Position of Gateway Drugs in the Spectrum of Adolescent Drug-Use Initiation in Indiana: Relevance of Market Basket Analysis of Data Mining in Detection of Common Substance-Use Initiation Sequences Ahmed H. YoussefAgha*(PhD), Wasantha P. Jayawardene*†(MD) * Department of Applied Health Science, School of Health, Physical Education, and Recreation, Indiana University Bloomington, Indiana, United States; † Corresponding Author Abstract Background: Empirical evidence confirms that the majority of individuals’ legal or illicit drug use begins after use of cigarettes, alcohol, or cannabis (marijuana) hence they are known as the gateway drugs. Objectives: To study the sequences of initiating use of gateway drugs and other drugs among adolescents, especially among 12th-grade students, in Indiana. Methods: Data from the 1993-2009 Annual Surveys of Alcohol, Tobacco, and Other Drug Use by Indiana Children and Adolescents was used for the current study. Market Basket Analysis was used to identify sequences and association among drug- use. Results: The majority of adolescents initiated drug use with gateway drugs followed by combining them with prescription drugs. Conclusion: Market Basket Analysis reveals that the gateway theory is applicable to Indiana adolescents because many of them initiate drug use with alcohol, tobacco, or marijuana or with a combination of them followed by use of prescription drugs. Keywords: Gateway drugs, Prescription drugs, Market basket analysis; Drug-use initiation 1 Introduction Empirical evidence confirms that the majority of individuals’ illicit drug use starts only after they use cigarettes, alcohol, or cannabis (marijuana). Because using these less deleterious drugs may lead to more dangerous hard drugs, these three substances are known as the gateway drugs [1]. The gateway pattern of drug initiation describes a normative sequence that begins with alcohol and tobacco use followed by cannabis use and then other illicit drugs [2]. In the United States also, this progression proceeds from the use of tobacco or alcohol to the use of marijuana and other illicit drugs. Kandel presented the first systematic discussion of the gateway hypothesis [3]. He explored the hypothesis from various perspectives ranging from developmental social psychology to prevention and intervention science, animal models, neurobiology and analytical methodology. Although this pattern is recognized, some evidence suggests that prevention strategies focusing on breaking this chain to chronic and severe drug abuse may be of limited value [4]. Considering the prevalence of these three substances among the U.S. adolescent population, the gateway drugs should still be given maximum attention when compared to all other substances that can cause abuse and/or dependence [5]. Some studies revealed that weekly alcohol consumption came after marijuana use and preceded all other illicit drug use for Hispanic, White, and Black youth, but followed use of hard drugs for Asians [6]. There was evidence supporting that both Hispanic ethnicity and gender was significant in determining adolescent polydrug use. In the first year of middle school, gender moderated the effect of Hispanic ethnicity on lifetime polydrug use, and more serious levels of polydrug use were observed in the second year of middle school [7]. According to data from a recent national survey, nonsmokers who were also nondrinkers were most likely to indicate they would never use drugs in the future; users of both substances were most likely to indicate they would use drugs. And while smokers and drinkers were most likely to indicate expected illegal substance use in the future, those who only smoke were more likely than those who only drink to indicate probable use of such substances [8]. Adolescents were not likely to use marijuana without first using alcohol and tobacco, and they will not use more serious drugs, such as cocaine and heroin, without first using marijuana [9, 10]. Some research revealed that Whites, both male and female, had a significantly stronger positive association between predictors and drug use when compared to Blacks, while White females were most likely to use gateway substances, followed by White males [11]. Positive alcohol and cannabis outcome expectancies were meaningfully related to expectancies of future substance use, and positive cannabis outcome expectancies increased, while negative cannabis outcome expectancies decreased, with increasing frequency of alcohol use [12]. In another research, it was revealed that alcohol intake has a positive correlation with drug abuse among smokers [13]. A South African study showed that the pattern of onset of drug-use was similar across genders. Black South African adolescents first used either alcohol or tobacco, followed by both, and then dagga (cannabis) [14]. A prospective study among male Japanese students found that those who had smoking experiences by 16 years of age were more likely to abuse alcohol by 22 years of age than those without smoking experiences, but females did not show such a correlation [15]. In another study in Japan, cigarette smoking in high school students showed a strong relationship with solvent inhalation, while drinking alcohol depended on the presence of adults [16]. Particularly strong associations were found between smoking and having many friends who smoke. Using marijuana was strongly correlated with having friends or older siblings who use marijuana. The self-esteem scale score was not correlated with substance use. Anomie and antisocial behavior were more strongly associated with substance use than depression was. These associations were stronger for females than for males. Therefore, when designing and implementing preventive interventions in schools, it is essential to simultaneously address the use of gateway drugs and deviant behavior with environmental risk factors [17]. Current prevention programs are often ineffective when substance use is normalized by schools, community, and family. In such situations, positive adult role models can deter use, and focus group discussions can strengthen the development of tailored prevention programs [18]. Between U.S. and European students, there were the similar trends in drug use of gateway drugs already in use by early adolescence and higher rates of drug use among males than among females. However, cross-national differences in cannabis use revealed that U.S. preadolescents were 25 times more likely to use it than those in Europe. In the same study, Arizona Latinos reported higher cannabis use than Arizona non-Latinos, which was probably influenced by their acculturation level [19]. Low physical activity is associated with cigarette smoking and marijuana use. For alcohol consumption, significant links were found to race/ethnicity or sex, suggesting that socio-cultural factors may affect the relationships between physical activity and substance use [20]. Data from the Youth Risk Behavior Survey revealed that those using cigarettes before age 13 were 3.3 times more likely to have used marijuana than those who had never smoked. And users of alcohol were 4.5 times more likely to have used marijuana than those who had never used alcohol. The same survey showed that youth using marijuana before the age of 14 were 7.4 times more likely to have used other drugs even after controlling for selected other risk and protective behaviors [21]. The use of both tobacco and illicit drugs appears to be strongly associated with alcohol use, which is more prevalent, and the risk of smoking and illicit drug use is particularly high in adolescents who report high levels of drunkenness [22]. Factors which tend to increase the probability of alcohol use by adolescents include the fact that their friends drink, their awareness of the risks associated with alcohol use, and ease in obtaining alcohol. Hence, the typical adolescent who uses alcohol seems to be a risk-taker who may enjoy the dangers surrounding alcohol use and have alcohol-using friends. Religion, gender, race, academic performance, and extracurricular school activities are not directly related to the use of alcohol [23]. In 2006, marijuana (cannabis) was the most commonly used illicit drug (14.8 million past-month users) among persons in the United States aged 12 or older Marijuana users were twice as likely to use illicit drugs as young adults than non-users. Shared environmental factors mediated much of the relationship between adolescent marijuana use and young adult drug use [24]. Individuals who used cannabis by age 17 had higher rates of other drug use, alcohol dependence, and drug abuse or dependence. The association may arise from the effects of the peer and social context within which cannabis is obtained and used. In particular, early access and use of cannabis may reduce perceived barriers against the use of other illegal drugs and provide more access to these drugs [25]. Although the gateway effects of marijuana exist, they are not required to explain the common-factor model of initiating illicit drug use [26]. Extensive research does suggest that marijuana use tends to precede the use of other illicit substances among adolescents; however, there is a simultaneous argument that the correlation between marijuana use and other drug use may be spurious, which reflects the influence of one or more confounding variables that can concurrently cause both behaviors. Even after adjusting for the influence of variables, National Youth Survey results confirm the hypothesis that marijuana use exerts a causal influence on one’s probability of using other illicit substances [27]. 2 Purpose of the Study This study aims to study the sequences of initiating use of gateway and other drugs among adolescents, especially among 12th-grade students, in Indiana. The study also explores the applicability of method “Market Basket Analysis” of data mining in detection of common drug-use initiation patterns in the community. 3 Methods 3.1 Study Design Data from the 1993 to 2009 Annual Surveys of Alcohol, Tobacco, and Other Drug Use by Indiana Children and Adolescents—a survey conducted by the Indiana Prevention Resource Center—was used for the current study [1]. The original survey’s purpose was to provide data for state and local planning in respect to alcohol, tobacco, and other drug use, gambling behaviors, and risk and protective factors. 3.2 Study Population Surveys were administered through local school officials to students in grades 6 through 12 in 557 schools throughout Indiana. Participation in the survey was voluntary. In 2009, a total of 202,091 youth (182,496 usable surveys) from both public and private schools completed it [1]. The current research sample was composed of 12thgraders. They are of particular interest because they are on the verge of entering college or the workforce. Their substance abuse habits can negatively impact their academic and career functioning, which has detrimental effects on society. 3.3 Instrument A closed-ended, self-administered written questionnaire that was anonymous and voluntary asked adolescents about their use of twenty different types of drugs or drug categories, including prescription medications. When necessary, examples and descriptions were provided next to the drug’s name or its classification. 3.4 Analysis In 1980s, Yamaguchi and Kandel [28] utilized the log linear model to study associations and sequences of drug use. Their model was applied to study sequences up to five drugs; alcohol, cigarette, marijuana, cocaine, and heroin. This model is selected for our research to study the preferences of using drugs. Annual prevalence, the percentage who report using a particular drug at least once in the past year, was used for the current study. Annual prevalence in abuse of prescription drugs, alcohol, tobacco, marijuana, cocaine, and heroin, including the self-reported age at first-time use of each drug, was obtained from the survey [1]. Prescription drug use among 12th-graders was identified using a sample of 260,183 students during the 3-year period of 2007-2009. For alcohol, cigarettes, and marijuana use among 12 thgraders during the same time period, a sample of 246,793 students was utilized. Adolescents were asked, in which grade he or she started using each drug. Based on this information, the chronological order of drugs use initiation was established. This study focuses on data derived from a single item of the survey. Both items were used to collect information about 11 different drugs or drug categories: alcohol, cigarettes, marijuana, OTC substances, tranquilizers, narcotics, Ritalin, cocaine, crack, steroids, and heroin. The question asks “At what age did you first use drug A” with the response options: 10 or younger, 11, 12, 13, 14, 15, 16, 17, and 18 or older. Data obtained from the second question was analyzed using a derivative of Market Basket Analysis (MBA), a model utilized in studying patterns of goods sales (here they are drugs) focusing associations and sequences. In general, Market Basket Analysis is utilized in business management and marketing to understand how particular items were bought in conjunction with other items. MBA is one of the data mining models under SAS Enterprise Minor. The term “market basket” means a basket of items purchased in a transaction. The purpose of such analyses typically is to find subsets of items that are purchased together in many transactions [29]. We used a modified version of MBA to determine sequences of drug-use initiation by suggesting that individuals initiate use (equivalent to purchase) of specific drugs from among a larger set of available drugs (equivalent to market). Specifically, we drew from the overall sample those adolescents who report ever having used a gateway substance. We used SAS 9.3 to identify the most frequent sequences of drug use initiation among those individuals by ordering the respondents’ age of first use for each substance. Analyses focused on the “confidence %” statistic, which represents the probability, that event “Y” occurred after event “X.” If 500 people abuse drug “X,” and then, in a subsequent year, 125 of those people abuse drug “Y,” the confidence for XY is 25%. For example, the sequence “Alcohol Cocaine” is assessed using a rule that identifies all alcohol users, and then determines, how many initiated cocaine use in a year subsequent to their initiation of alcohol use. 4 Results Of the 492,957 students who provided data on drug use, 45.3% used at least one category of drugs in the past year, and 44.8% of the total used non-prescription drugs in the past year. But, only 34.4% of the total used nonprescription drugs alone and 10.4% used them simultaneously with prescription drugs. Of the 48,227 twelfth-graders who provided data on drug use, 33,961 (70.4%) used non-prescription drugs and 25,652 (75.5%) of non-prescription drug-users used only gateway drugs, while 22.1% used gateway drugs with prescription medications and 2% used gateway drugs with hard drugs (heroin, cocaine, and crack). Of the 12th-graders, 19.3% used alcohol only; 8.4% used cigarettes only and 4.8% used marijuana only (Table 1). More males compared to females used prescription and nonprescription drugs simultaneously. In addition, more males compared to females used at least one category of drugs, either prescription or non-prescription. Both of these findings are unexpected differences (p<0.0001). Similarly, more Whites compared to other races used prescription and nonprescription drugs simultaneously. In addition, more Whites compared to other races used at least one category of drugs, either prescription or non-prescription. Both of these findings are also unexpected differences (p<0.0001). Drug User Category (In Grade 12) Use Prescr. Drugs Don’t Use Prescr. Drugs Total Use NonPrescription Drugs 9,174 (19.0%) 26,070 (54.1%) 35,244 (73.1%) Don’t Use Non-Prescr Drugs 104 (0.2%) 12,879 (26.7%) 12,983 (26.9%) Total 9,278 (19.2%) 38,949 (80.8%) 48,227 (100.0%) Table 1: Self-Reported Annual Use of Prescription and Non-Prescription Drugs Among Indiana Adolescents in Grade 12 During the Period 2007-2009. Gateway drug use and the use of prescription and over-the-counter (OTC) medications have strong relationships (Figure 1). Also evident is that there are always two groups of drugs representing two chains. The first group contains alcohol, cigarettes, marijuana, cocaine, and crack. The second group consists of psychostimulants, CNS depressants, narcotics, and OTC drugs (Figure 2). After adjusting for age, gender, and race, similar sequences of using gateway drugs followed by hard drugs and prescription drugs were observed (Figure 2). But analysis of normative beliefs revealed that a majority of peers and parents of adolescents do not pay attention to prevention of tobacco, alcohol, prescription drug, and cocaine use, although considerable attention is paid to marijuana use [1]. Figure 1: Relationships and Strength Between Use of Alcohol, Cigarettes, and Marijuana as Gateway Drugs and the Use of Psychostimulants, Narcotics, CNS Depressants, or OTC Drugs Among Indiana Adolescents (2007-2009) According to Market Basket Analysis. Thickness of the Lines is Proportionate to Strength of Association. Figure 2: Sequence and Strength Between Use of NonPrescription Drugs (Alcohol, Cigarettes, Marijuana, Cocaine, & Crack) and the Nonmedical Use of Medications (Psychostimulants, Narcotics, CNS Depressants, & OTC Drugs) Among Indiana Adolescents (2007-2009) According to Market Basket Analysis. Thickness of the Lines is Proportionate to Strength of Association. 5 Discussion Slightly less than half of Indiana adolescents use at least one category of prescription or non-prescription drugs, and almost 99% of these drug users report annual use of non-prescription drugs, including alcohol, cigarettes, and marijuana. This indicates that almost all prescription drug users simultaneously use non-prescription drugs as well. This pattern intensifies when 12th-graders are considered because almost three quarters of these adolescents use some kind of drug. In general, males are more prone to drug use and non-prescription drug use than females as males tend to report combined use of gateway drugs with other drugs, such as prescription drugs and hard drugs. The tendency of males to experiment may be the reason for this trend. Market Basket Analysis reveals that the gateway theory is applicable to Indiana adolescents because many of them initiate drug use with alcohol, tobacco, or marijuana or with a combination of them followed by use of prescription drugs. The majority of adolescents who use gateway drugs start using psychostimulants non-medically, follow them with CNS depressants, and then use prescription narcotics and OTC drugs. This model proves that the gateway theory is useful in explaining adolescent drug abuse. But many parents and peers of adolescents do not pay attention to the prevention of tobacco, alcohol, prescription drug, and cocaine use, although considerable attention is paid to marijuana use. In 2002, Yamaguchi and Kandel [28] applied two statistical models for analysis of drug use progression: quasi-independence model for the analysis of events sequence data [30] and the log linear model for analysis of parametric events sequences. Our future studies will focus on how the outputs of Market Basket Analysis can be integrated with the models of Yamaguchi and Kandel. 6 Limitations This is a self-administered anonymous survey on drug use. The responses of younger adolescents may not be very accurate. The percentage of missing data in some sections of the questionnaire is somewhat high. [3] [4] [5] [6] [7] 7 Conclusion Of the 48,227 twelfth-graders who provided data on their drug use, 70.4% use non-prescription drugs and 75.5% of those non-prescription drug-users abuse gateway drugs only while 22.1% use gateway drugs with prescription medications and 2% use gateway drugs with hard drugs. The majority of adolescents start drug use with gateway drugs followed by combining them with prescription drugs. Males, Whites, and older adolescents are more likely to report single-use of gateway drugs and combined use of gateway drugs with other drugs. Market basket analysis in data mining is found to be a useful technique in identifying common substance-use initiation patterns. 8 Conflict of Interests Statement [8] [9] [10] [11] This research is not fully or partially supported by any funding source and authors have no conflict of interests. [12] 9 Acknowledgement [13] The authors would like to thank Prof. Mohammad Torabi and Dr. Ruth Gasman of Dept. of Applied Health Science, Indiana University Bloomington for their help. [14] 10 References [1] [2] "Survey of Alcohol, Tobacco, and Other Drugs use by Indiana Children and Adolescents. Indiana Prevention Resource Center," 2010. L. Degenhardt, et al., "Does the 'gateway' matter? Associations between the order of drug use initiation and the development of drug dependence in the National Comorbidity Study Replication," Psychological Medicine, vol. 39, pp. 157-167, 2009. [15] [16] [17] D. B. Kandel, Stages and Pathways of Drug Involvement: Examining the Gateway Hypothesis Cambridge: Cambridge University Press, 2002. M. E. MackesyAmiti, et al., "Sequence of drug use among serious drug users: Typical vs atypical progression," Drug and Alcohol Dependence, vol. 45, pp. 185-196, 1997. (2010). Monitoring the Future Survey: A Continuing Study of American Youth. Available: http://monitoringthefuture.org/ P. L. Ellickson, et al., "Stepping Through the Drug-Use Sequence - Longitudinal Scalogram Analysis of Initiation and Regular Use," Journal of Abnormal Psychology, vol. 101, pp. 441-451, 1992. J. A. Epstein, et al., "Gateway polydrug use among Puerto Rican and Dominican adolescents residing in New York city: The moderating role of gender," Journal of Child & Adolescent Substance Abuse, vol. 12, pp. 33-46, 2002. P. B. Johnson, et al., "The relationship between adolescent smoking and drinking and likelihood estimates of illicit drug use," Journal of Addictive Diseases, vol. 19, pp. 75-81, 2000. P. Willner, "Alcohol and cannabis expectancies in adolescents: Relationship to alcohol and cannabis use, and a potential mechanism for an alcohol-drug gateway," Journal of Psychopharmacology, vol. 14, p. A10, 2000. R. J. Kane and G. S. Yacoubian, "Patterns of drug escalation among Philadelphia arrestees: An assessment of the gateway theory," Journal of Drug Issues, vol. 29, pp. 107-120, 1999. N. Graham, "A test of magnitude: Does the strength of predictors explain differences in drug use among adolescents?," Journal of Drug Education, vol. 27, pp. 83-104, 1997. P. Willner, "A view through the gateway: expectancies as a possible pathway from alcohol to cannabis," Addiction, vol. 96, pp. 691-703, 2001. L. M. Sanchez-Zamorano, et al., "Prevalence of illicit use in function of tobacco smoking in Mexican students sample," Salud Publica De Mexico, vol. 49, pp. S182-S193, 2007. M. E. Patrick, et al., "A Prospective Longitudinal Model of Substance Use Onset Among South African Adolescents," Substance Use & Misuse, vol. 44, pp. 647-662, 2009. K. Suzuki, et al., "Is adolescent tobacco use a gateway drug to adult alcohol abuse? A Japanese longitudinal prospective study on adolescent drinking," Japanese Journal of Alcohol Studies & Drug Dependence, vol. 43, pp. 44-53, 2008. K. Wada, "The concept of "Gateway Drug"," Japanese Journal of Alcohol Studies and Drug Dependence, vol. 34, pp. 95-106, 1999. A. Kokkevi, et al., "Psychosocial correlates of substance use in adolescence: A cross-national [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] study in six European countries," Drug and Alcohol Dependence, vol. 86, pp. 67-74, 2007. J. Peterson, "A Qualitative Comparison of Parent and Adolescent Views Regarding Substance Use," Journal of School Nursing, vol. 26, pp. 53-64, 2010. M. A. Luengo, et al., "A cross-national study of preadolescent substance use: exploring differences between youth in Spain and Arizona," Subst Use Misuse, vol. 43, pp. 1571-93, 2008. R. R. Pate, et al., "Associations between physical activity and other health behaviors in a representative sample of US adolescents," American Journal of Public Health, vol. 86, pp. 1577-1581, 1996. J. C. Merrill, et al., "Cigarettes, alcohol, marijuana, other risk behaviors, and American youth," Drug and Alcohol Dependence, vol. 56, pp. 205-212, 1999. I. Sutherland and P. Willner, "Patterns of alcohol, cigarette and illicit drug use in English adolescents," Addiction, vol. 93, pp. 1199-1208, 1998. B. M. Yarnold, "The use of alcohol by Miami's adolescent public school students 1992: Peers, risktaking, and availability as central forces," Journal of Drug Education, vol. 28, pp. 211-233, 1998. J. M. Lessem, et al., "Relationship between adolescent marijuana use and young adult illicit drug use," Behavior Genetics, vol. 36, pp. 498-506, 2006. M. T. Lynskey, et al., "Escalation of drug use in early-onset cannabis users vs co-twin controls," Jama-Journal of the American Medical Association, vol. 289, pp. 427-433, 2003. A. R. Morral, et al., "Reassessing the marijuana gateway effect," Addiction, vol. 97, pp. 1493-1504, 2002. C. J. Rebellon and K. Van Gundy, "Can social psychological delinquency theory explain the link between marijuana and other illicit drug use? Al ongitudinal analysis of the gateway hypothesis," Journal of Drug Issues, vol. 36, pp. 515-539, 2006. K. Yamaguchi and D. B. Kandel, Loglinear Sequence Analyses: Gender and Racial/Ethnic Differences in Drug Use Progression - Stages and Pathways of Drug Involvement: Examining the Gateway Hypothesis. Cambridge: Cambridge University Press, 2002. V. Ganti, et al., "Mining very large databases," Computer, vol. 32, pp. 38-45, 1999. L. A. Goodman, "A new model for scaling response patterns: An application of quasiindependent concept," journal of American Statistical Association, vol. 70, pp. 755-768, 1975.
© Copyright 2026 Paperzz