Childhood ADHD Symptoms and Future Illicit Drug Use: The Role of Adolescent Cigarette Use

Childhood ADHD Symptoms and Future Illicit Drug Use: The Role of Adolescent Cigarette Use Abstract Objectives The aim of this study is to understand how early cigarette use might predict subsequent illicit drug use, especially among individuals with attention-deficit hyperactivity disorder (ADHD) symptoms during childhood. Methods Data were drawn from the National Longitudinal Study of Adolescent Health (Waves I–IV). The analysis sample involves participants who had not used illicit drugs at Wave I, with no missing responses for studied predictors (N = 7,332). Results Smoking status at Wave I (ever regular vs. never regular) and childhood ADHD symptoms predicted subsequent illicit drug use at Waves II to IV. No interaction effect of smoking status at Wave I and childhood ADHD symptoms was found. However, an indirect effect from childhood ADHD symptoms on illicit drug use was identified, through smoking status at Wave I. Similar results were observed for predicting illicit drug dependence. Conclusions The findings support the notion that smoking status during early adolescence may mediate the association between childhood ADHD symptoms and risk of later adult drug use. Interventions to prevent smoking among adolescents may be particularly effective at decreasing subsequent drug use, especially among children with ADHD symptoms. ADHD symptoms, adolescent smoking, illicit drug use Introduction Attention-deficit hyperactivity disorder (ADHD) is one of the most common neurodevelopmental disorders in children and often persists into adulthood (Kessler et al., 2005; Visser et al., 2014; Visser, Lesesne, & Perou, 2007). Independent of a clinical diagnosis, ADHD symptoms are associated with a number of health problems including smoking and drug use during the transition into adulthood (Charach, Yeung, Climans, & Lillie, 2011; Fuemmeler, Kollins, & McClernon, 2007; Lee, Humphreys, Flory, Liu, & Glass, 2011; Molina et al., 2013; Molina & Pelham, 2014; Sundquist, Ohlsson, Sundquist, & Kendler, 2015; Szobot et al., 2007). However, the factors underlying risk for increased drug use among individuals with ADHD symptoms are not fully understood (Molina & Pelham, 2014). Clarity regarding salient risk factors among this population could advance our understanding of the etiology of drug use, as well as inform prevention strategies. One of the strongest predictors of persistent and continued drug use during adolescence and young adulthood is earlier initiation of “milder” legal drugs (Kandel & Kandel, 2015). For instance, cigarette use during early adolescence often precedes subsequent illicit drug use during late adolescence (Palmer et al., 2009). This progression of substance use is usually referred to as a gateway sequence/hypothesis (Hanna, Yi, Dufour, & Whitmore, 2001; Kandel & Kandel, 2015; Kandel, Yamaguchi, & Klein, 2006; Mayet, Legleye, Chau, & Falissard, 2011; Merrill, Kleber, Shwartz, Liu, & Lewis, 1999; Palmer et al., 2009). The gateway sequence could explain the pathway for illicit drug use among individuals with ADHD, given their propensity to initiate cigarette smoking at an earlier age (Biederman et al., 2006; Lambert, 2005). Two studies have specifically examined the relationship among ADHD symptoms, cigarette smoking, and illicit drug use. In one study, the likelihood of illicit drug use was increased among ADHD youth who smoked compared with their counterpart who did not smoke (Biederman et al., 2006). In other words, cigarette smoking seemed to strengthen (or moderate) the association between childhood ADHD symptoms and subsequent illicit drug use. Alternatively, Lambert (2005) showed that children with ADHD symptoms were more likely to report illicit drug dependence as adults; however, the association was null after adjusting for early adolescent cigarette use. Accordingly, these findings from Lambert’s study suggests that risk for illicit drug use among those with ADHD symptoms may be mediated by cigarette smoking during early adolescence. Although both studies support the importance of ADHD symptoms and early cigarette smoking as predictors of subsequent illicit drug use, it is not clear how these factors operate to confer risk. As such, we propose a few possible scenarios. First, based on a vast literature of individual vulnerability as well as the gateway hypothesis in the field of substance use, ADHD symptoms and cigarette smoking could be independent risk factors for illicit drug use. Second, according to Biederman et al. (2006), cigarette smoking could moderate the relationships between ADHD symptoms and future illicit drug use. That is, cigarette smoking influences the strength of the relationship between ADHD symptoms and future illicit drug use. Third, based on Lambert’s study (2005), cigarette smoking could mediate, or explain, the relationship between ADHD symptoms and future illicit drug use. The results of these associations have different implications for prevention (or intervention) efforts. If ADHD symptoms and cigarette smoking are independent risk factors, then the risk factor that is more strongly associated with subsequent illicit drug use should be the target of prevention efforts. If ADHD symptoms and cigarette smoking are interacting risk factors (e.g., moderation exists), whereby the risk for illicit drug use is highest among those who both have ADHD symptoms and initiate cigarette smoking early in adolescence, then prevention efforts should focus on youth with ADHD symptoms who also initiate smoking at an early age. If cigarette smoking mediates the relationship between ADHD symptoms and illicit drug use, then early education of adolescents with ADHD symptoms on strategies to stay away from using cigarettes to self-medicate should reduce the prevalence of future illicit drug use in the population. Thus, determining how these three factors (ADHD symptoms, early cigarette use, and subsequent illicit drug use) are related would be helpful in identifying the target populations for specific prevention strategies, in addition to broad ones (e.g., prevent underage cigarette/substance use). Using data from a large longitudinal cohort, the purpose of this study was to evaluate the manner in which childhood ADHD symptoms and early cigarette smoking potentially relate to subsequent illicit drug use. Specifically, in a series of statistical models, we tested (1) if ADHD symptoms and smoking were independently associated with subsequent illicit drug use (Figure 1A); (2) if smoking moderated the association between ADHD symptoms and illicit drug use (Figure 1B); or (3) if there was an indirect effect of cigarette smoking in the relationship between ADHD symptoms and illicit drug use, suggesting a potential mediating effect (Figure 1C). We hypothesized that greater ADHD symptoms are associated with a higher likelihood of early cigarette smoking, which is further related to a higher probability of illicit drug use and drug dependence during adulthood. Figure 1. View largeDownload slide Theoretical model of this current study (A–C). Figure 1. View largeDownload slide Theoretical model of this current study (A–C). Figure 2. View largeDownload slide ADHD symptoms classes. Figure 2. View largeDownload slide ADHD symptoms classes. Methods Study Sample The study population was drawn from 20,774 adolescents in the National Longitudinal Study of Adolescent Health (Add Health), a nationally representative longitudinal cohort. Add health recruited their participants from 80 high schools and 52 middle schools in the United States. These schools were selected based on systematic sampling methods to ensure the sample is representative of U.S. schools with respect to region, urbanity, school size, school type, and ethnicity. Respondents completed in-home surveys in 1995, as well as up to three additional in-home surveys (1996, 2001–2002, 2008–2009). Detailed study design and data collection have been described elsewhere (Harris et al., 2009; Resnick et al., 1997). In this current study, we excluded participants from the analysis who either indicated that they had used (n =1,056) or were missing data on illicit drug use at Wave I (n = 55) and who had missing data on one or more of the independent variables in the model. A number of independent variables had some missing data: ethnicity (n =3), parental education (n = 955), ADHD symptoms (n =7), early history of regular smoking (n =18). Participants with missing data on one or more of these variables (listwise cases = 978) were excluded from the analysis. Accordingly, the final analysis sample involved 7,332 participants. Measures Childhood ADHD Symptoms At Wave III, participants were asked to report on DSM-IV ADHD symptoms they experienced between the ages of 5 and 12 years. Each response to DSM-IV ADHD symptom questions were collected on a four-point scale: 1 = never or rarely, 2 = sometimes, 3 = often, or 4 = very often. One DSM-IV hyperactive-impulsive ADHD symptom, “Often interrupts or intrudes on others,” was not part of the survey, and could not be included in the analyses, resulting in responses to nine inattentive (IN) and eight hyperactive (HI) symptoms. Consistent with scoring convention, a symptom was considered present if it was experienced “often” or “very often” (Fuemmeler et al., 2007; Murphy & Barkley, 1996). The Kuder–Richardson Formula 20 (KR-20) was used as a measure of internal consistency reliability for measures with dichotomous responses (e.g., presence/absence of hyperactive-impulsiveness and inattention). Reliability coefficients for the hyperactive-impulsive symptoms and IN symptoms in this study were 0.75 and 0.80, respectively, indicating an adequate internal consistency for each set of symptoms. Moreover, we have previously shown that ADHD symptoms measured in this sample have good concurrent validity. That is, individuals reporting high levels of ADHD symptoms (six or more HI or IN symptoms) were more likely to have a learning disability or behavioral problems at Wave I and were more likely to indicate having taken medication for ADHD at Wave III (Kollins, McClernon, & Fuemmeler, 2005). Cigarette Smoking A computer-assisted survey instrument was used to reduce response bias and collect data during home visits, regarding self-reported use of various nicotine and nonnicotine substances. Specifically, participants were asked to indicate if they had ever (yes/no) smoked regularly (i.e., ever having smoked at least one cigarette every day for 30 days currently or in the past). Individuals who indicated having smoked regularly at some point during their lifetime herein are referred to as “ever-regular smokers.” All other individuals were classified as “never-regular smokers.” The never-regular smokers comprised individuals who had never tried smoking, had only taken one or two puffs, who had taken puffs but never smoked an entire cigarette, or who had smoked an entire cigarette but never smoked regularly. This classification was chosen because the primary aim of the present study was to evaluate whether participants with heavier smoking patterns during adolescence predict later drug use in adulthood. Accordingly, smoking status collected at Wave I (ever regular vs. never regular) was included as a predictor of subsequent illicit drug use during Waves II–IV. Illicit Drug Use and Dependence For each wave, illicit drug use was assessed by asking participants if they had ever used illicit drugs, including cocaine, inhalants (e.g., glue or solvents), crystal meth (ice), LSD (Lysergic acid diethylamide), PCP (phencyclidine), ecstasy, heroin, mushrooms, speed, or pills, without a doctor’s prescription. A positive response to any of the above eight items defined the participant as having used illicit drugs. During the Wave IV interview, participants were further asked to report drug dependence symptoms. Three or more positive DSM-IV drug dependence symptoms (e.g., tolerance, withdrawal, spent a lot of time using) in the past 12 months were defined as “dependent.” Although participants were queried on marijuana use, this was excluded from our analysis for two reasons: it was not uniformly assessed across waves, and the legal implications of marijuana use varies by state. Other Variables Participants were asked about level of engagement in 13 conduct problem behaviors (e.g., property damage, lying to guardians, fighting) at Wave I. Each of the items was dichotomized into 0 (never) and 1 (one or more incidents) and then summed. The scale exhibits adequate reliability (KR-20 = 0.69) and has been used in previously published studies with this sample to control for the effects of conduct problems (Kollins et al., 2005; McClernon, Fuemmeler, Kollins, Kail, & Ashley-Koch, 2008). Gender, age, race/ethnicity (White, Hispanic, African American, other), and parental education (less than high school, high school or equivalent, some college, and college degree or beyond) at Wave I were included in analytic models as control variables. Statistical Analyses Using latent class modeling, we classified individuals based on their level of self-reported HI and IN symptoms. This resulted in two subgroups representing an affected class of individuals who reported high childhood HI and IN symptoms versus an unaffected class representing individuals who reported few problems in these domains as children (See Figure 2). Bayesian information criterion (BIC; the lower, the better), entropy (the higher, the better), and the Lo-Mendell–Rubin adjusted likelihood ratio test informed decisions regarding number of latent classes (Lo, Mendell, & Rubin, 2001). The average posterior probability of class membership (the closer to 1, the better) and interpretability of the classes were also considered. We tested three models aimed at evaluating (1) the independent effects of ADHD symptoms and smoking status at Wave I (ever regular vs. never regular) on illicit drug use (Model 1); (2) the interacting effects of ADHD symptoms and smoking status at Wave I on illicit drug use (Model 2), and the indirect effect of smoking status at Wave I on the relationship between ADHD symptoms and illicit drug use (Model 3). Each model included the covariates of child age, race, gender, parental education, and conduct disorder symptoms. All analyses used sample weights to correct for selection probabilities and to obtain accurate standard errors. After identifying the latent classes of ADHD symptoms, the first model (Model 1) evaluated the survival risk of illicit drug use (from Waves II–IV) regressed on classes of ADHD symptoms, and smoking status at Wave I via discrete-time survival analysis in Mplus 7.4 (Muthén & Muthén, 1998–2005). Discrete-time survival analysis identifies the hazard probability of a nonrepeatable event (in this case, the initial use of illicit drugs) from Wave II to Wave IV. Unlike Cox regression, discrete-time survival analysis treats time as discrete units or chunks rather than as a continuous variable, which is appropriate for panel survey data. In addition, discrete-time survival analysis is an optimal analytical approach for evaluating the developmental nature of substance use in adolescence because it does not assume that the shape of the survival function over time is the same for all cases. Next, in Model 2, we tested the interaction effect of ADHD symptoms and smoking status at Wave I as related to the survival hazard of illicit drug use. Finally, in a statistical mediation model (Model 3), we used the MODEL INDIRECT command in Mplus to assess whether the effect of classes of ADHD symptoms on survival risk of illicit drug use is statistically mediated (or partially statistically mediated) by smoking status at Wave I. We also calculated the proportion statistically mediated by the smoking variable (i.e., the indirect effect/the total effect*100). In addition, using logistic regression, we evaluated whether the influences of ADHD symptoms and smoking status at Wave I have a similar effect on illicit drug dependence observed in adulthood (Wave IV). For handling missing data among ever-use illicit drug across waves, full information maximum likelihood (FIML) was conducted via Mplus. Conventionally, FIML is favored over listwise or pairwise deletion because it provides more robust and efficient estimations. The rationales can be found in Enders and Bandalos (2001). Results At Wave I, participants were on average 15.5 (SE = 0.11) years of age and 19.8% categorized as ever-regular smokers. Around 50.0% of the participants were female and 65.9% White. Approximately one quarter (23.4%) of the sample had parents who had earned a college degree. Additional details on the sample characteristics are presented in Table I. Table II shows the bivariate association between studied predictors and illicit drug use at Waves II–IV and drug dependence at Wave IV. Table I. Sample Characteristics (n = 7,332a) Variable  n  % or (M, SE)  Age (mean, SE)    (15.7, 1.61)   Wave 1  7,332  (15.7, 1.61)   Wave 2  7,332  (16.7, 1.60)   Wave 3  7,332  (21.9, 1.67)   Wave 4  7,332  (28.7, 1.61)  Gender (female)  4,028  54.9%  Race/ethnicity       Hispanic  1,071  14.6%   Black  1,601  21.8%   White  4,074  55.6%   Other  586  8.0%  Parental education—college degree  1,907  26.0%  Early history of regular smoking  1,039  14.2%  Conduct problems (M, SE)    (0.7, 1.16)  Hyperactive symptoms (M, SE)    (1.3, 0.03)   0 Symptoms  2,726  37.2%   1–3 Symptoms  3,529  48.1%   4–6 Symptoms  904  12.3%   6+ Symptoms  172  2.3%  Inattentive symptoms (M, SE)    (1.7, 0.03)   0 Symptoms  4,166  56.8%   1–3 Symptoms  2,352  32.1%   4–6 Symptoms  621  8.5%   6+ Symptoms  192  2.6%  Drug use—Wave II       Ever tried any kind of cocaine  1,429  23.7%   Ever tried or used inhalants  86  1.2%   Ever used illegal drugs  196  2.7%   Ever injected any illegal drugs  24  0.3%  Drug use—Wave III       Ever used any kind of cocaine  604  8.2%   Ever used crystal meth (ice)  298  4.1%   Ever used illegal drugs  1,015  13.8%   Ever injected any illegal drugs  52  0.7%  Drug use—Wave IV       Ever used cocaine  1,138  15.5%   Ever used crystal meth (ice)  499  6.8%   Ever used illegal drugs  1,307  17.8%   Ever injected any illegal drugs  24  0.3%  Variable  n  % or (M, SE)  Age (mean, SE)    (15.7, 1.61)   Wave 1  7,332  (15.7, 1.61)   Wave 2  7,332  (16.7, 1.60)   Wave 3  7,332  (21.9, 1.67)   Wave 4  7,332  (28.7, 1.61)  Gender (female)  4,028  54.9%  Race/ethnicity       Hispanic  1,071  14.6%   Black  1,601  21.8%   White  4,074  55.6%   Other  586  8.0%  Parental education—college degree  1,907  26.0%  Early history of regular smoking  1,039  14.2%  Conduct problems (M, SE)    (0.7, 1.16)  Hyperactive symptoms (M, SE)    (1.3, 0.03)   0 Symptoms  2,726  37.2%   1–3 Symptoms  3,529  48.1%   4–6 Symptoms  904  12.3%   6+ Symptoms  172  2.3%  Inattentive symptoms (M, SE)    (1.7, 0.03)   0 Symptoms  4,166  56.8%   1–3 Symptoms  2,352  32.1%   4–6 Symptoms  621  8.5%   6+ Symptoms  192  2.6%  Drug use—Wave II       Ever tried any kind of cocaine  1,429  23.7%   Ever tried or used inhalants  86  1.2%   Ever used illegal drugs  196  2.7%   Ever injected any illegal drugs  24  0.3%  Drug use—Wave III       Ever used any kind of cocaine  604  8.2%   Ever used crystal meth (ice)  298  4.1%   Ever used illegal drugs  1,015  13.8%   Ever injected any illegal drugs  52  0.7%  Drug use—Wave IV       Ever used cocaine  1,138  15.5%   Ever used crystal meth (ice)  499  6.8%   Ever used illegal drugs  1,307  17.8%   Ever injected any illegal drugs  24  0.3%  Note. Childs age, gender, race/ethnicity, parental education, smoking, and conduct problems variables were from Wave I; ADHD symptoms were obtained from Wave III; at Wave II, the category of “ever used illegal drugs” included crystal meth as an example; at both Waves III and IV, “ever used illegal drugs” included inhalants as an example. a Sample size varies slightly depending on missing data for some variables. Table II. Bivariate Relationship between Sample Characteristics and Illicit Drug Use and Dependence at Wave IV (N = 7,332) Variable  Illicit drug use   ÷2 or t-value  Cramer’s V or Cohen’s d  Illicit drug dependence   ÷2 or t-value  Cramer’s V or Cohen’s d  Non-Use (n = 5,259)  Use (n = 2,073)  Nondependent (n = 7,077)  Dependent (n = 255)  Gender (female)  58.8%  45.1%  114.0***  0.125  55.2%  47.8%  5.4*  0.027  Race/ethnicity                   Hispanic  14.7%  14.5%  310.1***  0.206  14.6%  14.1%  42.2***  0.076   Black  27.0%  8.8%      22.4%  6.7%       White  50.5%  68.4%      55.0%  72.5%       Other  7.9%  8.3%      8.0%  6.7%      Age (M, SE)  15.78 (1.62)  15.45 (1.55)  7.84***  0.208  15.70 (1.60)  15.42 (1.64)  2.69**  0.172  Parental education                   College degree  25.5%  27.3%  2.3  0.018  26.1%  22.7%  1.5  0.014  Early history of regular smoking  10.1%  24.4%  249.1***  0.184  13.4%  35.7%  100.6***  0.117  High ADHD symptoms class  21.8%  32.7%  94.0***  0.113  24.3%  40.4%  34.1***  0.068  Conduct problems (M, SE)  0.59 (1.04)  1.00 (1.35)  12.55***  0.340  0.68 (1.14)  1.20 (1.52)  5.41***  0.387  Variable  Illicit drug use   ÷2 or t-value  Cramer’s V or Cohen’s d  Illicit drug dependence   ÷2 or t-value  Cramer’s V or Cohen’s d  Non-Use (n = 5,259)  Use (n = 2,073)  Nondependent (n = 7,077)  Dependent (n = 255)  Gender (female)  58.8%  45.1%  114.0***  0.125  55.2%  47.8%  5.4*  0.027  Race/ethnicity                   Hispanic  14.7%  14.5%  310.1***  0.206  14.6%  14.1%  42.2***  0.076   Black  27.0%  8.8%      22.4%  6.7%       White  50.5%  68.4%      55.0%  72.5%       Other  7.9%  8.3%      8.0%  6.7%      Age (M, SE)  15.78 (1.62)  15.45 (1.55)  7.84***  0.208  15.70 (1.60)  15.42 (1.64)  2.69**  0.172  Parental education                   College degree  25.5%  27.3%  2.3  0.018  26.1%  22.7%  1.5  0.014  Early history of regular smoking  10.1%  24.4%  249.1***  0.184  13.4%  35.7%  100.6***  0.117  High ADHD symptoms class  21.8%  32.7%  94.0***  0.113  24.3%  40.4%  34.1***  0.068  Conduct problems (M, SE)  0.59 (1.04)  1.00 (1.35)  12.55***  0.340  0.68 (1.14)  1.20 (1.52)  5.41***  0.387  * p < .05, **p < .01, ***p < .001. Childhood ADHD Symptom Classes The results from latent class analysis for childhood ADHD symptoms suggested that a two-class model fit best (BIC = 117,535; entropy =0.88; Lo-Mendell-Rubin adjusted likelihood ratio test = 18,764.89, p < .001), when compared with a three-class model (BIC = 100,599.39; entropy = 0.83; Lo-Mendell-Rubin adjusted LRT test = 2,542.13, p = .166). The two-class average posterior probabilities for most likely class membership were 0.98 and 0.94, respectively. A two-class model was selected for further analysis because of the higher entropy, a significant Lo-Mendell-Rubin adjusted LRT test, and good posterior probability of each assigned class. The two classes were labeled as “high ADHD symptom endorsement” (27.7%) and “low ADHD symptom endorsement” (72.3%). Illicit Drug Use Model 1 (Table III) shows the effects of childhood ADHD symptoms (low vs. high symptom endorsement) and smoking status at Wave I (ever regular vs. never regular) on risk of illicit drug use, controlling for conduct problems and demographic variables. The results indicated that both ever-regular smoking at Wave 1 (OR = 2.40, p < .001; Cohen’s d = 0.48) and childhood ADHD symptom class (OR = 1.32, p = .001; Cohen’s d = 0.15) predicted subsequent illicit drug use. Table III. Risk Factors for Experimenting with Illicit Drugs Use from Wave II to Wave IV (N = 7,332)   Model 1   Model 2   Model 3   Est.  Odds ratio (95% CI)  P  Est.  Odds ratio (95% CI)  P  Est.  Odds ratio (95% CI)  p  Drug use on                     Age  −0.20  0.82 (0.78–0.86)  <.0001  −0.20  0.82 (0.78–0.86)  <.0001  −0.20  0.82 (0.78–0.86)  <.0001   Gender (female)  −0.32  0.73 (0.63–0.85)  <.0001  −0.32  0.73 (0.63–0.85)  <.0001  −0.32  0.73 (0.63–0.85)  <.0001   Race/ethnicity                      White (referent)                      Hispanic  −0.12  0.89 (0.68–1.16)  0.389  −0.12  0.89 (0.68–1.16)  0.385  −0.11  0.89 (0.68–1.17)  0.403    Black  −1.40  0.25 (0.19–0.32)  <.0001  −1.40  0.25 (0.19–0.32)  <.0001  −1.40  0.25 (0.19–0.32)  <.0001    Other  −0.15  0.86 (0.66–1.14)  0.300  −0.15  0.86 (0.66–1.14)  0.285  −0.15  0.87 (0.66–1.14)  0.303   Parental education  0.16  1.17 (1.05–1.30)  0.005  0.16  1.17 (1.05–1.30)  0.005  0.16  1.17 (1.05–1.31)  0.005   Conduct problems  0.29  1.34 (1.24–1.44)  <.0001  0.29  1.34 (1.24–1.44)  <.0001  0.29  1.33 (1.23–1.44)  <.0001   High (vs. low) ADHD symptom class  0.28  1.32 (1.13–1.54)  0.001  0.33  1.39 (1.16–1.67)  <.0001  0.27  1.31 (1.12–1.53)  0.001   Smoking status  0.88  2.40 (1.94–2.97)  <.0001  0.96  2.62 (2.04–3.35)  <.0001  0.88  2.41 (1.94–2.98)  <.0001  High ADHD symptom class × Smoking status    0.23  1.26 (0.88–1.81)  0.203    Smoking on                 ADHD symptom class          0.64  1.90 (1.59–2.27)  <.0001    Model 1   Model 2   Model 3   Est.  Odds ratio (95% CI)  P  Est.  Odds ratio (95% CI)  P  Est.  Odds ratio (95% CI)  p  Drug use on                     Age  −0.20  0.82 (0.78–0.86)  <.0001  −0.20  0.82 (0.78–0.86)  <.0001  −0.20  0.82 (0.78–0.86)  <.0001   Gender (female)  −0.32  0.73 (0.63–0.85)  <.0001  −0.32  0.73 (0.63–0.85)  <.0001  −0.32  0.73 (0.63–0.85)  <.0001   Race/ethnicity                      White (referent)                      Hispanic  −0.12  0.89 (0.68–1.16)  0.389  −0.12  0.89 (0.68–1.16)  0.385  −0.11  0.89 (0.68–1.17)  0.403    Black  −1.40  0.25 (0.19–0.32)  <.0001  −1.40  0.25 (0.19–0.32)  <.0001  −1.40  0.25 (0.19–0.32)  <.0001    Other  −0.15  0.86 (0.66–1.14)  0.300  −0.15  0.86 (0.66–1.14)  0.285  −0.15  0.87 (0.66–1.14)  0.303   Parental education  0.16  1.17 (1.05–1.30)  0.005  0.16  1.17 (1.05–1.30)  0.005  0.16  1.17 (1.05–1.31)  0.005   Conduct problems  0.29  1.34 (1.24–1.44)  <.0001  0.29  1.34 (1.24–1.44)  <.0001  0.29  1.33 (1.23–1.44)  <.0001   High (vs. low) ADHD symptom class  0.28  1.32 (1.13–1.54)  0.001  0.33  1.39 (1.16–1.67)  <.0001  0.27  1.31 (1.12–1.53)  0.001   Smoking status  0.88  2.40 (1.94–2.97)  <.0001  0.96  2.62 (2.04–3.35)  <.0001  0.88  2.41 (1.94–2.98)  <.0001  High ADHD symptom class × Smoking status    0.23  1.26 (0.88–1.81)  0.203    Smoking on                 ADHD symptom class          0.64  1.90 (1.59–2.27)  <.0001  Note. Model 1 represents the risk of both ADHD symptom class and smoking status at Wave I controlling for the listed covariates. Model 2 represents the interaction between ADHD symptom class and smoking status at Wave I controlling for the listed covariates. Model 3 further added the path from ADHD symptom class to smoking status at Wave I to test whether ever-regular smoking serves as a mediator between ADHD class and illicit drug use. Model 2 (Table III) further shows that there was no significant interaction effect of childhood ADHD symptom classes and smoking status in predicting risk of subsequent illicit drug use (OR =1.26, p = .203; Cohen’s d = 0.13). This indicates the association between smoking status at Wave I, and future illicit drug use was not moderated (i.e., was neither magnified nor weakened) by childhood ADHD symptom classes. In addition to the main effects of childhood ADHD symptom class and smoking status at Wave I on illicit drug use, Model 3 (Table III) shows that the path from childhood ADHD symptom class to smoking status at Wave I was also significant (OR = 1.90, p < .001; Cohen’s d = 0.35). This suggests that the association between ADHD symptom class on the risk of illicit drug was partially statistically mediated by smoking status at Wave I. Follow-up indirect effect test via Mplus through “IND” command confirmed that there was a significant indirect effect from childhood ADHD symptom class to risk of illicit drug use through smoking status at Wave I (Est = 0.11, SE = 0.03, p < .001) in addition to a significant direct effect (Est = 0.27, SE = 0.08, p < .001). The total effect was also significant (Est = 0.38, SE = 0.08, p < .001). The proportion of the total effect statistically mediated by ever-regular smoking at Wave I was 29.5%. Illicit Drug Dependence Model 1 (Table IV) presents the effect of childhood ADHD symptom class and smoking status at Wave I (ever regular vs. never regular) on emerging illicit drug dependence symptoms in adulthood (Wave IV). Similar results were observed as those for risk of illicit drug use. Both childhood ADHD symptom class (OR = 1.32, p = .001; Cohen’s d = 0.15) and smoking status (OR = 2.40, p < .001; Cohen’s d = 0.48) were significantly and independently related to illicit drug dependence during adulthood. However, there was no significant interaction effect of these two predictors on drug dependence (Model 2 in Table IV). The paths from childhood ADHD symptom class to smoking status at Wave I and from smoking status to illicit drug dependence were significant. A follow-up test for statistical mediation was conducted. Results showed that there was a significant indirect effect from childhood ADHD symptom class to illicit drug dependence via smoking status at Wave I (Est. = 0.03, SE = 0.01, p = .05) in addition to a marginally significant direct effect (Est. = 0.09, SE = 0.01, p = .075). The total effect was also significant (Est. = 0.12, SE = 0.05, p = .026). The proportion of the total effect statistically mediated by smoking status at Wave I was 23.8%. Table IV. Risk Factors for Predicting Illicit Drug Dependence at Wave IV (N = 7,332)   Model 1   Model 2   Model 3   Est.  Odds ratio (95% CI)  p  Est.  Odds ratio (95% CI)  p  Est.  Odds ratio (95% CI)  P  Drug dependence on                     Age  −0.16  0.85 (0.75–0.96)  0.008  −0.16  0.85 (0.75–0.96)  0.008  −0.16  0.85 (0.75–0.96)  0.008   Gender                     Gender (female)  −0.02  0.98 (0.69–1.38)  0.901  −0.02  0.98 (0.69–1.38)  0.901  −0.02  0.98 (0.70–1.38)  0.912   Race/ethnicity                      White (referent)                      Hispanic  −0.13  0.87 (0.51–1.50)  0.628  −0.13  0.87 (0.51–1.50)  0.627  −0.13  0.87 (0.51–1.50)  0.628    Black  −1.90  0.15 (0.08–0.27)  <.0001  −1.90  0.15 (0.08–0.27)  <.0001  −1.91  0.15 (0.08–0.27)  <.0001    Other  −0.28  0.75 (0.35–1.62)  0.471  −0.28  0.75 (0.35–1.62)  0.468  −0.28  0.76 (0.35–1.63)  0.475   Parental education  0.09  1.10 (0.84–1.44)  0.510  0.09  1.10 (0.84–1.44)  0.511  0.09  1.10 (0.84–1.44)  0.510   Conduct problem  0.26  1.29 (1.14–1.46)  <.0001  0.26  1.29 (1.14–1.46)  <.0001  0.25  1.29 (1.14–1.46)  <.0001   High (vs. low) ADHD symptom class  0.41  1.51 (1.00–2.26)  0.048  0.42  1.52 (0.98–2.36)  0.062  0.41  1.50 (1.00–2.25)  0.050   Smoking Status  0.93  2.52 (1.70–3.75)  <.0001  0.94  2.55 (1.52–4.27)  <.0001  0.93  2.52 (1.70–3.75)  <.0001  High ADHD symptom class × Smoking status    0.02  1.02 (0.51–2.04)  0.951    Smoking on                 ADHD symptom class          0.65  1.91 (1.59–2.28)  <.0001    Model 1   Model 2   Model 3   Est.  Odds ratio (95% CI)  p  Est.  Odds ratio (95% CI)  p  Est.  Odds ratio (95% CI)  P  Drug dependence on                     Age  −0.16  0.85 (0.75–0.96)  0.008  −0.16  0.85 (0.75–0.96)  0.008  −0.16  0.85 (0.75–0.96)  0.008   Gender                     Gender (female)  −0.02  0.98 (0.69–1.38)  0.901  −0.02  0.98 (0.69–1.38)  0.901  −0.02  0.98 (0.70–1.38)  0.912   Race/ethnicity                      White (referent)                      Hispanic  −0.13  0.87 (0.51–1.50)  0.628  −0.13  0.87 (0.51–1.50)  0.627  −0.13  0.87 (0.51–1.50)  0.628    Black  −1.90  0.15 (0.08–0.27)  <.0001  −1.90  0.15 (0.08–0.27)  <.0001  −1.91  0.15 (0.08–0.27)  <.0001    Other  −0.28  0.75 (0.35–1.62)  0.471  −0.28  0.75 (0.35–1.62)  0.468  −0.28  0.76 (0.35–1.63)  0.475   Parental education  0.09  1.10 (0.84–1.44)  0.510  0.09  1.10 (0.84–1.44)  0.511  0.09  1.10 (0.84–1.44)  0.510   Conduct problem  0.26  1.29 (1.14–1.46)  <.0001  0.26  1.29 (1.14–1.46)  <.0001  0.25  1.29 (1.14–1.46)  <.0001   High (vs. low) ADHD symptom class  0.41  1.51 (1.00–2.26)  0.048  0.42  1.52 (0.98–2.36)  0.062  0.41  1.50 (1.00–2.25)  0.050   Smoking Status  0.93  2.52 (1.70–3.75)  <.0001  0.94  2.55 (1.52–4.27)  <.0001  0.93  2.52 (1.70–3.75)  <.0001  High ADHD symptom class × Smoking status    0.02  1.02 (0.51–2.04)  0.951    Smoking on                 ADHD symptom class          0.65  1.91 (1.59–2.28)  <.0001  Note. Model 1 represents the risk of both ADHD symptom class and smoking status at Wave I controlling for the listed covariates. Model 2 represents the interaction between ADHD symptom class and smoking status at Wave I controlling for the listed covariates. Model 3 further added the path from ADHD symptom class to smoking status to test whether ever-regular smoking serves as a mediator between ADHD symptom class and illicit drug dependence. Discussion In this study, we aimed to determine the ways in which childhood ADHD symptoms and smoking status in adolescence are related to subsequent illicit drug by explicitly testing main effects as well as moderating and indirect effects of smoking status. High levels of ADHD symptom endorsement and adolescent cigarette use are two prevalent and established risk factors for subsequent illicit drug use (Molina & Pelham, 2014; Palmer et al., 2009; Sundquist et al., 2015). The findings of this study suggest that ADHD symptoms may predict a greater likelihood of illicit drug use by increasing the risk of early cigarette smoking. The findings here elucidate the relationships among childhood ADHD symptoms, smoking status during adolescence, and subsequent illicit drug use. The use of longitudinal panel data and discrete-time survival analysis further strengthens the findings by imposing temporal ordering on some of the studied variables. As was shown by Biederman et al. (2006), youth with ADHD symptoms who also smoked cigarettes were more likely to use illicit drugs at later time points compared with ADHD youth who did not smoke. In other words, cigarette smoking moderated the association between ADHD symptoms and illicit drug use. In contrast to Biederman et al. (2006), the present study did not find evidence of a statistical interaction effect between ADHD symptoms and smoking status on subsequent drug use. Rather, the effect found here supported the potential mediating effect of smoking and helping to identify a possible explanation for why this particular relationship might occur (Baron & Kenny, 1986). In the present study, nearly a third of the total effect of ADHD symptoms on future illicit drug use was statistically mediated by smoking status at Wave I. Our findings provide support for the mediation hypothesis and suggest that preventing or reducing adolescent smoking especially among individuals with childhood ADHD symptoms (or children who demonstrate general liabilities such as poor impulse control and attention problems) could decrease the future incidence of illicit drug use and dependence. The fact that we observed these associations in a large, population-based cohort further supports the potential generalizability of these findings. The etiology of substance use and addiction is multifaceted. Although there is a growing literature focused on the gateway sequence to substance use (Kandel & Kandel, 2015; Kandel, Yamaguchi & Klein, 2006), others suggest a general liability reflected by poor self-regulation, which increases the risk for substance use during late adolescence or early adulthood (Masse & Tremblay, 1997; Moffitt et al., 2011; Tarter, Kirisci, Reynolds, & Mezzich, 2004). While the gateway sequence focuses on the developmental progression of substance use, the common liability model suggests that a general liability increases the risk of progressing to using another drug (Vanyukov et al., 2012). Our findings seem to be in line with both frameworks. On one hand, our model demonstrated a significant link between an earlier “light” substance use (early history of cigarette smoking) and a later usage of “hard” drug (illicit drug use), regardless of ADHD symptoms. On the other hand, our data show that both smoking during early adolescence and illicit drug use during adulthood were predicted by a generalized liability manifested as ADHD symptoms. Replication in other prospective cohorts would help further strengthen support the associations observed here. A limitation of the current study was the use of retrospective self-report of ADHD symptoms. Research has shown that using adult retrospective data to derive population-based estimates of ADHD prevalence may lead to an overestimate of the point prevalence (Mannuzza, Klein, Klein, Bessler, & Shrout, 2002). Thus, prospective collection of data over the life course would be preferred in future studies. Another limitation of the study was that data regarding ADHD medication taken during childhood was not available. Therefore, we were unable to control for the medication use could have had on the associations between ADHD symptoms and later drug use. Another methodological caveat that should be considered in the interpretation of the results is that we excluded individuals who reported illicit drugs at Wave I (n =1,056). This was done intentionally, by design, to ensure temporal ordering of the variables. In so doing, however, the results represent the effect of smoking during early adolescence on subsequent, but not concurrent, illicit drug use. Further, we are not able to make claims about factors that may be associated with the small numbers of adolescents who began illicit drug use early during adolescence – meaning future research is needed to examine predictors among this high-risk group. Finally, the inconsistent assessment of marijuana use across the waves did not permit us to include marijuana as an illicit drug in this study. However, given that marijuana is one of the commonly used substances among U.S. adolescents (See Dubowitz, Thompson, Arria, English, Metzger, & Kotch, 2016; Johnston et al. 2016), future research is needed to evaluate it as a predictor, and as an outcome among youth with and without ADHD symptoms. Despite utilizing longitudinal modeling, causation cannot be determined. Thus, it is important to be cognizant that observational methods for testing statistical mediation here can help generate hypotheses about how risk factors might relate to one another, but establishing a mediation role of certain variables would only be possible using experimental methods. In addition, the effect size of some paths observed in this analysis would be considered small. For instance, Ferguson (2009) suggests odds ratios that are <2 represent small effects. However, this study does contribute to the literature, in that its findings support the hypothesis that early cigarette use might mediate the relationship between childhood ADHD symptoms and future illicit drug use. More research is needed to better understand why children with ADHD symptoms may be more prone to experiment with and maintain cigarette use during early adolescence. The findings of this study can be useful in the development of meaningful public health and prevention (or intervention) efforts. For example, reducing cigarette use during early adolescence could result in lowering future illicit drug use and dependence, especially among individuals with a high liability for substance use—such as children with ADHD symptoms. Illicit drug use prevention research should capitalize on this sensitive period of development (i.e., early adolescence) by focusing on cigarette use cessation and prevention in this population. Implications and Contributions Smoking during adolescence may play a role in explaining why some children with ADHD symptoms progress to other drugs. Reducing adolescent smoking is a prudent public health strategy. However, it is possible that a reduction in smoking among adolescents with ADHD symptoms could lead to a decrease in the incidence of drug use in this high-risk population. Acknowledgments This research uses data from Add Health, a program project directed by Kathleen Mullan Harris and designed by J. Richard Udry, Peter S. Bearman, and Kathleen Mullan Harris at the University of North Carolina at Chapel Hill, and funded by grant P01 HD31921 from the Eunice Kennedy Shriver National Institute of Child Health and Human Development, with cooperative funding from 23 other federal agencies and foundations. Special acknowledgment is owing to Ronald R. Rindfuss and Barbara Entwisle for assistance in the original design. Information on how to obtain the Add Health data files is available on the Add Health Web site (http://www.cpc.unc.edu/addhealth). No direct support was received from grant P01-HD31921 for this analysis. All authors have contributed significantly to the work. Funding For analyses and manuscript development was supported by National Institutes of Health grants R01 DA030487 (awarded to B.F.F.), K07CA124905 (awarded to B.F.F.), and K24DA023464 (awarded to K.S.H.). Conflicts of interest: None declared. References Baron R. M. , Kenny D. A. ( 1986). The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology , 51, 1173– 1182. Google Scholar CrossRef Search ADS PubMed  Biederman J. , Monuteaux M. C., Mick E., Wilens T. E., Fontanella J. A., Poetzl K. M., Kirk T., Masse J., Faraone S. V. ( 2006). Is cigarette smoking a gateway to alcohol and illicit drug use disorders? A study of youths with and without attention deficit hyperactivity disorder. Biological Psychiatry , 59, 258– 264. doi: 10.1016/j.biopsych.2005.07.009 Google Scholar CrossRef Search ADS PubMed  Charach A. , Yeung E., Climans T., Lillie E. ( 2011). Childhood attention-deficit/hyperactivity disorder and future substance use disorders: Comparative meta-analyses. Journal of the American Academy of Child and Adolescent Psychiatry , 50, 9– 21. doi: 10.1016/j.jaac.2010.09.019 Google Scholar CrossRef Search ADS PubMed  Dubowitz H. , Thompson R., Arria A. M., English D., Metzger R., Kotch J. B. ( 2016). Characteristics of Child Maltreatment and Adolescent Marijuana Use: A Prospective Study. Child Maltreatment , 21 1, 16– 25. http://doi.org/10.1177/1077559515620853 Google Scholar CrossRef Search ADS PubMed  Enders C. K. , Bandalos D. L. ( 2001). The relative performance of full information maximum likelihood estimation for missing data in structural equation models. Structural Equation Modeling , 8, 430– 457. Google Scholar CrossRef Search ADS   Ferguson C. J. ( 2009). An effect size primer: A guide for clinicians and researchers. Professional Psychology: Research and Practice , 40, 532– 538. Google Scholar CrossRef Search ADS   Fuemmeler B. F. , Kollins S. H., McClernon F. J. ( 2007). Attention deficit hyperactivity disorder symptoms predict nicotine dependence and progression to regular smoking from adolescence to young adulthood. Journal of Pediatric Psychology , 32, 1203– 1213. doi: 10.1093/jpepsy/jsm051 Google Scholar CrossRef Search ADS PubMed  Hanna E. Z. , Yi H. Y., Dufour M. C., Whitmore C. C. ( 2001). The relationship of early-onset regular smoking to alcohol use, depression, illicit drug use, and other risky behaviors during early adolescence: Results from the youth supplement to the third national health and nutrition examination survey. Journal of Substance Abuse , 13, 265– 282. Google Scholar CrossRef Search ADS PubMed  Harris K. M. , Florey F., Tabor J., Bearman P. S., Jones J., Udry J. R. ( 2009). The National Longitudinal Study of adolescent health: Research design. Retrieved from http://www.cpc.unc.edu/projects/addhealth/design Retrieved January 30, 2010. Johnston L. D. , O’malley P. M., Miech R. A., Bachman J. G., Schulenberg J. E. ( 2016). Monitoring the future national survey results on drug use, 1975-2015: Overview, key findings on adolescent drug use . Ann Arbor, MI: Institute for Social Research, The University of Michigan. Kandel D. , Kandel E. ( 2015). The Gateway Hypothesis of substance abuse: Developmental, biological and societal perspectives. Acta Paediatrica , 104, 130– 137. doi: 10.1111/apa.12851 Google Scholar CrossRef Search ADS PubMed  Kandel D. B. , Yamaguchi K., Klein L. C. ( 2006). Testing the gateway hypothesis. Addiction , 101, 470– 472; discussion 474–476. doi: 10.1111/j.1360-0443.2006.01426.x Google Scholar CrossRef Search ADS PubMed  Kessler R. C. , Adler L. A., Barkley R., Biederman J., Conners C. K., Faraone S. V., Greenhill L. L., Jaeger S., Secnik K., Spencer T., Ustün T. B., Zaslavsky A. M. ( 2005). Patterns and predictors of attention-deficit/hyperactivity disorder persistence into adulthood: Results from the national comorbidity survey replication. Biological Psychiatry , 57, 1442– 1451. doi: 10.1016/j.biopsych.2005.04.001 Google Scholar CrossRef Search ADS PubMed  Kollins S. H. , McClernon F. J., Fuemmeler B. F. ( 2005). Association between smoking and attention-deficit/hyperactivity disorder symptoms in a population-based sample of young adults. Archives of General Psychiatry , 62, 1142– 1147. doi: 10.1001/archpsyc.62.10.1142 Google Scholar CrossRef Search ADS PubMed  Lambert N. ( 2005). The contribution of childhood ADHD, conduct problems, and stimulant treatment to adolescent and adult tobacco and psychoactive substance abuse. Ethical Human Psychology and Psychiatry , 7, 197– 221. Lee S. S. , Humphreys K. L., Flory K., Liu R., Glass K. ( 2011). Prospective association of childhood attention-deficit/hyperactivity disorder (ADHD) and substance use and abuse/dependence: A meta-analytic review. Clinical Psychology Review , 31, 328– 341. doi: 10.1016/j.cpr.2011.01.006 Google Scholar CrossRef Search ADS PubMed  Lo Y. T. , Mendell N. R., Rubin D. B. ( 2001). Testing the number of components in a normal mixture. Biometrika , 88, 767– 778. doi: DOI 10.1093/biomet/88.3.767 Google Scholar CrossRef Search ADS   Mannuzza S. , Klein R. G., Klein D. F., Bessler A., Shrout P. ( 2002). Accuracy of adult recall of childhood attention deficit hyperactivity disorder. The American Journal of Psychiatry , 159, 1882– 1888. doi: 10.1176/appi.ajp.159.11.1882 Google Scholar CrossRef Search ADS PubMed  Masse L. C. , Tremblay R. E. ( 1997). Behavior of boys in kindergarten and the onset of substance use during adolescence. Archives of General Psychiatry , 54, 62– 68. Google Scholar CrossRef Search ADS PubMed  Mayet A. , Legleye S., Chau N., Falissard B. ( 2011). Transitions between tobacco and cannabis use among adolescents: A multi-state modeling of progression from onset to daily use. Addictive Behaviors , 36, 1101– 1105. doi: 10.1016/j.addbeh.2011.06.009 Google Scholar CrossRef Search ADS PubMed  McClernon F. J. , Fuemmeler B. F., Kollins S. H., Kail M. E., Ashley-Koch A. E. ( 2008). Interactions between genotype and retrospective ADHD symptoms predict lifetime smoking risk in a sample of young adults. Nicotine and Tobacco Research , 10, 117– 127. doi: 10.1080/14622200701704913 Google Scholar CrossRef Search ADS PubMed  Merrill J. C. , Kleber H. D., Shwartz M., Liu H., Lewis S. R. ( 1999). Cigarettes, alcohol, marijuana, other risk behaviors, and American youth. Drug and Alcohol Dependence , 56, 205– 212. Google Scholar CrossRef Search ADS PubMed  Moffitt T. E. , Arseneault L., Belsky D., Dickson N., Hancox R.J., Harrington H., Houts R., Poulton R., Roberts B. W., Ross S., Sears M. R., Thomson W. M., Caspi A. ( 2011). A gradient of childhood self-control predicts health, wealth, and public safety. Proceedings of the National Academy of Sciences of the United States of America , 108, 2693– 2698. doi: 10.1073/pnas.1010076108 Google Scholar CrossRef Search ADS PubMed  Molina B. S. , Hinshaw S. P., Eugene Arnold L., Swanson J. M., Pelham W. E., Hechtman L., Hoza B., Epstein J. N., Wigal T., Abikoff H. B., Greenhill L. L., Jensen P. S., Wells K. C., Vitiello B., Gibbons R. D., Howard A., Houck P. R., Hur K., Lu B., Marcus S.; MTA Cooperative Group. ( 2013). Adolescent substance use in the multimodal treatment study of attention-deficit/hyperactivity disorder (ADHD) (MTA) as a function of childhood ADHD, random assignment to childhood treatments, and subsequent medication. Journal of the American Academy of Child and Adolescent Psychiatry , 52, 250– 263. doi: 10.1016/j.jaac.2012.12.014 Google Scholar CrossRef Search ADS PubMed  Molina B. S. , Pelham W. E.Jr. ( 2014). Attention-deficit/hyperactivity disorder and risk of substance use disorder: Developmental considerations, potential pathways, and opportunities for research. Annual Review of Clinical Psychology , 10, 607– 639. doi: 10.1146/annurev-clinpsy-032813-153722 Google Scholar CrossRef Search ADS PubMed  Murphy K. , Barkley R. A. ( 1996). Attention deficit hyperactivity disorder adults: Comorbidities and adaptive impairments. Comprehensive Psychiatry , 37, 393– 401. Google Scholar CrossRef Search ADS PubMed  Muthén L. K. , Muthén B. O. ( 1998-2005). Mplus user’s guide  (7th ed.). Los Angeles, CA: Muthén & Muthén. Palmer R. H. , Young S. E., Hopfer C. J., Corley R. P., Stallings M. C., Crowley T. J., Hewitt J. K. ( 2009). Developmental epidemiology of drug use and abuse in adolescence and young adulthood: Evidence of generalized risk. Drug and Alcohol Dependence , 102, 78– 87. doi: 10.1016/j.drugalcdep.2009.01.012 Google Scholar CrossRef Search ADS PubMed  Resnick M. D. , Bearman P. S., Blum R. W., Bauman K. E., Harris K. M., Jones J., Tabor J., Beuhring T., Sieving R. E., Shew M., Ireland M., Bearinger L. H., Udry J.R. ( 1997). Protecting adolescents from harm. Findings from the National Longitudinal Study on Adolescent Health. JAMA , 278, 823– 832. Google Scholar CrossRef Search ADS PubMed  Sundquist J. , Ohlsson H., Sundquist K., Kendler K. S. ( 2015). Attention-deficit/hyperactivity disorder and risk for drug use disorder: A population-based follow-up and co-relative study. Psychological Medicine , 45, 977– 983. doi: 10.1017/S0033291714001986 Google Scholar CrossRef Search ADS PubMed  Szobot C. M. , Rohde L. A., Bukstein O., Molina B. S., Martins C., Ruaro P., Pechansky F. ( 2007). Is attention-deficit/hyperactivity disorder associated with illicit substance use disorders in male adolescents? A community-based case-control study. Addiction , 102, 1122– 1130. doi: 10.1111/j.1360-0443.2007.01850.x Google Scholar CrossRef Search ADS PubMed  Tarter R. E. , Kirisci L., Reynolds M., Mezzich A. ( 2004). Neurobehavior disinhibition in childhood predicts suicide potential and substance use disorder by young adulthood. Drug and Alcohol Dependence , 76 (Suppl), S45– S52. doi: 10.1016/j.drugalcdep.2004.08.006 Google Scholar CrossRef Search ADS PubMed  Vanyukov M. M. , Tarter R. E., Kirillova G. P., Kirisci L., Reynolds M. D., Kreek M. J., Conway K. P., Maher B. S., Iacono W. G., Bierut L., Neale M. C., Clark D. B., Ridenour T.A. ( 2012). Common liability to addiction and “gateway hypothesis”: Theoretical, empirical and evolutionary perspective. Drug and Alcohol Dependence , 123 (Suppl 1), S3– 17. doi: 10.1016/j.drugalcdep.2011.12.018 Google Scholar CrossRef Search ADS PubMed  Visser S. N. , Danielson M. L., Bitsko R. H., Holbrook J. R., Kogan M. D., Ghandour R. M., Perou R., Blumberg S.J. ( 2014). Trends in the parent-report of health care provider-diagnosed and medicated attention-deficit/hyperactivity disorder: United States, 2003-2011. Journal of the American Academy of Child and Adolescent Psychiatry , 53, 34– 46 e32. doi: 10.1016/j.jaac.2013.09.001 Google Scholar CrossRef Search ADS PubMed  Visser S. N. , Lesesne C. A., Perou R. ( 2007). National estimates and factors associated with medication treatment for childhood attention-deficit/hyperactivity disorder. Pediatrics , 119 (Suppl 1), S99– S106. doi: 10.1542/peds.2006-2089O Google Scholar CrossRef Search ADS PubMed  © The Author 2017. Published by Oxford University Press on behalf of the Society of Pediatric Psychology. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Pediatric Psychology Oxford University Press

Childhood ADHD Symptoms and Future Illicit Drug Use: The Role of Adolescent Cigarette Use

Loading next page...
 
/lp/ou_press/childhood-adhd-symptoms-and-future-illicit-drug-use-the-role-of-EoWtm6SV3U
Publisher
Oxford University Press
Copyright
© The Author 2017. Published by Oxford University Press on behalf of the Society of Pediatric Psychology. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com
ISSN
0146-8693
eISSN
1465-735X
D.O.I.
10.1093/jpepsy/jsx098
Publisher site
See Article on Publisher Site

Abstract

Abstract Objectives The aim of this study is to understand how early cigarette use might predict subsequent illicit drug use, especially among individuals with attention-deficit hyperactivity disorder (ADHD) symptoms during childhood. Methods Data were drawn from the National Longitudinal Study of Adolescent Health (Waves I–IV). The analysis sample involves participants who had not used illicit drugs at Wave I, with no missing responses for studied predictors (N = 7,332). Results Smoking status at Wave I (ever regular vs. never regular) and childhood ADHD symptoms predicted subsequent illicit drug use at Waves II to IV. No interaction effect of smoking status at Wave I and childhood ADHD symptoms was found. However, an indirect effect from childhood ADHD symptoms on illicit drug use was identified, through smoking status at Wave I. Similar results were observed for predicting illicit drug dependence. Conclusions The findings support the notion that smoking status during early adolescence may mediate the association between childhood ADHD symptoms and risk of later adult drug use. Interventions to prevent smoking among adolescents may be particularly effective at decreasing subsequent drug use, especially among children with ADHD symptoms. ADHD symptoms, adolescent smoking, illicit drug use Introduction Attention-deficit hyperactivity disorder (ADHD) is one of the most common neurodevelopmental disorders in children and often persists into adulthood (Kessler et al., 2005; Visser et al., 2014; Visser, Lesesne, & Perou, 2007). Independent of a clinical diagnosis, ADHD symptoms are associated with a number of health problems including smoking and drug use during the transition into adulthood (Charach, Yeung, Climans, & Lillie, 2011; Fuemmeler, Kollins, & McClernon, 2007; Lee, Humphreys, Flory, Liu, & Glass, 2011; Molina et al., 2013; Molina & Pelham, 2014; Sundquist, Ohlsson, Sundquist, & Kendler, 2015; Szobot et al., 2007). However, the factors underlying risk for increased drug use among individuals with ADHD symptoms are not fully understood (Molina & Pelham, 2014). Clarity regarding salient risk factors among this population could advance our understanding of the etiology of drug use, as well as inform prevention strategies. One of the strongest predictors of persistent and continued drug use during adolescence and young adulthood is earlier initiation of “milder” legal drugs (Kandel & Kandel, 2015). For instance, cigarette use during early adolescence often precedes subsequent illicit drug use during late adolescence (Palmer et al., 2009). This progression of substance use is usually referred to as a gateway sequence/hypothesis (Hanna, Yi, Dufour, & Whitmore, 2001; Kandel & Kandel, 2015; Kandel, Yamaguchi, & Klein, 2006; Mayet, Legleye, Chau, & Falissard, 2011; Merrill, Kleber, Shwartz, Liu, & Lewis, 1999; Palmer et al., 2009). The gateway sequence could explain the pathway for illicit drug use among individuals with ADHD, given their propensity to initiate cigarette smoking at an earlier age (Biederman et al., 2006; Lambert, 2005). Two studies have specifically examined the relationship among ADHD symptoms, cigarette smoking, and illicit drug use. In one study, the likelihood of illicit drug use was increased among ADHD youth who smoked compared with their counterpart who did not smoke (Biederman et al., 2006). In other words, cigarette smoking seemed to strengthen (or moderate) the association between childhood ADHD symptoms and subsequent illicit drug use. Alternatively, Lambert (2005) showed that children with ADHD symptoms were more likely to report illicit drug dependence as adults; however, the association was null after adjusting for early adolescent cigarette use. Accordingly, these findings from Lambert’s study suggests that risk for illicit drug use among those with ADHD symptoms may be mediated by cigarette smoking during early adolescence. Although both studies support the importance of ADHD symptoms and early cigarette smoking as predictors of subsequent illicit drug use, it is not clear how these factors operate to confer risk. As such, we propose a few possible scenarios. First, based on a vast literature of individual vulnerability as well as the gateway hypothesis in the field of substance use, ADHD symptoms and cigarette smoking could be independent risk factors for illicit drug use. Second, according to Biederman et al. (2006), cigarette smoking could moderate the relationships between ADHD symptoms and future illicit drug use. That is, cigarette smoking influences the strength of the relationship between ADHD symptoms and future illicit drug use. Third, based on Lambert’s study (2005), cigarette smoking could mediate, or explain, the relationship between ADHD symptoms and future illicit drug use. The results of these associations have different implications for prevention (or intervention) efforts. If ADHD symptoms and cigarette smoking are independent risk factors, then the risk factor that is more strongly associated with subsequent illicit drug use should be the target of prevention efforts. If ADHD symptoms and cigarette smoking are interacting risk factors (e.g., moderation exists), whereby the risk for illicit drug use is highest among those who both have ADHD symptoms and initiate cigarette smoking early in adolescence, then prevention efforts should focus on youth with ADHD symptoms who also initiate smoking at an early age. If cigarette smoking mediates the relationship between ADHD symptoms and illicit drug use, then early education of adolescents with ADHD symptoms on strategies to stay away from using cigarettes to self-medicate should reduce the prevalence of future illicit drug use in the population. Thus, determining how these three factors (ADHD symptoms, early cigarette use, and subsequent illicit drug use) are related would be helpful in identifying the target populations for specific prevention strategies, in addition to broad ones (e.g., prevent underage cigarette/substance use). Using data from a large longitudinal cohort, the purpose of this study was to evaluate the manner in which childhood ADHD symptoms and early cigarette smoking potentially relate to subsequent illicit drug use. Specifically, in a series of statistical models, we tested (1) if ADHD symptoms and smoking were independently associated with subsequent illicit drug use (Figure 1A); (2) if smoking moderated the association between ADHD symptoms and illicit drug use (Figure 1B); or (3) if there was an indirect effect of cigarette smoking in the relationship between ADHD symptoms and illicit drug use, suggesting a potential mediating effect (Figure 1C). We hypothesized that greater ADHD symptoms are associated with a higher likelihood of early cigarette smoking, which is further related to a higher probability of illicit drug use and drug dependence during adulthood. Figure 1. View largeDownload slide Theoretical model of this current study (A–C). Figure 1. View largeDownload slide Theoretical model of this current study (A–C). Figure 2. View largeDownload slide ADHD symptoms classes. Figure 2. View largeDownload slide ADHD symptoms classes. Methods Study Sample The study population was drawn from 20,774 adolescents in the National Longitudinal Study of Adolescent Health (Add Health), a nationally representative longitudinal cohort. Add health recruited their participants from 80 high schools and 52 middle schools in the United States. These schools were selected based on systematic sampling methods to ensure the sample is representative of U.S. schools with respect to region, urbanity, school size, school type, and ethnicity. Respondents completed in-home surveys in 1995, as well as up to three additional in-home surveys (1996, 2001–2002, 2008–2009). Detailed study design and data collection have been described elsewhere (Harris et al., 2009; Resnick et al., 1997). In this current study, we excluded participants from the analysis who either indicated that they had used (n =1,056) or were missing data on illicit drug use at Wave I (n = 55) and who had missing data on one or more of the independent variables in the model. A number of independent variables had some missing data: ethnicity (n =3), parental education (n = 955), ADHD symptoms (n =7), early history of regular smoking (n =18). Participants with missing data on one or more of these variables (listwise cases = 978) were excluded from the analysis. Accordingly, the final analysis sample involved 7,332 participants. Measures Childhood ADHD Symptoms At Wave III, participants were asked to report on DSM-IV ADHD symptoms they experienced between the ages of 5 and 12 years. Each response to DSM-IV ADHD symptom questions were collected on a four-point scale: 1 = never or rarely, 2 = sometimes, 3 = often, or 4 = very often. One DSM-IV hyperactive-impulsive ADHD symptom, “Often interrupts or intrudes on others,” was not part of the survey, and could not be included in the analyses, resulting in responses to nine inattentive (IN) and eight hyperactive (HI) symptoms. Consistent with scoring convention, a symptom was considered present if it was experienced “often” or “very often” (Fuemmeler et al., 2007; Murphy & Barkley, 1996). The Kuder–Richardson Formula 20 (KR-20) was used as a measure of internal consistency reliability for measures with dichotomous responses (e.g., presence/absence of hyperactive-impulsiveness and inattention). Reliability coefficients for the hyperactive-impulsive symptoms and IN symptoms in this study were 0.75 and 0.80, respectively, indicating an adequate internal consistency for each set of symptoms. Moreover, we have previously shown that ADHD symptoms measured in this sample have good concurrent validity. That is, individuals reporting high levels of ADHD symptoms (six or more HI or IN symptoms) were more likely to have a learning disability or behavioral problems at Wave I and were more likely to indicate having taken medication for ADHD at Wave III (Kollins, McClernon, & Fuemmeler, 2005). Cigarette Smoking A computer-assisted survey instrument was used to reduce response bias and collect data during home visits, regarding self-reported use of various nicotine and nonnicotine substances. Specifically, participants were asked to indicate if they had ever (yes/no) smoked regularly (i.e., ever having smoked at least one cigarette every day for 30 days currently or in the past). Individuals who indicated having smoked regularly at some point during their lifetime herein are referred to as “ever-regular smokers.” All other individuals were classified as “never-regular smokers.” The never-regular smokers comprised individuals who had never tried smoking, had only taken one or two puffs, who had taken puffs but never smoked an entire cigarette, or who had smoked an entire cigarette but never smoked regularly. This classification was chosen because the primary aim of the present study was to evaluate whether participants with heavier smoking patterns during adolescence predict later drug use in adulthood. Accordingly, smoking status collected at Wave I (ever regular vs. never regular) was included as a predictor of subsequent illicit drug use during Waves II–IV. Illicit Drug Use and Dependence For each wave, illicit drug use was assessed by asking participants if they had ever used illicit drugs, including cocaine, inhalants (e.g., glue or solvents), crystal meth (ice), LSD (Lysergic acid diethylamide), PCP (phencyclidine), ecstasy, heroin, mushrooms, speed, or pills, without a doctor’s prescription. A positive response to any of the above eight items defined the participant as having used illicit drugs. During the Wave IV interview, participants were further asked to report drug dependence symptoms. Three or more positive DSM-IV drug dependence symptoms (e.g., tolerance, withdrawal, spent a lot of time using) in the past 12 months were defined as “dependent.” Although participants were queried on marijuana use, this was excluded from our analysis for two reasons: it was not uniformly assessed across waves, and the legal implications of marijuana use varies by state. Other Variables Participants were asked about level of engagement in 13 conduct problem behaviors (e.g., property damage, lying to guardians, fighting) at Wave I. Each of the items was dichotomized into 0 (never) and 1 (one or more incidents) and then summed. The scale exhibits adequate reliability (KR-20 = 0.69) and has been used in previously published studies with this sample to control for the effects of conduct problems (Kollins et al., 2005; McClernon, Fuemmeler, Kollins, Kail, & Ashley-Koch, 2008). Gender, age, race/ethnicity (White, Hispanic, African American, other), and parental education (less than high school, high school or equivalent, some college, and college degree or beyond) at Wave I were included in analytic models as control variables. Statistical Analyses Using latent class modeling, we classified individuals based on their level of self-reported HI and IN symptoms. This resulted in two subgroups representing an affected class of individuals who reported high childhood HI and IN symptoms versus an unaffected class representing individuals who reported few problems in these domains as children (See Figure 2). Bayesian information criterion (BIC; the lower, the better), entropy (the higher, the better), and the Lo-Mendell–Rubin adjusted likelihood ratio test informed decisions regarding number of latent classes (Lo, Mendell, & Rubin, 2001). The average posterior probability of class membership (the closer to 1, the better) and interpretability of the classes were also considered. We tested three models aimed at evaluating (1) the independent effects of ADHD symptoms and smoking status at Wave I (ever regular vs. never regular) on illicit drug use (Model 1); (2) the interacting effects of ADHD symptoms and smoking status at Wave I on illicit drug use (Model 2), and the indirect effect of smoking status at Wave I on the relationship between ADHD symptoms and illicit drug use (Model 3). Each model included the covariates of child age, race, gender, parental education, and conduct disorder symptoms. All analyses used sample weights to correct for selection probabilities and to obtain accurate standard errors. After identifying the latent classes of ADHD symptoms, the first model (Model 1) evaluated the survival risk of illicit drug use (from Waves II–IV) regressed on classes of ADHD symptoms, and smoking status at Wave I via discrete-time survival analysis in Mplus 7.4 (Muthén & Muthén, 1998–2005). Discrete-time survival analysis identifies the hazard probability of a nonrepeatable event (in this case, the initial use of illicit drugs) from Wave II to Wave IV. Unlike Cox regression, discrete-time survival analysis treats time as discrete units or chunks rather than as a continuous variable, which is appropriate for panel survey data. In addition, discrete-time survival analysis is an optimal analytical approach for evaluating the developmental nature of substance use in adolescence because it does not assume that the shape of the survival function over time is the same for all cases. Next, in Model 2, we tested the interaction effect of ADHD symptoms and smoking status at Wave I as related to the survival hazard of illicit drug use. Finally, in a statistical mediation model (Model 3), we used the MODEL INDIRECT command in Mplus to assess whether the effect of classes of ADHD symptoms on survival risk of illicit drug use is statistically mediated (or partially statistically mediated) by smoking status at Wave I. We also calculated the proportion statistically mediated by the smoking variable (i.e., the indirect effect/the total effect*100). In addition, using logistic regression, we evaluated whether the influences of ADHD symptoms and smoking status at Wave I have a similar effect on illicit drug dependence observed in adulthood (Wave IV). For handling missing data among ever-use illicit drug across waves, full information maximum likelihood (FIML) was conducted via Mplus. Conventionally, FIML is favored over listwise or pairwise deletion because it provides more robust and efficient estimations. The rationales can be found in Enders and Bandalos (2001). Results At Wave I, participants were on average 15.5 (SE = 0.11) years of age and 19.8% categorized as ever-regular smokers. Around 50.0% of the participants were female and 65.9% White. Approximately one quarter (23.4%) of the sample had parents who had earned a college degree. Additional details on the sample characteristics are presented in Table I. Table II shows the bivariate association between studied predictors and illicit drug use at Waves II–IV and drug dependence at Wave IV. Table I. Sample Characteristics (n = 7,332a) Variable  n  % or (M, SE)  Age (mean, SE)    (15.7, 1.61)   Wave 1  7,332  (15.7, 1.61)   Wave 2  7,332  (16.7, 1.60)   Wave 3  7,332  (21.9, 1.67)   Wave 4  7,332  (28.7, 1.61)  Gender (female)  4,028  54.9%  Race/ethnicity       Hispanic  1,071  14.6%   Black  1,601  21.8%   White  4,074  55.6%   Other  586  8.0%  Parental education—college degree  1,907  26.0%  Early history of regular smoking  1,039  14.2%  Conduct problems (M, SE)    (0.7, 1.16)  Hyperactive symptoms (M, SE)    (1.3, 0.03)   0 Symptoms  2,726  37.2%   1–3 Symptoms  3,529  48.1%   4–6 Symptoms  904  12.3%   6+ Symptoms  172  2.3%  Inattentive symptoms (M, SE)    (1.7, 0.03)   0 Symptoms  4,166  56.8%   1–3 Symptoms  2,352  32.1%   4–6 Symptoms  621  8.5%   6+ Symptoms  192  2.6%  Drug use—Wave II       Ever tried any kind of cocaine  1,429  23.7%   Ever tried or used inhalants  86  1.2%   Ever used illegal drugs  196  2.7%   Ever injected any illegal drugs  24  0.3%  Drug use—Wave III       Ever used any kind of cocaine  604  8.2%   Ever used crystal meth (ice)  298  4.1%   Ever used illegal drugs  1,015  13.8%   Ever injected any illegal drugs  52  0.7%  Drug use—Wave IV       Ever used cocaine  1,138  15.5%   Ever used crystal meth (ice)  499  6.8%   Ever used illegal drugs  1,307  17.8%   Ever injected any illegal drugs  24  0.3%  Variable  n  % or (M, SE)  Age (mean, SE)    (15.7, 1.61)   Wave 1  7,332  (15.7, 1.61)   Wave 2  7,332  (16.7, 1.60)   Wave 3  7,332  (21.9, 1.67)   Wave 4  7,332  (28.7, 1.61)  Gender (female)  4,028  54.9%  Race/ethnicity       Hispanic  1,071  14.6%   Black  1,601  21.8%   White  4,074  55.6%   Other  586  8.0%  Parental education—college degree  1,907  26.0%  Early history of regular smoking  1,039  14.2%  Conduct problems (M, SE)    (0.7, 1.16)  Hyperactive symptoms (M, SE)    (1.3, 0.03)   0 Symptoms  2,726  37.2%   1–3 Symptoms  3,529  48.1%   4–6 Symptoms  904  12.3%   6+ Symptoms  172  2.3%  Inattentive symptoms (M, SE)    (1.7, 0.03)   0 Symptoms  4,166  56.8%   1–3 Symptoms  2,352  32.1%   4–6 Symptoms  621  8.5%   6+ Symptoms  192  2.6%  Drug use—Wave II       Ever tried any kind of cocaine  1,429  23.7%   Ever tried or used inhalants  86  1.2%   Ever used illegal drugs  196  2.7%   Ever injected any illegal drugs  24  0.3%  Drug use—Wave III       Ever used any kind of cocaine  604  8.2%   Ever used crystal meth (ice)  298  4.1%   Ever used illegal drugs  1,015  13.8%   Ever injected any illegal drugs  52  0.7%  Drug use—Wave IV       Ever used cocaine  1,138  15.5%   Ever used crystal meth (ice)  499  6.8%   Ever used illegal drugs  1,307  17.8%   Ever injected any illegal drugs  24  0.3%  Note. Childs age, gender, race/ethnicity, parental education, smoking, and conduct problems variables were from Wave I; ADHD symptoms were obtained from Wave III; at Wave II, the category of “ever used illegal drugs” included crystal meth as an example; at both Waves III and IV, “ever used illegal drugs” included inhalants as an example. a Sample size varies slightly depending on missing data for some variables. Table II. Bivariate Relationship between Sample Characteristics and Illicit Drug Use and Dependence at Wave IV (N = 7,332) Variable  Illicit drug use   ÷2 or t-value  Cramer’s V or Cohen’s d  Illicit drug dependence   ÷2 or t-value  Cramer’s V or Cohen’s d  Non-Use (n = 5,259)  Use (n = 2,073)  Nondependent (n = 7,077)  Dependent (n = 255)  Gender (female)  58.8%  45.1%  114.0***  0.125  55.2%  47.8%  5.4*  0.027  Race/ethnicity                   Hispanic  14.7%  14.5%  310.1***  0.206  14.6%  14.1%  42.2***  0.076   Black  27.0%  8.8%      22.4%  6.7%       White  50.5%  68.4%      55.0%  72.5%       Other  7.9%  8.3%      8.0%  6.7%      Age (M, SE)  15.78 (1.62)  15.45 (1.55)  7.84***  0.208  15.70 (1.60)  15.42 (1.64)  2.69**  0.172  Parental education                   College degree  25.5%  27.3%  2.3  0.018  26.1%  22.7%  1.5  0.014  Early history of regular smoking  10.1%  24.4%  249.1***  0.184  13.4%  35.7%  100.6***  0.117  High ADHD symptoms class  21.8%  32.7%  94.0***  0.113  24.3%  40.4%  34.1***  0.068  Conduct problems (M, SE)  0.59 (1.04)  1.00 (1.35)  12.55***  0.340  0.68 (1.14)  1.20 (1.52)  5.41***  0.387  Variable  Illicit drug use   ÷2 or t-value  Cramer’s V or Cohen’s d  Illicit drug dependence   ÷2 or t-value  Cramer’s V or Cohen’s d  Non-Use (n = 5,259)  Use (n = 2,073)  Nondependent (n = 7,077)  Dependent (n = 255)  Gender (female)  58.8%  45.1%  114.0***  0.125  55.2%  47.8%  5.4*  0.027  Race/ethnicity                   Hispanic  14.7%  14.5%  310.1***  0.206  14.6%  14.1%  42.2***  0.076   Black  27.0%  8.8%      22.4%  6.7%       White  50.5%  68.4%      55.0%  72.5%       Other  7.9%  8.3%      8.0%  6.7%      Age (M, SE)  15.78 (1.62)  15.45 (1.55)  7.84***  0.208  15.70 (1.60)  15.42 (1.64)  2.69**  0.172  Parental education                   College degree  25.5%  27.3%  2.3  0.018  26.1%  22.7%  1.5  0.014  Early history of regular smoking  10.1%  24.4%  249.1***  0.184  13.4%  35.7%  100.6***  0.117  High ADHD symptoms class  21.8%  32.7%  94.0***  0.113  24.3%  40.4%  34.1***  0.068  Conduct problems (M, SE)  0.59 (1.04)  1.00 (1.35)  12.55***  0.340  0.68 (1.14)  1.20 (1.52)  5.41***  0.387  * p < .05, **p < .01, ***p < .001. Childhood ADHD Symptom Classes The results from latent class analysis for childhood ADHD symptoms suggested that a two-class model fit best (BIC = 117,535; entropy =0.88; Lo-Mendell-Rubin adjusted likelihood ratio test = 18,764.89, p < .001), when compared with a three-class model (BIC = 100,599.39; entropy = 0.83; Lo-Mendell-Rubin adjusted LRT test = 2,542.13, p = .166). The two-class average posterior probabilities for most likely class membership were 0.98 and 0.94, respectively. A two-class model was selected for further analysis because of the higher entropy, a significant Lo-Mendell-Rubin adjusted LRT test, and good posterior probability of each assigned class. The two classes were labeled as “high ADHD symptom endorsement” (27.7%) and “low ADHD symptom endorsement” (72.3%). Illicit Drug Use Model 1 (Table III) shows the effects of childhood ADHD symptoms (low vs. high symptom endorsement) and smoking status at Wave I (ever regular vs. never regular) on risk of illicit drug use, controlling for conduct problems and demographic variables. The results indicated that both ever-regular smoking at Wave 1 (OR = 2.40, p < .001; Cohen’s d = 0.48) and childhood ADHD symptom class (OR = 1.32, p = .001; Cohen’s d = 0.15) predicted subsequent illicit drug use. Table III. Risk Factors for Experimenting with Illicit Drugs Use from Wave II to Wave IV (N = 7,332)   Model 1   Model 2   Model 3   Est.  Odds ratio (95% CI)  P  Est.  Odds ratio (95% CI)  P  Est.  Odds ratio (95% CI)  p  Drug use on                     Age  −0.20  0.82 (0.78–0.86)  <.0001  −0.20  0.82 (0.78–0.86)  <.0001  −0.20  0.82 (0.78–0.86)  <.0001   Gender (female)  −0.32  0.73 (0.63–0.85)  <.0001  −0.32  0.73 (0.63–0.85)  <.0001  −0.32  0.73 (0.63–0.85)  <.0001   Race/ethnicity                      White (referent)                      Hispanic  −0.12  0.89 (0.68–1.16)  0.389  −0.12  0.89 (0.68–1.16)  0.385  −0.11  0.89 (0.68–1.17)  0.403    Black  −1.40  0.25 (0.19–0.32)  <.0001  −1.40  0.25 (0.19–0.32)  <.0001  −1.40  0.25 (0.19–0.32)  <.0001    Other  −0.15  0.86 (0.66–1.14)  0.300  −0.15  0.86 (0.66–1.14)  0.285  −0.15  0.87 (0.66–1.14)  0.303   Parental education  0.16  1.17 (1.05–1.30)  0.005  0.16  1.17 (1.05–1.30)  0.005  0.16  1.17 (1.05–1.31)  0.005   Conduct problems  0.29  1.34 (1.24–1.44)  <.0001  0.29  1.34 (1.24–1.44)  <.0001  0.29  1.33 (1.23–1.44)  <.0001   High (vs. low) ADHD symptom class  0.28  1.32 (1.13–1.54)  0.001  0.33  1.39 (1.16–1.67)  <.0001  0.27  1.31 (1.12–1.53)  0.001   Smoking status  0.88  2.40 (1.94–2.97)  <.0001  0.96  2.62 (2.04–3.35)  <.0001  0.88  2.41 (1.94–2.98)  <.0001  High ADHD symptom class × Smoking status    0.23  1.26 (0.88–1.81)  0.203    Smoking on                 ADHD symptom class          0.64  1.90 (1.59–2.27)  <.0001    Model 1   Model 2   Model 3   Est.  Odds ratio (95% CI)  P  Est.  Odds ratio (95% CI)  P  Est.  Odds ratio (95% CI)  p  Drug use on                     Age  −0.20  0.82 (0.78–0.86)  <.0001  −0.20  0.82 (0.78–0.86)  <.0001  −0.20  0.82 (0.78–0.86)  <.0001   Gender (female)  −0.32  0.73 (0.63–0.85)  <.0001  −0.32  0.73 (0.63–0.85)  <.0001  −0.32  0.73 (0.63–0.85)  <.0001   Race/ethnicity                      White (referent)                      Hispanic  −0.12  0.89 (0.68–1.16)  0.389  −0.12  0.89 (0.68–1.16)  0.385  −0.11  0.89 (0.68–1.17)  0.403    Black  −1.40  0.25 (0.19–0.32)  <.0001  −1.40  0.25 (0.19–0.32)  <.0001  −1.40  0.25 (0.19–0.32)  <.0001    Other  −0.15  0.86 (0.66–1.14)  0.300  −0.15  0.86 (0.66–1.14)  0.285  −0.15  0.87 (0.66–1.14)  0.303   Parental education  0.16  1.17 (1.05–1.30)  0.005  0.16  1.17 (1.05–1.30)  0.005  0.16  1.17 (1.05–1.31)  0.005   Conduct problems  0.29  1.34 (1.24–1.44)  <.0001  0.29  1.34 (1.24–1.44)  <.0001  0.29  1.33 (1.23–1.44)  <.0001   High (vs. low) ADHD symptom class  0.28  1.32 (1.13–1.54)  0.001  0.33  1.39 (1.16–1.67)  <.0001  0.27  1.31 (1.12–1.53)  0.001   Smoking status  0.88  2.40 (1.94–2.97)  <.0001  0.96  2.62 (2.04–3.35)  <.0001  0.88  2.41 (1.94–2.98)  <.0001  High ADHD symptom class × Smoking status    0.23  1.26 (0.88–1.81)  0.203    Smoking on                 ADHD symptom class          0.64  1.90 (1.59–2.27)  <.0001  Note. Model 1 represents the risk of both ADHD symptom class and smoking status at Wave I controlling for the listed covariates. Model 2 represents the interaction between ADHD symptom class and smoking status at Wave I controlling for the listed covariates. Model 3 further added the path from ADHD symptom class to smoking status at Wave I to test whether ever-regular smoking serves as a mediator between ADHD class and illicit drug use. Model 2 (Table III) further shows that there was no significant interaction effect of childhood ADHD symptom classes and smoking status in predicting risk of subsequent illicit drug use (OR =1.26, p = .203; Cohen’s d = 0.13). This indicates the association between smoking status at Wave I, and future illicit drug use was not moderated (i.e., was neither magnified nor weakened) by childhood ADHD symptom classes. In addition to the main effects of childhood ADHD symptom class and smoking status at Wave I on illicit drug use, Model 3 (Table III) shows that the path from childhood ADHD symptom class to smoking status at Wave I was also significant (OR = 1.90, p < .001; Cohen’s d = 0.35). This suggests that the association between ADHD symptom class on the risk of illicit drug was partially statistically mediated by smoking status at Wave I. Follow-up indirect effect test via Mplus through “IND” command confirmed that there was a significant indirect effect from childhood ADHD symptom class to risk of illicit drug use through smoking status at Wave I (Est = 0.11, SE = 0.03, p < .001) in addition to a significant direct effect (Est = 0.27, SE = 0.08, p < .001). The total effect was also significant (Est = 0.38, SE = 0.08, p < .001). The proportion of the total effect statistically mediated by ever-regular smoking at Wave I was 29.5%. Illicit Drug Dependence Model 1 (Table IV) presents the effect of childhood ADHD symptom class and smoking status at Wave I (ever regular vs. never regular) on emerging illicit drug dependence symptoms in adulthood (Wave IV). Similar results were observed as those for risk of illicit drug use. Both childhood ADHD symptom class (OR = 1.32, p = .001; Cohen’s d = 0.15) and smoking status (OR = 2.40, p < .001; Cohen’s d = 0.48) were significantly and independently related to illicit drug dependence during adulthood. However, there was no significant interaction effect of these two predictors on drug dependence (Model 2 in Table IV). The paths from childhood ADHD symptom class to smoking status at Wave I and from smoking status to illicit drug dependence were significant. A follow-up test for statistical mediation was conducted. Results showed that there was a significant indirect effect from childhood ADHD symptom class to illicit drug dependence via smoking status at Wave I (Est. = 0.03, SE = 0.01, p = .05) in addition to a marginally significant direct effect (Est. = 0.09, SE = 0.01, p = .075). The total effect was also significant (Est. = 0.12, SE = 0.05, p = .026). The proportion of the total effect statistically mediated by smoking status at Wave I was 23.8%. Table IV. Risk Factors for Predicting Illicit Drug Dependence at Wave IV (N = 7,332)   Model 1   Model 2   Model 3   Est.  Odds ratio (95% CI)  p  Est.  Odds ratio (95% CI)  p  Est.  Odds ratio (95% CI)  P  Drug dependence on                     Age  −0.16  0.85 (0.75–0.96)  0.008  −0.16  0.85 (0.75–0.96)  0.008  −0.16  0.85 (0.75–0.96)  0.008   Gender                     Gender (female)  −0.02  0.98 (0.69–1.38)  0.901  −0.02  0.98 (0.69–1.38)  0.901  −0.02  0.98 (0.70–1.38)  0.912   Race/ethnicity                      White (referent)                      Hispanic  −0.13  0.87 (0.51–1.50)  0.628  −0.13  0.87 (0.51–1.50)  0.627  −0.13  0.87 (0.51–1.50)  0.628    Black  −1.90  0.15 (0.08–0.27)  <.0001  −1.90  0.15 (0.08–0.27)  <.0001  −1.91  0.15 (0.08–0.27)  <.0001    Other  −0.28  0.75 (0.35–1.62)  0.471  −0.28  0.75 (0.35–1.62)  0.468  −0.28  0.76 (0.35–1.63)  0.475   Parental education  0.09  1.10 (0.84–1.44)  0.510  0.09  1.10 (0.84–1.44)  0.511  0.09  1.10 (0.84–1.44)  0.510   Conduct problem  0.26  1.29 (1.14–1.46)  <.0001  0.26  1.29 (1.14–1.46)  <.0001  0.25  1.29 (1.14–1.46)  <.0001   High (vs. low) ADHD symptom class  0.41  1.51 (1.00–2.26)  0.048  0.42  1.52 (0.98–2.36)  0.062  0.41  1.50 (1.00–2.25)  0.050   Smoking Status  0.93  2.52 (1.70–3.75)  <.0001  0.94  2.55 (1.52–4.27)  <.0001  0.93  2.52 (1.70–3.75)  <.0001  High ADHD symptom class × Smoking status    0.02  1.02 (0.51–2.04)  0.951    Smoking on                 ADHD symptom class          0.65  1.91 (1.59–2.28)  <.0001    Model 1   Model 2   Model 3   Est.  Odds ratio (95% CI)  p  Est.  Odds ratio (95% CI)  p  Est.  Odds ratio (95% CI)  P  Drug dependence on                     Age  −0.16  0.85 (0.75–0.96)  0.008  −0.16  0.85 (0.75–0.96)  0.008  −0.16  0.85 (0.75–0.96)  0.008   Gender                     Gender (female)  −0.02  0.98 (0.69–1.38)  0.901  −0.02  0.98 (0.69–1.38)  0.901  −0.02  0.98 (0.70–1.38)  0.912   Race/ethnicity                      White (referent)                      Hispanic  −0.13  0.87 (0.51–1.50)  0.628  −0.13  0.87 (0.51–1.50)  0.627  −0.13  0.87 (0.51–1.50)  0.628    Black  −1.90  0.15 (0.08–0.27)  <.0001  −1.90  0.15 (0.08–0.27)  <.0001  −1.91  0.15 (0.08–0.27)  <.0001    Other  −0.28  0.75 (0.35–1.62)  0.471  −0.28  0.75 (0.35–1.62)  0.468  −0.28  0.76 (0.35–1.63)  0.475   Parental education  0.09  1.10 (0.84–1.44)  0.510  0.09  1.10 (0.84–1.44)  0.511  0.09  1.10 (0.84–1.44)  0.510   Conduct problem  0.26  1.29 (1.14–1.46)  <.0001  0.26  1.29 (1.14–1.46)  <.0001  0.25  1.29 (1.14–1.46)  <.0001   High (vs. low) ADHD symptom class  0.41  1.51 (1.00–2.26)  0.048  0.42  1.52 (0.98–2.36)  0.062  0.41  1.50 (1.00–2.25)  0.050   Smoking Status  0.93  2.52 (1.70–3.75)  <.0001  0.94  2.55 (1.52–4.27)  <.0001  0.93  2.52 (1.70–3.75)  <.0001  High ADHD symptom class × Smoking status    0.02  1.02 (0.51–2.04)  0.951    Smoking on                 ADHD symptom class          0.65  1.91 (1.59–2.28)  <.0001  Note. Model 1 represents the risk of both ADHD symptom class and smoking status at Wave I controlling for the listed covariates. Model 2 represents the interaction between ADHD symptom class and smoking status at Wave I controlling for the listed covariates. Model 3 further added the path from ADHD symptom class to smoking status to test whether ever-regular smoking serves as a mediator between ADHD symptom class and illicit drug dependence. Discussion In this study, we aimed to determine the ways in which childhood ADHD symptoms and smoking status in adolescence are related to subsequent illicit drug by explicitly testing main effects as well as moderating and indirect effects of smoking status. High levels of ADHD symptom endorsement and adolescent cigarette use are two prevalent and established risk factors for subsequent illicit drug use (Molina & Pelham, 2014; Palmer et al., 2009; Sundquist et al., 2015). The findings of this study suggest that ADHD symptoms may predict a greater likelihood of illicit drug use by increasing the risk of early cigarette smoking. The findings here elucidate the relationships among childhood ADHD symptoms, smoking status during adolescence, and subsequent illicit drug use. The use of longitudinal panel data and discrete-time survival analysis further strengthens the findings by imposing temporal ordering on some of the studied variables. As was shown by Biederman et al. (2006), youth with ADHD symptoms who also smoked cigarettes were more likely to use illicit drugs at later time points compared with ADHD youth who did not smoke. In other words, cigarette smoking moderated the association between ADHD symptoms and illicit drug use. In contrast to Biederman et al. (2006), the present study did not find evidence of a statistical interaction effect between ADHD symptoms and smoking status on subsequent drug use. Rather, the effect found here supported the potential mediating effect of smoking and helping to identify a possible explanation for why this particular relationship might occur (Baron & Kenny, 1986). In the present study, nearly a third of the total effect of ADHD symptoms on future illicit drug use was statistically mediated by smoking status at Wave I. Our findings provide support for the mediation hypothesis and suggest that preventing or reducing adolescent smoking especially among individuals with childhood ADHD symptoms (or children who demonstrate general liabilities such as poor impulse control and attention problems) could decrease the future incidence of illicit drug use and dependence. The fact that we observed these associations in a large, population-based cohort further supports the potential generalizability of these findings. The etiology of substance use and addiction is multifaceted. Although there is a growing literature focused on the gateway sequence to substance use (Kandel & Kandel, 2015; Kandel, Yamaguchi & Klein, 2006), others suggest a general liability reflected by poor self-regulation, which increases the risk for substance use during late adolescence or early adulthood (Masse & Tremblay, 1997; Moffitt et al., 2011; Tarter, Kirisci, Reynolds, & Mezzich, 2004). While the gateway sequence focuses on the developmental progression of substance use, the common liability model suggests that a general liability increases the risk of progressing to using another drug (Vanyukov et al., 2012). Our findings seem to be in line with both frameworks. On one hand, our model demonstrated a significant link between an earlier “light” substance use (early history of cigarette smoking) and a later usage of “hard” drug (illicit drug use), regardless of ADHD symptoms. On the other hand, our data show that both smoking during early adolescence and illicit drug use during adulthood were predicted by a generalized liability manifested as ADHD symptoms. Replication in other prospective cohorts would help further strengthen support the associations observed here. A limitation of the current study was the use of retrospective self-report of ADHD symptoms. Research has shown that using adult retrospective data to derive population-based estimates of ADHD prevalence may lead to an overestimate of the point prevalence (Mannuzza, Klein, Klein, Bessler, & Shrout, 2002). Thus, prospective collection of data over the life course would be preferred in future studies. Another limitation of the study was that data regarding ADHD medication taken during childhood was not available. Therefore, we were unable to control for the medication use could have had on the associations between ADHD symptoms and later drug use. Another methodological caveat that should be considered in the interpretation of the results is that we excluded individuals who reported illicit drugs at Wave I (n =1,056). This was done intentionally, by design, to ensure temporal ordering of the variables. In so doing, however, the results represent the effect of smoking during early adolescence on subsequent, but not concurrent, illicit drug use. Further, we are not able to make claims about factors that may be associated with the small numbers of adolescents who began illicit drug use early during adolescence – meaning future research is needed to examine predictors among this high-risk group. Finally, the inconsistent assessment of marijuana use across the waves did not permit us to include marijuana as an illicit drug in this study. However, given that marijuana is one of the commonly used substances among U.S. adolescents (See Dubowitz, Thompson, Arria, English, Metzger, & Kotch, 2016; Johnston et al. 2016), future research is needed to evaluate it as a predictor, and as an outcome among youth with and without ADHD symptoms. Despite utilizing longitudinal modeling, causation cannot be determined. Thus, it is important to be cognizant that observational methods for testing statistical mediation here can help generate hypotheses about how risk factors might relate to one another, but establishing a mediation role of certain variables would only be possible using experimental methods. In addition, the effect size of some paths observed in this analysis would be considered small. For instance, Ferguson (2009) suggests odds ratios that are <2 represent small effects. However, this study does contribute to the literature, in that its findings support the hypothesis that early cigarette use might mediate the relationship between childhood ADHD symptoms and future illicit drug use. More research is needed to better understand why children with ADHD symptoms may be more prone to experiment with and maintain cigarette use during early adolescence. The findings of this study can be useful in the development of meaningful public health and prevention (or intervention) efforts. For example, reducing cigarette use during early adolescence could result in lowering future illicit drug use and dependence, especially among individuals with a high liability for substance use—such as children with ADHD symptoms. Illicit drug use prevention research should capitalize on this sensitive period of development (i.e., early adolescence) by focusing on cigarette use cessation and prevention in this population. Implications and Contributions Smoking during adolescence may play a role in explaining why some children with ADHD symptoms progress to other drugs. Reducing adolescent smoking is a prudent public health strategy. However, it is possible that a reduction in smoking among adolescents with ADHD symptoms could lead to a decrease in the incidence of drug use in this high-risk population. Acknowledgments This research uses data from Add Health, a program project directed by Kathleen Mullan Harris and designed by J. Richard Udry, Peter S. Bearman, and Kathleen Mullan Harris at the University of North Carolina at Chapel Hill, and funded by grant P01 HD31921 from the Eunice Kennedy Shriver National Institute of Child Health and Human Development, with cooperative funding from 23 other federal agencies and foundations. Special acknowledgment is owing to Ronald R. Rindfuss and Barbara Entwisle for assistance in the original design. Information on how to obtain the Add Health data files is available on the Add Health Web site (http://www.cpc.unc.edu/addhealth). No direct support was received from grant P01-HD31921 for this analysis. All authors have contributed significantly to the work. Funding For analyses and manuscript development was supported by National Institutes of Health grants R01 DA030487 (awarded to B.F.F.), K07CA124905 (awarded to B.F.F.), and K24DA023464 (awarded to K.S.H.). Conflicts of interest: None declared. References Baron R. M. , Kenny D. A. ( 1986). The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology , 51, 1173– 1182. Google Scholar CrossRef Search ADS PubMed  Biederman J. , Monuteaux M. C., Mick E., Wilens T. E., Fontanella J. A., Poetzl K. M., Kirk T., Masse J., Faraone S. V. ( 2006). Is cigarette smoking a gateway to alcohol and illicit drug use disorders? A study of youths with and without attention deficit hyperactivity disorder. Biological Psychiatry , 59, 258– 264. doi: 10.1016/j.biopsych.2005.07.009 Google Scholar CrossRef Search ADS PubMed  Charach A. , Yeung E., Climans T., Lillie E. ( 2011). Childhood attention-deficit/hyperactivity disorder and future substance use disorders: Comparative meta-analyses. Journal of the American Academy of Child and Adolescent Psychiatry , 50, 9– 21. doi: 10.1016/j.jaac.2010.09.019 Google Scholar CrossRef Search ADS PubMed  Dubowitz H. , Thompson R., Arria A. M., English D., Metzger R., Kotch J. B. ( 2016). Characteristics of Child Maltreatment and Adolescent Marijuana Use: A Prospective Study. Child Maltreatment , 21 1, 16– 25. http://doi.org/10.1177/1077559515620853 Google Scholar CrossRef Search ADS PubMed  Enders C. K. , Bandalos D. L. ( 2001). The relative performance of full information maximum likelihood estimation for missing data in structural equation models. Structural Equation Modeling , 8, 430– 457. Google Scholar CrossRef Search ADS   Ferguson C. J. ( 2009). An effect size primer: A guide for clinicians and researchers. Professional Psychology: Research and Practice , 40, 532– 538. Google Scholar CrossRef Search ADS   Fuemmeler B. F. , Kollins S. H., McClernon F. J. ( 2007). Attention deficit hyperactivity disorder symptoms predict nicotine dependence and progression to regular smoking from adolescence to young adulthood. Journal of Pediatric Psychology , 32, 1203– 1213. doi: 10.1093/jpepsy/jsm051 Google Scholar CrossRef Search ADS PubMed  Hanna E. Z. , Yi H. Y., Dufour M. C., Whitmore C. C. ( 2001). The relationship of early-onset regular smoking to alcohol use, depression, illicit drug use, and other risky behaviors during early adolescence: Results from the youth supplement to the third national health and nutrition examination survey. Journal of Substance Abuse , 13, 265– 282. Google Scholar CrossRef Search ADS PubMed  Harris K. M. , Florey F., Tabor J., Bearman P. S., Jones J., Udry J. R. ( 2009). The National Longitudinal Study of adolescent health: Research design. Retrieved from http://www.cpc.unc.edu/projects/addhealth/design Retrieved January 30, 2010. Johnston L. D. , O’malley P. M., Miech R. A., Bachman J. G., Schulenberg J. E. ( 2016). Monitoring the future national survey results on drug use, 1975-2015: Overview, key findings on adolescent drug use . Ann Arbor, MI: Institute for Social Research, The University of Michigan. Kandel D. , Kandel E. ( 2015). The Gateway Hypothesis of substance abuse: Developmental, biological and societal perspectives. Acta Paediatrica , 104, 130– 137. doi: 10.1111/apa.12851 Google Scholar CrossRef Search ADS PubMed  Kandel D. B. , Yamaguchi K., Klein L. C. ( 2006). Testing the gateway hypothesis. Addiction , 101, 470– 472; discussion 474–476. doi: 10.1111/j.1360-0443.2006.01426.x Google Scholar CrossRef Search ADS PubMed  Kessler R. C. , Adler L. A., Barkley R., Biederman J., Conners C. K., Faraone S. V., Greenhill L. L., Jaeger S., Secnik K., Spencer T., Ustün T. B., Zaslavsky A. M. ( 2005). Patterns and predictors of attention-deficit/hyperactivity disorder persistence into adulthood: Results from the national comorbidity survey replication. Biological Psychiatry , 57, 1442– 1451. doi: 10.1016/j.biopsych.2005.04.001 Google Scholar CrossRef Search ADS PubMed  Kollins S. H. , McClernon F. J., Fuemmeler B. F. ( 2005). Association between smoking and attention-deficit/hyperactivity disorder symptoms in a population-based sample of young adults. Archives of General Psychiatry , 62, 1142– 1147. doi: 10.1001/archpsyc.62.10.1142 Google Scholar CrossRef Search ADS PubMed  Lambert N. ( 2005). The contribution of childhood ADHD, conduct problems, and stimulant treatment to adolescent and adult tobacco and psychoactive substance abuse. Ethical Human Psychology and Psychiatry , 7, 197– 221. Lee S. S. , Humphreys K. L., Flory K., Liu R., Glass K. ( 2011). Prospective association of childhood attention-deficit/hyperactivity disorder (ADHD) and substance use and abuse/dependence: A meta-analytic review. Clinical Psychology Review , 31, 328– 341. doi: 10.1016/j.cpr.2011.01.006 Google Scholar CrossRef Search ADS PubMed  Lo Y. T. , Mendell N. R., Rubin D. B. ( 2001). Testing the number of components in a normal mixture. Biometrika , 88, 767– 778. doi: DOI 10.1093/biomet/88.3.767 Google Scholar CrossRef Search ADS   Mannuzza S. , Klein R. G., Klein D. F., Bessler A., Shrout P. ( 2002). Accuracy of adult recall of childhood attention deficit hyperactivity disorder. The American Journal of Psychiatry , 159, 1882– 1888. doi: 10.1176/appi.ajp.159.11.1882 Google Scholar CrossRef Search ADS PubMed  Masse L. C. , Tremblay R. E. ( 1997). Behavior of boys in kindergarten and the onset of substance use during adolescence. Archives of General Psychiatry , 54, 62– 68. Google Scholar CrossRef Search ADS PubMed  Mayet A. , Legleye S., Chau N., Falissard B. ( 2011). Transitions between tobacco and cannabis use among adolescents: A multi-state modeling of progression from onset to daily use. Addictive Behaviors , 36, 1101– 1105. doi: 10.1016/j.addbeh.2011.06.009 Google Scholar CrossRef Search ADS PubMed  McClernon F. J. , Fuemmeler B. F., Kollins S. H., Kail M. E., Ashley-Koch A. E. ( 2008). Interactions between genotype and retrospective ADHD symptoms predict lifetime smoking risk in a sample of young adults. Nicotine and Tobacco Research , 10, 117– 127. doi: 10.1080/14622200701704913 Google Scholar CrossRef Search ADS PubMed  Merrill J. C. , Kleber H. D., Shwartz M., Liu H., Lewis S. R. ( 1999). Cigarettes, alcohol, marijuana, other risk behaviors, and American youth. Drug and Alcohol Dependence , 56, 205– 212. Google Scholar CrossRef Search ADS PubMed  Moffitt T. E. , Arseneault L., Belsky D., Dickson N., Hancox R.J., Harrington H., Houts R., Poulton R., Roberts B. W., Ross S., Sears M. R., Thomson W. M., Caspi A. ( 2011). A gradient of childhood self-control predicts health, wealth, and public safety. Proceedings of the National Academy of Sciences of the United States of America , 108, 2693– 2698. doi: 10.1073/pnas.1010076108 Google Scholar CrossRef Search ADS PubMed  Molina B. S. , Hinshaw S. P., Eugene Arnold L., Swanson J. M., Pelham W. E., Hechtman L., Hoza B., Epstein J. N., Wigal T., Abikoff H. B., Greenhill L. L., Jensen P. S., Wells K. C., Vitiello B., Gibbons R. D., Howard A., Houck P. R., Hur K., Lu B., Marcus S.; MTA Cooperative Group. ( 2013). Adolescent substance use in the multimodal treatment study of attention-deficit/hyperactivity disorder (ADHD) (MTA) as a function of childhood ADHD, random assignment to childhood treatments, and subsequent medication. Journal of the American Academy of Child and Adolescent Psychiatry , 52, 250– 263. doi: 10.1016/j.jaac.2012.12.014 Google Scholar CrossRef Search ADS PubMed  Molina B. S. , Pelham W. E.Jr. ( 2014). Attention-deficit/hyperactivity disorder and risk of substance use disorder: Developmental considerations, potential pathways, and opportunities for research. Annual Review of Clinical Psychology , 10, 607– 639. doi: 10.1146/annurev-clinpsy-032813-153722 Google Scholar CrossRef Search ADS PubMed  Murphy K. , Barkley R. A. ( 1996). Attention deficit hyperactivity disorder adults: Comorbidities and adaptive impairments. Comprehensive Psychiatry , 37, 393– 401. Google Scholar CrossRef Search ADS PubMed  Muthén L. K. , Muthén B. O. ( 1998-2005). Mplus user’s guide  (7th ed.). Los Angeles, CA: Muthén & Muthén. Palmer R. H. , Young S. E., Hopfer C. J., Corley R. P., Stallings M. C., Crowley T. J., Hewitt J. K. ( 2009). Developmental epidemiology of drug use and abuse in adolescence and young adulthood: Evidence of generalized risk. Drug and Alcohol Dependence , 102, 78– 87. doi: 10.1016/j.drugalcdep.2009.01.012 Google Scholar CrossRef Search ADS PubMed  Resnick M. D. , Bearman P. S., Blum R. W., Bauman K. E., Harris K. M., Jones J., Tabor J., Beuhring T., Sieving R. E., Shew M., Ireland M., Bearinger L. H., Udry J.R. ( 1997). Protecting adolescents from harm. Findings from the National Longitudinal Study on Adolescent Health. JAMA , 278, 823– 832. Google Scholar CrossRef Search ADS PubMed  Sundquist J. , Ohlsson H., Sundquist K., Kendler K. S. ( 2015). Attention-deficit/hyperactivity disorder and risk for drug use disorder: A population-based follow-up and co-relative study. Psychological Medicine , 45, 977– 983. doi: 10.1017/S0033291714001986 Google Scholar CrossRef Search ADS PubMed  Szobot C. M. , Rohde L. A., Bukstein O., Molina B. S., Martins C., Ruaro P., Pechansky F. ( 2007). Is attention-deficit/hyperactivity disorder associated with illicit substance use disorders in male adolescents? A community-based case-control study. Addiction , 102, 1122– 1130. doi: 10.1111/j.1360-0443.2007.01850.x Google Scholar CrossRef Search ADS PubMed  Tarter R. E. , Kirisci L., Reynolds M., Mezzich A. ( 2004). Neurobehavior disinhibition in childhood predicts suicide potential and substance use disorder by young adulthood. Drug and Alcohol Dependence , 76 (Suppl), S45– S52. doi: 10.1016/j.drugalcdep.2004.08.006 Google Scholar CrossRef Search ADS PubMed  Vanyukov M. M. , Tarter R. E., Kirillova G. P., Kirisci L., Reynolds M. D., Kreek M. J., Conway K. P., Maher B. S., Iacono W. G., Bierut L., Neale M. C., Clark D. B., Ridenour T.A. ( 2012). Common liability to addiction and “gateway hypothesis”: Theoretical, empirical and evolutionary perspective. Drug and Alcohol Dependence , 123 (Suppl 1), S3– 17. doi: 10.1016/j.drugalcdep.2011.12.018 Google Scholar CrossRef Search ADS PubMed  Visser S. N. , Danielson M. L., Bitsko R. H., Holbrook J. R., Kogan M. D., Ghandour R. M., Perou R., Blumberg S.J. ( 2014). Trends in the parent-report of health care provider-diagnosed and medicated attention-deficit/hyperactivity disorder: United States, 2003-2011. Journal of the American Academy of Child and Adolescent Psychiatry , 53, 34– 46 e32. doi: 10.1016/j.jaac.2013.09.001 Google Scholar CrossRef Search ADS PubMed  Visser S. N. , Lesesne C. A., Perou R. ( 2007). National estimates and factors associated with medication treatment for childhood attention-deficit/hyperactivity disorder. Pediatrics , 119 (Suppl 1), S99– S106. doi: 10.1542/peds.2006-2089O Google Scholar CrossRef Search ADS PubMed  © The Author 2017. Published by Oxford University Press on behalf of the Society of Pediatric Psychology. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com

Journal

Journal of Pediatric PsychologyOxford University Press

Published: Mar 1, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off