Psychology of Addictive Behaviors 2015, Vol. 29, No. 3, 541–551

© 2015 American Psychological Association 0893-164X/15/$12.00 http://dx.doi.org/10.1037/adb0000059

Internalizing and Externalizing Psychopathology as Predictors of Cannabis Use Disorder Onset During Adolescence and Early Adulthood Richard F. Farmer, John R. Seeley, Derek B. Kosty, Jeff M. Gau, and Susan C. Duncan

Michael T. Lynskey Institute of Psychiatry, Kings College

Oregon Research Institute

Peter M. Lewinsohn This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Oregon Research Institute Risk-related liabilities associated with the development of cannabis use disorders (CUDs) during adolescence and early adulthood are thought to be established well before the emergence of the index episode. In this study, internalizing and externalizing psychopathology from earlier developmental periods were evaluated as risk factors for CUDs during adolescence and early adulthood. Participants (N ⫽ 816) completed 4 diagnostic assessments between the ages 16 and 30, during which current and past CUDs were assessed as well as a full range of psychiatric disorders associated with internalizing and externalizing psychopathology domains. In unadjusted and adjusted time-to-event analyses, externalizing but not internalizing psychopathology from proximal developmental periods predicted subsequent CUD onset. A large proportion of adolescent and early adult cases, however, did not manifest any externalizing or internalizing psychopathology during developmental periods before CUD onset. Findings are consistent with the emerging view that externalizing disorders from proximal developmental periods are robust risk factors for CUDs. Although the identification of externalizing liabilities may aid in the identification of individuals at risk for embarking on developmental pathways that culminate in CUDs, such liabilities are an incomplete indication of overall risk. Keywords: cannabis use disorders, predictors, psychopathology, externalizing, internalizing

disorder covariation among children and adults in cross-sectional and longitudinal studies (e.g., Achenbach, 1966; Farmer, Seeley, Kosty, Olino, & Lewinsohn, 2013b; Kessler et al., 2011; Krueger & Markon, 2006; Lahey et al., 2008), and has been suggested as a viable model for guiding research on common causal pathways that account for disorder comorbidity (Kessler et al., 2011; Krueger, 1999). Externalizing tendencies and their precursors constitute a broad liability for disruptive behavior and substance use disorders inclusive of cannabis abuse or dependence disorders (collectively, cannabis use disorders or CUDs; Farmer, Seeley, Kosty, & Lewinsohn, 2009; Iacono, Malone, & McGue, 2008; Krueger & Markon, 2006). Prospective studies, for example, have documented that externalizing tendencies and associated attributes frequently precede the onset of cannabis initiation (Agrawal, Lynskey, Bucholz, Madden, & Heath, 2007; Griffith-Lendering, Huijbregts, Mooijaart, Vollebergh, & Swaab, 2011b; Kandel & Yamaguchi, 1993; King, Iacono, & McGue, 2004; van den Bree & Pickworth, 2005) and CUDs (Blanco et al., 2014; Brook, Lee, Finch, Koppel, & Brook, 2011; Fergusson, Horwood, & Ridder, 2007; Hayatbakhsh et al., 2008; Tarter, Kirisci, Ridenour, & Vanyukov, 2008; although see Wittchen et al., 2007, who did not find a significant prospective association with externalizing disorders). The contributions of internalizing tendencies and their precursors to the risk for cannabis initiation and CUDs are less clear. Although CUDs and internalizing disorders are highly comorbid and share common risk factors (Grant et al., 2004), prospective

Although adolescent and early adult developmental periods are associated with the greatest risk for cannabis experimentation and the subsequent progression to abuse or dependence (Degenhardt et al., 2013; Fergusson & Horwood, 2000; Johnston, O’Malley, Bachman, & Schulenberg, 2013), risk factors that influence these events are often evident earlier in development. A contemporary view on the epidemiology of substance use disorders among adolescents and young adults suggests that multiple factors related to processes that increase vulnerabilities to externalizing and internalizing psychiatric disorders contribute to overall risk (Vanyukov et al., 2012; Zucker, Donovan, Masten, Mattson, & Moss, 2008). The internalizing-externalizing organizational model of psychopathology (Achenbach, 1966; Krueger, 1999) has received considerable support for describing patterns of psychiatric symptom and

This article was published Online First March 23, 2015. Richard F. Farmer, John R. Seeley, Derek B. Kosty, Jeff M. Gau, and Susan C. Duncan, Oregon Research Institute; Michael T. Lynskey, Addictions Department, Institute of Psychiatry, Kings College; Peter M. Lewinsohn, Oregon Research Institute. National Institutes of Health Grants MH40501, MH50522, and DA12951 to P.M.L. and R01DA032659 to R.F.F. and J.R.S. supported this research. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. Correspondence concerning this article should be addressed to Richard F. Farmer, Oregon Research Institute, 1776 Millrace Drive, Eugene, OR 97403. E-mail: [email protected] 541

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

542

FARMER ET AL.

studies have been equivocal as to whether internalizing symptoms or disorders afford a significant risk for later cannabis use or CUDs. Some studies have found that internalizing symptoms or disorders predict later cannabis use (Henry et al., 1993; Paton, Kessler, & Kandel, 1977) or CUDs (Wittchen et al., 2007), whereas findings reported by Colder et al. (2013) imply that internalizing features in the absence of externalizing features may act as a protective factor against cannabis use. Other researchers found no prospective association between internalizing features or disorders and subsequent cannabis use (Griffith-Lendering et al., 2011b; Miller-Johnson, Lochman, Coie, Terry, & Hyman, 1998) or CUDs (Tarter et al., 2008). Mixed findings concerning internalizing features may be related to differences in statistical control of psychopathology-related confounders across studies. An earlier prospective analysis based on Oregon Adolescent Depression Project (OADP) data (Buckner et al., 2008), which constitutes the parent study for the present investigation, found that a diagnosis of social phobia at baseline was associated with an increased risk for cannabis dependence (but not cannabis abuse) at a 14-year followup. This study, however, exercised limited control over concurrent or lifetime externalizing psychopathology that might account for longitudinal associations between social phobia and cannabis dependence. Although there is ample research on internalizing and externalizing predictors of cannabis initiation and use, there is considerably less research on these domains as predictors of CUD onset. The distinction between cannabis initiation or use and CUDs is an important one when investigating the developmental pathways that constitute a heightened risk for hazardous cannabis use. Contemporary views suggest that CUD risk is related to underlying biologically based mechanisms, whereas cannabis initiation and occasional use are largely the product of shared and nonshared environmental factors (Vanyukov et al., 2012; Verweij et al., 2010). The biological underpinnings of CUDs likely arise from a common liability shared with other substance use disorders, externalizing behavioral disorders, and possibly internalizing disorders (Vanyukov et al., 2012; Zucker et al., 2008). To the extent that such risk liabilities are shared among families of related lifetime disorders that are initially expressed during childhood, their existence should be evident well before the onset of CUDs. Correspondingly, a primary objective of this research was to evaluate whether earlier manifestations of internalizing and externalizing psychopathology predict CUD onset. Prior research on predictors of cannabis initiation and CUD onset has generally not considered the possibility of the developmental specificity of predictors. In one exception, van den Bree and Pickworth (2005) found both common and developmental period-specific risk factors were associated with the progression from cannabis initiation to regular use. Blanco et al. (2014) further reported that proximal externalizing tendencies, or those that are manifest closer to the onset of problematic cannabis use, are generally more salient predictors of risk than similar distal (e.g., childhood) tendencies. Maslowsky, Schulenberg, and Zucker (2014), however, reported that conduct problems in the 8th grade (distal) assessed via a short self-report questionnaire were better predictors of cannabis use in the 12th grade than comparable problems reported during the 10th grade (proximal). Given the mixed pattern of findings concerning the temporal specificity of predictors, this study investigated the relative salience of adoles-

cent versus childhood internalizing and externalizing tendencies in the prediction of CUD onsets in early adulthood. As noted earlier, research on internalizing and externalizing predictors of cannabis initiation or CUD onset has generally not considered either domain following the control of the other. Given that internalizing and externalizing psychopathology covary (Farmer et al., 2013b; Krueger, 1999; Lahey et al., 2008), erroneous conclusions about the relative contributions of either domain as risk factors for CUDs might be reached when the other domain is not simultaneously considered and controlled. Consequently, in the present research, we evaluated unadjusted and adjusted prediction models, with the latter controlling for demographic and psychosocial variables as well as the psychopathology domain that was not the predictor variable modeled in the analysis. Finally, we evaluated whether gender moderates associations between internalizing and externalizing psychopathology and subsequent CUD onsets. Although reliable gender differences in cannabis use often begin to emerge during late adolescence and early adulthood, with males tending to use more frequently or at a higher rate than females (Agrawal & Lynskey, 2007; Kandel & Chen, 2000; Perkonigg et al., 2008), there has been little investigation of, or evidence for, systematic gender moderation of psychopathology-related risk factors associated with the onset of CUDs (Blanco et al., 2014; Costello, Erkanli, Federman, & Angold, 1999). Our tests of gender moderation of psychopathology domain effects were, therefore, exploratory.

Method Participants This study is based on the OADP (Lewinsohn, Hops, Roberts, Seeley, & Andrews, 1993), a longitudinal study of an adolescent cohort from the community. At the first wave of data collection (T1; ⬃ age 16), the sample consisted of 1,709 adolescents randomly selected from nine high schools that were representative of urban and rural districts in western Oregon. About 1 year later (T2), 1,507 (88%) of these persons were reassessed. At T3 (⬃ age 24), a sampling stratification procedure was introduced whereby eligible participants included all individuals self-identified as nonWhite to enhance ethnic diversity, all persons with a positive history of a psychiatric diagnosis by T2 (n ⫽ 644), and a randomly selected subset of participants with no history of a psychiatric or substance use disorder by T2 (n ⫽ 457 of 863 persons). Of these 1,101 eligible persons, 941 (85%) completed T3. The T4 assessment (⬃ age 30) included 816 T3 participants (87%) and was conducted ⬃ 6 years after T3. We previously reported period and lifetime prevalence rates for CUDs in the OADP sample (Lewinsohn et al., 1993; Farmer et al., 2015, respectively). Briefly, by age 30.0, 19.1% of the weighted T4 panel had a lifetime CUD diagnosis (22.5% male, 16.4% female, p ⬍ .05). For those with a lifetime CUD episode by age 30.0, the mean age at time of first CUD onset was 18.6 years (SD ⫽ 4.1; Mdn ⫽ 18.2). Analyses of participant attrition revealed only minimal sample bias related to study discontinuation (Farmer et al., 2015; Farmer, Kosty, Seeley, Olino, & Lewinsohn, 2013a; Lewinsohn et al., 1993). The reference sample for this study varied in accordance with the research question. Demographic predictors of CUD onset were

PREDICTORS OF CANNABIS USE DISORDER ONSET

evaluated using data from the complete T4 panel (n ⫽ 816). Psychiatric predictors of adolescent CUD onset (between ages 13.0 and 17.9) were evaluated using data from participants without an incidence of CUD between ages 0 and 12.9 (n ⫽ 804), in which CUD episodes occurring at or after age 18 were right-censored.1 Psychiatric predictors of early adult CUD onset (between ages 18.0 and 30.0) were evaluated using data from participants without an incidence of CUD before age 17.9 (n ⫽ 727).

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Diagnostic Assessments During T1, T2, and T3, participants were interviewed with the Schedule for Affective Disorders and Schizophrenia for SchoolAge Children (K-SADS) that combined features of the Epidemiologic and Present Episode versions (Chambers et al., 1985; Orvaschel, Puig-Antich, Chambers, Tabrizi, & Johnson, 1982). Follow-up assessments at T2 and T3 also involved the administration of the Longitudinal Interval Follow-Up Evaluation (LIFE; Keller et al., 1987) that, in conjunction with the K-SADS, provided detailed information related to the presence and course of disorders since participation in the previous interview. The T4 assessment included administration of the LIFE and the Structured Clinical Interview for Axis I DSM–IV (Diagnostic and Statistical Manual for Mental Disorders-Fourth Edition) Disorders–Non-Patient Edition (SCID-NP; First, Spitzer, Gibbon, & Williams, 1994). Diagnostic agreement was good to excellent across assessment waves (␬ ⫽ .56 to .84 for diagnostic categories other than CUD; median ⫽ .77; see Farmer et al., 2013a, for more details). Diagnostic interviews included the assessment of current episodes (i.e., within the last 12 months) and since the last interview (or, in the case of the T1 assessment, ever in the past), and disorder onsets and offsets were recorded as age in months. The presence versus absence of specific diagnostic categories, including cannabis abuse and cannabis dependence, was represented continuously in 1-month increments between childhood and age 30.0. For this study, cannabis abuse and dependence diagnoses were combined into a single category (CUDs) to indicate syndromal cannabis use that resulted in significant impairment in functioning or distress, with the timing of the episode onset denoted by age in months. Hazard rates (described below) were estimated for recovery in 1-month intervals commencing with the onset of the first CUD episode. Diagnostic agreement among raters for CUD diagnoses since the previous interview was good to excellent at each wave (␬s: T1 ⫽ .72, T2 ⫽ .93, T3 ⫽ .83, and T4 ⫽ .82).

Childhood and Adolescent Predictors of CUD Onset In a set of prospective analyses described below, we evaluated the degree to which internalizing and externalizing psychopathology were predictors of CUD onset, further differentiated into adolescent onset (ages 13 to 17.9) and early adult onset (ages 18.0 to 30.0). In models predicting adolescent CUD onset, we evaluated psychopathology occurring between ages 8.0 to 12.9 as childhood predictors. In models predicting early adult CUD onset, we evaluated psychopathology occurring from ages 8.0 to 12.9 and from 13.0 to 17.9 as childhood and adolescent predictors, respectively. Internalizing and externalizing disorder domain scores were categorically modeled, whereby a value of 0 was assigned if no disorder associated with a given domain was diagnosed and a

543

value of 1 assigned if one or more domain-related disorders was diagnosed within the timeframe specified. Diagnostic criteria and disorder threshold cut-offs are related to clinical impressions about the severity of a disorder that, in turn, warrants treatment when the diagnosis is reached (Fergusson, Horwood, & Lynskey, 1994). Based on our earlier research in explicating domains of lifetime psychopathology with the OADP sample (Farmer et al., 2009; Farmer et al., 2013b; Seeley, Kosty, Farmer, & Lewinsohn, 2011), DSM-defined disorders that contributed to the internalizing domain, further distinguished by DSM-defined subdomains, were: mood disorders (major depressive; dysthymia; bipolar spectrum inclusive of bipolar I, bipolar II, and cyclothymic disorders), anxiety disorders (separation anxiety, simple/specific phobia, generalized anxiety, obsessive— compulsive, panic, agoraphobia without panic, posttraumatic stress, and social phobia), and bulimia nervosa. Disorders that contributed to the definition of the externalizing domain, further distinguished by DSM-defined subdomains, were: disruptive behavior disorders (attention deficit/ hyperactivity, oppositional defiant, and conduct) and noncannabisrelated substance use disorders (alcohol use disorders, which includes alcohol abuse or dependence diagnoses; hard drug use disorders, which includes the abuse or dependence of substances other than cannabis and alcohol).

Statistical Analyses Most of the earlier predictive studies of cannabis initiation or CUD onset have utilized logistic regression methods. Cox proportional hazards (PH) regression methods, however, take into account time until an outcome event, and generally have more associated statistical power and yield more precise effect estimates than logistic regression models (Callas, Pastides, & Hosmer, 1998; van der Net et al., 2008). To fully exploit longitudinal information available from our data set, primary analyses utilized Cox PH models. Hazard ratios (HR) and the 95% confidence intervals produced from these analyses are based on the ratio of CUD onset probabilities, specifically differences in hazard rates as a linear function of the predictor variables. They are, therefore, relative indicators of risk, reflect any differences in hazard functions, and specify the increased likelihood of developing a CUD within a unit of time (months) as a function of the presence versus absence of a psychopathology-domain related diagnosis. The developmental period analyses presented below adhered to the following data analytic sequence, separately for those with initial adolescent and early adult CUD onsets: (a) descriptive data on the course of CUDs; (b) nonparametric tests of rate differences of individuals who had either internalizing and externalizing disorders (i.e., any psychopathology) in the preceding developmental period(s) as a function of the presence versus absence of an initial CUD diagnosis in the reference developmental period; (c) separate nonparametric tests of rate differences in internalizing and exter1 Initial CUD onsets before age 13 are rare (Boden, Fergusson, & Horwood, 2006; Chen & Kandel, 1995; Wittchen et al., 2007). To evaluate prospective associations between childhood psychopathology and later CUD onset, 12 cases with initial CUD onsets before age 13 were excluded from analyses involving predictors of onset (Tables 2 and 3). These participants, however, were included in analyses of demographic factors related to CUD onset (see Table 1), as these data were analyzed without reference to any specific developmental period.

FARMER ET AL.

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

544

nalizing disorder occurrences within the preceding developmental period(s) as a function of the presence versus absence of an initial CUD diagnosis in the reference developmental period; and, (d) unadjusted and adjusted time-to-event analyses that predicted CUD onset as a function of internalizing and externalizing psychopathology in the preceding developmental period(s). Time-toevent analyses were further distinguished at three levels: (a) unadjusted omnibus bivariate analyses, whereby psychiatric disorder domain variables (i.e., internalizing and externalizing) were individually evaluated as predictors of CUD onset without concurrent consideration of the possible influence of putative confounders; (b) adjusted omnibus multivariate analyses, whereby internalizing and externalizing domains were individually evaluated as predictors of CUD onset following control of putative confounders, including the superordinate psychopathology domain that was not the target variable of the analysis; and, (c) separate unadjusted and adjusted time-to-event analyses that evaluated DSM-defined disorder subgroupings within the internalizing and externalizing superordinate domains to better isolate whether subgroups of related disorders evidenced different predictive associations with CUD onset. Internalizing subdomains were mood disorders and anxiety disorders. Externalizing subdomains were disruptive behavior disorders and noncannabis-related substance use disorders (“other SUDs”). Time to CUD onset was expressed in month-length increments beginning with the start of the developmental period. As a result of the unequal stratified sampling strategy implemented at T3, White participants without a psychiatric diagnosis by T2 were undersampled at T3 and T4. To adjust for this sampling procedure, White participants with no lifetime diagnosis by T2 were assigned a weight that reflected the probability of this subgroup being sampled during T3 and T4 assessments. The Taylor series linearization method (Wolter, 1985) was implemented in SUDAAN version 11.0.0 (Research Triangle Institute, Research Triangle Park, NC) to appropriately adjust standard errors and confidence intervals for the parameter estimates. To avoid artificially inflating our sample size, we computed normalized sampling weights. Unless otherwise specified, all findings subsequently presented (e.g., rates, ratios) were based on weighted data, and references to the numbers of cases were based on unweighted data.

Results Demographic Predictors of CUD Onset By T4, 77 and 84 participants had initial CUD onsets during the adolescent and early adult developmental periods, respectively. Demographic variables were individually evaluated as predictors of CUD onset with Cox PH regression methods.2 Findings from these analyses, which were conducted with the entire participant sample, are presented in Table 1. Male participants and those who resided in a single parent household at T1 were significantly more likely to experience a CUD episode at any time interval through age 30 (Hazard ratios [HR] ⫽ 1.42 and 1.82, respectively). All demographic variables, regardless of their individual significance, were included among the covariates in adjusted analyses reported below.

Psychiatric Predictors of Adolescent CUD Onset For those with initial CUD episode onsets during adolescence, the mean onset age of the first episode was 15.5 years (SD ⫽ 1.4; Mdn ⫽ 15.6). The mean duration of the index episode was 32.2 months (SD ⫽ 44.4; Mdn ⫽ 13.0). Period prevalence rates of internalizing and externalizing disorders during childhood as a function of initial CUD onsets during adolescence. The proportions of participants with either internalizing or externalizing disorders during childhood were comparable among those without and with initial CUD onsets during adolescence (13.3% and 18.3%, respectively; Likelihood Ratio [LR] ␹2 ⫽ 1.13, p ⫽ .288). The left half of Table 2 presents the period prevalence rates of internalizing and externalizing psychopathology within the childhood developmental period among participants without and with an initial CUD onset during adolescence. The only statistically significant difference in childhood psychopathology between these two groups was found in the rates of disorders from the superordinate externalizing domain (3.1% and 9.1%, respectively; LR ␹2 ⫽ 4.54, p ⫽ .033). Childhood internalizing and externalizing disorders as predictors of CUD onset during adolescence. Table 2 also presents unadjusted and adjusted analyses that evaluated internalizing and externalizing domains as predictors of initial CUD onset during adolescence. Reported under the unadjusted analysis heading of the table are the total effects of psychiatric disorder domains in the prediction of CUD onset. As indicated in the right half of Table 2, childhood externalizing domain disorders significantly predicted initial CUD episodes in the adolescent onset group (HR [CI95] ⫽ 3.02 [1.42, 6.43]).3 This finding indicates that those with externalizing disorder histories during childhood had a three times greater odds of being diagnosed with a CUD during any monthlength interval within the adolescent developmental period when compared with those without such histories. Cumulative hazard functions for CUD onset during adolescence by psychopathology domain during childhood are presented in Figure 1. We also evaluated the relative contributions of childhood internalizing and externalizing subdomains in the prediction of CUD onset during adolescence. Only subdomains with a sufficient number of observed cases within the preceding developmental period 2 An assumption of Cox PH models is the absence of a significant time— by—predictor interaction. We initially included time-by-predictor interaction terms in each Cox model evaluated, and observed no instance where the interaction term was statistically significant. Interaction terms were consequently removed and the models rerun, with data from these latter analyses reported. Significant predictor effects, when observed, should therefore be regarded as consistent across the full developmental period examined rather than limited to a discrete portion of the interval. 3 In a series of additional analyses involving these data as well as those associated with index CUD onsets during early adulthood, we examined whether comorbidity within and across domains added significant information to the prediction of time to CUD onset beyond the consideration of “any externalizing” or “any internalizing.” These analyses, available from the first author, indicated that the consideration of disorder comorbidity did not account for significant additional variance in the prediction of CUD onset with the exception of models where internalizing disorders were the target predictor and externalizing disorders were included among the comorbid conditions considered. Extending prediction models to include consideration of comorbidity, therefore, does not change or otherwise modify our findings presented in Tables 2 and 3 or our conclusions about the data.

PREDICTORS OF CANNABIS USE DISORDER ONSET

545

Table 1 Descriptive Statistics for Demographic Variables and Bivariate Associations With Cannabis Use Disorders Onset Through Age 30 (n ⫽ 816)

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Predictor Participant characteristics Male, % [CI95] Non-White, % [CI95] Pubertal timing, % [CI95] Early (n ⫽ 153) vs. on-time (n ⫽ 567), % [CI95] Late (n ⫽ 96) vs. on-time (n ⫽ 567), % [CI95] Early (n ⫽ 153) vs. late (n ⫽ 96), % [CI95] Repeat grade before age 12, % [CI95] Family demographic variables Single versus dual parent household, % [CI95] At least one parent completed college, % [CI95] Mean parent age at T1, M (SD) Number of older siblings, M (SD)

No CUD (n ⫽ 643)

CUD (n ⫽ 173)

Test statistica

42.0 [38.3, 45.8] 7.4 [5.5, 9.5]

51.7 [43.8, 59.5] 9.2 [5.3, 14.4]

4.77 0.57

.030 .449

1.42 [1.03, 1.95] 1.29 [0.79, 2.08]

19.0 [15.9, 22.3] 13.6 [10.9, 16.7] 59.7 [52.6, 66.6] 10.1 [8.0, 12.6]

24.8 [18.0, 32.5] 17.8 [11.7, 25.2] 60.4 [47.2, 72.6] 13.9 [9.1, 19.9]

2.22 1.34 0.01 1.75

.136 .248 .928 .186

1.40 [0.94, 2.07] 1.33 [0.82, 2.14] 1.06 [0.61, 1.82] 1.38 [0.89, 2.15]

60.6 [56.8, 64.3] 44.9 [41.1, 48.7] 42.4 (5.9) 0.9 (1.1)

44.2 [36.6, 52.1] 41.0 [33.5, 48.8] 41.5 (5.6) 0.9 (1.1)

13.66 0.78 1.85 0.19

⬍.001 .376 .065 .851

1.82 [1.33, 2.50] 0.87 [0.63, 1.20] 0.97 [0.94, 1.00] 0.98 [0.85, 1.14]

p-value

Hazard ratio

Note. CI95 ⫽ 95% confidence interval. Statistically significant hazard ratios are bolded. a Likelihood ratio ␹2 and independent observation t-tests were conducted on categorical (as indicated by the % designation) and continuous variables (as indicated by the M designation), respectively.

(n ⱖ 20) were analyzed as predictors, resulting in the omission of other SUDs and bulimia nervosa in tests of predictors in time-toevent analyses. In unadjusted analyses, the disruptive behavior disorder subdomain emerged as a significant predictor (HR [CI95] ⫽ 2.90 [1.30, 6.47]) of adolescent CUD. Under the adjusted analysis heading of Table 2, we report the additive effects of individual disorder domains after variance associated with putative confounders and the nontargeted psychiatric disorder superordinate domain was controlled.4 When compared with unadjusted findings, adjusted effects were generally smaller, although the effects associated with the externalizing superordinate domain (HR [CI95] ⫽ 2.90 [1.30, 6.47]) and disruptive behavior disorders subdomain (HR [CI95] ⫽ 2.40 [1.03, 5.57]) remained significant.

Childhood and Adolescent Psychiatric Predictors of Early Adult CUD Onset For those with an initial CUD episode onset during early adulthood, the mean age of first CUD onset was 21.7 years (SD ⫽ 2.8; Mdn ⫽ 21.2). The mean duration of the index episode was 44.4 months (SD ⫽ 40.1; Mdn ⫽ 31.5). Period prevalence rates of internalizing and externalizing disorders during childhood and adolescence as a function of initial CUD onsets during early adulthood. Period prevalence rates for either internalizing or externalizing disorders during adolescence were compared among participants without and with initial CUD onsets during early adulthood. When compared with those who did not experience an index CUD episode within this developmental period, participants with an initial CUD onset during early adulthood experienced a significantly higher rate of disorders from either of these superordinate domains during adolescence (28.0% and 44.0%, respectively; LR ␹2 ⫽ 8.47, p ⫽ .004). The left half of Table 3 presents rates of internalizing and externalizing disorder occurrences for the early adult onset group, separately for adolescence and childhood. Rates of adolescent internalizing disorders at both the superordinate domain and subdomain levels, when examined separately from externalizing dis-

orders, did not significantly differ between those without and with initial CUD onsets during early adulthood. Rates of adolescent externalizing domain disorders, however, significantly differed between those without and with initial CUD onsets during early adulthood (7.2% vs. 24.2%, respectively; LR ␹2 ⫽ 19.68, p ⬍ .001). Similarly, when analyzed separately, both subdomains of externalizing disorders significantly differed between those without and with CUD onsets during early adulthood (for disruptive behavior disorders: 3.5% and 13.0%, respectively; LR ␹2 ⫽ 11.14, p ⬍ .001; for other SUDs: 4.2% and 15.6%, respectively; LR ␹2 ⫽ 13.34, p ⬍ .001). The proportions of participants with either internalizing or externalizing disorders during childhood were comparable among those without and with initial CUD onsets during early adulthood (13.0% and 15.5%, respectively; LR ␹2 ⫽ 0.38, p ⫽ .539). Rates of childhood internalizing and externalizing disorders, at both the superordinate domain and subdomain levels, similarly did not significantly differ between those without and with initial CUD onsets during early adulthood in any analysis (see Table 3). Adolescent internalizing and externalizing disorders as predictors of CUD onset during early adulthood. The upper right half of Table 3 presents unadjusted and adjusted analyses that evaluated the utility of internalizing and externalizing domains as proximal and distal predictors of early adult CUD onset. Reported under the unadjusted analysis heading of Table 3 are the total effects associated with adolescent psychiatric disorder domains. As indicated in the table, a diagnosis of either internalizing or externalizing domain psychopathology during adolescence significantly predicted CUD onset during early adulthood (HR [CI95] ⫽ 1.91 [1.22, 2.98]). When analyzed separately from externalizing disorders, the internalizing superordinate domain and associated subdomains did 4 The homogeneity of regression assumption was tested before performing each adjusted time-to-event analysis. In each instance, the effect of the interaction term was nonsignificant, and the term was subsequently removed from the final model.

FARMER ET AL.

546

Table 2 Period Prevalence Rates of Internalizing and Externalizing Psychopathology and Associations With Initial CUD Onset During Adolescence Hazard ratio [CI95]

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Period prevalence % [CI95] Predictor

No CUD (n ⫽ 727)

CUD (n ⫽ 77)

Likelihood ratio ␹

Childhood psychopathology Internalizing domain Mood disorders Anxiety disorders Bulimia nervosa Externalizing domain Disruptive behavior disorders Other substance use disorders Either internalizing or externalizing

11.0 [8.9, 13.4] 5.8 [4.3, 7.6] 6.9 [5.3, 8.9] 0.0 3.1 [2.0, 4.5] 2.8 [1.8, 4.1] 0.3 [0.1, 0.9] 13.3 [11.0, 15.9]

11.4 [5.1, 20.8] 8.0 [3.0, 16.4] 5.7 [1.7, 13.3] 0.0 9.1 [3.6, 17.9] 8.0 [3.0, 16.4] 1.1 [0.0, 6.0] 18.3 [10.1, 29.0]

0.01 0.47 0.14 NA 4.54 3.74 0.79 1.13

2

p-value

Unadjusted

Adjusted

.915 .494 .705 NA .033 .053 .374 .288

1.05 [0.53, 2.06] 1.39 [0.64, 3.04] 0.83 [0.33, 2.08] NA 3.02 [1.42, 6.43] 2.90 [1.30, 6.47] NA 1.24 [0.70, 2.22]

0.90 [0.45, 1.80] 1.17 [0.50, 2.73] 0.78 [0.31, 1.93] NA 2.44 [1.13, 5.29] 2.40 [1.03, 5.57] NA 1.45 [0.83, 2.54]

Note. CUD ⫽ cannabis use disorder; CI95 ⫽ 95% confidence interval; NA ⫽ not analyzed as predictors because of insufficient period prevalence rates. A value of 0.0 indicates the absence of diagnosed cases within the referenced developmental period. Statistically significant hazard ratios are bolded. Adjusted analyses control for variables in Table 1 and other psychopathology. Childhood was defined as ages 8.0 to 12.9 and adolescence was defined as ages 13.0 to 17.9.

not significantly predict CUD onset. Externalizing domain disorders between ages 13.0 and 17.9, however, were significant predictors of early adult CUD onset (HR [CI95] ⫽ 3.60 [2.20, 5.88]). This finding indicates that those with externalizing disorder histories during adolescence had a 3.6 times greater odds of being diagnosed with a CUD during any month-length interval within the early adult developmental period when compared with those without such histories. Cumulative hazard functions for CUD onset during early adulthood by psychopathology domain during adolescence are presented in Figure 2. When the two subdomains of externalizing psychopathology during adolescence were separately modeled, each emerged as a significant predictor of index CUD episodes during early adulthood (for disruptive behavior disorders: HR [CI95] ⫽ 3.47 [1.91, 6.31]; for other SUDs: HR [CI95] ⫽ 3.68 [2.04, 6.62]). Under the adjusted analysis heading of Table 3, we reported the additive effects of individual psychiatric disorder domains after variance associated with putative confounders and other domain psychopathology was controlled. When compared with unadjusted findings, adjusted effects were similar or smaller, but the pattern of

statistical significance remained the same as that observed for unadjusted analyses. Childhood internalizing and externalizing disorders as predictors of CUD onset during early adulthood. The lower right portion of Table 3 reports data on the association of childhood psychopathology and time to initial CUD onset during early adulthood. Reported under the unadjusted analysis heading of Table 3 are the total effects of childhood disorder domains in predicting early adult CUD onset. As indicated in Table 3, the predictive utility of internalizing and externalizing superordinate domains between ages 8.0 and 12.9, separately or in combination, did not achieve statistical significance, although externalizing domain disorders alone between ages 8.0 and 12.9 approached statistical significance (HR [CI95] ⫽ 2.16 [0.99, 4.71], p ⫽ .054). The disruptive behavior disorders subdomain of externalizing disorders, however, significantly predicted CUD onsets during early adulthood (HR [CI95] ⫽ 2.42 [1.11, 5.29], p ⬍ .05). Under the adjusted analysis heading of Table 3, we reported the additive effects of individual childhood psychiatric disorder domains after variance associated with putative confounders and other domain psychopathology within the same developmental period was controlled. The adjusted effects were comparable with the unadjusted findings. When the significant adjusted effect for childhood disruptive behavior disorders was reevaluated following the additional control of internalizing and externalizing domain psychopathology during the adolescent developmental period, disorders from this subdomain were no longer significant unique predictors of early adult CUD onset (HR [CI95] ⫽ 1.22 [0.53, 2.83]).

Gender Moderation

Figure 1. Cumulative hazard functions for cannabis use disorders onset during adolescence (ages 13.0 to 17.9) by psychopathology domain during childhood (ages 8.0 to 12.9).

Gender was evaluated as a moderator of the association between, separately, internalizing and externalizing psychopathology during childhood in predicting CUD onset during adolescence. Gender was similarly evaluated as a moderator of the association between internalizing and externalizing psychopathology during childhood and adolescence and CUD onset in the early adult onset

PREDICTORS OF CANNABIS USE DISORDER ONSET

547

Table 3 Period Prevalence Rates of Internalizing and Externalizing Psychopathology and Associations With Initial CUD Onset During Early Adulthood Hazard ratio [CI95]

Period prevalence % [CI95]

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Predictor Adolescent psychopathology Internalizing domain Mood disorders Anxiety disorders Bulimia nervosa Externalizing domain Disruptive behavior disorders Other substance use disorders Either internalizing or externalizing Childhood psychopathology Internalizing domain Mood disorders Anxiety disorders Bulimia nervosa Externalizing domain Disruptive behavior disorders Other substance use disorders Either internalizing or externalizing

No CUD (n ⫽ 643)

CUD (n ⫽ 84)

Likelihood ratio ␹

23.3 [20.2, 26.7] 19.6 [16.7, 22.8] 6.2 [4.5, 8.2] 1.0 [0.4, 1.9] 7.2 [5.4, 9.3] 3.5 [2.3, 5.1] 4.2 [2.9, 6.0] 28.0 [24.7, 31.5]

31.8 [22.5, 42.2] 25.7 [17.2, 35.8] 8.6 [3.9, 15.9] 1.7 [0.2, 6.1] 24.2 [15.9, 34.1] 13.0 [6.9, 21.2] 15.6 [8.9, 24.3] 44.0 [33.6, 54.7]

10.9 [8.6, 13.4] 6.0 [4.3, 7.9] 6.7 [5.0, 8.8] 0.0 2.7 [1.7, 4.1] 2.4 [1.4, 3.7] 0.3 [0.1, 1.0] 13.0 [10.6, 15.7]

12.1 [6.2, 20.2] 4.3 [1.2, 10.1] 8.6 [3.9, 15.9] 0.0 6.0 [2.2, 12.5] 6.0 [2.2, 12.5] 0.0 15.5 [8.8, 24.2]

2

p-value

Unadjusted

Adjusted

2.73 1.62 0.66 0.33 19.68 11.14 13.34 8.47

.099 .203 .416 .567 ⬍.001 ⬍.001 ⬍.001 .004

1.48 [0.95, 2.30] 1.38 [0.88, 2.18] 1.37 [0.68, 2.78] NA 3.60 [2.20, 5.88] 3.47 [1.91, 6.31] 3.68 [2.04, 6.62] 1.91 [1.22, 2.98]

1.36 [0.86, 2.16] 1.22 [0.75, 1.97] 1.30 [0.63, 2.68] NA 3.15 [1.87, 5.30] 2.32 [1.22, 4.43] 2.89 [1.53, 5.47] 1.98 [1.24, 3.16]

0.11 0.41 0.39 NA 2.20 2.83 0.51 0.38

.743 .521 .533

1.11 [0.61, 2.04] 0.72 [0.29, 1.79] 1.27 [0.63, 2.58] NA 2.16 [0.99, 4.71] 2.42 [1.11, 5.29] NA 1.20 [0.70, 2.07]

1.08 [0.59, 1.99] 0.56 [0.21, 1.49] 1.46 [0.68, 3.16] NA 1.97 [0.90, 4.33] 2.23 [1.01, 4.96] NA 1.18 [0.68, 2.05]

.138 .093 .474 .539

Note. CUD ⫽ cannabis use disorder; CI95 ⫽ 95% confidence interval; NA ⫽ not analyzed as predictors because of insufficient period prevalence rates. A value of 0.0 indicates the absence of diagnosed cases within the referenced developmental period. Statistically significant hazard ratios are bolded. Adjusted analyses control for variables in Table 1 and other psychopathology within the same developmental period. Childhood was defined as ages 8.0 to 12.9, adolescence was defined as ages 13.0 to 17.9, and early adulthood was defined as ages 18.0 to 30.0.

group. Gender did not emerge as a significant moderator in any analysis (all ps ⱖ .150).

Discussion We investigated the developmental pathways that constitute a heighted risk for future CUDs during two adjoining developmental periods when the risk for cannabis initiation and initial CUD onset is greatest: adolescence and early adulthood. Our research emphasized the predictive utility of indicants of internalizing and externalizing liabilities that, when observed, are thought to not only signal a heighted risk for future CUDs but also suggest common

Figure 2. Cumulative hazard functions for cannabis use disorders onset during early adulthood (ages 18.0 to 29.9) by psychopathology domain during adolescence (ages 13.0 to 17.9).

core processes that underlie sets of related disorders (Kessler et al., 2011; Krueger, 1999; Vanyukov et al., 2012; Zucker et al., 2008). Findings from the present research indicate that the externalizing domain is as relevant to the prediction of index CUD episode rates within units of time during early adulthood as it is for episodes that emerge during adolescence. Furthermore, externalizing psychopathology remained a robust prospective predictor of time to index CUD episode onset after demographic variables, participant family characteristics, and internalizing psychopathology were statistically controlled. Consistent with earlier studies which demonstrated that effects due to early life risk factors (distal influences) are often attenuated or eliminated once later life risk factors (proximal influences) are taken into account (Blanco et al., 2014; van den Bree & Pickworth, 2005), we found in adjusted analyses that adolescent but not childhood externalizing psychopathology were significant and moderately strong predictors of time to CUD onset within the early adulthood developmental period. We found no evidence that internalizing psychopathology was a significant risk or protective factor related to the development of CUDs during adolescence or early adulthood. Rather, our observations are consistent with other findings (Tarter et al., 2008) that internalizing features are unrelated to CUD onset risk through age 30 regardless of the developmental period within which the index episode occurs. We similarly found no evidence of gender moderation in any of the primary analyses, which suggests that the internalizing and externalizing risk factors operate similarly for male and female participants. Homotypic comorbidity between the externalizing disorders is substantial (Farmer et al., 2013b; Krueger & Markon, 2006; Lahey et al., 2008), and can be attributed to a shared liability that is

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

548

FARMER ET AL.

manifested as impulsivity or behavioral disinhibition (Beauchaine & McNulty, 2013; Iacono, Malone, & McGue, 2008). Common core processes likely include those associated with deficiencies in the functioning of mesolimbic dopamine networks (Beauchaine & McNulty, 2013). Low basal dopamine activity is associated with an increased risk for use and abuse of dopamine agonist drugs (Laine et al., 2001; Martin-Soelch et al., 2001), with cannabis being among those that facilitates dopamine release (Bossong et al., 2009). When viewed from this context, cannabis abuse could be regarded as an “attribute” or manifestation of the externalizing liability (Vanyukov et al., 2012) that, in turn, enhances or amplifies existing externalizing propensities, partially as a result of cannabis’ compromising effects on dopamine-related functions over time (Koob & Le Moal, 2008). Although the present study has several strengths, there are also limitations that suggest caution in the interpretation of study findings. Study strengths include OADPs longitudinal study design. Longitudinal studies allow for more accurate investigations of temporal relations between various psychiatric antecedents and later CUDs. Childhood and early adolescent diagnostic data in the present research, however, were based on retrospective assessments collected at T1. Retrospective assessments are known to introduce recall-related biases that generally favor the underreporting of psychopathology (Moffitt et al., 2010). The low base rates of certain disorders also precluded an examination of disorderspecific predictive association with CUD onset. This study was also limited by its emphasis on predictors of index CUD episodes during adolescent and early adult developmental periods. To our knowledge, internalizing and externalizing psychopathology have not been evaluated as predictors of index CUD episodes with onsets after age 30. With respect to alcohol use disorders, genetic pathways associated with adolescent and early adult onset of alcohol dependence disorder appear to differ from alcohol dependence episodes that occur after age 30 (Kendler, Gardner, & Dick, 2011). The extent to which the present findings generalize to cases with CUD onsets after age 30 is uncertain. Finally, this study did not consider disorder severity or duration in the prediction models. These features of psychiatric disorders might further contribute to the prediction of CUD onsets within the developmental periods examined.

Future Directions The common liability model (e.g., Vanyukov et al., 2012) suggests that environmental risk factors or underlying mechanisms are largely overlapping among highly comorbid substance use disorders and other psychiatric disorders. Indicators of the externalizing liability emerged as robust proximal predictors of CUD onset during adolescence and early adulthood in the present research. From a prevention perspective, findings presented here suggest that altering malleable environmental factors that contribute to or maintain externalizing tendencies (Bayer et al., 2011; DeaterDeckard, Dodge, Bates, & Pettit, 1998; Zucker et al., 2008) may substantially reduce the risk of subsequent CUDs. The externalizing liability, however, is a complex trait that encompasses a broad range of psychiatric disorders, personality features, and behavioral tendencies (Farmer et al., 2009; Krueger et al., 2002; Lahey et al., 2008) that are, in turn, predicted by common sets of environmental and biological risk factors (Kilpatrick et al., 2000; Shelton et al.,

2007). Future research might seek to establish which mechanisms or core processes associated with the externalizing liability are most strongly associated with the progression of cannabis initiation to CUDs, and from time-limited CUD episodes to more chronic or enduring forms of CUD. Although findings from the present research are generally supportive of the externalizing liability as a moderate risk factor, not all instances of CUD were explained by externalizing psychopathology. The significant proportion of disorder-free individuals in earlier developmental periods who subsequently developed CUDs suggests that psychopathology or externalizing liability theoretical frameworks, as organizing or explanatory models of the progression of cannabis experimentation to CUD, have limited predictive utility for identifying at-risk individuals. Additional work is clearly needed on the identification of mechanisms involved in the development of CUDs among those without significant externalizing histories. The identification of subgroups of persons with CUD histories may, in turn, suggest tailored interventions that better address the needs and characteristics of subgroup members. Because of the apparent limited explanatory power of internalizing and externalizing liability models for predicting CUD onset, multiple developmental pathways are suggested, and comprehensive models of CUD development must therefore consider a broader range of variables than were investigated here. Variables that have been shown to be associated with an elevated risk for cannabis use or CUD are parent (Andrews, Hops, & Duncan, 1997), sibling (Brook, Zhang, Koppel, & Brook, 2008), and peer (Creemers et al., 2010) cannabis use, as well as lower community cohesiveness (Lin, Witten, Casswell, & You, 2012). In addition to social context, personal factors such as deficits in social skills are known risk factors for early onset cannabis use (Griffith-Lendering et al., 2011a; Hampson, Tildesley, Andrews, Luyckx, & Mroczek, 2010). Explanatory models and interventions that take into account the social environment and the individual’s ability to effectively interact with its members may therefore be especially beneficial for informing theory and reducing future risk.

References Achenbach, T. M. (1966). The classification of children’s psychiatric symptoms: A factor-analytic study. Psychological Monographs, 80, (Whole No. 615). http://dx.doi.org/10.1037/h0093906 Agrawal, A., & Lynskey, M. T. (2007). Does gender contribute to heterogeneity in criteria for cannabis abuse and dependence? Results from the national epidemiological survey on alcohol and related conditions. Drug and Alcohol Dependence, 88, 300 –307. http://dx.doi.org/10.1016/j .drugalcdep.2006.10.003 Agrawal, A., Lynskey, M. T., Bucholz, K. K., Madden, P. A., & Heath, A. C. (2007). Correlates of cannabis initiation in a longitudinal sample of young women: The importance of peer influences. Preventive Medicine, 45, 31–34. http://dx.doi.org/10.1016/j.ypmed.2007.04.012 Andrews, J. A., Hops, H., & Duncan, S. C. (1997). Adolescent modeling of parent substance use: The moderating effect of the relationship with the parent. Journal of Family Psychology, 11, 259 –270. http://dx.doi .org/10.1037/0893-3200.11.3.259 Bayer, J. K., Ukoumunne, O. C., Lucas, N., Wake, M., Scalzo, K., & Nicholson, J. M. (2011). Risk factors for childhood mental health symptoms: National longitudinal study of Australian children. Pediatrics, 128, e865– e879. http://dx.doi.org/10.1542/peds.2011-0491 Beauchaine, T. P., & McNulty, T. (2013). Comorbidities and continuities as ontogenic processes: Toward a developmental spectrum model of

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

PREDICTORS OF CANNABIS USE DISORDER ONSET externalizing psychopathology. Development and Psychopathology, 25, 1505–1528. http://dx.doi.org/10.1017/S0954579413000746 Blanco, C., Rafful, C., Wall, M. M., Ridenour, T. A., Wang, S., & Kendler, K. S. (2014). Towards a comprehensive developmental model of cannabis use disorders. Addiction, 109, 284 –294. http://dx.doi.org/10.1111/ add.12382 Boden, J. M., Fergusson, D. M., & Horwood, L. J. (2006). Illicit drug use and dependence in a New Zealand birth cohort. Australian and New Zealand Journal of Psychiatry, 40, 156 –163. http://dx.doi.org/10.1080/ j.1440-1614.2006.01763.x Bossong, M. G., van Berckel, B. N., Boellaard, R., Zuurman, L., Schuit, R. C., Windhorst, A. D., . . . Kahn, R. S. (2009). ⌬ 9-tetrahydrocannabinol induces dopamine release in the human striatum. Neuropsychopharmacology, 34, 759 –766. http://dx.doi.org/10.1038/npp.2008.138 Brook, J. S., Lee, J. Y., Finch, S. J., Koppel, J., & Brook, D. W. (2011). Psychosocial factors related to cannabis use disorders. Substance Abuse, 32, 242–251. http://dx.doi.org/10.1080/08897077.2011.605696 Brook, J. S., Zhang, C., Koppel, J., & Brook, D. W. (2008). Pathways from earlier marijuana use in the familial and non-familial environments to self-marijuana use in the fourth decade of life. The American Journal on Addictions, 17, 497–503. http://dx.doi.org/10.1080/10550490802408373 Buckner, J. D., Schmidt, N. B., Lang, A. R., Small, J. W., Schlauch, R. C., & Lewinsohn, P. M. (2008). Specificity of social anxiety disorder as a risk factor for alcohol and cannabis dependence. Journal of Psychiatric Research, 42, 230 –239. http://dx.doi.org/10.1016/j.jpsychires.2007.01 .002 Callas, P. W., Pastides, H., & Hosmer, D. W. (1998). Empirical comparisons of proportional hazards, poisson, and logistic regression modeling of occupational cohort data. American Journal of Industrial Medicine, 33, 33– 47. http://dx.doi.org/10.1002/(SICI)1097-0274(199801)33: 1⬍33::AID-AJIM5⬎3.0.CO;2-X Chambers, W. J., Puig-Antich, J., Hirsch, M., Paez, P., Ambrosini, P. J., Tabrizi, M. A., & Davies, M. (1985). The assessment of affective disorders in children and adolescents by semistructured interview. Testretest reliability of the schedule for affective disorders and schizophrenia for school-age children, present episode version. Archives of General Psychiatry, 42, 696 –702. http://dx.doi.org/10.1001/archpsyc.1985 .01790300064008 Chen, K., & Kandel, D. B. (1995). The natural history of drug use from adolescence to the mid-thirties in a general population sample. American Journal of Public Health, 85, 41– 47. http://dx.doi.org/10.2105/AJPH.85 .1.41 Colder, C. R., Scalco, M., Trucco, E. M., Read, J. P., Lengua, L. J., Wieczorek, W. F., & Hawk, L. W., Jr. (2013). Prospective associations of internalizing and externalizing problems and their co-occurrence with early adolescent substance use. Journal of Abnormal Child Psychology, 41, 667– 677. http://dx.doi.org/10.1007/s10802-012-9701-0 Costello, E. J., Erkanli, A., Federman, E., & Angold, A. (1999). Development of psychiatric comorbidity with substance abuse in adolescents: Effects of timing and sex. Journal of Clinical Child Psychology, 28, 298 –311. http://dx.doi.org/10.1207/S15374424jccp280302 Creemers, H. E., Dijkstra, J. K., Vollebergh, W. A., Ormel, J., Verhulst, F. C., & Huizink, A. C. (2010). Predicting life-time and regular cannabis use during adolescence; the roles of temperament and peer substance use: The TRAILS study. Addiction, 105, 699 –708. http://dx.doi.org/ 10.1111/j.1360-0443.2009.02819.x Deater-Deckard, K., Dodge, K. A., Bates, J. E., & Pettit, G. S. (1998). Multiple risk factors in the development of externalizing behavior problems: Group and individual differences. Development and Psychopathology, 10, 469 – 493. http://dx.doi.org/10.1017/S0954579498001709 Degenhardt, L., Coffey, C., Romaniuk, H., Swift, W., Carlin, J. B., Hall, W. D., & Patton, G. C. (2013). The persistence of the association between adolescent cannabis use and common mental disorders into

549

young adulthood. Addiction, 108, 124 –133. http://dx.doi.org/10.1111/j .1360-0443.2012.04015.x Farmer, R. F., Kosty, D. B., Seeley, J. R., Duncan, S. C., Lynskey, M. T., Rohde, P., . . . Lewinsohn, P. M. (2015). Natural course of cannabis use disorders. Psychological Medicine, 45, 63–72. http://dx.doi.org/10.1017/ S003329171400107X Farmer, R. F., Kosty, D. B., Seeley, J. R., Olino, T. M., & Lewinsohn, P. M. (2013a). Aggregation of lifetime Axis I psychiatric disorders through age 30: Incidence, predictors, and associated psychosocial outcomes. Journal of Abnormal Psychology, 122, 573–586. http://dx.doi .org/10.1037/a0031429 Farmer, R. F., Seeley, J. R., Kosty, D. B., & Lewinsohn, P. M. (2009). Refinements in the hierarchical structure of externalizing psychiatric disorders: Patterns of lifetime liability from mid-adolescence through early adulthood. Journal of Abnormal Psychology, 118, 699 –710. http:// dx.doi.org/10.1037/a0017205 Farmer, R. F., Seeley, J. R., Kosty, D. B., Olino, T. M., & Lewinsohn, P. M. (2013b). Hierarchical organization of axis I psychiatric disorder comorbidity through age 30. Comprehensive Psychiatry, 54, 523–532. http://dx.doi.org/10.1016/j.comppsych.2012.12.007 Fergusson, D. M., & Horwood, L. J. (2000). Does cannabis use encourage other forms of illicit drug use? Addiction, 95, 505–520. http://dx.doi.org/ 10.1046/j.1360-0443.2000.9545053.x Fergusson, D. M., Horwood, L. J., & Lynskey, M. T. (1994). Structure of DSM–III–R criteria for disruptive childhood behaviors: Confirmatory factor models. Journal of the American Academy of Child & Adolescent Psychiatry, 33, 1145–1157. http://dx.doi.org/10.1097/00004583199410000-00010 Fergusson, D. M., Horwood, L. J., & Ridder, E. M. (2007). Conduct and attentional problems in childhood and adolescence and later substance use, abuse and dependence: Results of a 25-year longitudinal study. Drug and Alcohol Dependence, 88(Suppl. 1), S14 –S26. http://dx.doi .org/10.1016/j.drugalcdep.2006.12.011 First, M. B., Spitzer, R. L., Gibbon, M., & Williams, J. B. W. (1994). Structured clinical interview for Axis I DSM–IV disorders, non-patient edition. New York, NY: Biometrics Research Department. Grant, B. F., Stinson, F. S., Dawson, D. A., Chou, S. P., Dufour, M. C., Compton, W., . . . Kaplan, K. (2004). Prevalence and co-occurrence of substance use disorders and independent mood and anxiety disorders: Results from the National Epidemiologic Survey on Alcohol and Related Conditions. Archives of General Psychiatry, 61, 807– 816. http://dx.doi .org/10.1001/archpsyc.61.8.807 Griffith-Lendering, M. F., Huijbregts, S. C., Huizink, A. C., Ormel, H., Verhulst, F. C., Vollebergh, W. A., & Swaab, H. (2011a). Social skills as precursors of cannabis use in young adolescents: A TRAILS study. Journal of Clinical Child and Adolescent Psychology, 40, 706 –714. http://dx.doi.org/10.1080/15374416.2011.597085 Griffith-Lendering, M. F. H., Huijbregts, S. C. J., Mooijaart, A., Vollebergh, W. A. M., & Swaab, H. (2011b). Cannabis use and development of externalizing and internalizing behaviour problems in early adolescence: A TRAILS study. Drug and Alcohol Dependence, 116, 11–17. http://dx.doi.org/10.1016/j.drugalcdep.2010.11.024 Hampson, S. E., Tildesley, E., Andrews, J. A., Luyckx, K., & Mroczek, D. K. (2010). The relation of change in hostility and sociability during childhood to substance use in mid adolescence. Journal of Research in Personality, 44, 103–114. http://dx.doi.org/10.1016/j.jrp.2009.12.006 Hayatbakhsh, M. R., McGee, T. R., Bor, W., Najman, J. M., Jamrozik, K., & Mamun, A. A. (2008). Child and adolescent externalizing behavior and cannabis use disorders in early adulthood: An Australian prospective birth cohort study. Addictive Behaviors, 33, 422– 438. http://dx.doi.org/ 10.1016/j.addbeh.2007.10.004 Henry, B., Feehan, M., McGee, R., Stanton, W., Moffitt, T. E., & Silva, P. (1993). The importance of conduct problems and depressive symptoms

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

550

FARMER ET AL.

in predicting adolescent substance use. Journal of Abnormal Child Psychology, 21, 469 – 480. http://dx.doi.org/10.1007/BF00916314 Iacono, W. G., Malone, S. M., & McGue, M. (2008). Behavioral disinhibition and the development of early-onset addiction: Common and specific influences. Annual Review of Clinical Psychology, 4, 325–348. http://dx.doi.org/10.1146/annurev.clinpsy.4.022007.141157 Johnston, L. D., O’Malley, P. M., Bachman, J. G., & Schulenberg, J. E. (2013). Monitoring the Future national survey results on drug use (1975–2012): Volume I. Secondary school students. Ann Arbor: Institute for Social Research, University of Michigan. Kandel, D. B., & Chen, K. (2000). Types of marijuana users by longitudinal course. Journal of Studies on Alcohol and Drugs, 61, 367–378. Kandel, D., & Yamaguchi, K. (1993). From beer to crack: Developmental patterns of drug involvement. American Journal of Public Health, 83, 851– 855. http://dx.doi.org/10.2105/AJPH.83.6.851 Keller, M. B., Lavori, P. W., Friedman, B., Nielsen, E., Endicott, J., McDonald-Scott, P., & Andreasen, N. C. (1987). The Longitudinal Interval Follow-up Evaluation. A comprehensive method for assessing outcome in prospective longitudinal studies. Archives of General Psychiatry, 44, 540 –548. http://dx.doi.org/10.1001/archpsyc.1987 .01800180050009 Kendler, K. S., Gardner, C., & Dick, D. M. (2011). Predicting alcohol consumption in adolescence from alcohol-specific and general externalizing genetic risk factors, key environmental exposures and their interaction. Psychological Medicine, 41, 1507–1516. http://dx.doi.org/ 10.1017/S003329171000190X Kessler, R. C., Ormel, J., Petukhova, M., McLaughlin, K. A., Green, J. G., Russo, L. J., . . . Ustün, T. B. (2011). Development of lifetime comorbidity in the World Health Organization world mental health surveys. Archives of General Psychiatry, 68, 90 –100. http://dx.doi.org/10.1001/ archgenpsychiatry.2010.180 Kilpatrick, D. G., Acierno, R., Saunders, B., Resnick, H. S., Best, C. L., & Schnurr, P. P. (2000). Risk factors for adolescent substance abuse and dependence: Data from a national sample. Journal of Consulting and Clinical Psychology, 68, 19 –30. http://dx.doi.org/10.1037/0022-006X .68.1.19 King, S. M., Iacono, W. G., & McGue, M. (2004). Childhood externalizing and internalizing psychopathology in the prediction of early substance use. Addiction, 99, 1548 –1559. http://dx.doi.org/10.1111/j.1360-0443 .2004.00893.x Koob, G. F., & Le Moal, M. (2008). Addiction and the brain antireward system. Annual Review of Psychology, 59, 29 –53. http://dx.doi.org/ 10.1146/annurev.psych.59.103006.093548 Krueger, R. F. (1999). The structure of common mental disorders. Archives of General Psychiatry, 56, 921–926. http://dx.doi.org/10.1001/archpsyc .56.10.921 Krueger, R. F., Hicks, B. M., Patrick, C. J., Carlson, S. R., Iacono, W. G., & McGue, M. (2002). Etiologic connections among substance dependence, antisocial behavior, and personality: Modeling the externalizing spectrum. Journal of Abnormal Psychology, 111, 411– 424. http://dx.doi .org/10.1037/0021-843X.111.3.411 Krueger, R. F., & Markon, K. E. (2006). Reinterpreting comorbidity: A model-based approach to understanding and classifying psychopathology. Annual Review of Clinical Psychology, 2, 111–133. http://dx.doi .org/10.1146/annurev.clinpsy.2.022305.095213 Lahey, B. B., Rathouz, P. J., Van Hulle, C., Urbano, R. C., Krueger, R. F., Applegate, B., . . . Waldman, I. D. (2008). Testing structural models of DSM–IV symptoms of common forms of child and adolescent psychopathology. Journal of Abnormal Child Psychology, 36, 187–206. http:// dx.doi.org/10.1007/s10802-007-9169-5 Laine, T. P. J., Ahonen, A., Räsänen, P., & Tiihonen, J. (2001). Dopamine transporter density and novelty seeking among alcoholics. Journal of Addictive Diseases, 20, 95–100. http://dx.doi.org/10.1300/ J069v20n04_08

Lewinsohn, P. M., Hops, H., Roberts, R. E., Seeley, J. R., & Andrews, J. A. (1993). Adolescent psychopathology: I. Prevalence and incidence of depression and other DSM–III–R disorders in high school students. Journal of Abnormal Psychology, 102, 133–144. http://dx.doi.org/ 10.1037/0021-843X.102.1.133 Lin, E. Y., Witten, K., Casswell, S., & You, R. Q. (2012). Neighbourhood matters: Perceptions of neighbourhood cohesiveness and associations with alcohol, cannabis and tobacco use. Drug and Alcohol Review, 31, 402– 412. http://dx.doi.org/10.1111/j.1465-3362.2011.00385.x Martin-Soelch, C., Leenders, K. L., Chevalley, A. F., Missimer, J., Künig, G., Magyar, S., . . . Schultz, W. (2001). Reward mechanisms in the brain and their role in dependence: Evidence from neurophysiological and neuroimaging studies. Brain Research Reviews, 36, 139 –149. http://dx .doi.org/10.1016/S0165-0173(01)00089-3 Maslowsky, J., Schulenberg, J. E., & Zucker, R. A. (2014). Influence of conduct problems and depressive symptomatology on adolescent substance use: Developmentally proximal versus distal effects. Developmental Psychology, 50, 1179 –1189. http://dx.doi.org/10.1037/a0035085 Miller-Johnson, S., Lochman, J. E., Coie, J. D., Terry, R., & Hyman, C. (1998). Comorbidity of conduct and depressive problems at sixth grade: Substance use outcomes across adolescence. Journal of Abnormal Child Psychology, 26, 221–232. http://dx.doi.org/10.1023/A:1022676302865 Moffitt, T. E., Caspi, A., Taylor, A., Kokaua, J., Milne, B. J., Polanczyk, G., & Poulton, R. (2010). How common are common mental disorders? Evidence that lifetime prevalence rates are doubled by prospective versus retrospective ascertainment. Psychological Medicine, 40, 899 – 909. http://dx.doi.org/10.1017/S0033291709991036 Orvaschel, H., Puig-Antich, J., Chambers, W., Tabrizi, M. A., & Johnson, R. (1982). Retrospective assessment of prepubertal major depression with the Kiddie-SADS-e. Journal of the American Academy of Child Psychiatry, 21, 392–397. http://dx.doi.org/10.1016/S0002-7138(09) 60944-4 Paton, S., Kessler, R., & Kandel, D. (1977). Depressive mood and adolescent illicit drug use: A longitudinal analysis. The Journal of Genetic Psychology: Research and Theory on Human Development, 131, 267– 289. http://dx.doi.org/10.1080/00221325.1977.10533299 Perkonigg, A., Goodwin, R. D., Fiedler, A., Behrendt, S., Beesdo, K., Lieb, R., & Wittchen, H.-U. (2008). The natural course of cannabis use, abuse and dependence during the first decades of life. Addiction, 103, 439 – 449. http://dx.doi.org/10.1111/j.1360-0443.2007.02064.x Seeley, J. R., Kosty, D. B., Farmer, R. F., & Lewinsohn, P. M. (2011). The modeling of internalizing disorders on the basis of patterns of lifetime comorbidity: Associations with psychosocial functioning and psychiatric disorders among first-degree relatives. Journal of Abnormal Psychology, 120, 308 –321. http://dx.doi.org/10.1037/a0022621 Shelton, K., Lifford, K., Fowler, T., Rice, F., Neale, M., Harold, G., . . . van den Bree, M. (2007). The association between conduct problems and the initiation and progression of marijuana use during adolescence: A genetic analysis across time. Behavior Genetics, 37, 314 –325. http://dx .doi.org/10.1007/s10519-006-9124-1 Tarter, R. E., Kirisci, L., Ridenour, T., & Vanyukov, M. (2008). Prediction of cannabis use disorder between childhood and young adulthood using the Child Behavior Checklist. Journal of Psychopathology and Behavioral Assessment, 30, 272–278. http://dx.doi.org/10.1007/s10862-0089083-3 van den Bree, M. B., & Pickworth, W. B. (2005). Risk factors predicting changes in marijuana involvement in teenagers. Archives of General Psychiatry, 62, 311–319. http://dx.doi.org/10.1001/archpsyc.62.3.311 van der Net, J. B., Janssens, A. C., Eijkemans, M. J., Kastelein, J. J., Sijbrands, E. J., & Steyerberg, E. W. (2008). Cox proportional hazards models have more statistical power than logistic regression models in cross-sectional genetic association studies. European Journal of Human Genetics, 16, 1111–1116. http://dx.doi.org/10.1038/ejhg.2008.59

PREDICTORS OF CANNABIS USE DISORDER ONSET

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Vanyukov, M. M., Tarter, R. E., Kirillova, G. P., Kirisci, L., Reynolds, M. D., Kreek, M. J., . . . Ridenour, T. A. (2012). Common liability to addiction and “gateway hypothesis”: Theoretical, empirical and evolutionary perspective. Drug and Alcohol Dependence, 123(Suppl. 1), S3–S17. http://dx.doi.org/10.1016/j.drugalcdep.2011.12.018 Verweij, K. J., Zietsch, B. P., Lynskey, M. T., Medland, S. E., Neale, M. C., Martin, N. G., . . . Vink, J. M. (2010). Genetic and environmental influences on cannabis use initiation and problematic use: A metaanalysis of twin studies. Addiction, 105, 417– 430. http://dx.doi.org/ 10.1111/j.1360-0443.2009.02831.x Wittchen, H. U., Fröhlich, C., Behrendt, S., Günther, A., Rehm, J., Zimmermann, P., . . . Perkonigg, A. (2007). Cannabis use and cannabis use disorders and their relationship to mental disorders: A 10-year

551

prospective-longitudinal community study in adolescents. Drug and Alcohol Dependence, 88(Suppl. 1), S60 –S70. http://dx.doi.org/10.1016/ j.drugalcdep.2006.12.013 Wolter, K. M. (1985). Introduction to variance estimation. New York, NY: Springer-Verlag. Zucker, R. A., Donovan, J. E., Masten, A. S., Mattson, M. E., & Moss, H. B. (2008). Early developmental processes and the continuity of risk for underage drinking and problem drinking. Pediatrics, 121(Suppl. 4), S252–S272. http://dx.doi.org/10.1542/peds.2007-2243B

Received September 7, 2014 Revision received December 22, 2014 Accepted December 24, 2014 䡲

Members of Underrepresented Groups: Reviewers for Journal Manuscripts Wanted If you are interested in reviewing manuscripts for APA journals, the APA Publications and Communications Board would like to invite your participation. Manuscript reviewers are vital to the publications process. As a reviewer, you would gain valuable experience in publishing. The P&C Board is particularly interested in encouraging members of underrepresented groups to participate more in this process. If you are interested in reviewing manuscripts, please write APA Journals at [email protected]. Please note the following important points: • To be selected as a reviewer, you must have published articles in peer-reviewed journals. The experience of publishing provides a reviewer with the basis for preparing a thorough, objective review. • To be selected, it is critical to be a regular reader of the five to six empirical journals that are most central to the area or journal for which you would like to review. Current knowledge of recently published research provides a reviewer with the knowledge base to evaluate a new submission within the context of existing research. • To select the appropriate reviewers for each manuscript, the editor needs detailed information. Please include with your letter your vita. In the letter, please identify which APA journal(s) you are interested in, and describe your area of expertise. Be as specific as possible. For example, “social psychology” is not sufficient—you would need to specify “social cognition” or “attitude change” as well. • Reviewing a manuscript takes time (1– 4 hours per manuscript reviewed). If you are selected to review a manuscript, be prepared to invest the necessary time to evaluate the manuscript thoroughly. APA now has an online video course that provides guidance in reviewing manuscripts. To learn more about the course and to access the video, visit http://www.apa.org/pubs/authors/reviewmanuscript-ce-video.aspx.