Addictive Behaviors, Vol. 25, No. 5, pp.

683691, 2000
Copyright 2000 Elsevier Science Ltd.
Printed in the USA. All rights reserved
0306-4603/00/$see front matter

Pergamon

PII S0306-4603(00)00065-4

SCREENING OF DRUG USE IN A TEENAGE BRAZILIAN SAMPLE

USING THE DRUG USE SCREENING INVENTORY (DUSI)
DENISE DE MICHELI and MARIA LUCIA O. S. FORMIGONI
Federal University of So Paulo (UNIFESP)

Abstract The Brazilian translation of the Drug Use Screening Inventory (DUSI) was applied to 213 Brazilian teenagers who were classified according to their alcohol and/or drug dependence level (DSM-III-R) as: 71 nondrug users (Group 1), 71 with light/moderate dependence (Group 2) and 71 with severe dependence (Group 3). The DUSI was applied and the
absolute density in each of 10 areas was calculated. The three groups presented statistically
significant differences (p .001) in the substance use area, with the following values (medians interquartile range): Group 1: 0 7; Group 2: 20 33 and Group 3: 80 33. The
groups also presented significant differences in behavior pattern, social competency, family
system, work adjustment, peer relationships and leisure/recreation. Other differences detected among the groups indicated an important relationship between drug use and school delay. A good Spearman rank correlation (0.86, p .0001) was observed between Composite International Diagnostic Interview (CIDI) diagnosis and DUSI, indicating that this instrument
can be useful in the screening of substance use among Brazilian teenagers. 2000 Elsevier
Science Ltd.

Key Words. Adolescence, Drug use, Screening instrument.

Drugs have always fascinated young people with their ability to induce pleasure, new
sensations, and sometimes relief from unpleasant emotional states. Analyses of personal drug histories of drug dependent patients shows that, in most cases, the initial
consumption occurred in early adolescence or even childhood (Duncan, Tildesley,
Duncan, & Himan, 1995; Sloboda & David, 1997).
Needle, McCubbin, Wilson, Reineck, and Lazar (1986) reported that the most common age for initial drug use was about 10 to 11. In Brazil, data from a national survey
of 15,503 1012 year-old students indicated that 51.2% had used alcohol, 11% had
used tobacco, and 11.7% had used other drugs (Galdurz, Noto, & Carlini, 1997).
According to Porter-Serviss, Opheim, and Hindmarsh (1994) treatment success depends on how early intervention occurs. This highlights the importance of having good
screening and diagnosis instruments available. Although in developed countries there
are many standardized instruments specifically designed for this purpose, there are
few translated and validated instruments in Brazil. In particular there are no instruments available to screen substance use and related problems in teenagers, except for
a general instrument proposed by the World Health Organization (1987) (Smart et al.,
1980) and adapted by Carlini-Cotrim, Carlini, Silva-Filho, and Barbosa, (1989). The
need for such instruments has become particularly pressing given the increase in the
involvement of Brazilian teenagers with alcohol and drugs.
Since screening instruments are aimed at identifying individuals who may present
problems related to the use/abuse of alcohol and other drugs, sensitivity is a key criteThis research was supported by AFIP (Associao Fundo de Incentivo Psicofarmacologia), CNPq and
FAPESP (98/07512-5).
Requests for reprints should be sent to Maria Lucia O. Souza Formigoni, Department of Psychobiology,
Federal University of So Paulo, Rua Botucatu 862, 1 andar, So Paulo, SP, Brazil 04023-062; E-mail:
mlformig@psicobio.epm.br
683

684

D. DE MICHELI and M. L. O. S. FORMIGONI

rion. This is in contrast to diagnostic instruments, where high specificity is a fundamental attribute. Ideally, screening instruments should be easy to apply and should
not require highly trained staff.
Tarter (1990) proposed the use of the Drug Use Screening Inventory (DUSI), made
up of 149 questions on 10 areas, aimed at the detection of abusive use of alcohol and
other drugs, as well as related problems. This instrument was validated (Tarter, Laird,
Bukstein, & Kaminer, 1992; Tarter, Mezzich, & Kirisci, 1994) in a sample of 846
American teenagers including users, nonusers, and psychiatric patients. The average
internal reliability across the 10 domains was found to be 74% for males and 78% for
females. Testretest reliability averaged 88% for females and 95% for males (Kirisci,
Mezzich, & Tarter, 1995). The test correctly identified between 68% and 86% of the
users. This instrument was developed to be self-applied, taking between 15 and 20
minutes to complete.
Tarter (1990) proposed the utilization of the DUSI as a screening instrument for alcohol and/or drug use among teenagers, suggesting the use of the Absolute Density of
Problems to investigate the percentage of problems in each specific area. In addition
to being an efficient method to screen teenagers who might require treatment, the
DUSI can also provide important data for therapeutic planning and follow-up,
thereby aiding in matching therapeutic modality and patient profile.
O B J E C T I V E

The present work sought to compare data obtained with the DUSI, after its translation and adaptation to the Brazilian population, in three samples of young individuals
diagnosed, respectively, as non-drug dependent, lightly/moderately drug dependent,
and severely drug dependent.
M E T H O D O L O G Y

Subjects: Teenagers were invited to participate in the study either by being directly
approached by the researchers in public places, or in their place of work, study, or
treatment, or indirectly, at the invitation of other participants (snowball technique).
Inclusion criteria were: being between 11 and 19 years of age and having had at least 4
years of formal education. After being informed of the objectives of the study and assured of the confidentiality of the data collected, all the participants signed a consent
form approved by the Committee of Medical Ethics in Research of the Federal University of So Paulo (UNIFESP). The application of the interview and instruments
was always performed in an isolated place, where only the interviewer and the interviewee were present. First the interviewer applied a standardized interview for the
collection of social and demographic data. The interviewee was then left alone to fill
out the DUSI. After the instrument was completed, the interviewer applied the Brazilian version of the alcohol, tobacco, and drugs sections of the Composite International
Diagnostic Interview (CIDI) (Wittchen et al., 1991). The answers to this instrument
were entered into the CIDI software which classified them, using the Diagnostic and
Statistical Manual of Mental Disorders (DSM-III-R; American Psychiatric Association, 1987) criteria, into three groups:
Group 1: 71 teenagers who had never or rarely used alcohol and/or drugs and did not
meet DSM-III-R criteria;

Screening of drug use

685

Group 2: 71 teenagers who met DSM-III-R criteria for abuse or light/moderate dependence;
Group 3: 71 teenagers who meet DSM-III-R criteria for severe dependence; of these,
34 were under treatment.
Two hundred thirteen teenagers, out of 225 invited, participated in the study, giving
71 volunteers in each of the above-mentioned groups. Most of those invited to participate but who did not agree to do so, justified their decision by time schedule problems.
Instruments
DUSI (Tarter, 1990, prototype version)Made up of 149 questions on 10 areas:
Substance use; Behavior pattern; Health status; Psychiatric disorders; Social competency; Family system; School adjustment; Work adjustment; Peer relationships; and
Leisure/Recreation. The questions in the DUSI are answered with Yes or No,
with affirmative answers indicating the presence of problems. Three indices were calculated after the application of the DUSI: Absolute Density of Problems, Corrected
Relative Density of Problems, and Global Density of Problems.
The Absolute Density of Problems is a measure of the severity of problems in each
specific area. It is calculated by dividing the number of affirmative answers in each
area by the total number of questions in the area and multiplying by 100. The percentage obtained represents the severity of the problems in the area evaluated.
The Corrected Relative Density of Problems is our adaptation of the index proposed by Tarter (1990); it is obtained by dividing the absolute density of each area by
the sum of the absolute densities of all the areas and multiplying by 100. This percentage represents the contribution of each area to the total of problems.
The Global Density of Problems is the sum of the affirmative answers in all the areas divided by the total number of questions, multiplied by 100. This percentage is the
indicator of the severity of problems in general.
The adaptation of the DUSI proceeded through the following steps: translation of
the questionnaire; application to a pilot sample of 30 teenagers aimed at investigating
the intelligibility of the document; alteration of those questions which presented intelligibility problems. The author of the original version approved the final version of the
DUSI in Portuguese, as well as its back-translation into English.
CIDIBrazilian version, previously translated (Wittchen et al., 1991)sections on
Tobacco, Alcohol and Drugs, with the diagnosis established according to the DSMIII-R criteria (American Psychiatric Association, 1987; Cottler et al., 1991; Miranda,
Mari, Ricciardi & Arruda, 1990). The initial interview lasted for 3 minutes. The volunteers needed an average of 15 to 20 minutes to fill out the DUSI and 35 minutes to answer the questions of the CIDI.
Statistical Analysis
A comparison of the medians of the density indexes of the DUSI between the three
groups was performed by the Kruskal-Wallis analysis of variance, followed by MannWhitneys test. Nonparametric tests were used as most of the variables did not present
normal distribution. A 5% significance level was adopted. The correlation between
DUSI areas was tested by Spearmans correlation test.
Results
Table 1 summarizes the social and demographic characteristics of the three groups.
Some significant differences were observed between the group with severe depen-

686

D. DE MICHELI and M. L. O. S. FORMIGONI

dence (Group 3) and the others, in the variables currently studying and school delay. Group 3 included a higher percentage of school dropouts. Some differences were
also observed in social and economic level. A higher proportion of the youths in
Group 2 (abuse or light/moderate dependence) belonged to social/economic classes C/D
than in the others. The nonusers group and the group with severe dependence presented similar profiles.
Figure 1 refers to the Absolute Density in the various areas evaluated by the DUSI.
The area substance use clearly differentiated the three groups. Group 2 (users with
a diagnosis of abuse or light/moderate dependence) presented scores with a median of
20, that is, about 20 times higher than Group 1 (nonusers). Those with severe dependence scored 80 times higher than those without dependence, and four times higher
than the light/moderate dependence group.
Groups 1 and 2 presented similar Absolute Density of Problems in the areas of
health and psychiatric disorders with median levels half of those presented by Group
3. The scale of family system also detected a marked difference between the groups,
mainly when those with severe dependence were compared to the other groups. On
the school adjustment area, Groups 2 and 3 presented median levels about 2 to 3
times higher than the nonusers group. Regarding the work adjustment, peer relationships, and leisure/recreation areas, the three groups differed, with problems increasing in proportion to drug use. No differences were observed among the groups
regarding social competency.
The nonusers group presented median levels similar to those with light/moderate
dependence in the area behavior pattern, and both presented significantly lower
levels than the group with severe dependence.
Table 1. Social and demographic data of the three groups studied for alcohol/drug use

Group 1
Nonusers
(n 71)
Age (mean SD)
Sex (%)
Male
Female
Marital status (%)
Single
Married
Currently studying (%)
School grade (%)
1st to 8th grade
High school
College
School delay (%)
1 to 2 years
2 years
Currently working (%)
Social and economic
classification (%)
A/B (upper class)
C/D (middle class)
E (lower class)

Group 2
Abuse/light/
Group 3
moderate
Severe
dependence dependence
Total
(n 71)
(n 71)
(n 213)

2
(p)

Differences
between
groups
(p .05)

120.50 (0.0001)

1 (2 3)

15 2

16 2.5

16 1.5

42
58

46
54

58
42

49
51

3.63 (0.162)

ns

100
0
90

94
6
85

100
0
54

98
2
76

8.15 (0.017)

2 (3 1)

30.3 (0.0001)

123

39
52
9

41
45
14

60
40
0

44.5
47
8.5

13
3
23

47
17
35

32
42
41

30
17
33

53.2 (0.0001)

123

5.66 (0.59)

ns

71
25
3

36
63
0

60
24
15

56
37
6

54.4 (0.0001)

(1 3) 2

ns nonsignificant differences.

ns

Screening of drug use

687

Fig. 1. Absolute Densities in the 10 areas and Global Density percentages (median
interquartil range) in the three samples analyzed. *Differs from the other two groups (p
0.05). Group 1Nonusers;
Group 2Abuse or light/moderate dependence; Group
3Severe dependence.

The analysis of the Global Density of Problems detected by the DUSI revealed significant differences among the three groups. The median of global density was 19 in
the nonusers group, 25 in the group with light/moderate dependence, and 55 in the
group with severe dependence. It can thus be observed that the global density of

688

D. DE MICHELI and M. L. O. S. FORMIGONI

Group 3 was three times higher when compared to Group 1.

For the Corrected Relative Density of Problems (Figure 2), the areas with higher
levels of problems in the nonusers group were: leisure/recreation, peer relationships,
behavior pattern, social competency, psychiatric disorders, and school adjustment. In
this group, the area of family system was estimated to account for 7% of the total of
problems, followed by health status, substance use, and work adjustment. The order of
importance of problems was different in the group with light/moderate dependence,
where the most important problems were peer relationships, school adjustment, leisure/recreation, social competency, psychiatric disorders, and behavior pattern. Similar to the results for Group 1, family system accounted for 8% of the total of problems.
The least affected areas in Group 2 were health status and work adjustment. Group 3
(severe dependence) presented a more homogeneous distribution of problems: first
came substance use, followed by leisure/recreation, peer relationships, family system,
behavior pattern, psychiatric disorders, health status, school adjustment, and work adjustment. The least affected area was social competency.
The three groups differed between themselves in the absolute densities in most of
the areas studied (p .05), except in the area of social competency. A good correlation was observed between the diagnosis of abuse/dependence provided by the CIDI
and the scores in most of the areas of the DUSI, as can be seen in Table 2. The highest

Fig. 2. Relative Density (medians) of the 10 areas in the three groups of teenagers. Significant
differences were detected (see text) between nonusers and the other groups in substance use,
social, family, psychiatric, and behavior areas. The teenagers with severe dependence differ
from those with abuse/moderate dependence regarding peer relationship, school, social,
health, and substance use areas.

Screening of drug use

689

correlations were in the areas substance use, peer relationships, and leisure/recreation. It can be observed that there is a good correlation between severity of dependence and DUSI substance use area scores.
D I S C U S S I O N

The DUSI proved sensitive enough to distinguish between the three samples of
Brazilian teenagers, not only in relation to drug use but also in relation to levels of associated problems in the behavior, leisure/recreation, and peer relationships areas.
The groups did not differ regarding social competency, suggesting that drug use does
not affect sociability, or is acceptable to peer groups. In two other areas (health status
and psychiatric disorders) the drug user groups (2 and 3) were similar between themselves but different from Group 1, indicating that even an initial drug involvement
may affect these areas or that previous psychiatric disorders are associated with drug
abuse. However, these preliminary data do not make it possible to determine whether
these differences were pre-existing or attributable to the drug use.
The observation that the drug user groups presented higher levels of problems than
the nonusers group could suggest that drug consumption affects these areas, or that
pre-existing differences in these areas could cause a predisposition to drug abuse. Prospective studies are necessary to establish how and if these factors influence the development of drug dependence.
The analysis of the Global Density of Problems indicated that the nonusers group
had less intense problems than the others. As the level of dependence increased, the
relative density scores in the school adjustment, family system, and work adjustment
areas increased, indicating that these are the main areas affected by drug use, or that
problems in these areas are related to substance use.
Even though the samples studied may not be representative of the Brazilian teenage
population, it is possible to compare them to American samples to which the instrument
has also been applied. Kirisci et al. (1995) studied a sample of 846 American teenagers
between 12 and 18 years of age, in which 259 were users of psychoactive substances, 278
were nonusers or occasional users, and 309 had received a diagnosis of psychiatric disorders. These authors reported mean levels of Global Density of Problems close to 20%,

Table 2. Correlation between the diagnosis of severity by the

DSM-III-R and the Absolute Densities in each area
Absolute density areas
Substance use
Behavior pattern
Health status
Psychiatric disorder
Social competency
Family system
School adjustment
Work adjustment
Peer relationships
Leisure and recreation
Global density

Spearmans r

.86
.30
.47
.55
.02
.55
.32
.51
.72
.70

.0001
.0001
.0001
.0001
.7329
.0001
.0002
.0001
.0001
.0001

.77

.0001

The correlations were calculated using 213 cases.

690

D. DE MICHELI and M. L. O. S. FORMIGONI

very similar to the levels observed in our sample of nonusers. On the other hand, the
global density of the DUSI was 45% in the American sample of psychoactive substance
users. If we consider our total sample of users (Group 2 Group 3, i.e., light/moderate
and heavy users), the median global density is 42%, which is very close to the figure reported by Kirisci et al. (1995). In terms of Absolute Density of Problems, Kirisci et al.
observed mean levels of 5.5% in the nonusers group and 54.5% in the group of users.
These levels were, respectively, 4.5% and 53.5% in our Brazilian sample. As far as the
other areas are concerned, the levels were also very similar. Together, these data indicate that the Brazilian version of the instrument yielded very similar results to those observed in the American sample, showing that the two populations are relatively similar
and that the instrument kept its properties after the translation into Portuguese.
Considering the similarity of the samples, if we adopted the same cutoff point used
by those authors (absolute density 27% in the scale of substance use), the DUSI
would correctly classify 97% of those without dependence or nonusers, and 71% of
those with dependence on drugs.
The finding that the results closely parallel U.S. adolescents is intriguing from a
cross-cultural perspective. Many reasons could explain it. The globalization of information has promoted similar behavior patterns in many countries. In Brazil, American
culture has a strong influence, mainly on youths. Most of the movies and shows on TV
and in theaters, as well as music, clothes fashion, etc., come from the United States.
Many Brazilian teenagers, mainly those from the upper and middle classes, study English at regular or specialized schools and are very familiar with American values.
The good level of correlation between the substance use area and the DSM-III-R
diagnosis (0.86) suggests that the use of the DUSI as a diagnosis instrument should be
considered. Since the DUSI was developed based on the dependence syndrome criteria, the same construct is present in both instruments (DUSI and CIDI). Tarter,
Kirisci, and Mezzich (1996) have suggested that it is important that studies about the
predictive validity as well as temporal stability of DUSI scores are conducted.
It is possible to conclude that the DUSI can be useful in screening young Brazilians
with substance use problems. Its modular structure also allows for the isolated use of
the scale of substance use, which makes it an instrument of quick application in this
case (about 3 minutes). Even though it was individually applied in this study, there are
reports of its application to groups with no impairment in results (Kirisci et al., 1995). A
similar study is under way with a sample of 2,000 students from a school in So Paulo.
In addition to its use as a screening instrument, the DUSI can also be useful in therapeutic planning and follow-up. The various scales allow for identification of areas
that require more attention, making it possible to rank problems, and it can be used
after treatment or preventive intervention for medium and long-term follow-up, contributing to the process of outcome evaluation.
R E F E R E N C E S
American Psychiatric Association. (1987). Diagnostic and statistical manual of mental disorders (3rd ed.,
rev.). Washington, DC: Author.
Carlini-Cotrim, B., Carlini, E. A., Silva-Filho, A. R., & Barbosa, M. T. (1989). O uso de drogas psicotrpicas
por estudantes de primeiro e segundo graus da rede estadual, em dez capitais brasileiras, 1987. In Consumo de drogas psicotrpicas no Brasil em 1987 (pp. 984). Centro de Documentao do Ministrio da
Sade (Srie C: Estudos e Projetos 5).
Cottler, L. B., Robins, L. N., Grant, F., Blaine, J., Towle, L. H., Wittchen, H. U., & Sartorius, N. (1991). The
CIDI-core substance abuse and dependence questions: Cross-cultural and nosological issues. The
WHO/ADAMHA Field Trial. British Journal of Psychiatry, 159, 653658.

Screening of drug use

691

Duncan, T. E., Tildesley, E., Duncan, S. C., & Himan, H. (1995). The consistency of family and peer influences on the development of substance use in adolescence. Addiction, 90, 16471660.
Galdurz, J. C. F., Noto, A. R., & Carlini, E. A. (1997). IV Levantamento sobre Uso de Drogas entre
Estudantes de 1 e 2 graus em 10 Capitais Brasileiras. So Paulo, Brasil: Centro Brasileiro de Informaes sobre Drogas Psicotrpicas (CEBRID).
Kirisci, L., Mezzich, A., & Tarter, R. E. (1995). Norms and sensitivity of the adolescent version of the drug
use screening inventory Addictive Behaviors, 20, 149157.
Miranda, C. T., Mari, J. J., Ricciardi, A., & Arruda, M. E. (1990). Patient reactions to the CIDI in Brazil. In
C. N. Stefanis et al. (Eds.), Psychiatry: A world perspective: Vol. 1 (pp. 133137). Athens, Greece:
Elsevier Science Publishers B.V. Proceedings of the VIII World Congress of Psychiatry.
Needle, R., McCubbin, H., Wilson, M., Reineck, R., & Lazar, A. (1986). Interpersonal influences in adolescent drug usethe role of older siblings, parents, and peers. International Journal of the Addictions, 21,
739766.
Porter-Serviss, S., Opheim, E. E., & Hindmarsh, W. (1994). Perceptions and attitudes with respect to drug
use among grades 4 to 6 students: 1992. International Journal of the Addictions, 29, 225233.
Sloboda, Z., & David, S. I. (1997). Preventing drug use among children and adolescents [Brochure] (National
Institute on Drug Abuse Publication No. 97-4212). Bethesda, MD: National Institute on Drug Abuse.
Smart, R. G., Hughes, D. H. P., Johnston, L. D., Anumonye, A., Khant, U., Medina Mora, M. E., Navaratnam, V., Poshya-Chinda, V., Varma, V. K., & Walud, K. A. (1980). A methodology for students druguse surveys. Geneva, Switzerland: World Health Organization.
Tarter, R. E. (1990). Evaluation and treatment of adolescent substance abuse: A decision tree method.
American Journal of Drug and Alcohol Abuse, 16, 146.
Tarter, R. E., Kirisci, L., & Mezzich, A. (1996). The Drug Use Screening Inventory: School adjustment correlates of substance abuse. Measurement and Evaluation in Counseling and Development, 29, 2534.
Tarter, R. E., Laird, S. B., Bukstein, O., & Kaminer, Y. (1992). Validation of the Drug Use Screening Inventory: Preliminary findings. Psychology of Addictive Behaviors, 6, 233236.
Tarter, R. E., Mezzich, A., & Kirisci, L. (1994). Reliability of Drug Use Screening Inventory in adolescent
alcoholics. Journal of Child and Adolescent Substance Abuse, 3, 2536.
Wittchen, H. U., Robins, L. N., Cottler, L. B., Sartorius, N., Burke, J. D., & Regier, D. (1991). Cross cultural
feasibility, reliability and sources of variance of the Composite International Diagnostic Interview
(CIDI). The Multicentre WHO/ADAMHA, Field Trials. British Journal of Psychiatry, 159, 645653.
World Health Organization. (1987). Composite International Diagnostic Interview, CORE VERSION.
Geneva: Author.