Assessing Cannabis Use: The Case of Gabby

Sejmjl308

TheCannabisUseProblemsIdentificationTestdevelopmentreliabilityconcurrentandpredictivevalidityamongadolescentsandadults.pdf

Home >Psychology homework help >Assessing Cannabis Use: The Case of Gabby

The Cannabis Use Problems Identification Test (CUPIT): development, reliability, concurrent and predictive validity among adolescents and adultsadd_2859 615..625

Jan Bashford1, Ross Flett1 & Jan Copeland2

School of Psychology, Massey University, Palmerston North, New Zealand1 and National Cannabis Prevention and Information Centre, University of New South Wales, Sydney, Australia2

ABSTRACT

Aims To describe the empirical construction and initial validation of the Cannabis Use Problems Identification Test (CUPIT), a brief self-report screening instrument for detection of currently and potentially problematic cannabis use. Design In a three-phase prospective design an item pool of candidate questions was generated from a literature review and extensive expert consultation. The CUPIT internal structure, cross-sectional and longitudinal psychometric properties were then systematically tested among heterogeneous past-year users. Participants Volunteer participants were 212 high-risk adolescents (n = 138) and adults (n = 74) aged 13–61 years from multiple community settings. Measurements The comprehensive assessment battery included several established measures of cannabis-related pathology for CUPIT validation, with DSM-IV/ICD-10 diagnoses of cannabis use disorders as criterion standard. Findings Sixteen items loading highly on two subscales derived from principal components analysis exhibited good to excellent test–retest (0.89–0.99) and internal consistency reliability (0.92, 0.83), and highly significant ability to discriminate diagnostic subgroups along the severity continuum (non-problematic, risky, problematic use). Twelve months later, baseline CUPIT scores demonstrated highly significant longitudinal predictive utility for respondents’ follow-up diagnostic group membership. Receiver operating characteristic (ROC) analysis identified a CUPIT score of 12 to be the optimal cut-point for maximizing sensitivity for both currently diagnosable cannabis use disorder and those at risk of meeting diagnostic criteria in the following 12 months. Conclusions The CUPIT is a brief cannabis screener that is reliable, valid and acceptable for use across diverse community settings and consumers of all ages. The CUPIT has clear potential to assist with achievement of public health goals to reduce cannabis-related harms in the community.

Keywords Cannabis problems, cannabis screening, psychometrics, public health.

Correspondence to: Jan Bashford, 33 Brooklands Road, Nelson, New Zealand 7010. E-mail: jbashford@clear.net.nz Submitted 9 May 2009; initial review completed 21 July 2009; final version accepted 4 November 2009

INTRODUCTION

Cannabis use and misuse are serious public health con- cerns world-wide. Current estimates are around 4% of the global population aged from 15 years [1], with much higher rates reported among adolescents [2]. A corresponding increase in cannabis treatment seeking internationally and the whole spectrum of attendant health and psychosocial problems has prompted calls for stronger investment in public health initiatives to address cannabis-related harms [1,3]. The magnitude and potential societal ramifications of cannabis-related harms provide a mandate for opportunistic and targeted screening and early intervention (SEI) among users [4].

Cannabis screening is a proactive step in identifying users who may be at risk of developing use-related problems [5]. Screening information assists efforts to avert or arrest progression of a diagnosable or incubat- ing disorder by intervening as early as possible in its tra- jectory to dependence and more serious and chronic harms [6]. Given that most users with cannabis-related problems neither access nor seek specialist treatment [7], the high contact rates, accessibility and other characteristics of community-based health and social services make them ideal sites for opportunistic SEI for cannabis problems [4,8,9]. However, lack of a brief empirically verified self-report cannabis screen suitable for use across diverse community settings has

METHODS AND TECHNIQUES doi:10.1111/j.1360-0443.2009.02859.x

hampered implementation of routine SEI for cannabis problems.

There are a number of limitations of using existing lengthy, complex drug screening tools in primary care settings [10]. In addition, such general measures often take a life-time focus, vary widely in content and target group and are not clinically oriented with complex scoring algorithms. Even the few, briefer generic drug screens available are either inefficient or otherwise inad- equate as quick and accurate screens specifically for detecting a broad spectrum of harmful and risky can- nabis consumption. It has been estimated that clinicians would detect more than 80% of drug users if they limited their initial screening questionnaires to marijuana [11].

A small assortment of cannabis-specific screens have been developed internationally. These include the Can- nabis Abuse Syndrome Screening Test (CASST) [12] and Cannabis Use Disorders Identification Test (CUDIT) [13] in New Zealand, the Marijuana Screening Inventory (MSI-X) [14–16] in the United States, the Cannabis Abuse Screening Test (CAST) [17] in France and the Problematic Marijuana Use test (PUM) [18] in Poland. While closer to the model required for opportunistic can- nabis screening, in different ways and to varying extents these tools were developed and tested (1) for case-finding, and not sensitive to risky (pre-dependent) cannabis use, (2) among clinical or otherwise restricted samples, (3) not in generalist settings or (4) for use among both adolescent and adult populations or constituent ethnic/ cultural subgroups. Each has further idiosyncratic limi- tations. Still in their early developmental stages, with either limited or no evidence of their validity for use among at-risk younger users in the general population, these screens require further validation before their wide- spread use in their original countries or other cultures can be recommended [19].

This report describes the empirical construction and initial validation of the Cannabis Use Problems Identifi- cation Test (CUPIT), a brief self-report screen developed to expedite detection of currently and potentially problem- atic cannabis use. Over a 4-year period, the CUPIT was developed and validated systematically in a three-phase design: (1) generation of a pool of candidate questions; (2) examination of the CUPIT’s internal structure and psychometric properties among a heterogeneous com- munity sample of at-risk adolescent and adult cannabis users; and (3) 12-month predictive validity.

METHODS

Item pool generation

Following comprehensive review of the cannabis litera- ture, 90 items sourced from existing interviews and

scales, or constructed specifically for this research, were eliminated or refined systematically in successive rounds of an open-ended iterative expert panel (n = 20) consul- tation methodology at a local, then international, level. Forty-three items were selected.

The CUPIT specifications included: (1) a broad spec- trum of indicators of cannabis-related pathology for case-finding and detecting pre-clinical/risky use; (2) cov- erage of important content domains and variables (con- sumption, dependence, psychological, health and social consequences) relevant in screening; (3) items reflecting Diagnostic and Statistical Manual/International Classifi- cation of Diseases (DSM/ICD) criteria for screen valida- tion; (4) adolescent problem areas not represented adequately in DSM/ICD; (5) scoring formats that maxi- mize discriminatory power for diagnostic subgroups along the risk continuum; and (6) simple, acceptable, universal/culturally adaptable questions with rapid scoring and interpretation procedures.

Candidate questions addressed were: reasons for use; cannabis consumption (days used past 12 months, 3 months, times used per day, hours stoned) and potency (products used, smoking method); dependence (toler- ance, withdrawal symptoms, loss of control, preoccupa- tion, activities given up); psychological (regretted using, felt paranoid/anxious) and health (general health, nausea, respiratory, energy/motivation, memory/ concentration) effects; and abuse/harmful use (hazard- ous use, injury, failing to meet obligations, others’ concern, interpersonal, work/school, social, financial and legal problems). Predominantly five-point Likert-type [20] response categories (frequency, intensity, degree) were employed.

Development and initial validation of the CUPIT

Participants

Participants were 212 cannabis users recruited voluntar- ily from drug treatment clinics (n = 36), adult justice (n = 7), juvenile justice and alternative education pro- grammes for at-risk adolescents (n = 48), secondary school students (n = 52), tertiary students and job- seekers (n = 31) and the general population (n = 36). Inclusion criteria required having used cannabis at least once in the previous year, aged 13+ years, and English- literate to year eight level. Exclusion criteria included illiteracy, intoxication, acute psychiatric or cognitive impairment. Participants were recruited via counsellor, nurse or officer invitation, on-campus fliers and ‘snow- ball sampling’ [21,22].

Procedure and measures

Ethical approval for this project was granted by the appro- priate Ministry of Health Regional Ethics Committee,

616 Jan Bashford et al.

New Zealand, and all relevant institutions. Clinician cer- tification of competence, recruitment personnel and parent/guardian signatures, or school Principal/Board of Trustees’ formal permission was obtained for participants under 16 years. All participants signed informed consent and provided contact information.

Participants first completed the CUPIT item pool alone. Clinical participants (n = 40) provided urine samples. Shortly thereafter [mean = 6.30 days, standard deviation (SD) = 1.63, range: 4–14], the first author conducted comprehensive assessment interviews. First, most participants (93%, n = 197) completed the CUPIT again. The interview schedule took a mean of 60.19 (SD = 6.52, range: 50–80) minutes. Only the standard- ized measures incorporated for validation purposes are reported.

Current (90-day) cannabis use was measured using the Time-line Follow-back (TLFB) [23] interview vali- dated for users from age 14 years [24–26]. The Syva EMIT urinalysis procedure verified clinical participants’ self-reported consumption [27,28]. As criterion stan- dard, 12-month DSM-IV [29] and ICD-10 [30] cannabis dependence and abuse/harmful use diagnoses were obtained from the Drug Use module of the computerized Composite International Diagnostic Interview (CIDI)- Auto (version 2.1) [31]. This automatically generates symptom and diagnostic classifications [32–35]. The five-item Severity of Dependence Scale (SDS) [36,37] measured the degree of psychological dependence, while the adolescent or adult Cannabis Problems Questionnaire (CPQ, CPQ-A) [38–40] assessed cannabis-induced psy- chological, physical and social problems over the past 6 months. Current psychological health was assessed using the Brief Symptom Inventory 18 (BSI 18) [41], a short- ened BSI [42].

On individual baseline anniversaries (mean = 366 days, range: 318–407) participants were contacted (n = 200) and re-assessed (n = 194) on all baseline mea- sures except the CUPIT. Participants lost to attrition (8%, n = 17) either failed to return calls or declined an inter- view (n = 6), were in prison (n = 3) or had left the country (n = 8). Interviews were conducted face to face (77%), by telephone (20%) or post/e-mail (3%) from remote locations.

Data analysis

Screened data were analysed using SPSS for Windows (version 12.0) [43]. Descriptive statistics are presented with means, medians or percentages. Test–retest relia- bility was estimated using Pearson’s correlations. Spearman’s correlations assessed agreement between self-reported cannabis consumption and laboratory data. The CUPIT factor structure was investigated using

principal component analysis (PCA), and the internal consistency of retained components estimated using Cronbach’s alpha [44].

CUPIT scores were obtained by summating respon- dents’ raw scores [45,46]. Construct (criterion, concur- rent, convergent) validity was examined by comparing scores with measures of the same (CIDI-Auto, SDS, CPQ/ CPQ-A, TLFB) or associated (BSI) phenomena, employing two-tailed Pearson’s r (sample) and independent t-tests (adolescent/adult comparisons).

Discriminative validity was examined by comparing CUPIT subscale scores of the three diagnostic groups (dependence, abuse/harmful use, non-problematic use) using one-way analysis of variance (ANOVA). Signifi- cantly different between-groups scores would indicate good discriminative validity. Longitudinal predictive validity of baseline scores for follow-up diagnostic group was assessed similarly. A receiver operating characteristic (ROC) curve was plotted to determine the CUPIT’s capac- ity for discriminating cases (a diagnosis) from non-cases (no diagnosis) at baseline (n = 211) and follow-up (n = 194). A larger value of the area under the curve (AUC) represents greater sensitivity (baseline) or predic- tive efficacy (follow-up). Chi square (c2) values were cal- culated to determine the cut-point providing the best balance between sensitivity (proportion of true positives) and specificity (proportion of true negatives) for identify- ing currently diagnosable and at-risk users.

RESULTS

Participants

Participants were 138 adolescents (<=18 years) and 74 adults (19+ years) with a median age of 16 years (SD = 9.08, range: 13–61). Fifty-six per cent were male. The majority were European/Pakeha New Zealanders (70%), a large minority Maori (25%) or mixed race Maori/Pakeha (5%). Adolescents reported using can- nabis on a median of 45.5 (adults 80) and 15 (adults 29) days (both ranges: 0–90; 0–30) over the past 90 and 30 days, respectively.

Baseline 12-month DSM-IV/ICD-10 diagnoses were 72% cannabis dependence, 19% cannabis abuse and 68% both diagnoses. Only 8% (n = 17) had no diagnosis: 12 adolescents were ‘diagnostic orphans’ [47,48] (met one or two dependence criteria only), while five reported no symptoms. No significant adolescent/adult differences emerged in number of symptoms elicited (mean 6.4, SD = 3.26, range: 0–13), SDS (adolescent mean 4.05, SD = 2.42, range: 0–12; adult mean 4.64, SD = 3.23, range: 0–14 from a possible 15) or core CPQ/CPQ-A (mean 11.76, both ranges: 0–23) scores. BSI scores sug- gested 20% of this sample met Derogatis’ definition of

The CUPIT development 617

‘caseness’ or positive risk for psychiatric problems that warranted intervention [41]. While only 17% had ever sought treatment for, only 15% thought they currently had, a cannabis problem.

Twelve months later, adolescents (n = 128) reported significantly increased consumption while adults’ (n = 66) use remained at baseline levels. Endorsing sig- nificantly more symptoms (mean 7.52, SD = 3.47, range: 0–13), larger sample proportions (76%) obtained a DSM- IV/ICD-10 dependence diagnosis, abuse (17%) and both diagnoses (74%). Scores were significantly higher on the SDS (mean 5.52, SD = 2.96, range: 0–13), CPQ (mean 13.65 versus 11.75 baseline), CPQ-A (mean 16.8 versus 11.76 baseline) and the BSI, suggesting that a larger proportion (23% versus 20% baseline) had psychia- tric problems needing treatment. Again, however, 86% did not desire any assistance to help them cut down/ quit.

Dimensionality of the CUPIT

Removal or transformation of dichotomously/ trichotomously scored items left 39 of suitable metric scaling. A series of PCA was conducted, alternating Varimax and Oblique rotation on a combination of 14 original and 25 transformed items, and then on 39 untransformed items.

Factor retention criteria included Cattell’s scree test [49], eigenvalues, number of variables with significant loadings on a factor and interpretability [45,50]. Because

successive variable-technique combinations produced similar solutions with largely uncorrelated factors (less than 10% variance overlap), the solution using the untransformed variables and submitting two factors to Varimax rotation, and yielding the simplest structure (no cross-loadings) and several high-loading variables on both factors, was selected (Table 1).

Sixteen items loading on two primary components after rotation explained 38.62% of the total variance. While most items loaded above that deemed ‘excellent’ (>0.71) or ‘very good’ (>0.63), all loaded well above the minimum for interpretive purposes (0.30) [45,46]. The first component had significant loadings of 10 items: five were consumption variables, with the remaining five sug- gesting ‘impaired control’ over use. The second compo- nent comprised six items reflecting consequences of, or ‘problems’ caused by, cannabis use. The component scales were thus labelled.

Psychometric properties of the CUPIT

Reliability

One-week test–retest reliability estimates of CUPIT items (0.89–0.99) were all in the ranges deemed excellent (>0.80) [45,52] (Table 1). Internal consistency reliability estimates and descriptive statistics appear in Table 2. Cronbach’s alphas ranged from 0.79 to 0.92, well in excess of the recommended benchmark (0.70) for exploratory research and suggesting proper sampling of

Table 1 Cannabis Use Problems Identification Test (CUPIT) subscale items after principal components analysis with orthogonal rotation: item loadings, eigenvalues, percentage of variance explained (n = 212) and test–retest reliabilities (n = 197).

Item Impaired control Problems

Test–retest reliability (r)a

On how many days have you used cannabis during the past 12 months? 0.789 0.98 On how many days have you used cannabis over the past 3 months? 0.758 0.99 How many times would you use cannabis on a typical day when you were using? 0.748 0.91 How often have you used cannabis first thing in the morning? 0.726 0.95 How much of the average day do you feel/spend stoned? 0.706 0.90 How difficult do you think you would find it to stop using or go without cannabis altogether? 0.679 0.91 What was the longest time you went without using cannabis? 0.648 0.91 Have you felt you needed cannabis? 0.625 0.88 Have you been able to stop using cannabis when you wanted to? 0.612 0.89 Have you found it difficult to get through a day without using cannabis? 0.608 0.92 Did your use of cannabis ever interfere with your work at school, job or your home life? 0.753 0.90 Have you lacked the energy to get things done in the way you used to? 0.731 0.90 Have you given up things you used to enjoy or were important because of cannabis? 0.706 0.90 Has anything you had planned or were expected to do not happened after using cannabis? 0.669 0.89 Have you had problems concentrating and remembering things? 0.649 0.92 Did you ever use cannabis after you had decided not to? 0.606 0.93 Eigenvalue 8.199 6.861 Percentage of variance explained 21.024 17.592

aAll correlations were significant at the 0.01 level (two-tailed).

618 Jan Bashford et al.

content domains. Combined with the only moderate mean inter-item correlations of both components (range 0.40–0.59) and interscale correlation (r = 0.47, P = <0.001) this indicates that, with good discriminative validity (only 22% shared variance), the two subscales

were separate but related axes of a primary construct of cannabis-related problems in a heterogeneous commu- nity sample [51,52]. Spearman’s correlations between clinical participants’ urinalysis results (n = 40) and self- reported days (gs = 0.43, P = 0.01) and cones (gs = 0.37, P = 0.02) used over the past 30 days suggested a general absence of bias.

Construct validity

Criterion/concurrent validity. Adults’ significantly higher mean impaired control scores (t(201) = -2.13, P = 0.03) and adolescents’ significantly higher mean problems scores (t(201) = 2.00, P = 0.04) (Table 2) justified using age groups in further analyses. Pearson’s correlation coefficients between the CUPIT subscales and validation measures are displayed in Table 3.

As evident, both subscales converged in conceptually consistent, highly significant, moderate to strong corre- lation patterns with total DSM/ICD symptoms, the SDS, all CPQ-A subscales and the adult CPQ core and work scale. Impaired control correlated strongly with all TLFB variables. Adolescents largely accounted for the signifi- cant correlation between problems and consumption measures, and between impaired control and psychologi- cal distress (GSI).

Table 2 Standardized internal consistency reliability estimates and descriptives for the Cannabis Use Problems Identification Test (CUPIT) subscales.

Group/variable Impaired control Problems

Whole sample (n = 212) 0.92 0.83 Mean 28.14 6.96 SD 11.78 4.58 Mean inter-item correlation 0.53 0.46

Age groups Adolescents (n = 138) 0.92 0.79

Mean 26.98 7.36 SD 11.40 4.51 Mean inter-item correlation 0.55 0.40

Adults (n = 74) 0.92 0.90 Mean 30.30 6.20 SD 12.23 4.63 Mean inter-item correlation 0.52 0.59

SD: standard deviation.

Table 3 Correlation between the Cannabis Use Problems Identification Test (CUPIT) subscales and key validation measures.

Measure/variable

Impaired control Problems

Sample (n = 211)

Adolescents (n = 138)

Adults (n = 73)

Sample (n = 211)

Adolescents (n = 138)

Adults (n = 73)

CIDI-Auto Number of DSM/ICD symptoms 0.63*** 0.68*** 0.51*** 0.57*** 0.61*** 0.50*** Severity of Dependence Scale (SDS) 0.71*** 0.72*** 0.70*** 0.59*** 0.61*** 0.63***

TLFB interview: Days used past 90 days 0.80*** 0.81*** 0.76*** 0.25** 0.42** 0.08 Cones used past 90 days 0.60*** 0.60*** 0.65*** 0.19** 0.23** 0.27* Days used past 30 days 0.77*** 0.80*** 0.72*** 0.24*** 0.40** 0.10 Cones used past 30 days 0.60*** 0.55*** 0.67*** 0.19** 0.21* 0.28*

CPQ for adults Core scale (n = 76) 0.53*** 0.62*** Spouse scale (n = 26) 0.26 0.36 Children scale (n = 21) 0.43 -0.04 Work scale (n = 38) 0.46** 0.48**

CPQ for adolescents Core scale (n = 135) 0.59*** 0.63*** Parent scale (n = 129) 0.35*** 0.26** Partner scale (n = 64) 0.46*** 0.25* School scale (n = 117) 0.54*** 0.48*** Work scale (n = 26) 0.46* 0.47*

BSI-18 GSI score 0.18* 0.21* 0.09 0.42*** 0.36*** 0.56***

*P < 0.05, **P < 0.01, ***P < 0.001, all tests two-tailed. CIDI: Composite International Diagnostic Interview; DSM/ICD: Diagnostic and Statistical Manual/International Classification of Diseases; TLFB: time-line follow-back; CPQ: Cannabis Problems Questionnaire; BSI: Brief Symptom Inventory; GSI: Global Severity Index.

The CUPIT development 619

Discriminative validity. CUPIT subscale scores of the three independent CIDI-Auto-generated DSM-IV/ICD-10 diag- nostic groups (no diagnosis, n = 17; abuse and/or harmful use, n = 36; dependence, with or without abuse/ harmful use, n = 158) were compared using one-way ANOVA (Table 4). Given insufficient cases in the adult ‘no diagnosis’ group, Student’s t-tests were used for adult comparisons.

Mean scores differed significantly across sample diag- nostic groups, increasing in magnitude in the expected direction. Post-hoc Tukey’s HSD analyses revealed that while the sample and adolescent ‘dependent’ groups scored significantly higher than the ‘no diagnosis’ (P = <0.001) and ‘abuse/harmful use’ (P = <0.001) groups, the two latter groups’ means did not differ signifi- cantly, a pattern also evident in consumption scores. A mixed result emerged for adult groups.

Longitudinal predictive capacity. At follow-up, Pearson’s correlations were employed to examine the longitudinal relationship between baseline CUPIT subscale scores and re-administered validation measures. Again, impaired control exhibited a generally strong, highly significant correlation and problems a moderate to strong significant correlation, with most measures. While adolescents’ increased consumption and problems largely explained stronger associations, adult scores remained generally stable with some minor fluctuations.

Follow-up diagnostic groups from the CIDI-Auto algo- rithm (no diagnosis, n = 14; abuse/harmful use, n = 33; and dependence, with or without abuse/harmful use, n = 147) were compared using one-way ANOVAs fol- lowed by Tukey’s HSD (t-tests for adults) for any differ- ences in original CUPIT scores and baseline consumption as a function of group membership. Table 4 presents baseline/follow-up values.

Results replicated baseline patterns of significantly different between-group mean scores with post-hoc patterns of non-significant differences between ‘abuse/ harmful use’ and ‘no diagnosis’ groups. Baseline CUPIT subscale scores (particularly impaired control) had highly significant predictive power for diagnostic group membership 12 months later. Baseline consumption also exhibited highly significant ability to predict diagnostic group assignment prospectively.

Classification accuracy. Given no significant adolescent/ adult differences in total CUPIT scores (mean = 35.1, SD = 4.52, range: 5–72), the ROC curves were plotted using sample data at baseline (n = 211) and 12-month follow-up (n = 194). Two groups were formed: those with a DSM-IV/ICVD-10 diagnosis (baseline n = 194;

follow-up n = 180) defined a ‘case’; no diagnosis (baseline n = 17; follow-up n = 14) defined a ‘control’.

The AUC for CUPIT scores when discriminating respondents qualifying for a 12-month diagnosis from those who did not at baseline was 0.88 [95% confidence interval (CI) = 0.80, 0.97], indicating very good diagnos- tic performance. The closer the AUC is to 1, the more disparate the groups. Similarly, the AUC for CUPIT scores when predicting those who will qualify for a diagnosis 12 months later and those who will not was 0.89 (95% CI = 0.82, 0.96). The sensitivities, specificities, positive predictive value (PPV) and c2 values corresponding to potential score cut-points are presented in Table 5.

Table 5 suggests the optimal CUPIT cut-off score to be 20 at both baseline and follow-up (sensitivity 87%, speci- ficity 79%, PPV 98%). However, in this high-risk sample the more liberal cut-point of 12 classified almost all (176 of 180) true positives with increased sensitivity (98%), minor loss of PPV (95%), with a corresponding reduction in specificity (36%). Of critical importance in an oppor- tunistic screening context, this cut-point captured eight of the 12 ‘diagnostic orphans’ in the screening net.

DISCUSSION

The CUPIT is a highly acceptable, reliable and valid brief cannabis screener for use across community settings and consumers of all ages. Its novel capability is to reliably classify both currently diagnosable and potentially prob- lematic cannabis use among respondents. These esti- mable properties derive from the CUPIT’s systematic development informed by international expert opinion and empirical longitudinal validation among typically diverse cannabis-using adolescents and adults.

The CUPIT items (Table 1) were selected on the dual basis of statistical parameters and important operational standards. Questions had to be simple, readily under- standable and valid across adult and adolescent cannabis users universally (face/content validity). Respondents spontaneously confirmed that questions accurately cap- tured their experience of cannabis use and problems. The validation analyses clearly show that the CUPIT satisfies temporal and internal consistency reliability and con- struct, discriminative, diagnostic (sensitivity) and pre- dictive validity. Test–retest (0.89–0.99) and internal consistency reliability estimates (0.92 and 0.83) were well above the accepted benchmark for exploratory research (0.70) [46,53], while the construct validation data provided substantial evidence of the CUPIT’s specific validity properties. The particularly consistent, strong positive relationship of ‘impaired control’ with all key measures supports the dimensionality of the dependence syndrome and its existence in a broad spectrum of

620 Jan Bashford et al.

T a

bl e

4 C

an n

ab is

U se

P ro

bl em

s Id

en ti

fic at

io n

Te st

(C U

P IT

) su

bs ca

le an

d co

n su

m pt

io n

sc or

es by

D SM

-I V

/I C

D -1

0 di

ag n

os ti

c g

ro u

p at

ba se

li n

e (n

= 2

1 1

) an

d 1

2 -m

on th

fo ll

ow -u

p (n

= 1

9 4

V ar

ia bl

D ia

gn os

ti c

gr ou

N o

di ag

no si

s A

bu se

/h ar

m fu

l us

e D

ep en

de nc

M ea

n SD

M ea

n SD

M ea

n SD

B F

Im pa

ir ed

co n

tr ol

W h

ol e

sa m

pl e

(n =

2 1

1 ,1

9 4

) 1

4 .1

2 1

3 .1

4 7

.3 2

6 .7

8 1

7 .7

5 1

7 .5

5 7

.0 3

7 .5

8 3

2 .0

8 3

1 .5

4 1

0 .3

7 1

0 .4

0 5

1 .7

4 4

4 .7

5 **

* A

do le

sc en

ts (n

= 1

3 8

,1 2

8 )

1 3

.1 3

1 2

.1 7

6 .2

7 5

.0 1

1 9

.0 4

1 8

.5 7

6 .8

5 7

.9 6

3 1

.1 8

3 0

.6 3

1 0

.0 7

1 0

.1 7

3 7

.0 7

3 0

.7 2

** *

A du

lt s

(n =

7 3

,6 6

) 1

5 .1

7 1

5 .2

0 6

.9 5

6 .3

9 3

3 .5

5 3

3 .1

1 1

0 .7

6 1

0 .6

9 -5

.6 7

-5 .1

1 **

*( t)

P ro

bl em

s W

h ol

e sa

m pl

e 3

.1 8

3 .2

9 2

.7 7

2 .4

9 4

.5 6

4 .0

6 2

.8 0

2 .7

0 7

.9 1

7 .9

4 4

.6 8

4 .7

2 1

6 .1

0 1

6 .2

4 **

* A

do le

sc en

ts 3

.1 9

2 .9

2 2

.8 6

2 .2

7 4

.7 5

4 .2

6 2

.8 9

2 .9

3 8

.6 8

8 .7

1 4

.3 8

4 .4

6 1

9 .0

4 1

8 .8

8 **

* A

du lt

s 4

.1 7

3 .6

0 2

.6 9

2 .1

7 6

.6 3

6 .6

1 4

.9 0

4 .9

0 -1

.6 9

-1 .8

9 (t

)

C an

n ab

is co

n su

m pt

io n

W h

ol e

sa m

pl e

(n =

2 1

1 ,1

9 4

) D

ay s

u se

d pa

st 9

0 da

ys 1

5 .7

1 1

5 .0

0 1

9 .5

7 2

1 .1

4 2

9 .8

9 2

4 .0

3 2

3 .7

1 2

2 .1

0 6

1 .9

4 6

0 .5

4 2

7 .4

0 2

6 .9

3 4

0 .1

6 4

1 .6

6 **

* D

ay s

u se

d pa

st 3

0 da

ys 5

.0 6

4 .9

3 6

.9 2

7 .4

8 1

0 .5

3 8

.1 5

9 .1

5 8

.4 4

2 0

.8 6

2 0

.4 6

9 .7

1 9

.5 8

3 4

.5 4

3 7

.0 9

** *

A do

le sc

en ts

(n =

1 3

8 ,1

2 8

) D

ay s

u se

d pa

st 9

0 da

ys 1

1 .3

1 1

0 .3

3 7

.6 6

5 .0

5 3

0 .0

0 2

5 .2

2 2

2 .7

5 2

3 .1

3 5

5 .9

2 5

4 .2

2 2

6 .7

2 2

6 .4

3 2

8 .8

1 2

5 .9

3 **

* D

ay s

u se

d pa

st 3

0 da

ys 3

.5 0

3 .2

5 2

.6 6

1 .9

1 1

0 .1

7 8

.6 5

8 .2

6 8

.8 6

1 8

.6 6

1 8

.0 6

9 .5

4 9

.4 1

2 5

.4 7

2 1

.8 0

** *

A du

lt s

(n =

7 3

,6 6

) D

ay s

u se

d pa

st 9

0 da

ys 2

9 .6

7 2

1 .3

0 2

6 .5

8 2

0 .4

0 7

1 .7

7 7

1 .0

9 2

5 .7

9 2

4 .6

1 -5

.1 4

-6 .0

2 **

*( t)

D ay

s u

se d

pa st

3 0

da ys

1 1

.2 5

7 .0

0 1

1 .0

9 7

.6 9

2 4

.4 5

2 5

.5 7

8 .9

4 8

.4 7

-4 .4

8 -6

.0 1

** *(

B :b

as el

in e;

F: fo

ll ow

-u p;

(t )

St u

de n

t’ s

t- te

st ,7

0 df

(B ),

6 2

df (F

); **

*P <

0 .0

0 1

,a ll

te st

s tw

o- ta

il ed

.S D

:s ta

n da

rd de

vi at

io n

The CUPIT development 621

adolescent and adult cannabis users before they come to clinical attention [54].

For maximum utility in opportunistic cannabis screening, a screen should discriminate diagnostic groups along the risk continuum, from high risk (depen- dence diagnosis likely), through non-dependent but mod- erate risk of developing dependence (diagnostic orphans, abuse, risky use), to low risk [55–57]. The CUPIT’s dis- criminative ability was evident in significantly different between-group mean scores increasing with diagnostic severity. The non-significant difference between ‘abuse/ harmful use’ and ‘no diagnosis’ groups was explained by the 12 diagnostic orphans who, while distinct from ‘dependent’ groups, are similar to ‘abuse’ groups [47,48,58]. At follow-up four baseline diagnostic orphans had progressed to full-blown DSM-IV/ICD-10 disorder, while eight remained orphans. With cannabis problems just below the diagnostic threshold and on a trajectory towards dependence, these high-risk users are prime targets for an early intervention [7,9,59,60].

The CUPIT’s ROC diagnostic performance demon- strated its significant predictive capability over a 12-month period. The indicators at alternative cut-points (Table 5) illustrate the sensitivity/specificity trade-off decisions involved when attempting to detect at-risk and

possible cases at earlier stages [36,61]. Despite statisti- cally superior alternatives a cut-off score of 12 was the optimal criterion threshold for opportunistic screening for SEI purposes in this high-risk New Zealand sample, detecting almost all (98%) those who would receive a diagnosis 12 months later. Future research may show alternative CUPIT cut-points to be optimal in other cultures, contexts or settings (e.g. specialist, primary care settings, epidemiological research), among other cannabis-using populations with different prior probabili- ties for the disorder (general population, primary care, clinical) and for different purposes (preliminary screen- ing, assessment referral, treatment planning). Being directly proportional to disorder prevalence, PPVs will also vary in less clinical populations.

The 16 final CUPIT items (Table 1) measure the DSM/ ICD tridimensional concept of risky use (items 1 and 2), dependence/using behaviour (3–10), health and social problems (11–16), requiring a reading level of approxi- mately 7/8 years’ schooling and about 5 minutes admin- istration time. Scores can range from 0 (no use at all) to 82 (daily/more than daily use, severely problematic). Rapidly scored and interpreted, self- or other- administered, the CUPIT has potential as a population screener, research tool, pre–post measure in clinical out-

Table 5 Cannabis Use Problems Identification Test (CUPIT) sensitivity, specificity, c2 value and positive predictive value (PPV) at potential cut-off scores when discriminating between cannabis users with and without a DSM-IV/ICD-10 diagnosis of cannabis dependence or abuse/harmful use at baseline (n = 211) and follow-up (n = 194).

Baseline 12-Month follow-up

CUPIT score Sensitivity Specificity c2 Sensitivity Specificity c2 PPV (%)

5 1.00 0.06 11.47 1.00 0.07 12.92 93.2 6 1.00 0.06 11.47 1.00 0.07 12.92 93.2 7 0.99 0.06 4.80 0.99 0.07 5.52 93.2 8 0.99 0.12 9.68 0.99 0.14 11.17 93.6 9 0.99 0.24 28.64 0.98 0.21 16.93 93.7

10 0.98 0.24 23.55 0.98 0.21 13.78 94.1 11 0.98 0.29 33.27 0.98 0.29 22.81 94.6 12 0.98 0.35 43.59 0.98 0.36 32.94 95.1 13 0.96 0.41 35.61 0.96 0.43 28.63 95.5 14 0.96 0.47 41.12 0.95 0.50 34.76 96.0 15 0.96 0.47 41.12 0.95 0.50 34.76 96.1 16 0.95 0.53 43.56 0.94 0.57 38.29 96.6 17 0.94 0.53 38.13 0.93 0.57 35.79 96.6 18 0.92 0.59 39.06 0.91 0.57 27.90 96.5 19 0.90 0.71 44.14 0.88 0.71 34.55 97.5 20 0.89 0.76 47.92 0.87 0.79 38.90 98.1 21 0.88 0.76 44.41 0.86 0.79 35.96 98.1 22 0.84 0.82 41.04 0.82 0.79 26.88 98.0 23 0.81 0.82 34.15 0.78 0.79 21.99 97.9 24 0.78 0.82 28.72 0.75 0.79 18.16 97.8 25 0.75 0.82 24.33 0.72 0.86 20.05 98.5

622 Jan Bashford et al.

comes assessment, a discrete screener or component embedded within a life-style questionnaire, and is adapt- able for computerized administration. The CUPIT is avail- able at http://ncpic.org.au/assets/downloads/workforce/ cannabisinfo/assessment-tools/cannabis-use-problems- identification-test.pdf

The CUPIT has several advantages over existing can- nabis screens. The inclusive sampling strategy, DSM-IV/ ICD-10 criteria as criterion standard, and incorporating clinically assessed adolescent and adult users assured the CUPIT’s relevance to both clinical and general popula- tions. The CUPIT’s foremost advantage is its ability to reliably identify both currently problematic and risky use across adolescent and adult respondents.

There are some sampling and methodological caveats to this preliminary validation study. The uncontrolled, self-report, pre–post measures design raises possible reliability/validity issues. However, all indicators, includ- ing toxicology reports and (informal) internal ‘audits’, suggest over/under-reporting to be minimal, as found elsewhere [62–65]. Validating the CUPIT by reference to validation instruments incorporating the same key ele- ments may have inflated correlations, and hence validity estimates. While the data from a rather small adolescent/ adult community sample of predominantly high-risk users arguably represent those targeted primarily by a SEI approach, further studies among samples stratified by age, ethnicity, gender and consumption would extend information on the CUPIT’s psychometric properties among cannabis-using populations in other cultures. Research applying confirmatory factor analytical proce- dures and cross-validation is also desirable.

Cannabis use is adversely impacting the health and social functioning of a large and rapidly expanding cohort of the world’s adults, adolescents and, alarmingly, children. As exemplified in this sample, most are either unaware, or do not feel that they need, assistance [66–68]. Routine opportunistic and targeted screening with early intervention for cannabis problems has poten- tial for enormous gains in public health. Rapid, reliable, valid classification of a broad spectrum of cannabis- related problems and harms is the hallmark of a screener suitable for cannabis SEI. This report provides extensive evidence that the CUPIT satisfies these crucial criteria.

Declarations of interest

None.

Acknowledgements

This study would not be possible without the contribution of the many organizations and individuals who helped make the CUPIT project a reality, and we thank them for their valued contribution.

References

1. UNODC. (2009). World Drug Report 2009. Available at: http://www.unodc.org (accessed 25 July 2009).

2. Fergusson D. M., Horwood L. J. Cannabis use and depen- dence in a New Zealand birth cohort. NZ Med J 2000; 113: 156–8.

3. World Health Organization. The Neuroscience of Psychoactive Substance Use and Dependence. Geneva: Author; 2004.

4. World Health Organization. Alcohol, drugs, and tobacco pro- gramme. World Health Organization Phase III Collaborative Study on implementing and supporting early intervention strategies in primary care. Report on Strand I: General prac- titioner current practices and perceptions of preventive medicine and early intervention for hazardous alcohol use. A 16-country study. Copenhagen, Denmark: WHO; 1998.

5. Lader M., Edwards G., Drummond D. C. The demands which the problems perspective sets for research. In: Lader M., Edwards G., Drummond D. C., editors. The Nature of Alcohol and Drug-Related Problems. New York: Oxford University Press; 1992, p. 189–97.

6. Anthony J. C. Putting epidemiology and public health in needs assessment: drug dependence and beyond. In: Andrews G., Henderson S., editors. Unmet Need in Psychia- try: Problems, Resources, Responses. Cambridge: Cambridge University Press; 2000, p. 302–8.

7. Hall W., Swift W. Policy implications of cannabis depen- dence. In: Roffman R. A., Stephens R. S., editors. Cannabis Dependence: Its Nature, Consequences and Treatment. London: Cambridge University Press; 2006, p. 315–19.

8. Penrose-Wall J., Copeland J., Harris M. Shared Care of Illicit Drug Problems by General Practitioners and Primary Health Care Providers: a Literature Review. Sydney: Centre for General Practice Integration Studies, University of New South Wales; 2000.

9. Copeland J., Frewen A., Elkin K. Management of Cannabis Use Disorder and Related Issues: a Clinician’s Guide. Sydney: National Cannabis Prevention and Information Centre (NCPIC).

10. McPherson T. L., Hersch R. K. Brief substance use screening instruments for primary care settings: a review. J Subst Abuse Treat 2000; 18: 193–202.

11. Fleming M. Screening, assessment, and intervention for substance use disorders in general health care settings. Subst Abuse 2002; 23: 47–66.

12. Hannifin J. The cannabis abuse syndrome screening test: a brief report. Unpublished report. Wellington: Drugs Advisory Committee; 1990.

13. Adamson S., Sellman J. D. A prototype screening instrument for cannabis use disorder: the Cannabis Use Disorders Iden- tification Test (CUDIT) in an alcohol dependent clinical sample. Drug Alcohol Rev 2003; 22: 309–15.

14. Alexander D. A Marijuana Screening Inventory (experimen- tal version): description and preliminary psychometric properties. Am J Drug Alcohol Abuse 2003; 29: 619–46.

15. Alexander D., Leung P. The Marijuana Screening Inventory (MSI-X): reliability, factor structure, and scoring criteria with a clinical sample. Am J Drug Alcohol Abuse 2004; 30: 321–51.

16. Alexander D., Leung P. The Marijuana Screening Inventory (MSI-X): concurrent, convergent and discriminant validity with multiple measures. Am J Drug Alcohol Abuse 2006; 32: 351–78.

17. Legleye S., Karila L., Beck F., Reynaud M. Validation of the

The CUPIT development 623

CAST, a general population Cannabis Abuse Screening Test. J Subst Use 2007; 12: 233–42.

18. Okulicz-Kozaryn K. Psychometric properties of the Prob- lematic Marijuana Use (PUM) test for adolescents. Post Psychiatry Neurol 2007; 16: 105–11.

19. Piontek D., Kraus L., Klempova D. Short scales to assess cannabis-related problems: a review of psychometric prop- erties. Subst Abuse Treat Prev Policy 2008; 3: 25.

20. Likert R. A technique for the measurement of attitudes. Arch Psychol 1932; 140: 44–53.

21. Griffiths P., Gossop M., Powis B., Strang J. Reaching hidden populations of drug users by privileged access interviewers: methodological and practical issues. Addiction 1993; 88: 1617–26.

22. Watters J., Biernacki P. Targeted sampling: options for the study of hidden populations. Soc Probl 1989; 36: 416–30.

23. Sobell L. C., Sobell M. B. Timeline Follow-Back: a technique for assessing self-reported alcohol consumption. In: Litten R., Allen J., editors. Measuring Alcohol Consumption; Psycho- social and Biochemical Methods. Totowa, NJ: Humana Press Inc; 1992, p. 41–69.

24. Sobell L. C., Sobell M. B., Buchan G., Cleland P. A., Leo G. I. The reliability of the timeline followback method applied to drug, cigarette, and cannabis use. Paper presented at the 30th Annual Meeting of the Association for Advancement of Behavior Therapy, New York, November; 1996.

25. Fals-Stewart W., O’Farrell T. J., Freitas T. T., McFarlin S. K., Rutigliano P. The timeline followback reports of psychoac- tive substance use by drug-abusing patients: psychometric properties. J Consult Clin Psychol 2000; 68: 134–44.

26. Marijuana Treatment Project Research Group. Brief treat- ments for cannabis dependence: findings from a randomized multi-site trial. J Consult Clin Psychol 2004; 72: 455–6.

27. Leino A., Saaramies J., Gronholm M., Lillesunde P. Compari- son of eight commercial on-site screening devices for drugs of abuse testing. Scand J Clin Lab Invest 2001; 61: 325–31.

28. Lafolie P., Beck O., Blennow G., Boreus L., Borg S., Elwin C. E. et al. Importance of creatinine analyses of urine when screening for abused drugs. Clin Chem 1991; 37: 1927–31.

29. American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders, 4th edn. Washington: Author; 1994.

30. World Health Organization. International Classification of Diseases–Tenth Revision (ICD-10). Geneva: Author; 1992.

31. World Health Organization. Composite International Diagnos- tic Interview, 2.1, Computerized Version (CIDI-Auto). Geneva: Author; 1997.

32. Andrews G., Peters L. The psychometric properties of the Composite International Diagnostic Interview. Soc Psychia- try Psychiatr Epidemiol 1998; 33: 80–8.

33. Rubio-Stipec M., Peters L., Andrews G. Test–retest reliability of the computerized CIDI (CIDI-Auto): substance abuse modules. Subst Abuse 1999; 20: 263–72.

34. Wittchen H.-U., Lachner G., Wunderlich U., Pfister H. Test– retest reliability of the computerized DSM-IV version of the Munich-Composite International Diagnostic Interview (M-CIDI). Soc Psychiatry Psychiatr Epidemiol 1998; 33: 568– 78.

35. Kedzior K. K., Martin-Iverson M. T. Concurrent validity of cannabis misuse on CIDI-Auto 2.1 in low-level cannabis users from the general population. Aust J Psychol 2007; 59: 169–75.

36. Swift W., Copeland J., Hall W. Choosing a diagnostic cut-off for cannabis dependence. Addiction 1998; 93: 1681–92.

37. Martin G., Copeland J., Gates P., Gilmour S. The Severity of Dependence Scale (SDS) in an adolescent population of can- nabis users. J Subst Abuse Treat 2006; 29: 207–13.

38. Copeland J., Swift W., Roffman R., Stephens R. A random- ized controlled trial of brief cognitive-behavioral interven- tions for cannabis use disorder. J Subst Abuse Treat 2001; 21: 55–64.

39. Copeland J., Gilmour S., Gates P., Swift W. The Cannabis Problems Questionnaire: factor structure, reliability and validity. Drug Alcohol Depend 2005; 80: 313–9.

40. Martin G., Copeland J., Gilmour S., Gates P., Swift W. The Adolescent Cannabis Problems Questionnaire (CPQ-A): psy- chometric properties. Addict Behav 2006; 31: 2238–48.

41. Derogatis L. R. Brief Symptom Inventory 18 (BSI 18): Admin- istration, Scoring, and Procedures Manual. Minneapolis, MN: National Computing Systems Pearson, Inc; 2000.

42. Derogatis L. R. Symptom Checklist 90-R (SCL-90-R): Admin- istration, Scoring, and Procedures Manual, 3rd edn. Minne- apolis, MN: National Computing Systems Pearson, Inc; 1994.

43. SPSS, Inc. SPSS Base 12.0. User’s Guide. Chicago, IL: Author; 2003.

44. Cronbach L. J. Coefficient alpha and the internal structure of tests. Psychometrika 1951; 16: 297–334.

45. Comrey A. L., Lee H. B. A First Course in Factor Analysis. Hillsdale, NJ: Lawrence Erlbaum Associates; 1992.

46. Nunnally J. C. Psychometric Theory. New York: McGraw-Hill Inc; 1978.

47. Deas D., Roberts J., Grindlinger D. The utility of DSM-IV criteria in diagnosing substance abuse/dependence in ado- lescents. J Subst Use 2005; 10: 10–21.

48. Degenhardt L., Lynskey M., Coffey C., Patton G. ‘Diagnostic orphans’ among young adult cannabis users: persons who report symptoms but do not meet diagnostic criteria. Drug Alcohol Depend 2002; 67: 205–12.

49. Cattell R. B. The screen test for number of factors. Multivari- ate Behav Res 1966; 1: 245–76.

50. Kim J.-O., Mueller C. W. Factor Analysis: Statistical Methods and Practical Issues. Beverly Hills, CA: Sage; 1978.

51. Schmitt N. Uses and abuses of coefficient alpha. Psychol Assess 1996; 8: 350–3.

52. Kline P. The New Psychometrics: Science, Psychology, and Mea- surement. London: Routledge; 1998.

53. Streiner D. L., Norman G. R., editors. Health Measurement Scales: a Practical Guide to Their Development and Use, 2nd edn. Oxford: Oxford University Press; 1995.

54. Saunders J. B., Aasland O. G., Amunsden A., Grant M. Development of the Alcohol Use Disorders Indentification Test (AUDIT): WHO collaborative project on Early Detection of Persons with Harmful Alcohol Consumption: II. Addiction 1993; 88: 791–804.

55. Dawe S., Loxton N. J., Hides L., Kavanagh D. J., Mattick R. P. Review of Diagnostic Screening Instruments for Alcohol and Other Drug Use and Other Psychiatric Disorders, 2nd edn. Canberra: Australian Government Publishing Service; 2002.

56. Newcombe D. A. L., Humeniuk R. E., Ali R. Validation of the World Health Organization Alcohol, Smoking and Sub- stance Involvement Screening Test (ASSIST); report of results from the Australian site. Drug Alcohol Rev 2005; 24: 217–26.

57. Winters K. Assessment of alcohol and other drug use among adolescents. In: Allen J., Wilson V. B., editors. Assess- ing Alcohol Problems, 2nd edn. NIH Publication no.

624 Jan Bashford et al.

03-3745. Bethseda, MD: National Institute on Alcohol Abuse and Alcoholism; 2003.

58. Hasin D., Paykin A. Dependence symptoms but no diagno- sis: ‘diagnostic orphans’ in a 1992 national sample. Drug Alcohol Depend 1999; 53: 215–22.

59. Chen C.-Y., O’Brien M. S., Anthony J. C. Who becomes can- nabis dependent soon after onset of use? Epidemiological evidence from the United States: 2000–2001. Drug Alcohol Depend 2005; 79: 11–22.

60. Degenhardt L., Coffey C., Carlin J. B., Swift W., Patton G. C. Are diagnostic orphans at risk of developing cannabis abuse or dependence? Four-year follow-up of young adult can- nabis users not meeting diagnostic criteria. Drug Alcohol Depend 2008; 92: 86–90.

61. Conigrave K. M., Hall W., Saunders J. B. The AUDIT ques- tionnaire: choosing a cut-off score. Addiction 1995; 90: 1349–56.

62. Babor T. F., Steinberg K., McRee B., Vendetti J., Carroll K. C. Treating marijuana dependence in adults: a multi-site randomized clinical trial. New Dir Eval 2002; 94: 17– 30.

63. Buchan B. J., Dennis M. L., Tims F. M., Diamond G. S.

Cannabis use: consistency and validity of self-report, on-site urine testing and laboratory testing. Addiction 2002; 97: 98–108.

64. Gignac M., Wilens T. E., Biederman J., Kwon A., Mick E., Swezey A. Assessing cannabis use in adolescents and young adults: what do urine screens and parental report tell you? J Child Adolesc Pharmacol 2005; 15: 742–50.

65. Mennes C. E., Abdallah A. B., Cottler L. B. The reliability of self-reported cannabis abuse, dependence and withdrawal symptoms: multisite study of differences between general population and treatment groups. Addict Behav 2009; 34: 223–6.

66. Agosti V., Levin F. Predictors of treatment contact among individuals with cannabis dependence. Am J Drug Alcohol Abuse 2004; 30: 121–7.

67. Gates P., Taplin S., Copeland J., Swift W., Martin G. Barriers and facilitators to cannabis treatment. NCPIC Technical Report 1, NCPIC: Sydney; 2009.

68. Urbanoski K. A., Strike C. J., Rush B. R. Individuals seeking treatment for cannabis-related problems in Ontario: demo- graphic and treatment profile. Eur Addict Res 2005; 11: 115–23.

The CUPIT development 625

This document is a scanned copy of a printed document. No warranty is given about the accuracy of the copy.

Users should refer to the original published version of the material.