Predicting the transition from frequent cannabis use to cannabis ...

Drug and Alcohol Dependence 133 (2013) 352–359

Contents lists available at ScienceDirect

Drug and Alcohol Dependence journal homepage: www.elsevier.com/locate/drugalcdep

Predicting the transition from frequent cannabis use to cannabis dependence: A three-year prospective study夽 Peggy van der Pol a,∗ , Nienke Liebregts b , Ron de Graaf a , Dirk J. Korf b , Wim van den Brink c , Margriet van Laar a a

Trimbos Institute, Netherlands Institute of Mental Health and Addiction, PO Box 725, 3500 AS Utrecht, The Netherlands Bonger Institute of Criminology, Law Faculty, University of Amsterdam, PO Box 1030, 1000 BA Amsterdam, The Netherlands c Department of Psychiatry, Academic Medical Centre, University of Amsterdam, PO Box 22660, 1100 DD Amsterdam, The Netherlands b

a r t i c l e

i n f o

Article history: Received 6 February 2013 Received in revised form 27 May 2013 Accepted 14 June 2013 Available online 22 July 2013 Keywords: Cannabis dependence Frequent cannabis use Risk factors Longitudinal Exposure

a b s t r a c t Background: Frequent cannabis users are at high risk of dependence, still most (near) daily users are not dependent. It is unknown why some frequent users develop dependence, whereas others do not. This study aims to identify predictors of first-incidence DSM-IV cannabis dependence in frequent cannabis users. Methods: A prospective cohort of frequent cannabis users (aged 18–30, n = 600) with baseline and two follow-up assessments (18 and 36 months) was used. Only participants without lifetime diagnosis of DSM-IV cannabis dependence at baseline (n = 269) were selected. Incidence of DSM-IV cannabis dependence was established using the Composite International Diagnostic Interview version 3.0. Variables assessed as potential predictors of the development of cannabis dependence included sociodemographic factors, cannabis use variables (e.g., motives, consumption habits, cannabis exposure), vulnerability factors (e.g., childhood adversity, family history of mental disorders or substance use problems, personality, mental disorders), and stress factors (e.g., life events, social support). Results: Three-year cumulative incidence of cannabis dependence was 37.2% (95% CI = 30.7–43.8%). Independent predictors of the first incidence of cannabis dependence included: living alone, coping motives for cannabis use, number and type of recent negative life events (major financial problems), and number and type of cannabis use disorder symptoms (impaired control over use). Cannabis exposure variables and stable vulnerability factors did not independently predict first incidence of cannabis dependence. Conclusions: In a high risk population of young adult frequent cannabis users, current problems are more important predictors of first incidence cannabis dependence than the level and type of cannabis exposure and stable vulnerability factors. © 2013 Elsevier Ireland Ltd. All rights reserved.

1. Introduction Cannabis is one of the most commonly used illicit substances, yet only a minority of users becomes dependent (Lopez-Quintero et al., 2010). Although frequent cannabis users are at a much higher risk to become dependent (Coffey et al., 2003; Grant and Pickering, 1998; Noack et al., 2011), ‘only’ 20–50% of the (near) daily users are dependent (Coffey et al., 2002; EMCDDA, 2009). Until now, it is not understood why some frequent users develop dependence, whereas others do not. More knowledge is required concerning the predictors of cannabis dependence among frequent cannabis users,

夽 Supplementary material can be found by accessing the online version of this paper at http://dx.doi.org and by entering doi:. . .. ∗ Corresponding author. Tel.: +31 30 2959 290; fax: +31 30 297 1111. E-mail addresses: [email protected], [email protected] (P. van der Pol). 0376-8716/$ – see front matter © 2013 Elsevier Ireland Ltd. All rights reserved. http://dx.doi.org/10.1016/j.drugalcdep.2013.06.009

to target the prevention and treatment specifically at frequent users at a high risk of dependence. The distinction between non-dependent and dependent frequent use is important, because dependent users, by definition, experience significant psychosocial impairments related to reduced control over their cannabis use, whereas other frequent users do not develop such drug-related problems (Budney and Moore, 2002; Looby and Earleywine, 2007; Temple et al., 2011). This distinction is especially important since cannabis use per se is regarded to be relatively harmless (Nutt et al., 2010; Van Amsterdam et al., 2010). Furthermore, there are indications that the risk of comorbid disorders is substantially lower (often not exceeding the general population level) among non-dependent compared to dependent frequent cannabis users (Swift et al., 2001; Van der Pol et al., 2013b). Studies on predictors of cannabis dependence are predominantly performed in the general population, which generally

P. van der Pol et al. / Drug and Alcohol Dependence 133 (2013) 352–359

includes a minority of current frequent users (Bruns and Geist, 1984; Chen et al., 2005; EMCDDA, 2011; Grant and Pickering, 1998; Lopez-Quintero et al., 2010; McGee et al., 2000; Swift et al., 2008; Von Sydow et al., 2002). Therefore, these studies are limited in predicting the transition from non-dependent to dependent frequent use of cannabis and subsequently fail to detect the risk factors in this high risk population of cannabis users. Despite this limitation, some of these general population studies have identified predictors of dependence that may also apply to the transition from non-dependent frequent cannabis use to cannabis dependence. These include: younger in age, being male, deprived socioeconomic status, poor financial situation (Chen et al., 2005; Hayatbakhsh et al., 2009; Lopez-Quintero et al., 2010; Poulton et al., 1997; Von Sydow et al., 2002), early onset of cannabis use, use and dependence of other substances including nicotine, alcohol, and cocaine (Chen et al., 2005; Lopez-Quintero et al., 2010; Von Sydow et al., 2002), and the presence of cannabis dependence symptoms without meeting criteria for a DSM-IV diagnosis of cannabis dependence (Degenhardt et al., 2008). Some of these symptoms are predictive of the persistence of cannabis use (Perkonigg et al., 2008). However, they have not been assessed as predictors of incident cannabis dependence. Other risk factors for cannabis dependence that were identified in general population studies include a family history of substance use disorders, parents’ marital problems, early life events, impulsivity, aggressive or delinquent behaviour, and the presence of externalising disorders (Bruns and Geist, 1984; Coffey et al., 2002; Fergusson et al., 2007; Florez-Salamanca et al., 2013; Hayatbakhsh et al., 2006, 2009; Hyman and Sinha, 2009; Lopez-Quintero et al., 2010; Perkonigg et al., 2008; Pingault et al., 2012; Swift et al., 2008; Von Sydow et al., 2002; Wittchen et al., 2007). The role of internalising disorders is still unclear (Moore et al., 2007). Yet, there is increasing evidence for an association between mood/anxiety disorders and heavy or problematic cannabis use (Degenhardt et al., 2003). The onset of dependence has also been predicted specifically by the presence of social phobia (Buckner et al., 2008), whereas recurrence of dependence has been predicted by the presence of mood disorders (Florez-Salamanca et al., 2013). Therefore, it may be hypothesised that these disorders may be important for the transition from non-dependent frequent cannabis use to cannabis dependence (Florez-Salamanca et al., 2013). Together, these factors can be grouped as ‘vulnerability’ factors. Furthermore, current stressful factors may increase the risk of both cannabis use (Hyman and Sinha, 2009) and cannabis dependence (Harris, 2000). However, information on stress factors associated with the transition from non-dependent cannabis use to cannabis dependence is almost non-existent. Examples of acute stress factors include recent negative life events, and lack of social support (Bruns and Geist, 1984; Hyman and Sinha, 2009; Windle and Wiesner, 2004). Remarkably little is known about the intensity of cannabis exposure as a risk factor for the development of cannabis dependence. Even though prospective studies have shown that the number of days when using drugs is a strong predictor of dependence, most studies implicitly assume that a higher number days when using drugs equates to higher cannabis exposure. Only few studies also assessed the quantity of cannabis use such as the number of ‘joints’ per day (e.g., Chen et al., 1997; Grant and Pickering, 1998; Looby and Earleywine, 2007; Moss et al., 2012), while quantity has been shown to discriminate dependent and non-dependent users independently from the number of using days (Walden and Earleywine, 2008; Zeisser et al., 2011). The quantity and quality (potency) of the cannabis consumed may be of particular importance among (near) daily high risk users, since their cannabis consumption may vary widely from a single joint of low potent cannabis to numerous high potency joints per day (Temple et al., 2011; Zeisser et al., 2011).

353

Finally, potential predictors of the transition from nondependent frequent cannabis use to cannabis dependence are suggested from cross-sectional studies, but these have not yet been investigated prospectively. These factors include setting (of use), such as solitary use versus use with others and daytime versus evening cannabis use (Noack et al., 2009; Van der Pol et al., 2013b). In addition, cannabis users with coping-oriented reasons to use cannabis may be at a higher risk to develop dependence than other users (Bonn-Miller and Zvolensky, 2009; Bujarski et al., 2012; Chabrol et al., 2005; Fox et al., 2011; Johnson et al., 2010; Simons et al., 2005). To our knowledge, the only prospective study among frequent cannabis users investigating predictors of dependence is an exploratory study following long term (heavy) cannabis users for one year (Swift et al., 2000). However, most cannabis users in this sample were already cannabis dependent at baseline, and therefore little information is given about predictors of first incidence of cannabis dependence. It concluded that quantity (number of cones in water pipes per day) and severity of cannabis dependence at baseline were the main predictors of cannabis use and cannabis dependence (persistence) at follow-up. This study aims to identify predictors of the transition from non-dependent frequent cannabis use to cannabis dependence. A broad range of predictors identified in previous prospective studies (sociodemographic variables, substance use characteristics, and vulnerability factors) is investigated in a large sample of nondependent frequent cannabis users aged 18–30 who were followed for a period of three years. In addition, novel risk factors suggested by cross-sectional studies are considered, including stress indicators (e.g., recent life events, lack of social support), cannabis consumption variables (e.g., pattern of use, habits, motives) and symptoms of abuse and dependence (without full-blown diagnoses). The predictive value of the abuse and dependence symptoms will be assessed individually, but also cumulatively, because the presence of these symptoms may represent a subclinical underlying continuum of severity (Compton et al., 2009). 2. Methods Data were derived from the CanDep study, investigating a prospective cohort of 600 frequent cannabis users. The reader is referred to a previous publication (Van der Pol et al., 2011). In summary, 600 Dutch frequent cannabis users (≥3 days per week for ≥12 months) between 18 and 30 years were recruited from cannabis outlets (coffee shops) and through chain referral. Baseline measures (T0, September 2008–April 2009) included a face-to-face interview to assess the presence of DSMIV cannabis dependence and a range of potential predictors of cannabis dependence. In addition, there were two face-to-face follow-up interviews after 18 months (T1, March–November 2010) and after 36 months (T2, September 2011–March 2012) assessing DSM-IV cannabis dependence. To allow the investigation of first incidence of cannabis dependence, only participants who had never met DSM-IV criteria for cannabis dependence at baseline (n = 269) were included in the current study. Response rates relative to T0 were 80.3% (n = 216) at T1 and 74.0% at T2 (n = 199). 2.1. Outcome Participants who fulfilled a diagnosis of DSM-IV cannabis dependence for the first time between T0 and T2 were identified as incident cases. DSM-IV cannabis dependence was measured with the Composite International Diagnostic Interview (CIDI) version 3.0 (Kessler and Ustun, 2004), and required the presence of three or more of seven symptoms within the 12-month period since the previous interview (without requiring the presence of all symptoms at the same time). It should be noted that the CIDI includes a withdrawal symptom, which is not included in the DSV-IV manual. The Severity of Dependence Scale (SDS, 5 items on a 0–3 scale) was assessed at all interviews to address measurement error in the dependence diagnosis (Section 2.3.1; Gossop et al., 1995; Van der Pol et al., 2013a). 2.2. Baseline predictors of cannabis dependence 2.2.1. Sociodemographic variables. The following sociodemographic variables were included: age, sex, ethnicity (Western vs. non-Western), educational level (primary or lower secondary, higher secondary, higher professional or university),

354


employment situation (employed, unemployed or unable to work, student), and living arrangement (living alone vs. living with others). 2.2.2. Cannabis and other substance use. Baseline cannabis use was assessed in detail and included four categories of variables. First, variables related to cannabis use disorders (CUD; cannabis abuse and dependence) were assessed with the CIDI. These included the presence of 12-month cannabis abuse at baseline, the lifetime presence of individual CUD symptoms, and the total number of lifetime positively endorsed CUD symptoms at baseline. Second, besides age of onset of first cannabis use, cannabis variables related to (a typical cannabis using day in) the last four weeks. Exposure variables included: number of cannabis using days in the past four weeks, preference for herbal cannabis (yes, no), estimated potency (high, middle, low), number of joints per using day, dosage of cannabis per joint (measured as 1/number of joints from 1 g). In addition, total cannabis exposure (in grams) in the last four weeks was measured by multiplying the number of cannabis using days by the number of joints per day and the dose per joint. Third, the setting of cannabis use was defined by location (non-selective, home, coffee shop), solitary use or use with others, using at night or (also) at daytime, number of hours ‘high’ per using day, continual smoking a joint (rarely to never pause, sometimes to always pause), average intensity of perceived intoxication during cannabis consumption measured on a visual analogue scale ranging from 1 (light buzz) to 10 (very ‘high’). Fourth, motives for cannabis use were assessed with the Marijuana Motives Measure (MMM; Simons et al., 2005). The MMM is a 25-item self-report questionnaire with five internally consistent subscales of five items each (subscale range 5–25) measuring enhancement, conformity, expansion, coping, and social motives. Problematic use of other substances was also assessed. This included 12-month alcohol use and related problems measured with the Alcohol Use Disorders Identification Test (AUDIT), a screening instrument with excellent psychometric qualities (Bohn et al., 1995). Also, the Heaviness of Smoking Index (HSI) was used to measure smoking and related problems in the past four weeks (Burling and Burling, 2003). Finally, any 12-month use of ecstasy, cocaine, or amphetamines was assessed. 2.2.3. Vulnerability factors. We identified four types of vulnerability factors. First, family history of mental problems (anxiety, depression, psychosis) or substance use problems (alcohol, cannabis, other drugs). Second, two types of distress during childhood were evaluated (Hovens et al., 2010; Janssen et al., 2004; Konings et al., 2012; Zeisser et al., 2011): (i) the presence of one or more (dichotomous) of the following childhood family adversities before the age of 16 was assessed: parental divorce, parental death, and not being brought up by both parents; and (ii) a history of any child abuse before the age of 16 years: emotional, psychological, physical, and sexual abuse. Because these abuse experiences may not always occur in isolation and may have an additive effect, the cumulative number of these four types of child abuse was used as the main predictor (Van der Pol et al., 2013b). Third, the personality characteristic impulsivity was measured with the Dutch version of the Barratt Impulsiveness Scale (BIS-15). This is a 15-item reliable measure of impulsivity, which consists of three subscales: motor, attentional, and non-planning impulsivity (Spinella, 2007). Finally, the personal history of psychiatric disorders was assessed with the CIDI and was categorised in three major diagnostic categories of lifetime DSM-IV diagnoses: any mood disorder (major depression, bipolar disorder, dysthymia); any anxiety disorder (social phobia, panic disorder, agoraphobia, generalised anxiety disorder); any externalising disorder (childhood ADHD, conduct disorder). Lifetime psychotic symptoms such as hallucinations and delusions (yes, no) were measured with a selection of six items of the psychosis section of CIDI 1.1 (Van der Pol et al., 2011). 2.2.4. Stress factors: life events and social support. The sum of negative and positive life events were included as two count variables. Major negative recent life events that occurred in the previous 12 months were assessed with Brugha’s List of Threatening Experiences, and are itemised in Table 4 (Brugha et al., 1985). Positive life events included: establishing close friendships; commencing a steady relationship; having a baby; graduation or completion of a course/training; starting a new job; substantial financial gain. Finally, perceived lack of social support from network members was measured as two dichotomised variables: ‘heart unburdening below expectations’ and ‘practical help below expectations’. These were each based on two questions: the degree to which participants could count on their social network, and whether this support was below their expectations. 2.3. Statistical analysis To determine any bivariate association between aforementioned baseline predictors and the dependent variable, i.e., first incidence of cannabis dependence at some time during the three-year follow-up, odds ratios (ORs) and 95% confidence intervals (95% CI) were calculated using logistic regression analyses. Subsequently, to identify the unique contribution of predictors of cannabis dependence onset and to reduce the risk of missing unique predictors, all bivariate predictors with p ≤ 0.10 were entered into a multivariable logistic regression (Hosmer and Lemeshow, 2013). Yet, only variables significant at the 5% level in the multivariate model were considered unique predictors. Finally, McFadden’s pseudo R2 is reported for the multivariable model.

CUD symptoms are strong predictors of first incidence cannabis dependence (Compton et al., 2009; Degenhardt et al., 2008) and this may overshadow the effect of all other predictors that may be important for clinical practice (without diagnostic assessment). Therefore, the multivariable model was first tested with the CUD variables and then without the CUD variables in the model. Moreover, besides the sum of CUD symptoms as predictor of dependence, each criterion was assessed separately, both as a bivariate and as independent multivariable predictor. For all other aggregated variables (e.g., number of childhood adversities, number of recent life events) that significantly predicted first incidence cannabis dependence, the same procedure was applied, i.e., assessing each original item in the bivariate model and the combination of items in multivariable logistic regression. 2.3.1. Measurement error. Measurement errors are inherent to survey data. When several indicators of a concept are available, a latent class Markov model enables estimation of misclassification probabilities of ‘latent’ variables represented by the measured indicators (Van de Pol and Mannan, 2002). Using the CIDI dependence diagnosis (yes/no) and the total SDS score at T0, T1 and T2 as indicators, three latent variables of cannabis dependence were constructed with PanMark 3.2 software (Humphreys et al., 1994; Van de Pol et al., 2007). Both T0 dependent and non-dependent participants (with complete data N = 460) were included, because at least three waves with all possible transition options are needed. 2.3.2. Missing data. While missing data of predictor variables was limited (only four variables had a maximum of eight missing values), 26.0% of participants were lost during follow-up. Although this is not disconcerting in a population of young adult frequent substance users, performing complete case analyses may introduce bias. Imputing missing values with multiple imputation by chained equations is preferable when data are missing at random, which means that any systematic differences between missing and observed values can be explained by differences in observed data (Sterne et al., 2009). Therefore, imputation was preferred, as predictors associated with attrition included: non-Western ethnicity, lower education, being employed, being ‘high’ longer, using higher doses per joint, use at coffee shops, less illicit substance use, less (motor) impulsivity, and more negative life events. To impute missing data, 20 complete datasets were created, using 50 imputation cycles for each dataset. The analyses described above were conducted on the imputed datasets using ‘mi estimate’ in Stata 12.0, which combines results of analyses on the 20 completed datasets and adjusts standard errors using Rubin’s rules (Rubin, 1987). A detailed description of the imputation models and a discussion of the comparison with complete case analyses can be found in Supplementary Material.1

3. Results There were 73 incident cases of cannabis dependence within the three-year follow-up (n = 47 T0–T1, n = 26 T1–T2), which is 36.7% of the 199 complete cases. Similarly, the imputed (mean) incidence rate was 37.2% (95% CI = 30.7, 43.8). Bivariate logistic regression on the imputed data was performed to identify potential predictors of incident cannabis dependence. Table 1 shows that none of the sociodemographic variables predicted incident cannabis dependence, except living alone. Baseline cannabis use variables that predicted incident cannabis dependence included a diagnosis of 12-month abuse, number of lifetime CUD symptoms, (also) using at daytime, continual smoking, and coping motives for cannabis consumption. Remarkably, cannabis exposure variables were not predictive of the transition from nondependent to dependent cannabis use. Lifetime and current mental disorders were frequent in both family members and participants but these disorders were not predictive for the development of first incident cannabis dependence, and neither was family childhood adversity. However, the number of childhood adversities and the number of recent negative life events were associated with the development of dependence and so were two of the three impulsivity subscales. Finally, the lack of ‘heart unburdening’ social support was a significant predictor (p ≤ 0.10). All predictor variables that were significant at the 10% level were included in a multivariable logistic regression to identify independent predictors of the incidence of cannabis dependence. Living

1 Supplementary material can be found by accessing the online version of this paper at http://dx.doi.org and by entering doi:. . ..


355

Table 1 Bivariate predictors of three-year incidence of first cannabis dependence in young adult frequent cannabis users (N = 269). Predictors of first cannabis dependence

No (%/mean)

Yes (%/mean)

OR

95% CI

Socio-demographics

Mean age Male Non-Western (vs. Western) Education primary/lower secondary Higher secondary Higher professional Employment Employed Student Unemployed/unable to work Living alone (vs. with others)

21.9 75.7 31.1 27.9 38.9 33.2

21.8 78.0 32.4 33.9 38.1 27.9

0.99 1.13 1.06 1.00 0.81 0.69

[0.91, 1.08] [0.60, 2.16] [0.57, 1.98] [1.00, 1.00] [0.40, 1.62] [0.34, 1.40]

44.3 43.3 12.4 14.0

47.1 40.8 12.0 30.3

1.00 0.89 0.91 2.66**

[1.00, 1.00] [0.50, 1.57] [0.38, 2.16] [1.38, 5.14]

Last year abuse Mean number lifetime CUD symptoms Mean age of onset first use Mean last month days of use Preference for herbal cannabis Preferred potency Low Middle High Mean number of joints per day Mean dosage Mean total last month exposure Location Non selective Home settings Coffee shops Solitary use Using (also) at daytime Hours per day ‘high’ Continual smoking Mean level of intoxication Mean enhancement Mean conformity Mean expansion Mean coping Mean social Mean alcohol (AUDIT) score Mean smoking score (HSI) Any 12-month illicit substance use

21.8 2.48 14.2 21.8 54.3

36.1 3.65 14.6 23.1 65.2

2.03* 1.45*** 1.09 1.04 1.57

[1.10, 3.75] [1.21, 1.75] [0.97, 1.23] [0.99, 1.08] [0.88, 2.83]

13.1 33.8 53.1 3.24 0.33 27.6

7.0 30.7 62.4 3.49 0.33 28.8

1.00 1.72 2.22 1.05 1.04 1.00

[1.00, 1.00] [0.61, 4.80] [0.81, 6.10] [0.94, 1.18] [0.11, 10.04] [0.99, 1.01]

18.8 62.8 18.5 14.4 36.5 5.42 35.5 6.15 14.87 5.30 8.13 6.53 10.91 7.97 1.27 27.6

18.3 50.0 31.7 17.5 50.3 5.52 52.9 6.36 14.36 5.48 9.06 8.18 11.79 7.88 1.32 30.3

1.00 0.82 1.77 1.25 1.76* 1.01 2.04* 1.08 0.97 1.19 1.06 1.18*** 1.06 1.00 1.02 1.14

[1.00, 1.00] [0.38, 1.76] [0.67, 4.67] [0.60, 2.63] [1.01, 3.07] [0.93, 1.09] [1.16, 3.58] [0.92, 1.27] [0.90, 1.04] [0.90, 1.58] [0.99, 1.13] [1.07, 1.30] [0.99, 1.14] [0.94, 1.05] [0.86, 1.22] [0.63, 2.07]

Mental disorders Substance use problems Family adversity Mean number of child abuse BIS Motor impulsivity Attentional impulsivity Nonplanning Any lifetime mood disorder Any lifetime anxiety disorder Any lifetime externalising disorder Any lifetime psychosis symptom

27.2 21.3 51.4 0.76

34.0 29.0 45.0 1.05

1.38 1.50 0.77 1.30*

[0.78, 2.45] [0.82, 2.76] [0.44, 1.36] [1.00, 1.67]

9.52 9.92 11.83 11.6 8.6 47.3 4.6

10.47 10.73 12.04 19.4 11.5 54.0 9.2

1.15** 1.12* 1.03 1.83 1.39 1.31 2.10

[1.04, 1.28] [1.01, 1.24] [0.93, 1.14] [0.87, 3.86] [0.55, 3.52] [0.78, 2.19] [0.70, 6.26]

Mean number positive events Mean number negative events Unburden heart Practical help

2.12 2.64 18.1 15.3

2.30 3.73 28.4 23.1

1.15 1.35*** 1.80+ 1.66

[0.90, 1.48] [1.16, 1.58] [0.93, 3.50] [0.84, 3.28]

Cannabis use Cannabis use disorder Cannabis exposure

Setting

Motives

Other substances

Vulnerability Parents Childhood Personality Personality

Comorbidity

Stress Recent life events Lack of social support + * ** ***

p < 0.10. p < 0.05. p < 0.01. p < 0.001.

alone, baseline number of lifetime CUD symptoms, and number of negative life events were unique predictors (Table 2: Model 11), with a pseudo R2 of 23%. In a separate model without the CUD variables as predictors (Table 2: Model 21), coping as a motive for cannabis use was also an independent predictor. In this model, the pseudo R2 was 17%. Table 3 shows that the dependence symptoms “using larger amounts or longer than intended” and “tolerance” were the most

frequently reported CUD symptoms. While most baseline lifetime CUD symptoms were bivariately associated with the incidence of dependence, “impaired control over use” was the strongest and only independent multivariable predictor. Finally, Table 4 shows the specific negative life events associated with the incidence of cannabis dependence in a post hoc analysis: the frequently occurring event “major financial crisis” was the only independent multivariable predictor.

356


Table 2 Independent predictors of incidence of first cannabis dependence in young adult frequent cannabis users (n = 269): multivariable model including (1) and excluding cannabis abuse and dependence variables (2). Model 1. Including CUD

Sociodemographic Living alone (vs. with others) CUD Last year abuse Mean number CUD symptoms Cannabis use Using (also) at daytime No pauses while smoking joint Mean coping motive Vulnerability Mean number of child abuse Mean motor impulsivity Mean attentional impulsivity Stress Mean number of negative recent life events Lack ‘heart unburden’ social support

Model 2. Excluding CUD

OR

95% CI

OR

95% CI

3.85**

[1.69, 8.77]

3.38**

[1.55, 7.36]

1.37 1.42**

[0.65, 2.86] [1.14, 1.79]

– –

– –

1.82+ 1.70 1.07

[0.92, 3.62] [0.85, 3.38] [0.94, 1.21]

1.79+ 1.84+ 1.13*

[0.94, 3.41] [0.96, 3.54] [1.01, 1.26]

0.97 1.04 1.06

[0.69, 1.36] [0.91, 1.20] [0.92, 1.22]

0.99 1.05 1.08

[0.72, 1.36] [0.92, 1.20] [0.95, 1.23]

1.43*** 1.35

[1.18, 1.73] [0.58, 3.14]

1.35*** 1.36

[1.14, 1.60] [0.61, 3.01]

Model 1: multivariable logistic model predicting incidence of cannabis dependence with all variables in the model (pseudo R2 = 23%). Model 2: excludes last year abuse and the number of dependence and abuse symptoms (pseudo R2 = 17%). –: not included in the model. CUD: cannabis use disorder. + p < 0.10. * p < 0.05. ** p < 0.01. *** p < 0.001.

Table 3 Lifetime cannabis abuse and dependence symptoms at baseline predicting incidence of first cannabis dependence among young adult frequent users (n = 269). Dependence incidence Abuse Role impairment Hazardous use Legal problems Social problems Dependence Tolerance Withdrawal Larger/longer Impaired control over use Much time spent Reduced activities Use despite problems

Yes %

No %

40.1 28.0 – 22.7 55.9 30.3 88.4 43.2 25.5 27.7 20.1

Bivariate OR *

25.4 32.0 – 13.8 41.5 17.6 63.1 17.0 13.9 13.1 8.1

1.97 0.83 – 1.83+ 1.79+ 2.03* 2.38** 3.72*** 2.11* 2.54** 1.50

95% CI

Multivariable OR

95% CI

[1.10, 3.54] [0.46, 1.50] – [0.90, 3.72] [0.96, 3.32] [1.06, 3.90] [1.24, 4.57] [1.88, 7.36] [1.06, 4.20] [1.25, 5.16] [0.77, 2.94]

1.38 0.90 – 1.31 1.44 1.39 1.76+ 2.95** 1.48 1.53 1.28

[0.69, 2.76] [0.46, 1.77] – [0.58, 2.96] [0.72, 2.87] [0.66, 2.91] [0.93, 3.36] [1.41, 6.15] [0.67, 3.27] [0.66, 3.54] [0.79, 2.05]

–: legal problems omitted due to the low prevalence (1.9% of total sample) which led to perfect prediction of dependence. + p < 0.10. * p < 0.05. ** p < 0.01. *** p < 0.001.

Table 4 Negative life events predicting incidence of first cannabis dependence among young adult frequent users (n = 269) (post hoc). Dependence incidence

Separation from someone important Broke off steady relationship Death of a loved-one Dropped out of school Sacked from job or became unemployed Major financial crisis Problems with police Moved house Serious illness or injury Serious illness or injury to close relative Serious illness or injury to friend + * ** ***

p < 0.10. p < 0.05. p < 0.01. p < 0.001.

Yes

No

Bivariate

Multivariable

%

%

OR

95% CI

OR

95% CI

44.6 48.4 39.2 28.6 36.6 37.1 16.0 48.4 17.7 32.1 24.5

27.4 42.9 27.1 21.0 28.0 17.0 12.4 37.0 9.1 28.9 13.9

2.14** 1.25 1.73+ 1.51 1.48 2.88*** 1.34 1.60 2.16+ 1.16 2.01*

[1.24, 3.68] [0.73, 2.13] [0.96, 3.13] [0.80, 2.83] [0.82, 2.67] [1.57, 5.29] [0.61, 2.97] [0.91, 2.80] [0.98, 4.75] [0.65, 2.09] [1.04, 3.91]

1.72+ 0.99 1.75 1.15 1.14 2.44** 1.12 1.38 1.58 0.92 1.52

[0.95, 3.11] [0.54, 1.82] [0.87, 3.51] [0.56, 2.37] [0.59, 2.22] [1.27, 4.71] [0.47, 2.67] [0.74, 2.55] [0.62, 3.99] [0.47, 1.79] [0.72, 3.18]


3.1. Measurement error Compared to the latent dependence diagnose, false positive rates of the CIDI dependence diagnose at T0, T1 and T2 were estimated at 12.6 (SD = 0.06), 10.6 (SD = 0.03) and 4.9% (SD = 0.02), respectively. False negative rates were 25.6 (SD = 0.05), 25.2 (SD = 0.05) and 24.8% (SD = 0.05), respectively. 4. Discussion 4.1. Findings To our knowledge, this study is the first prospective exploration of factors predicting the incidence of cannabis dependence in a high risk population of frequent cannabis users. Six hundred (near) daily cannabis users were recruited to reach an adequate sample of 269 frequent users without a lifetime diagnosis of cannabis dependence. The response rate after three years was 74%. During this period, approximately one in three frequent users developed a first diagnosis of cannabis dependence (37.2%, 95% CI = 30.7, 43.8), emphasising both the importance and the potential of prevention targeted specifically at high risk frequent users. Interestingly, most risk factors for the development of cannabis dependence from general population studies had no predictive value in our sample of frequent cannabis users, including most of the sociodemographic factors, the presence of internalising and externalising mental disorders, childhood family adversity, and family history of substance use or mental health problems. Moreover, while impulsivity and child abuse were bivariately associated, they did not survive in multivariable prediction models. This finding suggests that in a group of high risk frequent cannabis users other factors are involved in the development of cannabis dependence than in a general population sample of cannabis users. It may also indicate that predictors established in general population studies actually predict progression to frequent cannabis use rather than the development of cannabis dependence. This interpretation is supported by our baseline comparison of dependent and non-dependent frequent users from the CanDep cohort with a representative general population sample (Van der Pol et al., 2013b). While dependent and non-dependent users were similar in this study, this comparison showed that both groups of frequent cannabis users differed from the general population particularly on risk factors previously reported in prospective general population studies. Together with the finding that the incidence of cannabis dependence in general population studies is strongly associated with frequent cannabis use, this may indicate that predictors from general population studies are mainly related to the transition from incidental to frequent cannabis use and not to the transition from frequent cannabis use to cannabis dependence. It is also remarkable that neither total levels of cannabis exposure or several detailed exposure measures were predictive of the development of cannabis dependence among frequent cannabis users. Similarly, only weak associations between alcohol consumption and the diagnosis of alcohol dependence have been reported (Tuithof et al., 2013). However, cannabis use may be dynamic, while only recent cannabis exposure at baseline was investigated as a predictor in the current study. Therefore, in future studies, transitions in exposure should be added as predictors of cannabis dependence incidence. Notably, coping as the motive for cannabis use was among the less commonly investigated variables (i.e., not inspired by previous prospective general population studies) which emerged as a predictor of the transition from non-dependent to dependent frequent cannabis use. In addition, cannabis consumption patterns and habits such as daytime cannabis use and continual smoking were significant bivariate and borderline significant multivariable

357

predictors of cannabis dependence incidence. Together, this suggests that among frequent cannabis users cannabis consumption motives and habits are more important predictors of future dependence than the actual cannabis (THC) exposure level. The number of negative life events was also an independent predictor; the risk of dependence increased with 43% with every additional negative event. This implies that in a young adult population, acute stressful events are stronger predictors of dependence than the presence of relatively stable vulnerability factors. A major financial problem was the strongest predictor among all negative life events. Also, living alone was a strong predictor, which corresponds very well with general population studies indicating that not living with a partner is associated with (any drug) dependence (Grant et al., 2009). The role of parents and partners in the development of cannabis dependence will be analysed in depth on the basis of qualitative interviews among 47 CanDep participants in a future report of our research group. Finally, it is not surprising that presence of last year cannabis abuse and lifetime number of CUD symptoms at baseline were important predictors of the onset of full-blown cannabis dependence. Of all CUD symptoms, “impaired control over use” was the strongest predictor of dependence. However, it should be noted that cultural differences may be important. For example, legal problems due to cannabis use are rare in the Netherlands (1.9% in the current sample) and no OR could be calculated, but legal problems might be an important predictor in countries with more restrictive regulations (Perkonigg et al., 2008). Similarly, interviewers reported that hazardous use in this age group was mainly restricted to riding a bike under the influence of cannabis, a CUD criterion that was not associated with the risk of the future development of cannabis dependence in the current study. 4.2. Strengths and limitations The most important strengths of the current study are the large sample size, the impressive follow-up rate and the broad range of predictors that were assessed using adequate instruments. However, some limitations have also emerged. Primarily, those with a lifetime diagnosis of cannabis dependence were excluded to assess first incidence of dependence. However, the minimum age of the current sample was relatively high (18 years) considering the fact that cannabis dependence on average has been shown to emerge at age 18 (for males) (Wittchen et al., 2008). It can therefore not be excluded that those frequent users with a lifetime diagnosis of cannabis dependence at baseline (who were excluded in the current analyses) were more vulnerable and that the risk profile of these youngsters is different from the ones who developed cannabis dependence only after the age of 18, i.e., after the baseline assessment. Second, although power was sufficient to identify a series of unique predictors, some of the bivariate predictors did not survive the competition in the multivariable models either because a lack of power or due the overlapping variance. However, when backward selection was manually applied (although this is a data driven method not recommended after multiple imputation (Wood et al., 2008)), the results were very similar, suggesting that overlapping variance is the more plausible explanation. Finally, although a broad range of predictor variables were assessed, including CUD symptoms, the pseudo R2 was only 23% in the multivariable model. Other factors that were not considered in this study may be important in the prediction of cannabis dependence incidence in this high risk population. For example role transition has shown to be of great relevance in the development of alcohol dependence (Copeland et al., 2012). It should also be noted that the current study was mainly restricted to phenotypic predictors and that intermediate phenotypes/endophenotypes (e.g., Cousijn et al., 2012a, 2012b) and (epi)genetic factors (e.g., Agrawal et al., 2011;

358


Jutras-Aswad et al., 2012) were not included as potential predictors for the development of cannabis dependence. Nevertheless, 23% is significant considering that the model has to predict the onset of a complex multi-factorial disorder over a substantial period of time and includes baseline variables of which some may change during follow-up. However, it should be considered that inaccuracy or bias caused by self-report may partly explain the limited pseudo R2 . Specifically, the estimated 25% false negative rate using latent Markov modelling suggests that part of our sample should have been excluded and a similar proportion of incident cases were missed. The estimated proportion false positive was much smaller and the effect would be opposite. Altogether, misclassification may have diluted associations, resulting in underestimated or even missed associations. Unfortunately, analyses could not be adjusted for this measurement error, because baseline dependent users had to be included in the Markov model in order to perform the analyses. However, measurement error may also have been overestimated because only the CIDI diagnosis and the SDS score were available to construct the latent cannabis dependence variable, and the SDS has previously been reported to poorly correlate with cannabis dependence (Van der Pol et al., 2013a). Therefore, it is most likely that the current findings are a reasonable representation of the real world possibly with some underestimation of the real associations between predictors and outcomes. 4.3. Conclusion In summary, our findings imply that young adult nondependent frequent cannabis users are at considerable risk (37%) to develop cannabis dependence in the near future, especially when they show the following risk factors: CUD symptoms, coping motives for their (frequent) use of cannabis, acute stressful negative life events and living alone. Unexpectedly, the actual level of cannabis exposure and the presence of stable vulnerability factors seem to be less significant. Role of funding source The CanDep study was financially supported by a grant of the Netherlands Organisation for Health Research and Development (NWO, grant number 31160009). The NWO had no further role in study design, in the collection, analysis and interpretation of data, in the writing of the report, or in the decision to submit the paper for publication. Contributors M. van Laar, W. van den Brink, D.J. Korf and R. de Graaf designed the study and wrote the protocol. N. Liebregts and P. van der Pol collected the data, and P. van der Pol managed the literature searches and summaries of previous related work, undertook the statistical analysis, and wrote the first draft of the manuscript. All authors contributed to and have approved the final manuscript. Conflict of interest All authors declare that they have no conflicts of interest. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.drugalcdep. 2013.06.009.

References Agrawal, A., Lynskey, M.T., Hinrichs, A., Grucza, R., Saccone, S.F., Krueger, R., Neuman, R., Howells, W., Fisher, S., Fox, L., Cloninger, R., Dick, D.M., Doheny, K.F., Edenberg, H.J., Goate, A.M., Hesselbrock, V., Johnson, E., Kramer, J., Kuperman, S., Nurnberger Jr., J.I., Pugh, E., Schuckit, M., Tischfield, J., Rice, J.P., Bucholz, K.K., Bierut, L.J., 2011. A genome-wide association study of DSM-IV cannabis dependence. Addict. Biol. 16, 514–518. Bohn, M.J., Babor, T.F., Kranzler, H.R., 1995. The Alcohol Use Disorders Identification Test (AUDIT): validation of a screening instrument for use in medical settings. J. Stud. Alcohol. 56, 423–432. Bonn-Miller, M.O., Zvolensky, M.J., 2009. An evaluation of the nature of marijuana use and its motives among young adult active users. Am. J. Addict. 18, 409–416. Brugha, T., Bebbington, P., Tennant, C., Hurry, J., 1985. The list of threatening experiences: a subset of 12 life event categories with considerable long-term contextual threat. Psychol. Med. 15, 189–194. Bruns, C., Geist, C.S., 1984. Stressful life events and drug use among adolescents. J. Human Stress 10, 135–139. 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. J. Psychiatr. Res. 42, 230–239. Budney, A.J., Moore, B.A., 2002. Development and consequences of cannabis dependence. J. Clin. Pharmacol. 42, 28S–33S. Bujarski, S.J., Norberg, M.M., Copeland, J., 2012. The association between distress tolerance and cannabis use-related problems: the mediating and moderating roles of coping motives and gender. Addict. Behav. 37, 1181–1184. Burling, A.S., Burling, T.A., 2003. A comparison of self-report measures of nicotine dependence among male drug/alcohol-dependent cigarette smokers. Nicotine Tob. Res. 5, 625–633. Chabrol, H., Duconge, E., Casas, C., Roura, C., Carey, K.B., 2005. Relations between cannabis use and dependence, motives for cannabis use and anxious, depressive and borderline symptomatology. Addict. Behav. 30, 829–840. Chen, C.Y., O‘Brien, M.S., Anthony, J.C., 2005. Who becomes cannabis dependent soon after onset of use? Epidemiological evidence from the United States: 2000–2001. Drug Alcohol Depend. 79, 11–22. Chen, K., Kandel, D.B., Davies, M., 1997. Relationships between frequency and quantity of marijuana use and last year proxy dependence among adolescents and adults in the United States. Drug Alcohol Depend. 46, 53–67. Coffey, C., Carlin, J.B., Degenhardt, L., Lynskey, M., Sanci, L., Patton, G.C., 2002. Cannabis dependence in young adults: an Australian population study. Addiction 97, 187–194. Coffey, C., Carlin, J.B., Lynskey, M., Li, N., Patton, G.C., 2003. Adolescent precursors of cannabis dependence: findings from the Victorian Adolescent Health Cohort Study. Br. J. Psychiatry 182, 330–336. Compton, W.M., Saha, T.D., Conway, K.P., Grant, B.F., 2009. The role of cannabis use within a dimensional approach to cannabis use disorders. Drug Alcohol Depend. 100, 221–227. Copeland, W.E., Angold, A., Shanahan, L., Dreyfuss, J., Dlamini, I., Costello, E.J., 2012. Predicting persistent alcohol problems: a prospective analysis from the Great Smoky Mountain Study. Psychol. Med. 42, 1925–1935. Cousijn, J., Goudriaan, A.E., Ridderinkhof, K.R., Van den Brink, W., Veltman, D.J., Wiers, R.W., 2012a. Approach-bias predicts development of cannabis problem severity in heavy cannabis users: results from a prospective FMRI study. PLoS One 7, e42394. Cousijn, J., Wiers, R.W., Ridderinkhof, K.R., Van den Brink, W., Veltman, D.J., Porrino, L.J., Goudriaan, A.E., 2012b. Individual differences in decision making and reward processing predict changes in cannabis use: a prospective functional magnetic resonance imaging study. Addict. Biol., http://dx.doi.org/10.1111/j.1369-1600.2012.00498.x. Degenhardt, L., Coffey, C., Carlin, J.B., Swift, W., Patton, G.C., 2008. Are diagnostic orphans at risk of developing cannabis abuse or dependence? Four-year followup of young adult cannabis users not meeting diagnostic criteria. Drug Alcohol Depend. 92, 86–90. Degenhardt, L., Hall, W., Lynskey, M., 2003. Exploring the association between cannabis use and depression. Addiction 98, 1493–1504. EMCDDA, 2009. Cannabis. Annual Report 2009: The State of the Drugs Problem in Europe. EMCDDA, Lisbon, pp. 38–47. EMCDDA, 2011. Cannabis. Annual Report 2011: The State of the Drugs Problem in Europe. Publications Office of the European Union, Luxembourg, pp. 40–48. 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 Alcohol Depend. 88 Suppl. 1, S14–S26. Florez-Salamanca, L., Secades-Villa, R., Budney, A.J., Garcia-Rodriguez, O., Wang, S., Blanco, C., 2013. Probability and predictors of cannabis use disorders relapse: results of the National Epidemiologic Survey on Alcohol and Related Conditions (NESARC). Drug Alcohol Depend., http://dx.doi.org/10.1016/j.drugalcdep. Fox, C.L., Towe, S.L., Stephens, R.S., Walker, D.D., Roffman, R.A., 2011. Motives for cannabis use in high-risk adolescent users. Psychol. Addict. Behav. 25, 492–500. Gossop, M., Darke, S., Griffiths, P., Hando, J., Powis, B., Hall, W., Strang, J., 1995. The Severity of Dependence Scale (SDS): psychometric properties of the SDS in English and Australian samples of heroin, cocaine and amphetamine users. Addiction 90, 607–614. Grant, B.F., Goldstein, R.B., Chou, S.P., Huang, B., Stinson, F.S., Dawson, D.A., Saha, T.D., Smith, S.M., Pulay, A.J., Pickering, R.P., Ruan, W.J., Compton, W.M., 2009. Sociodemographic and psychopathologic predictors of first incidence of DSM-IV

P. van der Pol et al. / Drug and Alcohol Dependence 133 (2013) 352–359 substance use, mood and anxiety disorders: results from the Wave 2 National Epidemiologic Survey on Alcohol and Related Conditions. Mol. Psychiatry 14, 1051–1066. Grant, B.F., Pickering, R., 1998. The relationship between cannabis use and DSM-IV cannabis abuse and dependence: results from the National Longitudinal Alcohol Epidemiologic Survey. J. Subst. Abuse 10, 255–264. Harris, T., 2000. Where Inner and Outer Worlds Meet: Psychosocial Research in the Tradition of George W. Brown. Routledge, London. Hayatbakhsh, M.R., Najman, J.M., Bor, W., O‘Callaghan, M.J., Williams, G.M., 2009. Multiple risk factor model predicting cannabis use and use disorders: a longitudinal study. Am. J. Drug Alcohol Abuse 35, 399–407. Hayatbakhsh, M.R., Najman, J.M., Jamrozik, K., Mamun, A.A., Alati, R., 2006. Do parents’ marital circumstances predict young adults’ DSM-IV cannabis use disorders? A prospective study. Addiction 101, 1778–1786. Hosmer, D.W., Lemeshow, S., 2013. Applied Logistic Regression. Wiley, New York, p. 95. Hovens, J.G., Wiersma, J.E., Giltay, E.J., van, O.P., Spinhoven, P., Penninx, B.W., Zitman, F.G., 2010. Childhood life events and childhood trauma in adult patients with depressive, anxiety and comorbid disorders vs. controls. Acta Psychiatr. Scand. 122, 66–74. Humphreys, K., Skinner, C.J., van de Pol, F., Langeheine, R., 1994. PANMARK: Version 2.2.6 by Frank van de Pol; Rolf Langeheine; Wil de Jong. J. Roy. Stat. Soc. Series C (Appl. Stat.) 43, 415–417. Hyman, S.M., Sinha, R., 2009. Stress-related factors in cannabis use and misuse: implications for prevention and treatment. J. Subst. Abuse Treat. 36, 400–413. Janssen, I., Krabbendam, L., Bak, M., Hanssen, M., Vollebergh, W., De, G.R., van, O.J., 2004. Childhood abuse as a risk factor for psychotic experiences. Acta Psychiatr. Scand. 109, 38–45. Johnson, K., Mullin, J.L., Marshall, E.C., Bonn-Miller, M.O., Zvolensky, M., 2010. Exploring the mediational role of coping motives for marijuana use in terms of the relation between anxiety sensitivity and marijuana dependence. Am. J. Addict. 19, 277–282. Jutras-Aswad, D., Jacobs, M.M., Yiannoulos, G., Roussos, P., Bitsios, P., Nomura, Y., Liu, X., Hurd, Y.L., 2012. Cannabis-dependence risk relates to synergism between neuroticism and proenkephalin SNPs associated with amygdala gene expression: case-control study. PLoS One 7, e39243. Kessler, R.C., Ustun, T.B., 2004. The World Mental Health (WMH) Survey Initiative Version of the World Health Organization (WHO) Composite International Diagnostic Interview (CIDI). Int. J. Methods Psychiatr. Res. 13, 93–121. Konings, M., Stefanis, N., Kuepper, R., De, G.R., ten, H.M., van, O.J., Bakoula, C., Henquet, C., 2012. Replication in two independent population-based samples that childhood maltreatment and cannabis use synergistically impact on psychosis risk. Psychol. Med. 42, 149–159. Looby, A., Earleywine, M., 2007. Negative consequences associated with dependence in daily cannabis users. Subst. Abuse Treat. Prev. Policy 2, 3. Lopez-Quintero, C., Cobos, J.P., Hasin, D.S., Okuda, M., Wang, S., Grant, B.F., Blanco, C., 2010. Probability and predictors of transition from first use to dependence on nicotine, alcohol, cannabis, and cocaine: results of the National Epidemiologic Survey on Alcohol and Related Conditions (NESARC). Drug Alcohol Depend. 115, 120–130. McGee, R., Williams, S., Poulton, R., Moffitt, T., 2000. A longitudinal study of cannabis use and mental health from adolescence to early adulthood. Addiction 95, 491–503. Moore, T.H., Zammit, S., Lingford-Hughes, A., Barnes, T.R., Jones, P.B., Burke, M., Lewis, G., 2007. Cannabis use and risk of psychotic or affective mental health outcomes: a systematic review. Lancet 370, 319–328. Moss, H.B., Chen, C.M., Yi, H.Y., 2012. Measures of substance consumption among substance users, DSM-IV abusers, and those with DSM-IV dependence disorders in a nationally representative sample. J. Stud. Alcohol Drugs 73, 820–828. Noack, R., Höfler, M., Gründler, R., Becker, A., Schulz, R., Paul, K., 2009. Konsummuster als kovariablen von DSM-IV cannabisabhängigkeit. Sucht: Zeitschrift für Wissenschaft und Praxis 55, 137–147. Noack, R., Hofler, M., Lueken, U., 2011. Cannabis use patterns and their association with DSM-IV cannabis dependence and gender. Eur. Addict. Res. 17, 321–328. Nutt, D.J., King, L.A., Phillips, L.D., 2010. Drug harms in the UK: a multicriteria decision analysis. Lancet 376, 1558–1565. 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. Pingault, J.B., Cote, S.M., Galera, C., Genolini, C., Falissard, B., Vitaro, F., Tremblay, R.E., 2012. Childhood trajectories of inattention, hyperactivity and

359

oppositional behaviors and prediction of substance abuse/dependence: a 15-year longitudinal population-based study. Mol. Psychiatry, http://dx.doi.org/10.1038/mp.2012.87. Poulton, R.G., Brooke, M., Moffitt, T.E., Stanton, W.R., Silva, P.A., 1997. Prevalence and correlates of cannabis use and dependence in young New Zealanders. N. Z. Med. J. 110, 68–70. Rubin, D., 1987. Multiple Imputation for Nonresponse in Surveys. Wiley, New York. Simons, J.S., Gaher, R.M., Correia, C.J., Hansen, C.L., Christopher, M.S., 2005. An affective-motivational model of marijuana and alcohol problems among college students. Psychol. Addict. Behav. 19, 326–334. Spinella, M.A.R.C., 2007. Normative data and a short form of the Barratt Impulsiveness Scale. Int. J. Neurosci. 117, 359–368. Sterne, J.A., White, I.R., Carlin, J.B., Spratt, M., Royston, P., Kenward, M.G., Wood, A.M., Carpenter, J.R., 2009. Multiple imputation for missing data in epidemiological and clinical research: potential and pitfalls. BMJ 338, b2393. Swift, W., Coffey, C., Carlin, J.B., Degenhardt, L., Patton, G.C., 2008. Adolescent cannabis users at 24 years: trajectories to regular weekly use and dependence in young adulthood. Addiction 103, 1361–1370. Swift, W., Hall, W., Copeland, J., 2000. One year follow-up of cannabis dependence among long-term users in Sydney, Australia. Drug Alcohol Depend. 59, 309–318. Swift, W., Hall, W., Teesson, M., 2001. Cannabis use and dependence among Australian adults: results from the National Survey of Mental Health and Wellbeing. Addiction 96, 737–748. Temple, E.C., Brown, R.F., Hine, D.W., 2011. The ‘grass ceiling’: limitations in the literature hinder our understanding of cannabis use and its consequences. Addiction 106, 238–244. Tuithof, M., Ten Have, M., Van den Brink, W., Vollebergh, W., De Graaf, R., 2013. The relationship between excessive alcohol consumption and alcohol use disorders according to DSM-IV and DSM-5. Alcohol. Clin. Exp. Res., Epub ahead of print. Van Amsterdam, J., Opperhuizen, A., Koeter, M., Van den Brink, W., 2010. Ranking the harm of alcohol, tobacco and illicit drugs for the individual and the population. Eur. Addict. Res. 16, 202–207. Van de Pol, F., Langeheine, R., De Jong, W., 2007. “PanMark Panel Analysis with Markov Chains, a Latent Class Analysis program”, 3.2 edn. Van de Pol, F., Mannan, H., 2002. Questions of a novice in latent Markov modelling. Methods of Psychological Research Online 7, 1–18. Van der Pol, P., Liebregts, N., De Graaf, R., Korf, D.J., Van den Brink, W., Van Laar, M., 2011. The Dutch Cannabis Dependence (CanDep) study on the course of frequent cannabis use and dependence: objectives, methods and sample characteristics. Int. J. Methods Psychiatr. Res. 20, 169–181. Van der Pol, P., Liebregts, N., De Graaf, R., Korf, D.J., Van den Brink, W., Van Laar, M., 2013a. Reliability and validity of the Severity of Dependence Scale for detecting cannabis dependence in frequent cannabis users. Int. J. Methods Psychiatr. Res., http://dx.doi.org/10.1002/mpr.1385. Van der Pol, P., Liebregts, N., De Graaf, R., Ten Have, M., Korf, D.J., Van den Brink, W., Van Laar, M., 2013b. Mental health differences between frequent cannabis users with and without cannabis dependence and the general population. Addiction, http://dx.doi.org/10.1111/add.12196. Von Sydow, K., Lieb, R., Pfister, H., Hofler, M., Wittchen, H.U., 2002. What predicts incident use of cannabis and progression to abuse and dependence? A 4-year prospective examination of risk factors in a community sample of adolescents and young adults. Drug Alcohol Depend. 68, 49–64. Walden, N., Earleywine, M., 2008. How high: quantity as a predictor of cannabisrelated problems. Harm Reduct. J. 5, 20. Windle, M., Wiesner, M., 2004. Trajectories of marijuana use from adolescence to young adulthood: predictors and outcomes. Dev. Psychopathol. 16, 1007–1027. Wittchen, H.U., Behrendt, S., Hofler, M., Perkonigg, A., Lieb, R., Buhringer, G., Beesdo, K., 2008. What are the high risk periods for incident substance use and transitions to abuse and dependence? Implications for early intervention and prevention. Int. J. Methods Psychiatr. Res. 17 Suppl. 1, S16–S29. Wittchen, H.U., Frohlich, C., Behrendt, S., Gunther, A., Rehm, J., Zimmermann, P., Lieb, R., Perkonigg, A., 2007. Cannabis use and cannabis use disorders and their relationship to mental disorders: a 10-year prospective-longitudinal community study in adolescents. Drug Alcohol Depend. 88 Suppl. 1, S60–S70. Wood, A.M., White, I.R., Royston, P., 2008. How should variable selection be performed with multiply imputed data? Stat. Med. 27, 3227–3246. Zeisser, C., Thompson, K., Stockwell, T., Duff, C., Chow, C., Vallance, K., Ivsins, A., Michelow, W., Marsh, D., Lucas, P., 2011. A ‘standard joint’? The role of quantity in predicting cannabis-related problems. Addict. Res. Theory 20, 82–92.