Adolescence is considered to be a significant period in the development of health risk behaviors, such as alcohol drinking or tobacco smoking (Doran et al. 2012). The transition from middle school to high school, along with the transition to adolescence from young adolescence, increases the likelihood of the development of adjustment problems (Ge et al. 2001) and substance use (Costa et al. 1999; Crews et al. 2007; Liao et al. 2013). This transitional process typically occurs in ninth grade. Data suggest that ninth grade is considered to be the most difficult one in the secondary school, since students have to navigate in an environment which is very different from the elementary school (Black 2004). The stress stemming from the transition to secondary school contributes to the perceived stress of the biological and psychological changes in this period (Costa et al. 1999). Moreover, based on the results of previous studies, ninth grade age (14–16 years) is particularly important in the development of smoking (Liao et al. 2013) and the co-occurring smoking and drinking (Johnson et al. 2009; Lotrean et al. 2009).

The use of alcohol and tobacco in this adolescent transition period are typically studied separately (Latvala et al. 2009; SAMSHA 2005); and the concurrent use of these substances is a relatively new area of research (Anthony and Echeagaray-Wagner 2000; Chun and Chung 2013; Grucza and Bierut 2006; Jackson et al. 2002; Johnson et al. 2009). The joint use of alcohol and tobacco in adolescence is more likely to persist into adulthood than the single use (Anthony and Echeagaray-Wagner 2000). Moreover, the health problems (such as oral cancer) caused by the joint use of these substances are more severe by adulthood than the problems caused when alcohol and tobacco is used alone (Ferreira Antunes et al. 2013; Muwonge et al. 2008). This highlights the public health importance of joint alcohol drinking and tobacco smoking in adolescence.

Studies on the prevalence of joint use of alcohol and tobacco are available (Grucza and Bierut 2006; Jackson et al. 2002; Lotrean et al. 2009; Wetzels et al. 2003). These studies, however, investigated the potential associations of alcohol and tobacco use with aggressive behaviors in isolation. Data from these investigations indicate that alcohol or tobacco use can be associated with the development of aggressive behaviors (Whalen et al. 2001; Lightowlers et al. 2014). An association between the use of alcohol and tobacco and aggressive behavior in adolescence has been proposed in the literature (OAS 2005; Roberts et al. 2010; White et al. 2013). Specifically, in the pertinent studies the development of adolescent problem behavior complex (OAS 2005) and of impulsive-aggressive behaviors (Walther et al. 2012) were reported in association with alcohol drinking and tobacco smoking, respectively. According to findings of recent longitudinal studies, the association can be bidirectional (White et al. 2013). In terms of aggressive behaviors, physical aggression was the most commonly studied type of aggressive behavior that was investigated in research into adolescent aggression (Skara et al. 2008).

The associations between adolescent drinking and physical aggression co-vary with socio-cultural factors, such as race, family, academic performance and gender (Mercado-Crespo and Mbah 2013). Results show a varying pattern (Chun and Chung 2013): physical aggression was associated with drinking for boys, and with smoking for girls (Chun and Chung 2013). An association was demonstrated between physical aggression and smoking among adolescent girls; furthermore, association between aggression and drinking was also reported for both genders (Skara et al. 2008).

Our aim was to investigate the relative importance of the single use of alcohol and tobacco and the joint use of these substances in the association with physical aggression in the transition period from early adolescence to adolescence among 14–16 years old students, currently attending ninth grade in secondary school.

Methods

Study Sample and Data Collection

In order to investigate these associations among students in the transition period from early adolescence to adolescence, a cross sectional survey was performed in a representative sample of ninth grade classes of the state-run public secondary schools in Budapest. There were 469 state-run secondary schools in Budapest, with 9th grade classes in the school year of 2009/2010; out of these we randomly selected 55 schools for further investigation. Thirty-nine of these, with a total of 40 classes comprising 1206 students, agreed to participate. Out of the 1206 students, 944 yielded data for the current investigation; data were not available for the rest of the sample due to absence from school at the date of data collection (n = 182) or lack of parental consent (n = 80).

Data collection with self-report questionnaires was conducted in the classrooms with the permission of the school principals. Data were collected by experienced professionals, who participated in a special training prior to data collection to ensure validity and to reduce biases, which may occur in classroom contexts (Burke and Sass 2008; Creswelll 2012). Teachers were not present during the data collection. The administration of questionnaires took approximately for 45 min. Approval to the study was granted by the central competent authority supervising research in educational institutions in Hungary. All subjects participated voluntarily and all data were collected anonymously. After being informed of the nature of the study, parents provided written informed consent.

Measures

Buss-Perry Aggression Questionnaire (AQ) (Buss and Perry 1992)/Physical Aggression

The AQ is a self-report instrument for assessing the aggressive traits of personality (Gerevich et al. 2007). We used the Physical Aggression subscale of the AQ. Both laboratory and real life studies demonstrated the construct-validity of the subscale. Physical Aggression subscale scores displayed a significant correlation with the manifest real-life occurrences of physical aggression (Bushman and Wells 1998). The instrument is suitable for assessing adolescents (Ang 2007; Miotto et al. 2008; Ongen 2010). A validated version of the scale is available in Hungarian language (Gerevich et al. 2007). The subscale consisted of 9 items. The alpha coefficient was generally high for the summary score in our sample (α = 0.90).

European Version of the Adolescent Assessment Dialogue (Euro ADAD)/Substance Use

EuroADAD is a semi-structured interview designed to assess the multidimensional profile of problems of adolescents (Czobor et al. 2011). We used the items measuring alcohol drinking and tobacco smoking from the Substance use problem area and the variable last year academic performance from the School problem area that we used for independent variables and covariates, respectively (see below).

Variables

Based on the above data, the presence of current drinking and smoking were established (yes/no) for the past 30 days prior to data collection.

Trait-aggression was investigated in two different ways. First, we used the subscale score of the Physical Aggression subscale of the Buss-Perry Aggression Questionnaire continuously. Second, we used the subscale score categorically, based on dichotomously recoded values. Specifically, we considered scale values below 27 points as ‘low’, and 27 points or more as ‘high’ level of physical aggression. The rationale for the 27-points threshold was established a priori. In particular, for the Buss-Perry Aggression Questionnaire the mid-point of the item score-range has a value of 3 (‘Neither uncharacteristic nor characteristic’). Considering a uniform distribution of ratings across the 9 constituting items with medium severity (i.e., 3 points) would result in a total score of 27 for this subscale; we therefore used this cut-off point (27) as described above.

Statistical Analyses

For descriptive statistical analyses, we applied analysis of variance (ANOVA) for continuous- and Chi-square test for categorical variables.

Our primary analysis investigated the relationship between trait-aggression as a continuous variable and the concurrent use of alcohol and tobacco. In our first analysis we used a Generalized Linear Integrated Mixed Model (GLIMMIX); the principal dependent variable (Buss-Perry Aggression Questionnaire AQ subscale score) was applied as a continuous measure, considered as the primary endpoint of the current study. Variables of current smoking and current drinking served as explanatory (independent) variables in the model. The interaction between the main effects of smoking and drinking was also included in the GLIMMIX model. Gender was used as a covariate.

In the secondary analysis we applied physical aggression as a categorical variable, considered as a secondary endpoint to investigate the association between the use of alcohol and tobacco and physical aggression. The rationale for applying the physical aggression score as a categorical variable was twofold: First, we expected that using this approach we can capture potential nonlinear associations; second, we also wanted to examine the robustness of the findings. The interaction between the main effects of the variables of smoking and drinking was also included as an independent variable in this analysis, using the log-link function, since our dependent variable, physical aggression represented a dichotomous variable (PA‘low’/‘high’). For effect sizes we used the odds ratio (OR) statistic.

In our ancillary analysis we included the effect of academic performance as an additional covariate in the model.

The Statistical Analysis System (SAS) for Windows (version 9.1; SAS Institute, Cary, NC) and the Statistical Product and Service Solutions (SPSS) for Windows (version 20.0; IBM) were used for all statistical analyses. The data management and descriptive statistical analyses were conducted in SPSS, the SAS was used for inferential statistical analyses. The (Type I Error) level of 0.05 was adopted to all analyses for statistical significance.

Results

Descriptive Statistics

The mean age was 15.03 (SD = 0.77) years, and 48.5 % of the participants were boys. Approximately half of the pupils (50.7 %; N = 478) did not use alcohol or tobacco in the past 30 days. Altogether 29.6 % (N = 279) of the students smoked at least one cigarette in the past month and 41.4 % (N = 391) of them were current drinkers. Within the total sample, 7.9 % (N = 75) of the students used only tobacco, 19.8 % (N = 187) used only alcohol and 21.7 % (N = 204) used both substances in the last month (Table 1).

Table 1 Basic demographic characteristics of the sample
Full size table

Relationships Between Smoking and Drinking and Physical Aggression (As a Continuous Variable)

Results of the primary GLIMMIX analysis, controlled for gender, showed that both drinking and tobacco smoking had a relationship with physical aggression (F = 36.54, df = 937, p < 0.001 (smoking) and F = 8.15, df = 937, p = 0.004 (alcohol)). Both current smoking and drinking were associated with higher levels of physical aggression (Table 2).

Table 2 Severity of trait-aggression as a function of smoking and drinking
Full size table

The interaction between smoking and drinking failed to reach statistical significance (F = 1.14, df = 937, p = 0.285). The joint use of these two substances was associated with a numerically higher average on the Physical Aggression subscale compared to the groups of abstainers, current drinkers and current smokers (Table 2). Thus the highest physical aggression scores were present in the group of concurrent users (26.43 (SE = 0.025) and the lowest values in the group of current abstainers (19.60 (SE = 0.016). The mean PA score in the current smokers’ group was 25.73 (SE = 0.042) points. This result indicates that the association between current smoking and physical aggression was considerably stronger than the association between alcohol and physical aggression (the least square mean difference between smokers and non-smokers is 0.26, while this difference is 0.07 between drinkers and non-drinkers). The main effect of gender – used as a covariate in this model – also reached statistical significance (F = 110.82, df = 937, p < 0.0001) (Table 2).

Relationships Between Smoking and Drinking and Physical Aggression (As a Categorical Variable)

In secondary analysis the dichotomized version of the physical aggression variable served as a dependent variable. Similar to the analysis of the continuous variable, GLIMMIX showed a significant main effect on current smoking (p < 0.0001) with no interaction. The main effect of current drinking reached marginal significance (p = 0.051). The interaction between the main effects of current drinking and current smoking and the main effect of gender did not reach statistical significance (p > 0.05). The estimated odds ratio for the occurrence of the high physical aggression among current smokers was 3.63 (CI: 2.55–5.18) (p < 0.0001). Regarding current drinking (p = 0.051) the OR was 1.42 (CI = 0.99–2.02), which represents a moderate effect size (OR < 2).

We conducted further analyses to explore the prevalence of high physical aggression values across groups of current abstainers, current smokers, current drinkers and joint users. Figure 1 shows the comparison of these groups by the estimated prevalence values of high physical aggression. Among current abstainers the prevalence of students with high physical aggression scores was 12 % (SE = 0.015). The analogous value was 39 % (SE = 0.056) among current smokers and 20 % among current drinkers (SE = 0.029). In the case of joint users the prevalence increased to 41.6 % (SE = 0.034) (Fig. 1).

Fig. 1
figure 1

Prevalence of high Physical Aggression scores based on the GLIMMIX model analysis broken down by substance use

Full size image

Pairwise comparisons indicated that the differences in the prevalence of high physical aggression scores between the four groups were significant in all cases (p < 0.005) with the exception of the groups of current drinkers and current smokers (p < 0.05).

Ancillary Analysis

In the ancillary analysis we investigated academic performance in the models we used for the primary and secondary analyses. With the inclusion of this variable the main effects remained statistically significant while academic performance showed a significant interaction with alcohol drinking (p = 0.0082). This variable showed no significant interaction either with smoking, or with the combined effect of smoking and drinking (p > 0.05).

Discussion and Conclusion

The principal finding of the present investigation was that current drinking and smoking are additively associated with physical aggression, and that this association was present regardless whether the physical aggression was examined as a continuous or as a categorical variable. The comparison of odds ratios for current smoking (3.63) and current drinking (1.42) also indicate the importance of tobacco smoking as a potential risk factor for physical aggression. This finding may reflect the fact that heavier tobacco use is more likely to develop by the age of 15, while the development of alcohol use occurs later. This interpretation is consistent with an earlier finding (Skara et al. 2008) which shows that the strength of the association between substance use and physical aggression varies with age and the specific substances used by adolescents.

Regarding the increased prevalence of aggressive behavior among joint users, one can hypothesize a possible neurobiological interaction of the two substances (i.e., additive psychoactive effects). Supporting this hypothesis, the associations between alcohol and aggression (Miczek et al. 2004; von der Pahlen et al. 2008; Ziherl et al. 2007) and nicotine and aggression (Picciotto et al. 2015; von der Pahlen et al. 2008) are well documented in the literature. Moreover, nicotine and alcohol has been shown to play a significant role in mood regulation and aggression-related behavioral states in preclinical and clinical studies (Picciotto et al. 2015). Since accumulating evidence show significant cross-talk between the cholinergic and GABAergic systems (Hendrickson et al. 2013), their additive effect on aggression is a potential explanation of our data. Our results may point out a potential aggression-provoking effect of alcohol drinking in students with high aggressive tendencies and tobacco use.

Poor academic performance had a significant interaction with current alcohol drinking in the relationship with physical aggression. This result is consistent with prior findings: drinking has been shown to be associated with poor academic performance, and development of physical aggression and delinquency and use of illegal substances (Guilamo-Ramos et al. 2005).

The fact that the prevalence of the joint alcohol and tobacco use was 21.7 % is in accordance with previous studies, which reported a prevalence of 19.8 (Johnson et al. 2009) and 22.6 % (Lotrean et al. 2009) in samples of ninth graders. According to the results of prior studies (Chung and Chun 2010; Johnson et al. 2009; Lotrean et al. 2009) there is a significant increase in the prevalence of the joint use of tobacco and alcohol in ninth grade. Furthermore, transition to ninth grade represents a significant change in the adolescents’ life. Transition to high school from the middle school co-occurs with a need for accommodation to a greater, impersonal and competitive system that requires a quick adjustment from ninth graders. This new environment poses far more complicated academic challenges, and results in an insufficient adjustment and a high rate of school dropouts in this year. This phenomenon is commonly called the ‘ninth grade bottleneck’ (Black 2004). Stress and socialization pressure and certain maladaptive behaviors, such as tobacco smoking, alcohol drinking, as well as anti-social behavior or anxiety problems and depression can develop at a higher rate during this period (Black 2004; Ge et al. 2001).

This study is subject to some limitations. First, the study was designed as cross-sectional and therefore it cannot indicate causality. Similar to other studies (Johnson et al. 2009; Lotrean et al. 2009), we defined current use on the basis of the past month period. At the age of 15 years both alcohol drinking and tobacco smoking are in their developmental phase. Therefore applying a weekly or daily time frame would not have provided a robust estimate, while the lifetime use would not provide a valid measure for this age. Furthermore, questionnaires were administered in a self-report format. The controlled conditions of data collection and the reliability and validity these questionnaires could counterweight the potential biases in self-report instruments. Moreover, prior studies demonstrated that prevalence of smoking in adolescence assessed by self-report questionnaires are nearly identical to the prevalence assessed by biological markers (Dolcini et al. 2003). Based on earlier results, Physical Aggression subscale of the Buss-Perry Aggression Questionnaire showed a close relationship with the observed manifestations of physical aggression (Bushman and Wells 1998), which ensures an established validity to that measure.

Our results pinpoint the potential importance of initiation of prevention efforts to address alcohol use and tobacco smoking, which should be implemented prior to this transition period. A dual, system-level approach on the joint use of alcohol and tobacco and aggression is expected to yield benefits (Lotrean et al. 2009), since the relationship between smoking and drinking is reciprocal in adolescence and the effectiveness of the combined approach was demonstrated in young adulthood (Wetzels et al. 2003). Furthermore, the academic and adjustment problems of ninth graders should be taken into account in order to prevent the development and stabilization of maladaptive behavior patterns including substance use and aggression.