Introduction

Psychoactive substance use represents a severe public health problem in many countries. A study evaluating the global burden of disease (GBD) between 1990 and 2016 showed significant increases in the GBD for several psychoactive substances on different continents. Main increases were observed in the consumption of alcohol (44%) and other drugs (56.9%) (Degenhardt & GBD 2016 Alcohol and Drug Use Collaborators, 2018). Furthermore, the GBD was significantly higher among low and middle-income countries. Particularly in Brazil, there was an increase in the GBD for alcohol, particularly among males (12%) (Marinho et al., 2018).

Substance-related disorders (SRD) are among the main public health problems in many countries due to their high prevalence in the general population, mainly among young people. In a report published by the World Health Organization, approximately three million premature deaths are due to excessive alcohol consumption, with one-third of deaths occurring in individuals aged 15–29 years (WHO, 2018). These findings align with epidemiological data that indicates rapid growth in SRD among young people (Drope & Schluger, 2018). This increase is significantly higher in low- and middle-income countries with poor prevention programs and specialized assistance.

The losses generated by SRD are significant, especially among youths (Bedendo et al., 2013, 2017). Several authors have reported that adolescents with SRD experience significantly more significant losses than adults in areas such as family, criminal offenses, problems with studying, and violent acts (Carvalho et al., 2017). In addition, adolescents have neurobiological and psychosocial differences that may increase the vulnerability of both SRD and other behavioral addictions (Bedi et al., 2021). Some authors also observed that dysregulation of dopamine reward pathways is more common among those adolescents exposed to reward stimuli than adults (Kim et al., 2011, 2020), and the fact that adolescents do not have a fully developed frontal cortex (i.e., executive functioning) may increase vulnerabilities to multiple addictions (Bedi et al., 2021). These findings indicate the possibility of a cross-addiction, in which specific neurochemical pathways are activated from environmental cues for both chemical (substance use) and behavioral stimuli (gambling, video games, smartphones, Internet, among others).

There is an increase in problematic Internet use (PIU) prevalence, especially in countries with high levels of connectivity (Andrade et al., 2020a, 2021a, b; Masaeli & Farhadi, 2021). Because of technological advances, it is easier to engage in many daily tasks associated with more extended Internet and smartphone use as they provide more types of entertainment that can act as reinforcing stimuli (Marín-Díaz et al., 2020). Although PIU is not considered a specific pathology in mental disorder manuals, several studies suggest similarities between the clinical symptoms of PIU and SRD (Berdin & Saules, 2019; Durkee et al., 2016; Lanthier-Labonté et al., 2020; Morioka et al., 2017; Razjouyan et al., 2018; Savolainen et al., 2020; Vázquez-Nava et al., 2020; Vranken et al., 2020). For example, neural areas are activated similarly to SRD and PIU stimuli (Kim et al., 2011). Other studies have also identified that problematic Internet users present alterations in the neural regions commonly associated with the clinical symptoms of SRDs, such as changes in impulsivity levels (Li et al., 2019), motivation (Brand et al., 2019), and cognitive control (Cudo et al., 2020). In addition, some studies have observed lower baseline levels of dopamine release and lower density of dopaminergic D2 receptors (Kim et al., 2011) in the ventral striatum in users with PIU, like those found in individuals with SRD (Nall et al., 2021).

Several clinical studies have been conducted to identify a possible association between SRD and PIU. In a study with 73,000 adolescents, smokers had almost three times the greater chance of developing PIU (Lee et al., 2013). In a study with more than 100,000 Japanese adolescents, a significant association was observed between SRD and PIU, with more than 55% of adolescents with PIU smoking more than 21 cigarettes daily (Morioka et al., 2016). In an innovative study, Berdin and Saules (2019) sought to identify behavioral and emotional problems in individuals who used the Internet while drinking alcoholic beverages. More than 60% of these people reported regret in some online behavior due to consuming alcohol, compared to only 30% of participants who accessed the Internet without consuming alcohol. In another study (Rücker et al, 2015), adolescents with PIU were more likely to consume tobacco (adjusted odds ratio (aOR) = 2.05), alcohol (aOR = 1.72), marijuana (aOR = 1.94), or other illegal substances (aOR = 2.73).

These findings indicate that addictions may share both clinical and behavioral symptoms, such as the feeling of anxiety due to deprivation of the reinforcing stimulus (drugs, Internet, games, smartphones, etc.). The search for behaviors that lead to the feeling of pleasure reduces time on other tasks unrelated to addictive behavior and increases the amount of time spent on addiction-related behaviors (Jin Jeong et al., 2020). Individuals with SRD and PIU also report more significant emotional problems than the general population and lower quality of life in various domains (Swendsen et al., 2010; Wartberg et al., 2016). In a 10-year longitudinal study with more than 5000 participants, those with a previous mental disorder (mainly anxiety or mood disorders) had a greater probability of experimental substance use and increased intensity of use (Swendsen et al., 2010). The authors also identified that those with mental disorders and who had already used alcohol or tobacco had a significantly higher risk of excessively consuming these substances. More recently, some authors have reported that individuals with severe levels of depression and stress are three times more likely to present PIU, and those with anxiety had an increased probability of 2.79 (Andrade et al., 2020a).

In sum, preliminary studies suggest a potential link between SRD and PIU, particularly among adolescents. However, there is a need for further research in this domain. For example, in a recent systematic review, the authors failed to identify a consistent association between PIU and the use of alcohol and marijuana among young people (Lanthier-Labonté et al., 2020). It should be noted that the number of studies assessing the relationship between SRD and PIU was rather small, which indicates the need for further research. In addition, no study has been conducted in Brazil to assess a possible association between SRD and PIU. The present study aimed to examine the possible association between PIU and substance use (alcohol and tobacco) in Brazilian adolescents. Additionally, we aimed to identify the possible influence of emotional problems (depression, anxiety, and stress) in the relationship between SRD and PIU. Based on previous research, we hypothesized that (i) there would be a significant association between PIU and substance use, (ii) adolescents who used alcohol and tobacco problematically would have a higher frequency of behaviors related to the use of digital media (time spent using the Internet, sending and receiving messages, among others), and (iii) quality of life and emotional problems would present higher levels of centrality (influence) on the use of substances and the Internet using network analysis.

Methods

This is an exploratory cross-sectional study using stratified and probabilistic sampling. The study was conducted with adolescent students from five schools randomly chosen from a pool of all schools in a municipality in the state of São Paulo, Brazil.

Sample

The inclusion criteria were as follows: (i) adolescents must have studied at one of the five schools selected for the study, (ii) they must have accessed the Internet at least once a day from any device (smartphone, computer, or tablet), (iii) participants must have provided a signed informed consent, and (iv) parents must have agreed with their child participating in the study provided a signed assent form. From an initial total of 1837 students, adolescents who had not consumed alcohol or tobacco in the last month and those classified by the DUSI instrument as risk-free users were excluded (n = 968). The final sample consisted of 869 adolescents aged 9 to 17 years (M = 13.3; SD = 2.10), with 322 consuming only alcohol (experimental or misuse), 150 consuming only tobacco (experimental or misuse), and 397 consuming both alcohol and tobacco according to the DUSI instrument.

Instruments

Sociodemographic Questionnaire

The questionnaire collected information about age, sex, education, income, and specific questions about Internet and smartphone use (time of use, number of messages sent and received, most accessed social networks, etc.).

Internet Addiction Test

The IAT was used to assess Internet use patterns. The IAT consists of 20 Likert items ranged from 0 (Never) to 5 (Always), with scores ranging from 0 to 100 points. Scores above 50 points indicate PIU. In Brazil, the IAT has been validated (Conti et al., 2012) with satisfactory internal consistency (α = 0.87).

Drug Use Screening Inventory

The DUSI was designed for adolescents with 15 specific dichotomous questions (yes/no) that assess problems associated with the use of various classes of drugs (De Micheli & Formigoni, 2002). The consumption of alcohol and tobacco refers to the last month, and participants were classified into three possible groups according to the number of affirmative responses: nonusers (no score on the instrument), experimental users (2 affirmative answers in substance use), and abusers (3 to 7 affirmative responses in substance use). As previously described in “Methods,” participants with no DUSI scores were excluded from this study. According to previous studies (Nascimento & De Micheli, 2015), we classified “experimental users” as individuals who used a substance 1 or 2 times in the last month and “abusive users” as those who consumed a substance up to 9 times a month. De Micheli and Formigoni (2002) validated the DUSI in Brazil and demonstrated high internal consistency (α = 0.86).

Depression, Anxiety, and Stress Scale-21

The Depression, Anxiety, and Stress Scale-21 (DASS-21) assesses the prevalence and intensity of depression, anxiety, and stress symptoms with seven questions for each domain. In Brazil, the DASS-21 was adapted and validated for adolescents by Patias et al. (2016). Adolescents may be classified into clinical and non-clinical groups based on their scores for each domain. The authors found high internal consistency for all assessed measures (depression, α = 0.90; anxiety, α = 0.83; stress α = 0.86).

Data Analysis

Raw data (continuous variables) were transformed into z-score (standardized), and data with three or more standard deviations were removed from the analysis. The Student’s t-test was used for independent samples, and nominal or categorical variables were analyzed using Fisher’s exact or the chi-square test (χ2).

We used a graphic exploratory technique, network analysis (NA), to estimate the grouping of different variables associated with sociodemographic and emotional measures and their influence on alcohol and tobacco consumption with PIU and other variables. The differences between boys and girls were also analyzed. This procedure enabled the analysis of the level of influence (centrality) of each variable in the system and estimated whether this process occurs directly or indirectly. NA is increasingly used to study mental disorders through direct or indirect interaction between symptoms (Spritzer et al., 2022; Andrade et al., 2021c; Borsboom, 2017).

The network items (nodes) connect using edges (partial correlations), with the green and red edges representing positive and negative correlations. Thicker edges indicate stronger correlations (Oliveira Pinheiro et al., 2020). The influence of variables on the system was assessed based on four centrality criteria: (i) the number of connections that a node has with the others (betweenness centrality); (ii) the strength of these connections (degree centrality); (iii) the mean weight of the correlations of each item with the others (closeness centrality); (iv) the weighted measure of the item’s influence in the system as if it were excluded from the network (expected influence). The analyses were performed using the open-source software JASP®, with a significance level set at 0.05.

Results

Regarding the participants’ sociodemographic characteristics (Table 1), girls had a lower rate of experimental use of alcohol and cigarettes than boys. However, no differences were found in alcohol and cigarette use patterns in terms of income, race, and type of Internet access (smartphone, computer, or tablet).

Table 1 Descriptive analysis of sociodemographic characteristics according to the patterns of substance use (alcohol and tobacco)
Full size table

Table 2 presents Internet use characteristics and between-group comparisons of the experimental and abusive use of alcohol. Participants with a higher consumption pattern of alcohol (abusive group) had a more significant duration of Internet use, time on cell phones, number of times they checked their cell phone, and time on WhatsApp compared to participants in the experimental or risk-free group.

Table 2 Internet usage pattern according to the patterns of substance use-alcohol
Full size table

For tobacco, there was a higher frequency of Internet time and total daily hours spent on the smartphone in the abusive group concerning the other groups (Table 3). Also, participants who abused nicotine had higher IAT scores and a higher frequency of all symptoms assessed by the IAT than the other participants.

Table 3 Internet usage pattern according to patterns of substance use-tobacco
Full size table

Figure 1 presents the NA related to alcohol use in girls (Fig. 1A) and boys (Fig. 1B). The data indicated that the use of alcohol (node 21) is directly related to the time spent on the Internet (positive correlation) and the frequency of checking cell phones in both boys and girls. However, the strength of these correlations was greater among girls. On the other hand, the score of the IAT instrument (node 20) was not directly connected with alcohol use among all participants. Time spent on the Internet was one of the variables with the most significant influence on the system as it presented the highest levels of centrality based on the criteria used (Fig. 1C).

Fig. 1
figure 1

Gaussian graphical model to detect the connections for substance use-alcohol (purple node) regarding sociodemographic (pink nodes), Internet usage (brown nodes), DASS-21 scores (green water nodes), PedsQL scores (blue nodes), and IAT total scores (green node) with smartphone use behaviors (blue nodes), SAS-SV (yellow nodes), and IAT scores (purple nodes) between girls (pink box) and boys (blue box). The last image reprtesents the network centrality measures of the variables between boys (blue line) and girls (red line). The more relevant nodes of the study (alcohol use and internet addiction) have been highlighted with a dotted square for better visualization

Full size image

Regarding the use of tobacco (Fig. 2), the graphs indicated a direct correlation between the score on the IAT instrument (node 20) and the use of tobacco (node 21) in both girls (Fig. 2A) and boys (Fig. 2B). Among girls, tobacco consumption was directly associated with the time spent on the Internet and the age of the participants. In the latter case, the older the age, the greater the frequency of consumption. On the other hand, a strong correlation between Internet use and tobacco consumption in boys was not identified. We also found a negative correlation between tobacco use and the frequency that boys check their smartphones. Concerning tobacco consumption, the variables most influenced the network (Fig. 2C) for both boys and girls were the participants’ income and anxiety levels.

Fig. 2
figure 2

Gaussian graphical model to detect the connections for substance use-tobacco (purple node) regarding sociodemographic (pink nodes), Internet usage (brown nodes), DASS-21 scores (green water nodes), PedsQL scores (blue nodes), and IAT total scores (green node) with smartphone use behaviors (blue nodes), SAS-SV (yellow nodes), and IAT scores (purple nodes) between girls (pink box) and boys (blue box). The last image reprtesents the network centrality measures of the variables between boys (blue line) and girls (red line). The more relevant nodes of the study (alcohol use and internet addiction) have been highlighted with a dotted square for better visualization

Full size image

Discussion

In this study, we tested a possible association between substance-related disorder (alcohol and tobacco) and PIU among adolescents. Using the NA approach, we also sought to identify whether alcohol and tobacco consumption were directly associated with PIU. To the best of our knowledge, this is the first study to assess the association between substance use and the Internet, by means of NA approach. Our main findings indicated an association between substance use and PIU. However, this association was more robust among smokers. The NA also indicated that time spent on the Internet had the most significant influence (centrality) on alcohol use, while income and anxiety had the highest centrality levels among smokers. These findings align with previous findings reported in the literature (Andrade et al., 2020b; Cruz et al., 2018; Swendsen et al., 2010). Notably, participants who abused alcohol used smartphones and the Internet more and used WhatsApp for more extended periods than the experimental group. These findings indicate that the association between alcohol and Internet use may be more specific to the use of smartphones and not the Internet itself. In a study conducted with young Belgians, there was a strong association between alcohol use and access to digital media (Vranken et al., 2020). Also, the authors detected a strong positive correlation between binge drinking and access to Snapchat and WhatsApp. In another study conducted in countries on different continents, alcohol abuse was associated with different digital media services and Facebook and Instagram were the digital media most associated with alcohol abuse among individuals in Spain, Finland, and South Korea (Savolainen et al., 2020).

These findings are relevant as they indicate that social media may maintain or increase the pattern of substance use in different ways. For instance, viewing advertisements for substances (such as alcohol and tobacco) can increase the probability of consumption (Hoffman et al., 2014). Platforms with more dynamic interactions (WhatsApp and Facebook) may increase the consumption risk due to the exchange of instant group messages and viewing content related to the topic under discussion, such as substances. In a study conducted with Mexican adolescent smokers, 32% of participants who used WhatsApp and Facebook daily reported that tobacco consumption was influenced by digital networks (Vázquez-Nava et al., 2020).

In our findings, the prevalence of adolescents with PIU was significantly higher among smokers, and there were no differences in this prevalence among drinkers. The NA indicated that tobacco use was directly related to PIU. Although our findings are in line with some studies in the literature (Lee et al., 2013; Morioka et al., 2016, 2017; Razjouyan et al., 2018), these results should be interpreted with caution for a few reasons. First, the problematic use of smartphones and the Internet are not synonymous, and relevant clinical and behavioral differences may be associated differently with each type of substance (De-Sola Gutiérrez et al., 2016). Second, tobacco is a substance commonly consumed at any time and can be associated with individuals’ daily routine such as work and study, thus increasing the likelihood of using cigarettes while on the Internet (Liu et al., 2019; Rücker et al., 2015).

Regarding emotional problems, the NA indicated that anxiety was associated with the highest levels of centrality for substance use, especially tobacco. Additionally, this influence was even more robust among girls. Our data confirm previous studies showing that higher anxiety levels are linked with SRD and the use of digital media (Berdin & Saules, 2019; Durkee et al., 2016; Lanthier-Labonté et al., 2020; Wartberg et al., 2016). Anxiety disorders are significantly higher among females (Fawcett et al., 2020), and tobacco use has significantly increased among girls in several countries (GBD 2016 Alcohol and Drug Use Collaborators, 2018; Drope & Schluger, 2018) which may help to explain the present findings.

Some limitations of our study must be considered. First, we conducted an exploratory study with a transversal collect data, which it was not possible to establish causal relationships or assess the mediating or moderating factors of emotional problems in Internet use. Second, a standardized instrument was not used to assess the problematic smartphone use, which reduces the possibility of identifying specific behaviors of adolescents using the Internet by smartphones. Third, the generalizability of our findings may be limited as participants consisted of adolescents living in a single Brazilian city. The study also had strengths, including the sample size, the data collection method (probabilistic sampling), and the field that has not yet been studied in Brazil.

Conclusion and Future Directions

The results from this study will contribute for future studies, in which we intend to (i) test the moderating or mediating role of anxiety for Internet use between alcohol and tobacco users through the structural equation modeling technique, (ii) assess the association between Internet use and other substances, and (iii) access the prevalence of the Internet and drug use over time, using new waves of data collection.

In summary, our data indicated an association between substance use, mobile devices, and the Internet, especially among smokers. The NA indicated that tobacco was directly related to PIU, especially among girls, and that anxiety was one of the most influential variables in the system. These findings are relevant as they clarify adolescents’ complex relationship between substance use and problematic Internet use.