Introduction

Gambling is a global problem that affects individuals of all ages. However, the expansion of the commercial gambling industry, the increased use of new technologies and emergence of new gambling and betting formats (e.g., online casinos) have made it easier for gambling to become part of many adolescents’ life experiences1,2. An estimated 159.6 million adolescents worldwide engage in some form of gambling-related activity, while the number of adolescents affected by problematic gambling is approximately 528,500 in Western Europe and 3,518,000 in North America1.

Gambling, along with other behavioural addictions, has been previously described3,4,5, demonstrating an association between gambling and addiction and/or use of illegal substances (e.g. cocaine, stimulants, amphetamine, tranquillizers, hallucinogens, psychedelics)6 as well as the overconsumption of legal substances such as alcohol, coffee, and energy drinks among adolescents3. A recent study in Canada on the early occurrence of gambling and substance use (SU) in a cohort of 1,594 adolescents aged 12–17 years found that adolescents who gambled with money were at higher risk of SU and exposure to other addictive behaviours6.

Furthermore, evidence shows relationships and associations between behavioural addictive disorders, such as gambling, gaming disorder, and problematic internet use in adolescents7,8,9,10, leading to new and different risk profiles among adolescents (e.g. non-gamers, gamers, recreational video gamers, participants or spectators of Esports)11,12,13. Previous studies show how digital media and internet use can lead to the adoption of risky behaviours (gambling) and the consumption of alcohol and marijuana14, finding an association between internet gaming activities and illegal SU10. Similarly, previous studies12,15,16 in adolescents showed how being an addictive gamer was associated with polysubstance use (alcohol together with medication and strong painkillers), illegal drug use (ecstasy, MDMA, amphetamines, marijuana, designer drugs), gambling12,15 and addictive use of social media16.

During the COVID-19 pandemic, social restriction measures were implemented to prevent the spread of the SARS-CoV-2 virus, which had a significant impact on adolescent mental health17. UNICEF described how the confinement of families was particularly harsh and prolonged in the case of Spain compared to other countries18, because during the first wave of the pandemic (January 31, 2020-June 21, 2020), home confinement was mandatory and enforced by security forces, and the reopening of educational institutions, cultural centres, parks, sports centres, and businesses did not return to normal until the end of the first state of emergency in June 202019. Pre-pandemic studies on adolescent SU conducted in Europe, North America and Canada20,21,22,23,24,25,26, including Spain27, –28 showed a decrease in the use of alcohol and cannabis, and other illegal drugs21,29. However, there are data suggesting that the COVID-19 pandemic may have led to an increase in adolescent substance use in Europe, Central Asia, Canada, and Australia30,31,32,33,34,35.

To the best of the present authors’ knowledge, no previous studies have reported and compared the prevalence of SU in adolescent gamblers before and after COVID-19 pandemic social restrictions, nor examined its association with socio-demographic factors, internet-and gaming-related addictive behaviours, risk perception, or substance accessibility. This study aimed to: (a) describe the prevalence of SU among Spanish adolescent gamblers, according to their demographic characteristics, videogame and internet variables and risk perception and substance accessibility, (b) determine the factors associated with SU in adolescent gamblers, (c) compare SU prevalence in the period from 2018/2019 (before the COVID-19 pandemic) to 2021 (after the COVID-19 pandemic social restrictions) using the Spanish National Survey on Drug Use in Secondary Education in Spain (ESTUDES).

Materials and methods

Study setting and participants

We conducted a cross-sectional study using individualised data taken from the 2018/2019, and 2021 ESTUDES Survey (ES). The ES is a Nationwide Population-Based Study performed by the Ministry of Health covering a representative sample of Spanish adolescents within the population aged between 14 and 18 years old. The data collection period ranged from February 2019 to April 2019 for the 2018/2019 ES, and from March 2021 to May 2021 for the 2021 ES.

The ES uses multi-stage cluster sampling, with proportional random selection of primary and secondary sampling units (towns and sections, respectively), with the final units (individuals) being selected by means of random routes and sex- and age-based quotas. There were 38,010 subjects from the 2018/2019 ES, and 22,321 from the 2021 ES. More details of the ES methodology are described elsewhere36,37. We used the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cross-sectional studies38. The Universidad Rey Juan Carlos Research Ethics Committee has confirmed that no ethical approval is required.

Measures

The sample was determined by the answers from adolescents aged 14 to 18 years old from 2018/2019 to 2021 ESTUDES who self-reported gambling with money in all forms (land-based gambling, and/or online) in the last 12 months. The main (dependent) variable was SU, which included: (a) alcohol, (b) marijuana (weed, pot) or hashish (hash, hash oil), (c) tranquillizer, sedative, and/or sleeping pill (TSSp), (d) cocaine (rock form and in any other form), (e) other illicit psychoactive drugs (LSD, non-LSD, hallucinogenic, amphetamine, metamphetamine, GHB), or (f) novel psychoactive substances (Ketamine, Spice, Mephedrone, and Ayahuasca) during the previous 12 months. The information used for creating the dependent variables (considered to be dichotomous variables) was obtained from “yes” or “no” answers to the following question: “Have you taken a [specific substance] during the last 12 months?”

The independent variables included the socio-demographic characteristics such as age (14 to 16 years, 17 to 18 years), identifying as male or female, country of birth (Spain or other), parents’ employment status (both unemployed, one employed, both employed), parents’ education level (no studies, primary education completed, secondary education completed, higher education), monthly family income (above average >€1200, around average €600-€1200, below average <€600) and size of town or city (< 10,000 habitants versus ≥ 10,000 habitants).

We also included independent variables related to video games and internet use. The type of video game user was determined based on the question: ‘In the past 12 months have you engaged in any of the following activities; a) playing video games (recreational use), b) spectating Esports, and c) participating as an Esports gamer?’ We included 9 questions based on internet gaming disorder (IGD) criteria from DSM V (American Psychiatric Association, 2013)39. We defined presence of IGD when subjects answered “Yes”, to 5 or more questions40. Also, the Compulsive Internet Use (CIU) Scale included in ES was used41. This consists of a total of 56 items, where a score ≥ 28 indicates a risk for possible compulsive internet use.

Finally, we included independent variables associated with the perceived risk of SU and availability of the substances. For the perceived risk variable, subjects were asked to give their opinions about the health effects and other problems that could result from substance use. This variable was categorized into no /few problems or quite a few/many problems. The perceived availability of the substances was categorized as “impossible/very difficult to obtain” or “easy /very easy to obtain”.

Statistical analysis

The characteristics of the sample and prevalence of each SU was calculated separately for 2018/2019 and 2021 ES dataset according to the study variables. We compared the reported prevalence for each aspect of SU from 2018/2019 ES to data from 2021 ES. To perform bivariate comparisons, we used the Chi square statistics for categorical variables. Subsequently, the logistic regression multivariate model was built using the “enter modeling” method of STATA 16.1. To build the model the recommendations by Hosmer et al.42 were followed. The process included three steps: 1º Univariate analysis of each variable; 2º Selection of variables for the multivariate analysis. We included all the variables whose univariate test was significant and those we considered scientifically relevant, according to the references reviewed; and 3º Following the fit of the multivariate model, the importance of each variable included in the model was verified. This included the examination of the Wald statistic for each variable and comparison of each estimated coefficient with the coefficient from the univariate model containing only that variable. Variables that did not contribute to the model based on these criteria were eliminated and a new model was fitted. The new model was compared to the old model using the LR test. Furthermore, estimated coefficients for the remaining variables were compared to those from the full model. This process of deleting, refitting and verifying continued until all the important variables were included in the model. There was no evidence of multicollinearity among the variables included in the models (supplementary file 1).

We repeated the abovementioned process to build the logistic regression multivariate model for each of the six substance groups of interest: alcohol, marijuana, TSSp, cocaine, other illicit psychoactive drugs and novel psychoactive drugs. All models started with all variables presented on Table 1; however, the variables that were kept in the final model (due its significance as associated factor or confounder) were different for each substance. To simplify the data presentation, we will only display on the table the variables that were kept on the final model for each substance. The level of significance adopted in all analysis was 0.05.

Table 1 Sociodemographic features, substance consumption, internet, and videogame variables among male and female adolescent gamblers (ESTUDES 2018/2019–2021).
Full size table

Results

The total number of subjects was 7,960 (1,781 female and 6,179 male). Table 1 shows the distribution by socio-demographic features, internet and video games and SU variables among male and female adolescent gamblers included in the ES conducted in 2018/2019 and 2021.

Most of the adolescent gamblers who reported substance use were males (77.23% 2018/2019; 78.52% 2021) and between 14 and 16 years of age (58.0% 2018/2019; 56.27% 2021) (Table 1). Alcohol was the most prevalent substance reported (83.55% 2018/2019; 80.15% 2021) followed by marijuana (41.32% 2018/2019; 37.38% 2021). There were no significant differences between this distribution regarding the sex and age stratification when comparing 2018/2019 to 2021 data (Table 1).

Compared to 2018/2019 dataset, the 2021 dataset presented a higher prevalence of adolescent gamblers who were Spanish-born (90.71% vs. 93.32%, P < .001); lived in cities with more than 10,000 inhabitants (86.22% vs. 90.97%, P < .001); played video games (recreational use) and were Esports spectators (25.31% vs. 28.02% and 3.6% vs. 5.35% respectively, P < .001) (Table 1). In contrast, the 2021 dataset showed a lower prevalence of adolescent gamblers with family income above average (18.79% vs. 16.76%, P = .045), those who reported CIU (69.35% vs. 65.6%, P < .001), alcohol use (83.55% vs. 80.15%, P < .001) and marijuana use (41.32% vs. 37.38%, P < .001) (Table 1).

The prevalence of alcohol consumption among adolescent gamblers decreased significantly across all age groups, both sexes, all categories of country of birth, among those with both parents employed, those with parents who completed the secondary education, those with an average family income and those from both small or large cities when comparing data from 2018/2019 to 2021 (Table 2). This decline in alcohol consumption prevalence was also observed for adolescent gamblers who were recreational video gamers, who were spectators or participated in Esports, who had no IGD, those with or without CIU, those who perceived easy access to substances and those with or without perceived health risk. The greatest reductions in prevalence were observed for Esports spectators (84.23% vs. 75.01%; change = 9.22% points; P = .020), adolescents who were born in other countries (73.21% vs. 64.45%; change = 8.76% points, P = .017) and those who lived in cities with < 10,000 habitants (86.67% vs. 79.64; change = 7.03% points, P = .021).

Table 2 Prevalence of alcohol, marijuana and TSSp use among male and female adolescent gamblers (ESTUDES 2018/2019–2021) according to sociodemographic factors, internet-videogame use, and perceived risk and availability.
Full size table

Marijuana use among adolescent gamblers decreased significantly among those aged 14–16, both females and males, individuals of Spanish origin, those with one or both parents employed, those with an average household income, and those living in large cities when comparing 2018/2019 to 2021 data. In addition, the prevalence of marijuana use reduced significantly in adolescent gamblers who played Esports, had no IGD, had no CIU and perceived no health risk for consumption. The greatest reductions in prevalences were observed for those adolescent gamblers that perceived no/few health risks associated with consumption (69.60% vs. 61.35%; change = 8.25% points, P = .001), those without CIU (43.87% vs. 31.85; change = 8.02% points, P < .001) and Esport players (41.15% vs. 34,11%; change = 7.04% points, P = .002).

The change in the prevalence of TSSp use between 2018/2019 and 2021 data was not significant, except for an increase among adolescents with an average household income (16.36% vs. 18.32%; change = 2.04% points, P = .013).

The prevalence of cocaine use decreased only among specific subgroups of adolescent gamblers between 2018/2019 and 2021 (Table 3). A significant decrease was observed among adolescent gamblers who perceived no or low health risks for consumption (25.76% vs. 15.36%; change = 10.4% points; P = .009), whose parents had completed only primary education (7.56% vs. 1.86%; change = 5.7% points, P = .028), with above-average family income (9.16% vs. 6.94%; change = 2.22% points, P = .034) and those who played video games recreationally (6.79% vs. 4.80%; change = 1.99% points, P = .018). The use of other psychoactive drugs or new substances did not show significant changes when comparing 2018/2019 to 2021 data (Table 3).

Table 3 Prevalence of cocaine, other illicit psychoactive drug and novel psychoactive substance use among male and female adolescent gamblers (ESTUDES 2018/2019–2021) according to sociodemographic data, internet-videogame use and perceived risk and availability.
Full size table

Table 4 shows the logistic regression models results for the 6 substances groups of interest among adolescent gamblers in the two study periods (2018/2019 and 2021). ‘Monthly family income’ and ‘Internet gaming disorder’ were kept in only one final model each: cocaine and novel psychoactive substances, respectively. Meanwhile, age and sex were kept in all six final models.

Table 4 Factors associated with substance use among adolescent gamblers (ESTUDES 2018/2019–2021).
Full size table

Older age, perceived ease of substance availability and no perceived health risk for consumption were all significantly associated with an increased likelihood of SU in most of the substances analysed. Moreover, the predictor that presented the greater likelihood for alcohol (OR = 2.92, 95%CI 1.95–4.37), marijuana (OR = 5.98, 95%CI 4.72–7.57) and TSSp use (OR = 2.12, 95%CI 1.70–2.65) was the perception of easy availability. Meanwhile, for the use of cocaine (OR = 5.44, 95%CI 3.57–8.28) or other illicit psychoactive drugs (OR = 3.20, 95%CI 2.09–4.90) the greatest association was attributed to no perceived risk (Table 4).

For the other predictors, the factors associated with SU among adolescent gamblers varied by substance. For example, being aged 17 years or older was associated with use of alcohol (OR 2.71, 95%CI 2.15–3.41), marijuana (OR 1.34, 95%CI 1.17–1.54), cocaine (OR 1.37, 95%CI 1.03–1.82), and other illicit psychoactive drugs (OR 1.28, 95%CI 1.02–1.59), although not with TSSp or novel psychoactive substances. Not playing videogames was associated with a higher likelihood of alcohol (OR 1.42, 95%CI 1.12–1.80) and marijuana use (OR 1.28, 95%CI 1.28 (1.07–1.53). Being female was associated with increased risk only for TSSp use (OR 2.04, 95%CI 1.65–2.52) (Table 4).

Being born in Spain was associated with a higher likelihood of alcohol use (OR 1.68, 95%CI 1.21–2.35) but appeared to be a protective factor against the use of cocaine (OR 0.63, 95%CI 0.42–0.96), other illicit psychoactive drugs (OR 0.54, 95%CI 0.38–0.77) and novel psychoactive substances (OR 0.58, 95%CI 0.40–0.85) (Table 4).

By contrast, several variables were associated with protection against SU. All levels of parental education were associated with reduced use of novel substances compared to no formal education (OR range 0.26 to 0.46). Also, having an average income was associated with protection from cocaine use (OR 0.51, 95%CI 0.26–0.99); having CIU was associated with some protection against the use of TSSp (OR 0.67, 95%CI 0.55–0.82) and new substances (OR 0.71, 95%CI 0.55–0.93). Lastly, decreased probabilities of alcohol and marijuana use were found after post-pandemic social restrictions (alcohol: OR 0.72, 95%CI 0.59–0.87; marijuana: OR 0.85, 95%CI 0.74–0.98) in adolescent gamblers, compared to pre-pandemic data (2018/2019-ES) (Table 4).

Discussion

This study described the prevalence of SU in adolescent gamblers. Our results show that the most prevalent substances are alcohol and marijuana. There has been a significant decrease in alcohol and marijuana use among adolescent gamblers between 2018/2019 and 2021. This decrease in prevalence is particularly noticeable among adolescent gamblers who attend Esports events, were born in other countries, and live in small cities in the case of alcohol; who participate in Esports, do not have an internet addiction, and do not perceive a risk in SU in the case of marijuana. There was also a significant decrease in prevalence of cocaine use but only for specific subgroups of adolescent gamblers: those who played video games recreationally, do not perceive a risk in SU, with above-average family income and whose parents have primary education. Finally, the risk factors most likely to be associated to SU in adolescent gamblers are older age, low perception of risk of consumption, and easy access to substances. Meanwhile, the risk factors associated with alcohol and marijuana use are older age, not playing video games, having little or no perception of risk, and easy access to substances.

It is worthy to note that these results of SU are related to any use in the last 12 months, so they should be interpreted accordingly. These associated factors do not consider neither predicts anything about the quantity, frequency or associated problems related to SU. Distinct scenario that accounts for further characteristics of the SU could result in distinct set of associated factors. However, it could be a starting point to delineate other studies or to plan preventive strategies.

Prior to the pandemic, the 2019 Youth Risk Behavior Survey in Connecticut43 showed that among adolescent gamblers (25%), the highest prevalence of SU was among 16–17 year olds (46%) and the highest prevalence of substances consumed were alcohol (39%) and marijuana (49%). Our results in adolescent gamblers show a decrease in alcohol and marijuana consumption between 2019 and 2021. However, there is still insufficient evidence to discuss prevalence trends among adolescent gamblers before and after the pandemic. More information is needed to propose hypotheses that could explain the causes of changes in SU consumption patterns among adolescent gamblers.

Evidence indicates both a decrease and an increase in SU among the general adolescent population. Specifically, the decline in SU—particularly alcohol and marijuana—among adolescents following the pandemic is documented in The Youth Risk Behavior Survey Data Summary & Trends Report 2013–202344, and the Monitoring the future study National Survey Results on Drug Use, 1975–202445. Conversely, the HBSC international report from the 2021/202231,32 shows an upward trend in alcohol consumption between 2018 and 2022. In addition, a study conducted in three German-speaking countries (Austria, Germany, Switzerland) showed a downward trend during the two years following the pandemic, followed by a rebound in 2022 to pre-pandemic levels or even higher, especially among adolescents aged 14–18, with an increase in the consumption of illegal substances such as cannabis30.

Previous studies6,46 have identified distinct patterns and/or trajectories of gambling and SU among adolescents aged 12–17 years with shared risk factors including family adversity (e.g., maternal age at first childbirth), maternal alcohol use, presence of hyperactivity-impulsivity, problematic behaviours, disinhibition, and lack of fear. Meanwhile, in Sweden, André et al.12 showed that the use of alcohol, illicit drugs, sedatives and strong prescription painkillers and gambling appeared in 50% of addicted gamers. In the USA, Noel et al.10 described how app-based gambling, and internet-based gambling were associated with alcohol use. Casino gambling was associated with marijuana and illicit drugs. In Connecticut (USA)43 a study of 1,807 adolescents on the prevalence of risk factors for SU and other risk behaviours described how adolescents who reported past-year gambling were at greater risk of alcohol use, marijuana use, and prescription drug use than adolescents who did not report past-year gambling. In Italy, a study involving 15,833 adolescents aged 14–17 reported a higher likelihood of consumption of energy drinks and new psychoactive substances among gamblers compared to non-gamblers3. In contrast, a Canadian study6 involving 1594 adolescents aged 12–17 found no evidence that gambling and SU influenced each other in terms of when they began or how they developed through adolescence. Among 17-year-olds specifically, no significant associations were found between the amount and type of gambling and SU involvement or abuse.

Our findings indicate that being over 17 years of age is associated with an increased likelihood for SU, aligning with previous research showing that the risk of use tends to rise with age6,47. Carbonneau et al.6 demonstrated in explanatory models that concurrent gambling and SU were influenced by both the timing of onset and developmental progression between the ages of 12 and 17, with the peak of SU occurring at age 17. Among adolescent gamblers specifically, SU was more frequent at older ages.

Previous studies have already described the association between gambling and alcohol, cannabis and illicit drug use5,9,48. Studying the presence of SU and co-occurrence of behavioural addictive disorders (gambling, gaming, internet use) and their risk factors in adolescents is relevant because these behaviours may have different origins and triggers, but may converge, leading to addiction problems2,12. Interestingly, our results show that IGD is not an associated factor with the likelihood for SU. These results do not agree with previous studies that show a positive association between gaming disorders and SU15,16,47 (alcohol, marijuana, amphetamines, sedatives, tranquillisers) and even point out how gaming could be considered a gateway to gambling49. One possible explanation in the Spanish context for the increased likelihood of alcohol and marijuana use among non-video game players could be due to increased contact with their peers and greater exposure to SU following post-pandemic restrictions.

In Spain, Fernández-Aliseda et al.50 showed that adolescents with CIU had a higher prevalence and were more likely to consume alcohol, alcohol combined with energy drinks, cannabis, sedatives, and new substances. In contrast, our results in adolescent gamblers have shown that CIU was associated with a lower likelihood of TSSp and new substances use. This lower likelihood of CIU could be because the time spent using devices would be devoted to other types of digital leisure activities unrelated to gambling or gaming, such as the use of social networking applications, including online chatting and social networking2.

In terms of potential interventions, addressing adolescents’ perception of the risks associated with SU and the perceived ease of access to substances is essential. Our findings show that low risk perception and high perceived accessibility are associated with the use of all substances—except new substances. Similar patterns are reported in relation to the use of TSSp with alcohol51, and the co-use of cannabis with TSSp in adolescents52. This increased the likelihood of co-use of TSSp with alcohol, the perceived no health risk of TSSp use, and perceived easy access to TSSp, and alcohol51. While adolescents were more likely to use marijuana with TSSp52, they were more likely to have low perceived risk and perceived easy access.

Information provision and education are usually proposed as the main intervention measures. However, research carried out in Spain53,54 on alcohol and cannabis use among adolescents showed that there is an ‘information paradox’. Adolescents who perceived themselves to be better informed and more knowledgeable were those who consumed more and had a lower sense of risk. Alternatively, participatory models could be applied where adolescents reflect on the pros and cons and negotiate realistic goals to reduce and/or eliminate SU. Whereas the degree of information in itself would not influence consumption so much, the sources from which the information is obtained would. Accordingly, de Andrés-Sánchez et al.53 showed that higher polydrug use was associated with unsupervised or informal sources, while the use of supervised sources to obtain information was associated with a reduction in consumption.

This study has some limitations. First, biases may arise from a survey design, related to questionnaire design (e.g. unclear wording, leading questions), questionnaire construction (e.g. open-ended responses, jumps between questions within the questionnaire, extended length), and derived from the use of the questionnaire (e.g. social acceptance, cultural or social rejection, recall bias). The current study may have a recall bias since the survey ask for data from the last 12 months. However, this time frame for collecting data on consumption and gambling in the last 12 months is commonly used in validated international surveys such as the European School Survey Project on Alcohol and Other Drugs (ESPAD)21. The National Survey on Drug Use and Health55 and the National Epidemiological Survey on Alcohol and Related Conditions (NESARC)56 in the United States; and national surveys such as the one used in this study, the Spanish National Survey on Drug Use in Secondary Education in Spain (ESTUDES)37. Another potential bias is that SU was assessed through self-report, which may have led to an underestimation of prevalence due to adolescents’ potential reluctance to disclose SU, influenced by sociocultural factors. This could counteract the increased probability of type I error caused by the large sample size, the lack of correction for multiple comparisons and the stepwise modelling approach. However, due to type I error, some of the observed differences may have occurred by chance, and the magnitude of the associations should be interpreted along with their confidence interval with caution. Second, the data were collected cross-sectionally and included a limited range of associated factors. The study design prevents us from determining the direction of the associations found; consequently, no causality interpretation can be drawn based on our results. Moreover, there might be others associated factors that were not included in the ESTUDES survey. Future research should collect longitudinal data in order to make more robust inferences of risk. Also, the surveys only allow adolescents to self-report whether they are male or female; there is no other option to analyse other genders. Finally, future studies should incorporate a comparative analysis of the factors associated with SU in adolescent gamblers and non-gamblers to better understand how gambling status modifies SU risk.

Conclusions

Our results show a reduction in alcohol and marijuana use between 2018/2019 and 2021 among adolescent gamblers, however, there is an increased likelihood of SU among 17-18-year-old gamblers who do not perceive the risk of SU and who can easily access SU. They do not present significant association with variables related to other addictive behaviours associated with video game and internet use.

These results have important implications for the prevention and management of substance use in adolescent gamblers. They highlight alcohol and marijuana as the most prevalent substances and underscore the need to develop targeted programs addressing higher-risk substances. Additionally, there is a clear need to continue and strengthen awareness campaigns about the health and social risks of drug use. Finally, these findings call for incorporating into scientific, political, and security discussions the urgent need to reevaluate measures controlling adolescents’ access to both legal and illegal substances.