رفتار ضد اجتماعی در نوجوانان و مصرف مواد: تجزیه و تحلیل طولی

Abstract This study explores how antisocial behavior among adolescents at age 14 is related longitudinally to their daily smoking, heavy alcohol use, and illicit drug use (hashish and amphetamines) at age 17. The sample of 9th graders (n=1293) attending compulsory schools in Reykjavik, Iceland participated in the study and in the follow-up 3 years later. The focus is on a subgroup of 17-year-old adolescents who had not experimented with cigarette smoking, alcohol consumption, or illicit drug use at age 14. Even after eliminating from the study those who had experimented with smoking at age 14 and those whose peers smoked, the adolescents who showed more signs of antisocial behavior at age 14 were more likely to smoke daily at age 17. Similar findings were revealed for illicit drug use at age 17. Further, with regard to alcohol use, adolescents who had not experimented with alcohol but showed indications of antisocial behavior at age 14 were more likely to drink heavily at each episode at age 17 if their parents drank.

Introduction Among the several risk factors for adolescent substance use (see overview by Hawkins, Catalano, & Miller, 1992) is antisocial behavior (e.g., Durant et al., 1997, Elliott et al., 1985, Jessor & Jessor, 1977 and Kandel et al., 1986). In fact, aggressive behavior in childhood predicts antisocial behavior, which relates to substance use in adolescence Block et al., 1988, Kellam et al., 1983 and Pulkkinen & Pitkänen, 1994. However, the direction and nature of the causal relationships between antisocial behavior and substance use remain obscure (e.g., Brook et al., 1992, Brook et al., 1996, Otero-Lopez et al., 1994, Robins & Wish, 1977, Rutter et al., 1998 and Windle, 1990). Moreover, few studies have explored whether antisocial behavior among adolescents who have not experimented with drugs is associated with their later substance use. In the literature about alcohol expectancy, some have argued that in order to avoid a potential artifact of concurrent associations between variables in longitudinally designed studies, we must explore expectancies in a sample of nonusers to predict later substance use Goldman et al., 1991 and Leigh, 1989. From both a theoretical and a methodological perspective, this focus is apparently lacking in the literature on adolescent antisocial behavior and substance use. Accordingly, the main purpose of this study is to explore antisocial behavior among nonusers at age 14 to predict their substance use at age 17. By excluding adolescents who have already experimented with substances at age 14, we can see more clearly whether and how antisocial behavior during early adolescence relates to becoming a substance user. Parental and peer influences have been shown to be important in adolescent substance use Ary et al., 1993, Chassin et al., 1986, Newcomb et al., 1983 and Peterson et al., 1994, as well as in adolescent antisocial behavior Jessor & Jessor, 1977, Patterson, 1996, Rutter et al., 1998 and Vuchinich et al., 1992. As recent studies claim that environmental factors, such as parental and peer cigarette smoking and alcohol use, are more effective than personal factors in predicting adolescent use of these substances (Doherty & Allen, 1994), surprisingly few longitudinal studies have focused simultaneously on adolescent antisocial behavior and the parental and peer use of substances. Therefore, we feel it important to take these factors into account in our model. In addition, we explore this relationship for the use of different substances, namely tobacco, alcohol, hashish, and amphetamines, as research has shown that antisocial behavior relates differently to the use of various drugs Otero-Lopez et al., 1994 and Windle, 1990. For example, Windle's study indicates that antisocial behavior (theft, fighting, vandalism) at age 14–15 predicts the use of alcohol and marijuana at age 18–19 but does not predict either cigarette smoking or illicit drug use. Moreover, in our previous analyses of Icelandic data, we have found that both environmental (Adalbjarnardottir & Hafsteinsson, 2000) and personal factors (Adalbjarnardottir & Rafnsson, 2001) serve as different risk factors with regard to the substance in question. Our main hypothesis is that adolescents who report more antisocial behavior at age 14 will be more likely to smoke daily, to drink heavily at each episode, and to use illicit drugs at age 17. We expect each of these findings even after eliminating from our analyses those participants who had experimented with the substance in question at age 14. Furthermore, we anticipate these findings even after taking into account the influential factors of both parental and peer smoking and drinking.

Results 3.1. Descriptive statistics Table 2 shows the proportion of adolescents reporting each substance use at age 17 by their own previous substance use, as well as those of peers and parents at age 14. Those who had tried each of the substances in question were eliminated from these analyses. In addition, those who had peers who smoked cigarettes or drank alcohol at age 14 were eliminated from these analyses. As Table 2 reveals, at age 17 heavy drinking (five or more glasses at each episode) had the highest prevalence (30.0%), followed by daily smoking (15.7%), hashish use (28.4%), and amphetamine use (14.2%). The mean of the antisocial behavior measure at baseline was 7.07 (S.D.=6.07). Table 2. Proportions (%) or mean values for predictor variables at age 14 as a function of case or noncase status of substance use at age 17 for nonusers of each substances at age 14 Results for hashish and amphetamine use are based on the number of 17-year-olds who had tried each substance in the previous 12 months. Variables at age 14 Substance use at age 17 Sample (%) Daily smoker Heavy drinker Hashish use Amphetamine use Yes No Yes No Yes No Yes No Sample – 15.7 84.3 30.0 70.0 28.4 71.6 14.2 85.8 Gender Boys 49.3 20.1 79.9 38.9 61.1 37.0 63.0 17.1 82.9 Girls 50.7 10.7 89.3 21.2 78.8 21.7 78.3 11.9 88.1 Smoking Yes 44.4 –a – 42.2 57.8 49.5 50.5 23.9 76.1 No 44.4 – – 28.2 71.8 15.6 84.4 8.0 92.0 Drinking Yes 58.8 23.3 76.7 –b – 44.2 55.8 21.7 78.3 No 12.1 87.9 76.7 – – 10.4 89.6 5.7 94.3 Peers smoking Yes 23.0 –a – 28.6 71.4 59.6 40.4 32.7 67.3 No 77.0 – – 30.3 69.7 21.9 78.1 10.1 89.9 Parent smoking Yes 55.0 20.3 79.7 37.9 62.1 35.7 64.3 19.0 81.0 No 45.0 12.7 87.3 25.0 75.0 22.5 77.5 9.8 90.2 Peers drinking Yes 15.0 31.2 68.8 –b – 59.3 40.7 32.0 68.0 No 85.0 14.9 85.1 – – 25.0 75.0 11.8 88.2 Parent drinking Yes 75.7 15.6 84.4 32.8 67.2 32.8 67.2 15.9 84.1 No 24.3 16.9 83.1 25.4 74.6 17.7 82.3 10.1 89.9 Antisocial behaviora 7.06 7.04 4.69 6.14 3.76 9.29 4.78 9.75 5.63 a These cases were eliminated from the analysis of daily smoking at age 17. b These cases were eliminated from the analysis of heavy drinking at age 17. Table options In order to explore relationships between the demographic characteristics and antisocial behavior at age 14, we conducted analyses of variance (ANOVAs) for Gender×Family structure (lives with both biological parents or other arrangement)×SES (low, medium, high). Statistically significant main effects were not found for family structure or SES. Nor was there an interaction between the variables in the analysis. However, a significant difference was found for gender, F(1, 978)=117.05, P<.001; η2=0.11, with boys reporting more antisocial behavior (Mean=9.13, S.D.=6.92) than girls (Mean=5.16, S.D.=4.35). 3.2. Antisocial behavior and substance use Multiple logistic regression was conducted to investigate the influences of antisocial behavior on adolescent use of each of the substances. Table 3 illustrates the results of logistic regressions on the four models of substance use at age 17 (daily smoking, heavy alcohol use, hashish use, and amphetamine use) for the various predictors at age 14. All the results are reported with regard to B, the standard error of B, and the associated log odds ratio. In addition, we drew 95% confidence intervals (CIs) around the estimated effects of the independent variables on substance use in order to avoid some of the problems involved in statistical significance testing (Schmidt, 1996). Results for gender are presented in Table 3, but results for SES and family structure were not included in the analyses as these variables did not relate significantly to adolescent substance use. Table 3. Longitudinal analyses: multiple logistic regression predicting adolescent substance use at age 17 for nonusers at age 14 The Wald chi-square was used to test the significance of individual parameters. Boys were coded as 0 and girls as 1. CI=confidence interval; lower=lower bound of 95% CI; upper=upper bound of 95% CI. Model and variables B S.E. Exp(B) Lower (95% CI) Upper (95% CI) Daily smoking a Gender −0.17 0.32 0.84 0.45 1.57 Parental smoking 0.62 0.28 1.86 1.07 3.23 Antisocial behavior 0.08 0.03 1.08 1.02 1.15 Heavy drinking b Gender −0.30 0.29 0.74 0.42 1.30 Parental drinking 0.84 0.34 2.31 1.20 4.47 Antisocial behavior −0.02 0.06 0.98 0.88 1.10 Parental drinking×Antisocial behavior 0.22 0.07 1.25 1.08 1.43 Hashish use c Gender −1.04 0.22 0.35 0.23 0.55 Drinking at age 14 1.29 0.25 3.63 2.24 5.89 Smoking at age 14 1.20 0.23 3.31 2.10 5.21 Antisocial behavior 0.09 0.02 1.10 1.05 1.14 Amphetamines use d Gender −0.25 0.25 0.78 0.47 1.28 Drinking at age 14 1.10 0.32 2.99 1.60 5.58 Smoking at age 14 0.68 0.27 1.97 1.16 3.34 Antisocial behavior 0.07 0.02 1.08 1.03 1.12 a Model χ2(4, N=417)=16.62, P<.001. b Model χ2(4, N=323)=33.65, P<.0001. c Model χ2(4, N=741)=207.25, P<.0001. d Model χ2(4, N=753)=78.05, P<.0001. Table options 3.3. Cigarette smoking As Table 3 shows, antisocial behavior predicted daily smoking at age 17, even after controlling for the effects of parental smoking and eliminating those adolescents who had tried smoking at least once in the previous 12 months and had peers who smoked at age 14. Special attention was devoted to exploring predicted probabilities for those who reported many antisocial acts (1 S.D. above mean) and those who reported few of them (1 S.D. below mean) at age 14 when other variables were held constant. The results indicated that the probability that a 17-year-old boy will smoke daily is 18% (95% CI=11–27%) if he reported a high level of antisocial behavior at age 14, but the probability is 7% (CI=4–12%) that he will smoke daily if he showed a low level of antisocial behavior. Predicted probabilities of daily smoking for a girl with the above attributes are 21% (CI=13–32%) and 9% (CI=5–15%), respectively. 3.4. Alcohol use As Table 3 indicates, the model for heavy alcohol drinking at age 17 suggested that the relationship between antisocial behavior and drinking at age 17 varied with parental drinking. To depict this interaction, probabilities were calculated on antisocial behavior one standard deviation below the mean, at the mean, and one standard deviation above the mean. The significance of the simple slopes of each regression lines were then examined (Aiken & West, 1991). As Fig. 1 illustrates, those who reported a high level of antisocial behavior at age 14 and had parents who drank were more likely to drink heavily 3 years later than those who reported a low level of antisocial behavior at age 14 (B=0.20, Wald(1)=17.79, P<.0001). By contrast, the antisocial behavior at age 14 of those whose parents did not drink was unrelated to heavy alcohol use at age 17 (B=0.00, Wald(1)=ns). Predicted probabilities of adolescent heavy drinking at age 17 as a function of ... Fig. 1. Predicted probabilities of adolescent heavy drinking at age 17 as a function of antiosocial behavior and parental drinking at age 14. Figure options 3.5. Illicit drug use Finally, as Table 3 shows, there was a main effect of antisocial behavior for both hashish and amphetamine use, indicating that 14-year-old adolescents who reported a high level of antisocial behavior were more likely to have tried hashish and amphetamines 3 years later. For a boy who reported a higher level of antisocial behavior, the probability was 23% (CI=17–29%) that at 17, he would have tried hashish and 9% (CI=6–12%) that he would have tried amphetamines. The probabilities were 9% (CI=6–12%) and 4% (CI=3–6%), respectively, if he showed a lower level of antisocial behavior. Predicted probabilities for a girl with higher levels of antisocial behavior were 9% (CI=7–13%) for hashish and 7% (CI=5–10%) for amphetamine use and 3% (CI=2–5%) and 3% (CI=2–4%), respectively, for a lower level of antisocial behavior.