Indian students and white students attending schools on or near reservations. Am J Orthopsychiatry. 2013;83(2, pt 3):422–429. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. May PA. Alcohol policy considerations for Indian reservations and bordertown communities. Am Indian Alsk Native Ment Health Res. 1992;4(3):5–59; discussion 71-75, 95-100. [DOI] [PubMed] [Google Scholar]
- 25. Herman-Stahl M, Chong J. Substance abuse prevalence and treatment utilization among American Indians residing on-reservation. Am Indian Alsk Native Ment Health Res. 2002;10(3):1–23. [DOI] [PubMed] [Google Scholar]
- 26. Walls M, Sittner Hartshorn KJ, Whitbeck LB. North American indigenous adolescent substance use. Addict Behav. 2013;38(5):2103–2109. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Camera L. Native American students left behind. US News and World Report. November 6, 2015 https://www.usnews.com/news/articles/2015/11/06/native-american-students-left-behind. Accessed March 26, 2018.
- 28. Swaim RC, Beauvais F, Chavez EL, Oetting ER. The effect of school dropout rates on estimates of adolescent substance use among three racial/ethnic groups. Am J Public Health. 1997;87(1):51–55. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Marsiglia FF, Ayers S, Gance-Cleveland B, Mettler K, Booth J. Beyond primary prevention of alcohol use: a culturally specific secondary prevention program for Mexican heritage adolescents. Prev Sci. 2012;13(3):241–251. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Hendriks V, van der Schee E, Blanken P. Matching adolescents with a cannabis use disorder to multidimensional family therapy or cognitive behavioral therapy: treatment effect moderators in a randomized controlled trial. Drug Alcohol Depend. 2012;125(1-2):119–126. [DOI] [PubMed] [Google Scholar]
- 31. Lamont AE, Woodlief D, Malone PS. Predicting high-risk versus higher-risk substance use during late adolescence from early adolescent risk factors using latent class analysis. Addict Res Theory. 2014;22(1):78–89. [DOI] [PMC free article] [PubMed] [Google Scholar]