Substance use risk profiles and associations with early substance use in adolescence

Substance use risk profiles and associations with early substance use in adolescence

Monique Malmberg • Geertjan Overbeek •

Karin Monshouwer • Jeroen Lammers •

Wilma A. M. Vollebergh • Rutger C. M. E. Engels

Received: November 9, 2009 / Accepted: June 30, 2010 / Published online: July 13, 2010

� The Author(s) 2010. This article is published with open access at Springerlink.com

Abstract We examined whether anxiety sensitivity,

hopelessness, sensation seeking, and impulsivity (i.e.,

revised version of the Substance Use Risk Profile Scale)

would be related to the lifetime prevalence and age of onset

of alcohol, tobacco, and cannabis use, and to polydrug use

in early adolescence. Baseline data of a broader effec-

tiveness study were used from 3,783 early adolescents aged

11–15 years. Structural equation models showed that

hopelessness and sensation seeking were indicative of ever-

used alcohol, tobacco or cannabis and for the use of more

than one substance. Furthermore, individuals with higher

levels of hopelessness had a higher chance of starting to

use alcohol or cannabis at an earlier age, but highly anxiety

sensitive individuals were less likely to start using alcohol

use at a younger age. Conclusively, early adolescents who

report higher levels of hopelessness and sensation seeking

seem to be at higher risk for an early onset of substance use

and poly substance use.

Keywords Alcohol use � Tobacco use � Cannabis use � Personality � Early adolescence

Introduction

Dutch adolescents are one of the leaders in terms of

drinking frequency and binge drinking in Europe and they

usually start drinking in early adolescence (Hibell et al.

2009). Also, their use of tobacco and cannabis increases

rapidly during this period (Monshouwer et al. 2008). This

is disturbing in that early initiation of substance use has

many detrimental consequences, like distortion of brain

development (e.g., Tapert et al. 2002) and elevated risk for

later dependence and misuse (e.g., Andersen et al. 2003).

Further, early initiation increases the likelihood of poly

substance use (Ellickson et al. 2003) that, in turn, leads to

more damaging health effects (Feigelman et al. 1998).

Thus, identifying risk profiles of early adolescent girls and

boys is of crucial importance, because it may facilitate

adequate prevention efforts targeted at youths who are at

risk for an early onset of substance use or abuse (e.g.,

Conrod et al. 2008, 2010).

It is well known that personality is associated with

substance use (e.g., Flory et al. 2002) and in general, per-

sonality dimensions involving neurotic tendencies or defi-

cits in behavioral inhibition are found to best predict

substance (mis)use (e.g., Barrett et al. 1998; Cloninger

et al. 1991). Furthermore, personality dimensions con-

cerning specific, rather than general personality disposi-

tions are of most interest for substance related behaviors

(Caspi et al. 1996; Comeau et al. 2001; Jackson and Sher

2003; Woicik et al. 2009). One instrument that specifically

taps specific personality dimensions involving neurotic

tendencies and inhibition deficits is the Substance Use Risk

M. Malmberg (&) � R. C. M. E. Engels

Behavioural Science Institute, Radboud University Nijmegen,

P.O. Box 9104, 6500 HE Nijmegen, The Netherlands

e-mail: m.malmberg@pwo.ru.nl

G. Overbeek

Developmental Psychology, Utrecht University, Utrecht,

The Netherlands

K. Monshouwer � J. Lammers

Trimbos Institute (Netherlands Institute of Mental Health

and Addiction), Utrecht, The Netherlands

K. Monshouwer � W. A. M. Vollebergh

Department of Interdisciplinary Social Science,

Utrecht University, Utrecht, The Netherlands

123

J Behav Med (2010) 33:474–485

DOI 10.1007/s10865-010-9278-4

Profile Scale (SURPS; Woicik et al. 2009). This instrument

measures four distinct and independent personality traits

(i.e., anxiety sensitivity, hopelessness, sensation seeking,

and impulsivity) that are hypothesized and actually ap-

peared to be related to high and problematic substance use

behaviors (Conrod et al. 1998; Jackson and Sher 2003;

Pulkkinen and Pitkänen 1994; Shall et al. 1992; Sher et al.

2000; Stewart et al. 1995) and other risk behaviors (e.g.,

delinquency; Woicik et al. 2009).

The first trait (i.e., anxiety sensitivity) refers to the fear

of symptoms of psychical arousal (e.g., feeling dizzy or

faint; Reis et al. 1986) and the second (i.e., hopelessness) is

identified as a risk factor for the development of depression

(Joiner 2001). Both anxiety sensitivity and hopelessness

relate to increased levels of drinking and problem drinking

(Stewart et al. 1995; Conrod et al. 1998). The third trait

(i.e., impulsivity) involves difficulties in the regulation

(controlling) of behavioral responses (Spoont 1992) and is

related to an increased risk for early alcohol and drug

(mis)use (Pulkkinen and Pitkänen 1994). Finally, the fourth

trait (i.e., sensation seeking) is characterized by the desire

for intense and novel experiences (Zuckerman 1994) and

sensation seekers have been found to drink more and to be

more at risk for heavy alcohol use (Shall et al. 1992; Sher

et al. 2000). The four SURPS’ personality traits are based

on extended personality measures (e.g., ASI; Peterson and

Reiss 1992) and show stronger associations with these

measures than with scales measuring broader dimensions

of personality (e.g., NEO-FFI; Costa and McCrae 1992).

Sensation seeking is, for instance, related to measures of

openness and extraversion, but is more strongly related to

scales measuring venturesomeness (Eysenck and Eysenck

1978; Woicik et al. 2009).

The SURPS personality traits show some overlap with

traits of temperament (TCI; Cloninger 1998). Novelty

seeking, for example, concerns the tendency to actively

respond to new stimuli and thus reflects elements of

impulsivity and sensation seeking. Further, the SURPS

personality traits are relevant for more neuropsychological

orientations. Different reinforcement processes are as-

sumed to mediate the relationship between the SURPS

personality traits and substance use in that the personality

traits are susceptible to different types of reinforcement

(e.g., Brunelle et al. 2004; Conrod et al. 1998). Individuals

with high levels of anxiety sensitivity or hopelessness are

more sensitive for the negative reinforcement processes of

substance use (i.e., the ability of substances to relieve

negative affective states). Individuals who score high on

sensation seeking and impulsivity on the other hand are

more sensitive for the positive reinforcement processes of

substance use (i.e., the positive hedonic effects of a sub-

stance).

According to Carver et al. (2009) these processes are

even more apparent in case of low serotonergic function.

It is argued that individual differences in serotonergic

function are important for personality dispositions in that

individuals with low serotonergic function are especially

susceptible for (affective) cues of the moment (Spoont

1992), like reinforcement processes. In accordance, low

serotonergic function is related to personality dispositions

as sensation seeking, impulsivity, and depression (Carver

et al. 2009). Considering the possible contribution of the

SURPS to many different fields (e.g., neuropsychology),

the fact that a more clinical orientation (i.e., the use of

more clinical instruments like the TCI) seems less obvi-

ous for early adolescents who are in the beginning stage

of substance use, and bearing in mind that specific rather

than general personality traits are most interesting, the

SURPS is a potentially important measurement for

examining the role of personality on substance use

behaviors.

Recall that the SURPS-based personality profiles are

useful in identifying individuals who are at risk for

alcohol use and alcohol-related problems in already

using samples. However, to our knowledge no previous

study examined whether these personality profiles are

indicative of an early onset of alcohol, tobacco, canna-

bis, and poly substance use. This is unfortunate, because

on the one hand early initiation is one of the strongest

identified risk factors for alcohol (De Wit et al. 2000),

tobacco (Breslau et al. 1993), and cannabis problems

(Chen et al. 2005) in later life. Further, poly substance

use in adolescence is a significant predictor of poly

substance use in adulthood (Galaif and Newcomb 1999).

On the other hand, the developmental role of personality

dispositions is important. The lower order personality

dispositions might be overruled by higher order systems

(i.e., rational or cognitive), but only if and once the

capacity for behavioral control develops (i.e., through

maturation of the pre-frontal cortex; Carver et al. 2009).

Thus, one might argue that especially early adolescents

are vulnerable for these lower order personality predis-

positions. To conclude, focusing on early onset of sub-

stance use in early adolescence, and identifying the

specific personality profiles related to these risk behav-

iors, might help us to identify youngsters at an early age

who are at risk for developing future substance misuse

patterns.

The present study examines a SURPS-based, four-factor

personality model in relation to early onset substance use

and poly substance use. A total of 3,783 adolescents in the

ages of 11–15 participated in the first wave of the ongoing

Healthy School and Drugs (HSD) effectiveness study in

which they filled out a digital questionnaire. Participants

J Behav Med (2010) 33:474–485 475

123

answered questions about alcohol, tobacco, and cannabis

use and their personality traits. Based on previous

research on personality, we expected to find strongest

associations with substance use for sensation seeking.

Specifically, we hypothesize sensation seekers to have an

increased risk for an early initiation of alcohol, tobacco,

and cannabis use. Hence, we expected to find that anxiety

sensitive adolescents have an increased risk for an early

onset of alcohol use, adolescents reporting higher levels

of hopelessness to have an increased risk for an early

onset of alcohol and tobacco use, and impulsive adoles-

cents to have an increased risk for an early onset of

alcohol and cannabis use. Following these expectations

we also expected to find associations between the SURPS

personality profiles and poly substance use. However,

considering the lack of knowledge so far in adolescence,

no concrete expectations were formulated on poly sub-

stance use.

Method

Sample and procedure

The cross-sectional data for this study were collected as

part of a broader effectiveness study on a national school

prevention program ‘‘The Healthy school and drugs.’’ A

total of 23 schools were included from seven regions in

The Netherlands. We visited participating schools and

during these visits we provided further information about

the research project. In collaboration with the schools’

headmasters, we informed the students’ parents about the

goals of the study by a letter in which parents were also

explained they could refuse participation of their child in

the study. Approval for the design and data collection

procedures was obtained from the ethic committee of the

Radboud University Nijmegen. All data were collected

between January and March 2009. All first grade students

independently filled out a digital questionnaire during

school hours in the presence of a teacher and a research

assistant. The questionnaires were counterbalanced on

alcohol, tobacco, and cannabis, thus six different versions

were administrated.

In total, 3,783 first-grade students took part in the study

of whom 231 (6.1%) were absent (i.e., illness) during data-

collection and three participants were declined participa-

tion by their parents. The total sample included 1,856 boys

(49.1%) and 31.5% (n = 1,192) of all participants pursued

lower secondary vocational education, 46.6% (n = 1,764)

pursued pre-university education, and 21.9% (n = 827) of

the students pursued a mixed educational program. Of the

participants who completed the questionnaire 3,375 par-

ticipants (96.2%) were of Dutch ethnic origin. Students

ranged in age from 11 to 15 years (M = 13.01, SD = .49).

For the question on lifetime prevalence of alcohol use,

2,103 (59.9%) reported to have at least once used alcohol

in the past. With regard to smoking, 768 (22.1%) partici-

pants had ever smoked, and with regard to cannabis 75

(2.1%) participants reported to have at least once used

cannabis. Finally, 670 (19.6%) stated that they already had

tried more than one substance.

Measures

Personality profiles

The Substance Use Risk Profile Scale (SURPS; Woicik

et al. 2009) distinguishes four personality dimensions,

namely anxiety sensitivity, hopelessness, sensation seek-

ing, and impulsivity. Each dimension was assessed using

five to seven items that could be answered on a 4-point

scale, ranging from 1 = ‘strongly agree’ to 4 = ‘strongly

disagree.’ Anxiety sensitivity refers to the fear for physical

arousal and an example item is: ‘It’s frightening to feel

dizzy or faint.’ Hopelessness concerns negative thinking

which might lead to depression proneness and ‘I feel that

I’m a failure’ is an example item. Sensation seeking is

characterized by wanting to try out new things and an

example of such an item is ‘I like doing things that frighten

me a little.’ Finally impulsivity refers to having difficulties

in controlling behavioral responses, and ‘I usually act

without stopping to think’ is an example item. Factor

structure, internal consistency and test–retest reliability, as

well as construct, convergent, and discriminant validity of

this instrument were shown to be adequate in studies

among college students and adult samples (e.g., Krank

et al. submitted). Because the instrument was translated in

Dutch and used for the first time the factor structure was

examined using Exploratory Factor Analysis (EFA) on a

randomly selected sample that consisted of the first half of

the original sample using Mplus (Muthén and Muthén

1998–2007). The Weighted Least Square parameter esti-

mator with Mean- and Variance adjusted chi-square test

statistic (WLSMV) was used because the metric of the

items is more ordered categorical than interval level. The

sample was randomly divided into two subsamples. Two

items were removed. The first item (i.e., I feel that I’m a

failure) had substantial loadings (.38 and .42, respectively)

on the factors anxiety sensitivity and hopelessness. The

second item (i.e., I feel I have to be manipulative to get

what I want) showed an almost zero loading on the factor

impulsiveness. A Confirmatory Factor Analysis (CFA) was

performed on the remaining 21 SURPS items on the other

half of the sample and confirmed the four-factor structure

of the SURPS. The final model had a satisfactory fit to

the data (v2 (54) = 611.315, P \ .001, RMSEA = .055,

476 J Behav Med (2010) 33:474–485

123

CFI = .943). Cronbach’s alphas were .69 for anxiety

sensitivity (factor loadings between .42 and .72), .85 for

hopelessness (loadings between .72 and .96), .68 for sen-

sation seeking (loadings between .38 and .72), and .67

for impulsivity (loadings between .48 and .72). These

reliability estimates converge with those from previous

research (e.g., Jaffee and D’Zurilla 2009) and are satis-

factory for short scales (Loewenthal 1996).

Substance use

We assessed adolescents’ alcohol use in terms of lifetime

prevalence, or whether participants had ever consumed

alcohol in their life. Lifetime prevalence was measured by

asking: ‘‘Have you ever drunk alcohol?’’ Participants could

answer this question with yes (=1) or no (=0). To determine

the age of onset of participants’ alcohol use we asked how

old they were when they had first drunk alcohol (Kuntsche

et al. 2009).

Lifetime prevalence of tobacco use was measured by a

single item on a 9-point scale ranging from 1 = ‘I never

smoked, not even a puff’ to 9 = ‘I smoke at least once a

day’ (Kremers et al. 2001). To tap lifetime prevalence of

smoking, adolescents who responded in the categories 2–9

were categorized as tried smoking before (=1), and the

adolescents who responded in category 1 were categorized

as never tried smoking (=0) following Kremers (2002). In

order to assess age of onset, participants who had ever

smoked were asked how old they were when they smoked

their first puff.

We assessed the lifetime prevalence of cannabis use

through a single item, namely: ‘‘Have you ever used can-

nabis?’’ (Monshouwer et al. 2005). Participants could an-

swer with yes (=1) or no (=0). Subsequently, participants

who ever used cannabis were asked how old they were

when they first used cannabis.

Finally, poly substance use was operationalized by the

use of more than one substance, regardless of the combi-

nation or amount of substances used. A new variable was

created in which all adolescents who used more than

one substance were categorized as poly substance users

(=1) and all other adolescents as non-poly substance users

(=0).

Strategy of analyses

First, descriptive analyses and Pearson correlations of age

of onset of alcohol, tobacco, and cannabis use and the

personality profiles (i.e., anxiety sensitivity, hopelessness,

sensation seeking, and impulsivity) were calculated

between model variables. Second, to investigate whether

participants’ sex and educational level should be specified

as covariates in the model, a MANOVA was conducted to

compare responses on the SURPS personality profiles be-

tween males and females and between different educational

levels. Another MANOVA was carried out to investigate

sex and educational differences on substance use. Also,

separate ANOVA’s were conducted to examine sex and

educational level differences on age of onset of alcohol,

tobacco, and cannabis use. The effect sizes (i.e., partial eta

squared) are reported for the analyses of variance. With

respect to the effect size, values around .02 are considered

small effects, values around .15 medium effects, and values

around .35 large effects (Cohen 1992). Post-hoc tests with

Bonferroni corrections were carried out to investigate the

significant differences in educational level on the outcome

variables.

Next, to investigate the relationships between person-

ality profiles and lifetime prevalence of alcohol, tobacco,

and cannabis use, we specified and tested a first model

(see Fig. 1) with structural equation modeling (SEM) in

Mplus (Muthén and Muthén 1998–2007). In this model,

lifetime prevalences of alcohol, tobacco, and cannabis

were included as observed variables and personality

profiles were added as latent constructs, with separate

scale items as indicators. Sex and educational level were

specified as covariates in the model. We used the

weighted least square method (WLSMV) to estimate

parameters in the model. The Chi-square and the p-value,

the Comparative Fit Index (CFI: Bentler 1989), and the

Anxiety Sensitivity

Hopeless- ness

Sensation Seeking

Alcohol use

Tobacco use

Cannabis use