The prevalence of PG among adolescents (in the last 12 months) was higher among boys than girls (7.6% versus 1.7%) in the age group of 14 to 18 years (p < 0.001). The observations from the current study concur with the findings from previous reports about a higher prevalence in boys (7.6%) versus girls (2.0%) in Spanish adolescents within this age range (OEDA. Spanish Drug Observatory, 2019) as well as in almost all the European countries (6.3% for boys vs. 2.4% for girls on average) among from 15- to 16-year-olds who had gambled in the last 12 months (ESPAD Group, 2019). Furthermore, the prevalence of PG is in line with another US survey, being significantly higher in boys (3.3%) than in girls (0.9%) (Welte et al., 2008). Unsurprisingly, the risk of PG increases with age and is higher in males (the odds are 5 times higher); these results are consistent with prior research (Calado et al., 2017; Ekholm et al., 2014).

With respect to family characteristics, a higher economic status perceived of the household was a risk factor for adolescent PG. Contradictory evidence exists about this association. While a study found that adolescent problem gamblers are more likely to come from the lower social class (Griffiths & Wood, 2000), in another longitudinal study from the USA, Barnes et al. (2005)  found that lower socioeconomic status (SES) was weakly associated with higher levels of gambling only in females. Such outcomes could be due to the greater accessibility of gambling in neighborhoods with lower socioeconomic conditions. A higher prevalence of gambling opportunities, as well as neighborhood environmental risk, was associated with PG (Lussier et al., 2014; Pearce et al., 2008). Nevertheless, previous studies (Buja et al., 2019; Welte et al., 2008) found a dose–response association between having money (e.g., higher weekly income) and gambling among adolescents, and the study led by Welte et al. (2008) did not establish a relationship between PG and low SES. Besides, Turner et al. (2011) noted that subjects who were involved in heavy gambling were the ones who could spend the money freely or without any economic constraints. Considering these researchers, our results could be explained by the hypothesis that teenagers with higher economic status perception are likely to spend money on gambling because they perceive having enough money to spend.

Meanwhile, the hypothesis for the reason behind a working mother as a protective factor might be that these mothers being informed about adolescent risky behaviors monitor their children carefully. Although, conversely, the mother who works outside the home may not protect her children of risky behaviors, such as adolescents’ smoking, alcohol intake, and cannabis use, due to reduced supervision at home (Buja et al., 2019; Moñino-García et al., 2018). Parental monitoring may reduce the risk of adolescent gambling (Castrén et al., 2021) and may lead to long-term effects on a decreased likelihood of substance use (Kristjansson et al., 2010). The risk factor concerned with the father working as a homemaker should be considered as the possible effect of the economic status of the family as the father is unemployed. A study by Livazović and Bojčić (2019) did not establish parent employment as a significant predictor of adolescent PG, while a mother’s employment status was only found to be associated with satisfaction in family life. This study with multivariate models has shown that parent employment status could be significant for increasing or decreasing gambling risk in adolescents. As the literature is scarce, we highlight the need for additional research exploring the influence of the family environment on gambling patterns in adolescents. Thus, the family role should be further investigated (McComb & Sabiston, 2010).

According to the multivariate model, the likelihood of PG increases with cannabis use in the previous month. This result is consistent, given that the association between cannabis use and pathological gambling in adolescents is well established in the literature (Canale et al., 2016). Similar conclusions were drawn from a metanalysis by Dowling et al. (2017). Cannabis use was identified as a longitudinal risk factor for the development of gambling problems among adolescents, as well as alcohol use frequency. Nonetheless, contrary to our predictions, alcohol use was not independently related to PG. However, considerable evidence links PG to adolescent alcohol use (Buja et al., 2019; Riley et al., 2021).

To our knowledge, after controlling all the other potential predictors, this is the first study showing shopping (online or offline) and compulsive internet use are risk factors for PG in adolescents, and practicing creative activities is protective, such as playing an instrument, painting, singing, and writing (1–7 times per week). Adolescents with a compulsive pattern of the internet have a threefold higher vulnerability to PG. The new technologies play a crucial role in facilitating the development of PG in youth. Olason et al. (2011) highlight that a higher proportion of problem gamblers is found among internet gamblers in contrast to non-internet gamblers. And in an era where the youth have easier access to a range of gambling influences (Welte et al., 2008), the change from an offline gambling pattern to an online one is foreseeable (Abbott, 2017) due to its accessibility, affordability, and anonymity (Calado et al., 2017). Oksanen et al. (2019) revealed that competent gambling (online gambling, card games in general, and betting on games involving personal skill, such as billiards) was associated with compulsive internet use and further stated that the new forms of online gambling are potential risks for younger generations. Estevez et al. (2015) found significant differences between groups with and without pathological gambling with elevated scores for compulsive shopping among gamblers. This result is also supported by Kausch (2003), but the association was among adult gamblers. Conversely, no differences were found between internet addiction and gambling (Estevez et al., 2015).

This study was not without limitations. One of the most important limitations is its cross-sectional design, which means that causality among the variables cannot be assured. On the other hand, we have a small sample of girls with problem gambling (n = 18) for this study, so it is difficult to obtain a statistically significant contrast. The adjusted analysis model was for both sexes, but there were differences between boys and girls, as well as the effect of the sex variable and interaction effects (see figures in Appendix A) . Although anonymity to participants was guaranteed, social desirability bias might affect self-reporting and hence data accuracy. The study has the advantage of being a large probability sample from our region, which allows comparison with other results, although differences in the study designs for PG complicate making comparisons across studies. Also, the sample design was adapted to the universe (the target population). In addition, part of the variables studied is based on the ESPAD questionnaire that has been validated, especially for substance use questions and behavioral addiction scales, as anticipated. The ESTUDES questionnaire has not been cross-culturally adapted in the Spanish language. Finally, another advantage is establishing a pre-COVID-19 baseline for future studies on gambling in adolescent samples.

Various studies conducted in Iceland have shown that accessible and structured leisure-time opportunities contributed to an extraordinary decrease in substance use and abuse among adolescents, similar to our results. Long-term preventive interventions in family, school, peers, and a healthy leisure environment, such as parental monitoring of evening and night entertainment, increased adolescent involvement in free time activities, participation in sports supervised by adults, and encouraging children and adolescents to spend time with their families also helped in better well-being (Kristjansson et al., 2016, 2020). However, no association was observed between the practice of sports and PG in the present study. In the words of Frieden (2010), “changing the context to make individual’s default choices healthy” facilitates implementing long-term primary prevention of adolescent substance use. In a Canadian cross-sectional study (14–17 years), participation in non-sport activities (including clubs or groups in music, art, drama, photography, Scouts, or student council activities) was associated with positive academic outcomes and with a reduced likelihood of having tried alcohol, smoking, and marijuana (Guèvremont et al., 2014). It is known that alcohol or substance use and gambling share analogous characteristics, and their social consequences are quite similar (Leeman & Potenza, 2012). Canale et al. (2016) proposed that theoretical models for substance use might also apply to gamble. Likewise, as hypothesized, our results suggest that leisure habits may play a key role in reducing the probability of developing a possible problem of gambling in adolescents, which is an age in which peer pressure is high and maladaptive behavioral patterns can be molded (Monaghan & Wood, 2010). Therefore, facilitating a healthier environment could have health and social benefits in preventing gambling-related harms and future risk-taking behaviors.

As mentioned earlier, our results provide useful information for the prevention of gambling addiction, and they represent an opportunity to examine the influence of modern life leisure activities, including information and communication technologies, on problem gambling, and their repercussions in adolescents. Indeed, our results provide for the first time evidence of an association between adolescent problem gambling and several leisure time factors. They also point out the important role of parents in monitoring and paying attention to potential compulsive patterns of internet use and shopping (1–7 times per week) of their adolescent children. Further research is warranted to confirm the potential influence of leisure-time behaviors on attitudes towards gambling in teenagers in prospective study settings.

In conclusion, prevention and health promotion programs geared toward cultural hobbies, monitoring shopping, compulsive internet use, and cannabis consumption may lead to a decrease in the levels of PG in teenagers, especially in boys. Regulatory and preventive strategies are needed which should focus on environmental actions as well as individual behavioral changes at an early age. Consequently, based on our findings, we recommend regional authorities and stakeholders responsible of health promotion policies to invest in cultural activities (e.g., music activities, or story-writing and short film competitions) as potential interventions to avoid or reduce the engagement of adolescents in gambling activities, to which they are prone when they exhibit unhealthy behaviors, such as compulsive internet use or cannabis consumption. Future studies of these factors would be useful for deepening the comprehension of the problems of gambling among adolescents. Additional research is also required for gauging the causes of interregional differences in the risks associated with PG (Huang & Boyer, 2007).