INTRODUCTION
Adolescence is marked by the intensification of physical and neuropsychological maturational processes as well as abrupt social and relational changes. Such processes, while raising new capacities for a better adaptation to the demands of social life, also generate vulnerability since they foster behaviors that sometimes imply the involvement in activities harmful to physical and mental health (Kaufmann et al. Reference Kaufmann, Alnæs, Doan, Brandt, Andreassen and Westlye2017; Lee et al. Reference Lee, Heimer, Giedd, Lein, Šestan, Weinberger and Casey2014). One of such activities consists of the use of licit and illicit substances, a practice that tends to start in adolescence. Motives for the use of psychoactive substances among youth are varied: in order to feel accepted by their peers (peer pressure use), to feel pleasure (recreational use), to feel well (therapeutic use), to improve their performance (use for sportive or academic reasons), or to experience new sensations (exploratory use) (NIDA (National Institute on Drug Abuse) 2014). For some, these practices may become habits that persist beyond adolescence (Meier et al. Reference Meier, Caspi, Ambler, Harrington, Houts, Keefe, McDonald, Ward, Poulton and Moffitt2012). In extreme cases, an abusive use may lead to dependence (Atkinson et al. Reference Atkinson, Anderson, Hughes, Bellis, Sumnall and Syed2009).
Furthermore, substance abuse relates to other mental and physical health problems as well as other antisocial types of conduct, especially the violent ones, which poses a great challenge to public health and security policies in various countries (Atkinson et al. Reference Atkinson, Anderson, Hughes, Bellis, Sumnall and Syed2009; Komatsu, Estevão, and Bazon Reference Komatsu, Estevão and Bazon2018). The criminological literature has been reporting these relationships between the use of substances and other problems of various magnitudes. It is known, for example, that substance abuse increases the risk of an individual becoming the victim or the aggressor in a situation of violence (Duke et al. Reference Duke, Smith, Oberleitner, Westphal and McKee2018). It is also known that being inserted in a context of repeated exposure to violence increases the risk of substance abuse as well as the development of other mental health problems (Löfving-Gupta et al. Reference Löfving-Gupta, Willebrand, Koposov, Blatný, Hrdlička, Schwab-Stone and Ruchkin2017; Vermeiren et al. Reference Vermeiren, Schwab-Stone, Deboutte, Leckman and Ruchkin2003). Additionally, co-morbidity between substance abuse and mental health problems has been observed even in the absence of exposure to violence (NIDA 2018a).
It is also known that, even though the relationship between psychoactive substance abuse and mental health and conduct problems exists and is well documented in the literature, many individuals only display one of the problems (Chalub and Telles Reference Chalub and Telles2006; Håkansson and Jesionowska Reference Håkansson and Jesionowska2018). This observation points to the existence of heterogeneity regarding the way each of these conditions manifests itself during the course of the life of each individual, hence the importance of identifying the different deviant trajectories and risk factors related to each one, as highlighted by Piquero et al. (Reference Piquero, Farrington, Nagin and Moffitt2010) as being one of the core aims of Developmental Criminology. Regarding this, many studies have resorted to methodological approaches centered on the person, such as latent class analysis (LCA), since they are more flexible than analytical techniques that focus on the variable (Collins and Lanza Reference Collins and Lanza2010). Following this trend, some studies have resorted to LCA in order to identify patterns of substance use and their relationship with other areas such as academic performance, violence involvement, among others (for a review of these studies, see Tomczyk, Isensee, and Hanewinkel Reference Tomczyk, Isensee and Hanewinkel2016). In Brazil, however, there have been no investigations of this nature, aiming to describe typologies of psychoactive substance abuse that may exist among the adolescent population in association with other variables of interest, be it from a phenomenon comprehension point of view, or a public policies point of view.
Therefore, this study aimed to identify patterns of psychoactive substance abuse in a sample of the population of Brazilian adolescent students and estimate the magnitude of the relationship between each identified pattern and indicators of involvement in violence, victimization and mental and general health. LCA was employed in order to identify subgroups of adolescents regarding conducts of use of substances like alcohol, tobacco, marijuana and cocaine/crack and, afterwards, these subgroups (latent classes) were compared in regard to their prevalence in: (1) four categories concerning involvement in violence (bullying practices, involvement in fights, involvement in fights with bladed weapons and fights with fire arms); (2) five categories of victimization (humiliation by a colleague, submission to bullying, physical aggression by an adult family member, serious injury and sexual abuse); (3) two categories related to mental health problems (loneliness/frequently feeling alone and worry/stress); (4) one category related to general health (general evaluation of overall well-being).
METHODS
The data used in this study come from the National Survey of School Health (PeNSE) conducted in 2015 in Brazil by the Brazilian Institute of Geography and Statistics (Instituto Brasileiro de Geografia e Estatística) (IBGE 2016a) through a partnership with the Ministry of Health and support from the Ministry of Education. In this cross-sectional survey, data were collected from two independent representative samples of public and private education in the country: one formed by schoolchildren attending the 9th grade (former 8th grade) of elementary school (sample A) and another one formed by schoolchildren attending the 6th grade of elementary school to the 3rd grade of high school (sample 2). PeNSE 2015 was approved by the National Research Ethics Committee (Conep; Comissão Nacional de Ética em Pesquisa) of the National Health Council, which regulates and approves health studies involving human beings through legal advice (no. 1,006,467 of Conep).
For this investigation, only data from sample 1 were used, that is, data from school children attending the 9th grade of elementary school. This choice was based on the recommendation made by the World Health Organization (2014) to consider children under 15 years of age as a reference in the conduct of surveys involving schoolchildren as well as the fact that this grade amasses, in Brazil, 80% of students between the ages of 13 and 15 years (IBGE 2015).
Therefore, we can affirm that this study was carried out with data from a representative sample of adolescents from the 9th grade of elementary school, from public and private teaching institutions, from urban and rural areas of all states of Brazil. Considering the aims of this study, only data from adolescents who had answered all the questions regarding the investigated substances (alcohol, tobacco, marijuana and crack cocaine) were included in the analyses, so that the final sample was 6702 adolescents averaging 14.9 years of age (standard deviation = 1.2). The majority were males (53%). The frequency of participants by country region assumed the following distribution: 23% in the Northeast, 22% in the North, 19% in the Southeast, 18% in the Midwest and 16% in the South. Regarding race, 32% claimed to be white, 44% brown, 15% black, 5% yellow and 4% indigenous. Regarding family, 42% said they lived with their father and mother, 40% said they lived only with their mother, 8% only with their father and 10% said they lived with someone else rather than their mother or father.
Instruments and Measures
The data were collected thorough a structured self-administered questionnaire with the help of a Personal Digital Assistant, a smartphone that allowed the answers to be recorded directly into an electronic database without the need for interference by the interviewer (IBGE 2016a). The questionnaire included questions about family, its socio-economic level; dietary habits; practice of physical activities; sexual and reproductive health; use of cigarettes, alcohol and other drugs; violence, safety and accidents; use of health services, among other aspects (IBGE 2016a). However, for the purposes of this study, only the questions related to the use of substances, the involvement (as aggressor and/or victim) insituations of violence and mental and general health were considered.
The questions within these themes generated the following measures:
Use of psychoactive substances: This consisted of four questions regarding each one of the investigated substances: (1) “In the past 30 days, how many days did you drink at least one glass or one dose of an alcoholic drink? (one dose is equivalent to one can of beer or a glass of wine or a dose of cachaça or whisky, etc.)”; (2) “In the past 30 days, how many days did you smoke cigarettes?”; (3) “In the past 30 days, how many days did you smoke marijuana?”; and (4) “In the past 30 days, how many days did you use crack?”. Answers to the questions could range from zero to 30 days. In this study, answers were dichotomized between “didn’t make use in the past 30 days” and “made use in the past 30 days”. We opted for using the 30-day period (instead of the prevalence of use throughout their lives) because we consider this measure to more faithfully represent the current life moment of each adolescent regarding substance use patterns.
Sociodemographic characteristics: This measure involved a group of questions related to age, gender, race, region of the country where they lived and the type of teaching institution attended (public/private).
Involvement insituations of violence: Involvement insituations of violence was investigated by four questions: (1) Practice of bullying: “In the past 30 days, did you roast, mock, ridicule, intimidate or make fun of any of your school colleagues so much that they felt upset, angry, offended or humiliated?” (yes/no); (2) Involvement in fights: “In the past 12 months, how many times did you get into a fight (physical fight)?” (not one/1 to 3 times/more than 3 times); (3) Involvement in fights with bladed weapons: “In the past 30 days, did you get into a fight in which someone used some other weapon such as a knife, jackknife, fishmonger knife, rock, piece of wood or glass bottle?” (yes/no); and (4) Involvement in fights with fire arms: “In the past 30 days, did you get into a fight in which someone used a fire arm such as a revolver or a shotgun?” (yes/no).
Victimization: This measure, concerning the exposure of the adolescent to violence as a victim, consisted of five questions: (1) Humiliation by colleague: “In the past 30 days, how often were you roasted, mocked, ridiculed, intimidated or made fun of by school colleagues so much that you felt upset, uncomfortable, angry, offended or humiliated?” (not one/1 to 3 times/more than 3 times); (2) Bullying: “Have you ever been bullied?” (yes/no); (3) Physical violence inside the family: “In the past 30 days, how many times were you physically hurt by an adult family member?” (not one/1 to 3 times/more than 3 times); (4) Serious physical injury: “In the past 12 months, how many times were you seriously injured?” (not one/1 to 3 times/more than 3 times); and (5) Sexual abuse: “Have you ever been forced to have sexual relations?” (yes/no).
General and mental health: Two questions were used as indicators of mental health problems among adolescents: (1) Loneliness: “In the past 12 months, how often did you feel alone?” (never/rarely or sometimes/always or almost always); and (2) Worry/stress: “In the past 12 months, how often were you unable to sleep because you were too worried about something?” (never/rarely or sometimes/always or almost always). One question was used as an indicator of general health/well-being: “How would you rate your health condition?” (good or very good/regular/bad or very bad).
Data Collection and Plan of Analysis
Data collection was collectively performed in school classrooms between April and September 2015. As stated before, students answered the questions directly into a smartphone provided by the investigators. Participation was voluntary and the adolescents were free not to answer any of the questions. The compiled data were stored in digital spreadsheets.
In order to analyze the data, at first, we used LCA to identify subgroups within the sample in terms of different patterns of substance use in the past 30 days. Four dichotomous variables were used as indicators for LCA: use of alcohol, tobacco, marijuana and crack cocaine. In addition to these four variables, a fifth descriptive indicator of use pattern, relative to the number of substances used, was incorporated into the analysis; this one could range from zero (abstemious) to four substances. LCA was performed using the poLCA package (Linzer and Lewis Reference Linzer and Lewis2011) of the statistical software R 3.5.2 (R Core Team 2018). Models consisting of two to seven classes were generated. Firstly, a two-class model was generated. Then, new classes were added to the model until no statistical or theoretical improvements were no longer observed. Models were compared based on the Bayesian information criterion (BIC). Additionally, visual and tabular representations of the classes were used in order to identify the conceptual adjustment of each model. The model with higher parsimony and interpretability was chosen, to ensure that it did not estimate more parameters than necessary in order to adequately represent the data. Then, bivariate analyses (χ 2 test) were used in order to verify whether or not belonging to each class was associated with sociodemographic variables and measures of involvement in violence, victimization and mental/general health problems.
In order to calculate the effect size on the association measures, adjusted Pearson residuals were used (Agresti Reference Agresti2002:81) based on the following equation:
$${r_{ij}} = {{{O_{ij}} - {E_{ij}}} \over {\sqrt {{E_{ij}}\left( {1 - {{{m_i}} \over N}} \right)\left( {1 - {{{n_j}} \over N}} \right)} }}$$in which O is the value observed in line i and column j, E is the expected value in line i and column j, m i is the sum in line i and n j is the sum of column j. With the premise that the adjusted Pearson residuals follow a standardized normal distribution, i.e. N~(0,1), the magnitude of the residues was interpreted based on normal distribution parameters, assuming, therefore, that r values above 2 and below –2 are significantly outside of the expected value, so that the size of the effect be comparable with the magnitude of the calculated residual value.
RESULTS
The model with five latent classes, represented in Figure 1, showed better statistical adjustment than the models with fewer classes (BIC2 = 43.616; BIC3 = 41.450; BIC4 = 41.188; BIC5 = 39.691; BIC6 = 39.156; BIC7 = 38.606) and better theoretical adjustment than the models with more classes. Therefore, it was the model elected to represent the findings of this study. Figure 1 shows the approximate statistical probability in each group of using each of the investigated substances as well as the probability, conditional to belonging to each class, of having used zero to four substances (maximum value) in the past 30 days prior to the investigation.
Figure 1. Probability of a “yes” response to each substance item conditional on latent class membership. Items with probability above .50 are highlighted in bold.
Based on the conditional probabilities shown in Figure 1 and the conceptual significance they represent together, class 1 was named Abstainers because adolescents in this class display low probability of using any of the investigated substances. Class 2, due to its high probability of alcohol use and low probability of using other substances, was named Drinkers. Class 3 was named Conventional Drug Users, based on the group found by Bohnert et al. (Reference Bohnert, Walton, Resko, Barry, Chermack, Zucker, Zimmerman, Booth and Blow2014) with a similar use pattern: high probability of alcohol use associated with another substance: tobacco or marijuana.
Class 4 was named Polysubstance Users, based on the group with similar characteristics identified by Choi et al. (Reference Choi, Lu, Schulte and Temple2018), which shows high probability of using three substances together: alcohol, tobacco and marijuana. Lastly, Class 5 was named Hard Drug Users because it groups adolescents with high probability of using all four investigated substances together: alcohol, tobacco, marijuana and crack.
Table 1 complements the descriptions of the groups by showing how the classes differ regarding the categories of frequency of use of each substance, making it possible to assess the problematic use of drugs, especially considering that the frequent use of psychoactive substances at this age represents a potential risk to development.
Table 1. Five-Latent-Class Model of Substance Use in the Last 30 Days
Source: Data from the National Survey of School Health (PeNSE) conducted in 2015, with analyses carried out by the authors.
Table 2 displays the sociodemographic characteristics of the participants. Abstainers had a slightly higher proportion of inhabitants in the Northeast region (r = 2.2) and much higher in the North region (r = 4.6) and a sub-representation in the Southeast (r = –3.4) and South (r = –3.0), while the number of adolescents in the Midwest region was close to what was expected (r = –1.1). The proportions of male (r = .2) and female adolescents (r = –.2) were also within the expected range, considering the existing proportions in the total sample. Regarding race, Abstainers were predominantly brown (r = 3.0) and rarely white (r = –3.2). Regarding the type of educational institution, Abstainers were more highly present in public schools (r = 3.4) and, consequently, less present in private schools (r = –3.4).
Table 2. Sociodemographic Characteristics of the Groups
Source: Data from the National Survey of School Health (PeNSE) conducted in 2015, with analyses carried out by the authors.
a Proportions significantly above or below the overall average (χ 2 test).
* p < .001.
Drinkers had fewer adolescents in the North region (r = –3.9) and slightly more in the Southeast region (r = 2.0), keeping the proportions for the Northeast (r = .4), South (r = .4) and Midwest (r = .5) regions close to the average. In this group, females (r = 2.1) were slightly more prevalent and males slightly less prevalent (r = –2.1) than in the overall sample. Indigenous people were underrepresented in this group (r = –2.8), while the other races had proportions within the expected range (r ranging from –1.3 to 1.5). The proportions of students in public (r = 4) and private (r = –4) schools also remained within the average range. Conventional Drug Users were the fewest in the North region (r = –4.1) and the most in the South (r = 2.5). Regarding gender, race and type of educational institutions, the percentages of this group remained within the expected (r ranging from –1.2 to 1.3).
Polysubstance Users had the lowest number of adolescents in the Northeast region (r = –3.2) and the highest number of adolescents in private schools (r = 2.7), keeping the remaining proportions within the average of the overall sample. Hard Drug Users had the highest number of adolescents living in the North region (r = 2.6) and the lowest in the Southeast (r = –3.2), in comparison with all the other classes, while the other regions displayed average proportions. This group also had, proportionally, the highest number of males (r = 4.6) and indigenous people (r = 6), substantially above the overall average, while females (r = –4.6) and brown adolescents (r = –2.6) were the lowest of all classes.
Table 3 presents the percentage of adolescents in each class who got involved with violence. Abstainers were significantly less involved with all types of violence than the average: bullying practice (r = –3.5), involvement in fights (r = –4.5 for “1 to 3 times” and r = –8.8 for “more than 3 times”), involvement in fights with bladed weapons (r = –11.3) and involvement in fights with fire arms (r = –10.5). The proportion of Drinkers who got involved in violent situations was also lower than the average, but generally larger than Abstainers: offending/humiliating a colleague (r = –4.5), involvement in fights (r = –7.5 for “more than three times”), involvement in fights with bladed weapons (r = –7.8) and involvement in fights with fire arms (r = –6.3). Conventional Drug Users had a prevalence of involvement in violence within the expected for all investigated modalities, except for involvement in a fight in the past 12 months, with a ratio above the average for “1 to 3 involvements” (r = 3.2). Polysubstance Users had significantly higher prevalence of involvement in violent situations than the three previous groups: bullying practice (r = 3.6), involvement in fight (r = 2.3 for “1 to 3 times” and “r = 7.2 for “more than three times”), involvement in fights with bladed weapons (r = 10.4) and involvement in fights with fire arms (r = 7.4). Finally, the proportion of adolescents from the Hard Drug Users group in all four violence modalities was substantially higher than all other classes: bullying practice (r = 8.2), involvement in fights (r = 17.1 for “more than three times”), involvement in fights with bladed weapons (r = 15.2) and involvement in fights with fire arms (r = 17.7).
Table 3. Prevalence of Violence Involvement in Each Latent Class
Source: Data from the National Survey of School Health (PeNSE) conducted in 2015, with analyses carried out by the authors.
a Proportions significantly above or below the overall average (χ 2 test).
* p < .001.
Table 4 presents the percentages of adolescents from each group of each of the five investigated victimization modalities. The proportion of Abstainers who reported having been bullied was higher than the average (r = 2.7), but for all other forms of victimization, this group was the one with the lowest proportion of adolescents: frequency with which they felt offended/humiliated by colleagues (r = .3 for “rarely or sometimes” and r = –2.5 for “always or almost always”), frequency with which they were physically hurt by a family member (r = –3.5 for “1 to 3 times” and r = –5.5 for “more than three times”), frequency with which they were seriously hurt (r = –5.1 for “1 to 3 times” and r = –6.3 for “more than three times”) and having suffered sexual abuse (r = –4.9). The proportion of Drinkers who reported being bullied was also higher than the average (r = 2.6) but, as well as the Abstainers, the proportion of Drinkers who suffered other forms of violence was significantly below average: frequency with which they felt offended/humiliated by colleagues (r = –2.9 for “always or almost always”), frequency with which they were physically hurt by a family member (r = .2 for “1 to 3 times” and r = –6.5 for “more than three times”), frequency with which they were seriously hurt (r = –2.8 for “1 to 3 times” and r = –6.2 for “more than three times”) and having suffered sexual abuse (r = –4.3). Among Conventional Drug Users, the proportion of those who reported having been bullied was below average (r = –2.1). Regarding other forms of violence, this class displayed prevalence equivalent to the overall averages. The proportion of adolescents in the Polysubstance Users class who reported being bullied was also below average (r = –3.2), as well as those who reported being offended/humiliated by a colleague (r = –2.2 for “always or almost always”). In this class, the proportion of adolescents who reported being physically hurt by a family member was above average (r = 2.6 for “1 to 3 times” and r = 3.2 for “more than three times”) as well as adolescents from this class who reported having been seriously hurt (r = 5.5). Regarding sexual abuse, the proportion of adolescents who reported having suffered from this form of violence was equivalent to the overall average (r = 1.1). Among Hard Drug Users, the percentage of those who reported having been bullied was equivalent to the average (r = .6), although the proportion of those who reported having been offended/humiliated by a colleague in the past 30 days was significantly higher than the other groups (r = 11.7 for “always or almost always”). In this class, the proportion of adolescents was substantially higher than all the other classes for other forms of violence: physical violence within the family (r = 11.7 for “more than three times”), having been seriously hurt (r = 24.1 for “more than three times”) and having suffered sexual abuse (r = 17.1).
Table 4. Prevalence of Victimization in Each Latent Class
Source: Data from the National Survey of School Health (PeNSE) conducted in 2015, with analyses carried out by the authors.
a Proportions significantly above or below the overall average (χ 2 test).
* p < .001.
Table 5 shows the percentages of adolescents from each class regarding general and mental health issues. Abstainers had the smallest proportion among the adolescents who reported “always or almost always” feeling lonely (r = –3.9) and “always or almost always” feeling stressed to the point of not being able to sleep (r = –4.5) for the past 12 months prior to the survey. Regarding general health, Abstainers had the highest frequency of adolescents who rated their health condition as “good or very good” (r = 2.4), being the only class in which the percentage of adolescents who rated themselves within this category was above average. Drinkers found themselves within the average regarding adolescents who reported “always or almost always” feeling lonely (r = .5) or stressed to the point of not being able to sleep (r = –1.8). Regarding general health, Drinkers were also within the average regarding a “good or very good” health condition (r = 1.8). Similarly to Drinkers, Conventional Drug Users were also within the average in all categories whose frequency could indicate problems. Polysubstance Users were within the average in all categories related to feeling lonely and general health condition, but displayed a prevalence higher than the three previous groups regarding “always or almost always” being stressed/worried to the point of not being able to sleep (r = 2.8). Finally, Hard Drug Users represented the class with the highest proportion of adolescents who reported “always or almost always” feeling lonely (r = 2.9) or stressed/worried to the point of not being able to sleep (r = 4.4) and having a “bad or very bad” health condition (r = 8.8).
Table 5. Prevalence of General and Mental Health Issues in Each Latent Class
Source: Data from the National Survey of School Health (PeNSE) conducted in 2015, with analyses carried out by the authors.
a Proportions significantly above or below the overall average (χ 2 test).
* p < .001.
DISCUSSION
Firstly, we can see that the analysis technique employed was able to identify a model with five classes that adequately adjusted itself to the data statistically and theoretically. The classes displayed a high degree of separation (heterogeneity among themselves) and internal homogeneity, revealing an empirical reality in which Brazilian adolescents can clearly be grouped in terms of substance use patterns. The characteristics of each class show that there is a tendency to have a combined use of specific substances. Thus, although 16 (24) possible combinations of use of the four substances could exist, four of them stood out in terms of prevalence: alcohol use only (Drinkers, 26%), alcohol use associated with tobacco or marijuana (Conventional Drug Users, 28%), combined use of alcohol, tobacco and marijuana (Polysubstance Users, 23%) and the use of all four substances (Hard Drug Users, 5%). Other patterns of use such as alcohol–crack, tobacco–marijuana, tobacco–crack or marijuana–crack showed relatively low probability of occurrence. These patterns are similar to those identified in other studies in different sociocultural contexts (Bohnert et al. Reference Bohnert, Walton, Resko, Barry, Chermack, Zucker, Zimmerman, Booth and Blow2014; Choi et al. Reference Choi, Lu, Schulte and Temple2018; Goldstick et al. Reference Goldstick, Stoddard, Carter, Zimmerman, Walton and Cunningham2016; Liu et al. Reference Liu, Elliott, Serdarevic, Leeman and Cottler2019).
In addition to the qualitative differences in terms of use of certain substances by some groups and not others, strong quantitative differences could also be observed, so that Hard Drug Users stood out for displaying the highest frequencies of use of all substances, followed by Polysubstance Users (second), Conventional Drug Users (third) and Drinkers (fourth). Knowledge of the patterns of substance use in adolescence is especially important since it allows us to learn about and describe the existing heterogeneity among adolescents regarding this problem and identify the behavioral characteristics that may develop in its early stages as well as to identify their similarities to other problems that affect adolescents, such as violence and those related to physical and mental health.
In terms of sociodemographic characterization, some differences deserve attention. Regarding the regions of the country, the North shows the most intriguing distribution. If, on the one hand, this region is the most frequent for Abstainers, on the other hand, it is also the most frequent for Hard Drug Users. Because this region is part of the Latin American route of international cocaine trafficking (Couto and Oliveira Reference Couto and Oliveira2017), it is possible that access to this substance in the North is facilitated, partially justifying the higher concentration of adolescents from the Hard Drug Users class in this region. However, the reasons why the North is also the region that concentrates most Abstainers are not clear. This region, along with the Northeast, are the ones with the highest Gini index in Brazil (IBGE 2016b), thus being characterized as a region marked by enormous social inequalities, which may impose an economic restriction to the use of substances by certain groups of adolescents.
Regarding gender, the disparity between males (66%) and females (34%) in the Hard Drug Users class is striking, while in the other classes distributions were closer to that of the overall sample (53% males and 47% females). There is evidence that prevalence and frequency of substance use tend, in general, to be higher among men (NIDA 2018b), although women tend to make more use of prescribed substances (Loikas et al. Reference Loikas, Wettermark, von Euler, Bergman and Schenck-Gustafsson2013; Skoog et al. Reference Skoog, Midlöv, Borgquist, Sundquist and Halling2014). The European Monitoring Centre for Drugs and Drug Addiction (2005) reports that the ratio between men and women regarding the use of substances is higher when we talk about adults rather than adolescents between 15 and 16 years of age, which may explain the relative parity of the two genders in the four classes that display lower substance use. The high concentration of males among Hard Drug Users, however, may be due to the use of cocaine in the form of crack, a substance that sets this class apart from the others. The use of crack is associated with the search for more intense effects (Jorge et al. Reference Jorge, Quinderé, Yasui and Albuquerque2013). It is also a substance characterized by a higher risk of addiction and short-term side effects (Noto et al. Reference Noto, Galduróz, Nappo, Fonseca, Carlini, Moura and Carlini2003) as well as greater reactivity and social stigma (McNeil et al. Reference McNeil, Kerr, Lampkin and Small2015; Ribeiro, Sanchez, and Nappo Reference Ribeiro, Sanchez and Nappo2010), aspects that may restrain its consumption by females.
Regarding race distribution among the groups, the proportion of 10 fewer percentage points of brown adolescents and the almost three times higher proportion of indigenous adolescents in the Hard Drug Users class, compared with those in the overall sample, are striking. Some studies have indicated a growing problematic use of substances associated with violence and mental health-related problems for indigenous populations (Duarte, Stempliuk, and Barroso Reference Duarte, Stempliuk and Barroso2009; Indigenist Missionary Council 2014; Netto Reference Netto2018). The various historical processes of marginalization of this population, associated with the increase in the spread of trafficking within indigenous communities (Indigenist Missionary Council 2014; Netto Reference Netto2018), may favor the use of substances as a mechanism to deal with the adversities faced by this population. These results are in line with the study by Stanley et al. (Reference Stanley, Harness, Swaim and Beauvais2014), which found that Native Americans from 8th, 10th and 12th grades have a significantly higher consumption rate than the national rate for all substances among schoolchildren. As pointed out by the authors themselves, a comprehensive understanding of the causes of this relationship has not been established yet, which requires special attention not only in terms of prevention policies and specific interventions directed towards this population, but also to the need to study these groups more deeply.
Regarding the correlated problems analyzed – involvement in violence as an aggressor or as a victim and aspects of mental and general health – results indicate that Abstainers would be better protected from them in all situations investigated. The number of adolescents in this class that reported being involved with violence, as a victim or as an aggressor, and that reported feeling lonely or stressed/worried to the point of not being able to sleep, was frequently or almost always the lowest among all classes. Although this study does not allow the identification of a correlation between mental health and the use of substances, we emphasize that the literature indicates that this relationship may be initiated in two ways: (1) the use of substances may harm the mental health and cause symptoms of depression and anxiety, which may lead to excessive worries; (2) a below-average performance, caused by internal mechanisms (personal) and/or external (social, relational and contextual), may favor the use of substances as a coping mechanism (NIDA 2018a; Ross Reference Ross2004; Stewart and Conrod Reference Stewart, Conrod, Stewart and Conrod2008). In both cases, these mechanisms may create a feedback loop. In terms of general health, the frequency of Abstainers who rated their health conditions as good or very good was higher than the other classes. Therefore, Abstainers were characterized by being less involved insituations of violence and displaying better general and mental health conditions.
Drinkers, who had a use pattern significantly different from Abstainers, especially when it came to alcohol, were, however, similar to Abstainers regarding the other problems: most were not involved with violence, neither as an aggressor nor as a victim. The only exception, as well as the Abstainers, was the high prevalence of adolescents from both classes who reported having been bullied. However, when the bullying is described in behavioral terms, few claimed to have been offended strongly enough in order to feel embarrassed/humiliated, which may indicate a varied and inaccurate understanding of the concept of “bullying” among the investigated adolescents. Regarding mental health aspects, the number of Drinkers with a frequency of worrisome problems (always or almost always) was below the other groups with a more accentuated use pattern. Additionally, most rated their general health as good or very good. These results suggest that Drinkers are not at a higher psychosocial risk than Abstainers. Therefore, we can speculate that the use of substances in this class would be within a normative pattern, exploratory/experimental, without representing significant personal or social–relational harm (Brown et al. Reference Brown, McGue, Maggs, Schulenberg, Hingson, Swartzwelder, Martin, Chung, Tapert, Sher, Winters, Lowman and Murphy2008). However, there is indeed a need for longitudinal studies in order to verify whether this trend will persist over time.
Conventional Drug Users, who displayed a pattern of alcohol use similar to Drinkers, differed from them regarding the use of tobacco or marijuana, reporting the use of these substances at a relatively higher frequency. Conventional Drug Users tended to combine the use of two substances, usually alcohol–tobacco and alcohol–marijuana and, also, tobacco–marijuana. Regarding the correlated problems, we could observe average proportions in most of the investigated variables, in comparison with those of the other groups. This class encompassed the largest number of adolescents in the sample, representing, maybe, an average profile of adolescents within the age group covered in the study. However, it is important to mention that, despite being the majority of the investigated adolescents, this class did not necessarily refer to a non-harmful (“healthy”) pattern of substance use. In this class, the use of the investigated substances happened at a high frequency (more than ten times in the past 30 days) for many of the adolescents: alcohol (18%), tobacco (12%), marijuana (12%) and crack (2%). This frequency of use certainly poses risks to their full development (Meier et al. Reference Meier, Caspi, Ambler, Harrington, Houts, Keefe, McDonald, Ward, Poulton and Moffitt2012; Meruelo et al. Reference Meruelo, Castro, Cota and Tapert2017). Moreover, involvement insituations of violence, as an aggressor or as a victim, was also higher within this class, in comparison with Abstainers and Drinkers, indicating an association between the more frequent use of substances and a higher exposure to violent situations. However, in terms of health, both mental and general, Conventional Drug Users were always within the average, reinforcing the idea that this class represents the “average adolescent”, with risk behaviors, but without letting them significantly affect their mental and/or physical health.
Regarding Polysubstance Users, it is important to note that adolescents within this class tended to make combined use of three specific substances, alcohol–tobacco–marijuana, at a significantly higher frequency than the three previous groups, which points to a more problematic pattern of use, characterizing 23% of the sample. Classes with a similar pattern of use were also identified in previous studies (Silveira et al. Reference Silveira, Green, Iannaccone, Kimmel and Conway2019; Tomczyk et al. Reference Tomczyk, Isensee and Hanewinkel2016). A significantly higher involvement in violence, as an aggressor or a victim, was associated with this pattern of use, in comparison with Abstainers, Drinkers and Conventional Users. In addition to the involvement in violence, studies indicate that problematic patterns of substance use are also associated with other delinquent behaviors (Komatsu et al. Reference Komatsu, Estevão and Bazon2018; Le Blanc and Bouthillier Reference Le Blanc and Bouthillier2003). With regard to mental and general health, this class was similar to Conventional Drug Users, showing that a more accentuated use of substances does not automatically correlate to worse health conditions, at least not for the investigated age group. However, in comparison with Abstainers and Drinkers, Polysubstance Users displayed worse health conditions. Therefore, it is important to point out that if this pattern of use persists over time, there is a risk of deterioration of the overall health condition, both physical and mental (Nelson et al. 2019). Future longitudinal studies must be carried out in order to verify this hypothesis.
Lastly, the Hard Drug Users class encompassed adolescents with a high probability of having used the four substances simultaneously in the past 30 days. Relatively to the other classes, Hard Drug Users stood out for the high probability of using crack, a drug with a stronger and more immediate effect than the other investigated substances. Additionally, the Hard Drug Users class displayed a higher proportion of adolescents with a higher frequency of use in the past 30 days (“10 or more times”). Moreover, the proportion of adolescents from this class that reported being involved in violent situations also attracts attention, being this the only class in which more than half the adolescents were involved in fights with bladed weapons (69%) or fire arms (64%). This was also the only class in which the majority (62%) reported some offense/humiliation from a colleague. These results suggest that the Hard Drug Users class represents a profile of adolescents who get actively involved with violence, being similar to other groups described in the literature, highly engaged in juvenile delinquency and characterized by personal and social deficits associated with substance abuse (Farrington, Piquero, and Jennings Reference Farrington, Piquero and Jennings2013; Komatsu, Bono, and Bazon Reference Komatsu, Bono and Bazonforthcoming; Le Blanc and Bouthillier Reference Le Blanc and Bouthillier2003; Moffitt Reference Moffitt2018).
Regarding victimization, the Hard Drug Users class also encompassed the highest proportion of adolescents at risk, being the only class in which most adolescents reported having been physically hurt by an adult family member, with almost half (45%) having suffered from violence more than three times in the past 30 days, and having suffered sexual abuse (44%). These results are in line with other studies which show a positive correlation between victimization and substance abuse in adolescence (Kobulsky et al. Reference Kobulsky, Minnes, Min and Singer2016; Vermeiren et al. Reference Vermeiren, Schwab-Stone, Deboutte, Leckman and Ruchkin2003). Additionally, many of these adolescents reported frequently feeling lonely (always or almost always) and feeling excessively stressed/worried to the point of not being able to sleep, denoting some harm to their mental health, as has been identified by other studies (Komatsu et al. Reference Komatsu, Bono and Bazonforthcoming; NIDA 2018a).
Finally, more than half of the Hard Drug Users rated their own health conditions as regular, bad or very bad, with the proportion of those who rated it as bad or very bad (31%) representing more than double the average of the sample. Therefore, the Hard Drug Users class was characterized by the highest prevalence and frequencies of use of multiple psychoactive substances and the most significant negative psychosocial conditions, being similar to classes identified in other sociocultural realities (Goldstick et al. Reference Goldstick, Stoddard, Carter, Zimmerman, Walton and Cunningham2016; Nelon et al. Reference Nelon, De Pedro, Gilreath, Patterson, Holden and Esquivel2019). Adolescents within this profile find themselves under a high risk of several negative events in their lives, requiring specific primary and secondary prevention policies.
LIMITATIONS
This study included only school adolescents, regularly attending public or private educational institutions, leaving out adolescents who have abandoned school (a relatively common phenomenon in Brazil, especially at the beginning of high school). In addition, the sample was composed of schoolchildren who completely answered the questions for all four of the investigated substances, omitting those who for various reasons did not answer the questions. Therefore, the accuracy of the results presented here depends to a certain extent on whether the non-answers were due to random reasons, in such a way that the data were not significantly biased. It should be noted, however, that no systematic error was observed in the sample, so that the large number of participants encompassed in this study should be enough to dilute non-response biases.
It should also be mentioned that other substances with relatively widespread use among Brazilian adolescents were not investigated. Although the questionnaire utilized for PeNSE 2015 touched on the use of other substances such as powdered cocaine, loló and lança perfume (a combination of ether, chloroform, ethyl chloride and perfume essence), ecstasy and oxycodone, this was done through a single question, in which marijuana and crack cocaine were repeated, not allowing us to separately investigate the exclusive use of each one.
CONCLUSION
This was the first study in Brazil to work with a considerable sample of the population and to systematically identify a typology of use of psychoactive substances during adolescence. This was also the first study to characterize each identified profile regarding involvement in violence, as an aggressor or a victim, and regarding indicators of physical and mental health, reiterating studies performed in other countries and thus contributing in a significant way to Brazilian criminological knowledge. In addition, results show clear implications for the creation of public policies. Findings highlight the importance of specific prevention and damage reduction policies and programs directed towards each target audience, considering each group displays specific characteristics.
Finally, it should be noted that most adolescents reported the use of at least one substance in the past 30 days, even though the sale of alcohol and tobacco is prohibited to anyone below 18 years of age, and marijuana and crack are illicit substances in Brazil. This evidence, in addition to showing the worrisome psychoactive substance use situation of many children still in the first half of adolescence, increasing their psychosocial vulnerability, denotes the fragility of social controls and, within this, the fragility of public health and security policies in the country, which are incapable of stopping an important part of its children from having access to these substances.
Acknowledgements
Data for this study come from the National Survey of School Health (PeNSE) conducted by the Brazilian Institute of Geography and Statistics (IBGE). We would also like to thank the United Nations Office on Drugs and Crime (Education for Justice Initiative), the International Society of Criminology, and São Paulo Research Foundation (FAPESP; process numbers 2019/09360-6 and 2019/17288-3) for their financial and logistical support to present this paper at the XIX World Congress of Criminology in Doha, Qatar.
André Vilela Komatsu is a researcher at the Center for the Study of Violence at the University of São Paulo (NEV-USP). He is also a member of the Group of Studies and Research on Development and Psychosocial Intervention (GEPDIP-USP), where he carried out his Master’s dissertation about antisocial behavior among adolescents and his PhD thesis about the development of violent behavior during adolescence. He was a visiting professor/researcher at the Group of Advanced Studies on Violence at the University of Barcelona (GEAV-UB, 2018–2019).
Rafaelle Carolynne Santos Costa is a Master’s student in Psychology at the University of São Paulo, with an internship at the Group of Advanced Studies on Violence at the University of Barcelona. She is also a member of GEPDIP-USP, where she is carrying out her Master’s dissertation about psychological assessment of juvenile offenders.
Lais Sette Galinari earned a Master’s in Psychology at the University of São Paulo. She is also a member of GEPDIP-USP, where she carried out her Master’s dissertation about psychosocial profiles among juvenile offenders.
Renato Carpio de la Torre is a Professor of Psychology at the National University of San Agustin. He is also a member of GEPDIP-USP, where he carried out his Master’s dissertation about Brazilian and Peruvian families in vulnerable situations.
Marina Rezende Bazon is Professor of Psychology at the University of São Paulo. She received her PhD in Psychology from the University of São Paulo and a Master’s in Psycho-education from the University of Montreal. She is also director of GEPDIP-USP.
