Prevalence of Personality Disorders Among

Clients in Treatment for Addiction

Louise Nadeau, PhD*, Michel Landry, PhD^ Stephane Racine, MPs'

This study determined the prevalence ofpersonality disorders among clients in treatmentfor addiction; this prevalence was compared with thosefound in similar studies and in clinical samples of individuals sufferingfrom other
Axis I disorders. Our sample comprised 255 subjects. The first edition of the Millon Clinical Multiaxial Inventory
(MCMI) was used. Only 11.8% of the subjects did not score over 84 on any of the 11 Axis II scales. Over one-half
had a score of 84 or higher on the passive-aggressive and dependent-personality scales. The mean number of
scales in the 84+ category was 2.68. Comparisons show that this sample was more severe in most cases.
(Can J Psychiatry 1999;44:592-596)
Key Words:personality disorder, substance disorder, Millon Clinical Multiaxial Inventory, treatment, gender

Objectives

ersonality disorders (PDs) among clients with substance

disorders (SDs) have been less studied than have Axis I
disorders, but instruments that are valid and inexpensive to
administer, such as the Millon Clinical Multiaxial Inventory
(MCMI) (1) used in the present study, have allowed their
prevalence to be documented.

This study determines the prevalence of PDs among individuals admitted to public treatment centres in Quebec and compares these results with findings of similar studies and with
those of clinical samples suffering from other Axis I disorders.

Clinical studies have found that the proportion of individuals

presenting both SDs and PDs varied from 53% to 100%
(2-10). The most frequently diagnosed disorders are antisocial, borderline, narcissistic, and dependent personality disorders. The simultaneous presence of Axis I and II disorders
increases the severity of all disorders (11,12). In cases of Axis
II comorbidity, addiction can be successfully controlled in
specialized treatment centres. However, the needs of these
clients often exceed the capabilities of the services normally
provided (13,14). There is an increased likelihood of premature termination and reduced effectiveness of treatment
(15-17). For those who complete treatment, remission is
comparable to that observed among those without PDs, although the level of psychological distress remains high (18).

Method
Our sample comprised 255 clients182 men and 73
womenin 8 SD treatment centres in Quebec. The subjects
were francophone and aged 18 years or over. The mean age
was 34.5 years: 12% aged 18-24 years; 41% aged 25-34
years; 34% aged 35-44 years; and 13% aged 45 years or
older. Testing took place after 7 days of treatment in order to
exclude individuals suffering from acute withdrawal. Participation was voluntary, and all subjects signed a consent form.
We compared our results with those of 5 studies (4,8,19-21)
that had used the MCMI-I (Table 1).
To better understand the specificity of SD, our results were
also compared with those of 3 clinical samples from Quebec.
The first comprised 180 subjects treated for erectile or orgasmic disorders. The other 2 samples included individuals
treated for sexual impulse problems at a medicolegal clinic:
one comprised 44 men who had committed a rape or had intrusive rape fantasies, and in the other were 87 men who had
committed pedophilic acts. All subjects were francophone.
All subjects signed a consent form.

Manuscript received April 1998, revised, and accepted November 1998.

'Associate professor, Ddpartement de psychologite. University de Montreal,
Montreal, Quebec.
^Director of Professional Services, Centre Dollard-Connier, Montreal, Quebec.
'Scientist, Recherche et Intervention sur les Substances psychoactives - Quebec (RISQ), Montreal, Quebec.
Address for correspondence: Dr L Nadeau, D^partement de psychologie.
University de Montreal, CP 6128, succ Centre-ville , Montreal, QC H3C
3J7
email: louise.nadeau.2@umontreal.ca

We used the first edition of the MCMI (1), a self-report inventory asking 175 true or false questions. Our results focused on
the 11 personality scales (PSs) described in Table 2. The
MCMI was translated into French (22). Based on

CanJ Psychiatry, Vol 44, August 1999

592

August 1999

Personality Disorders and Addiction

Table 1. Comparison of 5 studies that used the Millon Clinical

Multiaxial Inventory (MCMI-I)
Author

Diagnosis at
admission

Type of
treatment

Time of testing

Ness and
others (20)

30

Opiate abuse
or dependence

Brown (19)

50

All substance
dependencies

12-step
treatment from
21-30 days
detoxification

After 21-30
days of
treatment

Craig and
others (4)

86

Opiate abuse
or dependence

Detoxification

After
detoxification

Craig and
others(4)

107

Cocaine abuse

Detoxification

After
detoxiHcation

Marsh and
others (8)

159

Craig and
others (21)
Craig and
others (21)

Axis 11
No score +
Schizoid
Avoidant

After 48-72
hours of
abstinence

593

MCMI validity index (1 or more on the Y scale); 9 were rejected because of exaggeration of symptoms (sum of scales 1
to 8 greater than 164); 6 were rejected because of missing
data; 3 were rejected because the subjects were under age 18
years. In the other clinical samples, 25 tests (8%) were rejected.
Statistical tests designed to compare means and proportions
were used. The types of tests used are specified below. Differences between genders were calculated when the number of
subjects allowed it.
Results

The first column of Table 2 shows the results by scale. Subjects are classified in this category for each scale on which
their score is 84+. Only 11.8% did not score over 84 on any
scale, over one-half scored on the passive-aggressive (56.9%)
106
Alcohol abuse
and dependent (52.9%) PS, and 0% scored on the compulsive
or dependence
PS. For the 74 cutoff point, the third column of Table 2 shows
Opiate abuse
100
98.8% of subjects to be above this threshold. Over 50%
or dependence
scored on either the passiveaggressive, dependent, avoidant, borTable 2. Millon Clinical Multiaxial Inventory base rate score!
derline, or schizoid PS. None scored
By scale
Highest score
Average score
more that 74 on the compulsive PS.
Base rate score > 84
Base rate score > 74
Significantly more women than men
scored 84+ on the histrionic, schizo%
Rank
%
Rank
%
Rank
Rank
%
typal,
borderline, and paranoid PS.
11.8
1.2
0

11.8

Using the 74 cutoff point, significant

33.3
4
52.5
5
3.1
8
70.61
5
gender differences remained only for
48.6
3
68.6
3
14.5
3
81.05
2
the histrionic and borderline PS.
Opiate abuse
or dependence

Methadone
treatment

After
admission to
treatment

Dependent

52.9

69.4

21.2

75.85

Histrionic

10.6

24.7

2.7

58.63

Narcissistic

12.5

20.8

10

4.3

56.58

10

Antisocial

12.5

26.3

3.5

60.90

Compulsive

0.0

11

0.4

0.0

11

39.66

11

Passive-aggressive

56.9

76.5

20.8

82.30

Schizotypal

7.1

10

22.0

0.4

10

65.85

Borderline

21.2

55.3

3.9

75.45

Paranoid

12.5

31.4

4.7

69.05

Mixed

9.0

psychometric qualities (that is, internal consistency, testretest reliability, and discriminant validity), this translation is
reliable and valid (23). Millon's taxonomy parallels that of
the Diagnostic and Statistical Manual of Mental Disorders
(DSM-III).
Base rate (BR) scores were computed for each scale. A BR
score of 74 or more (74+) means that the subject shows signs
of the disorder; a score of 84 or more (84+) signals prominence of a disorder. Fifteen tests were rejected because of the

The fifth column of Table 2 presents

the highest score obtained on a scale.
Each subject is assigned to a single
category. If the same score is obtained on 2 or more scales, the mixed
category is used. Over one-fifth of
subjects obtained their highest score
on the passive-aggressive (20.8%) or
the dependent (21.2%) PS. Comparisons by gender were performed using
chi-square or Fisher's test when more
than 20% of the cells contained less
than 5 elements. Women were underrepresented on the passiveaggressive scale.

In the seventh column of Table 2, mean scores are presented

for each scale. Mean scores for the avoidant, dependent,
passive-aggressive, and borderline PSs are all in the 74+
range. As shown in Table 3, comparisons by gender were
made using Ntests. Women had a higher mean score than men
on the borderline PS.
For each subject, we summed the number of scales in the 84+
category. The mean number of scales was 2.68, with 71% of

594

The Canadian Joumal of Psychiatry

Table 3. Mean score on personality scales by gender

(substance abuse disorder, Quebec sample)
Mean score (SD)
Male

Scale

Female

Total

Schizoid

70.37 (23.44)

71.19(25.59)

70.61 (24.03)

Avoidant

81.04(21.11)

81.10(23.99)

81.05(21.92)

Dependent

75.34 (24.66)

77.15(25.83)

75.85 (24.97)

Histrionic

59.82(19.23)

55.64 (28.70)

58.63(22.38)

Narcissistic

56.95(21.29)

55.66(24.92)

56.58(22.35)

Antisocial

61.68(22.10)

58.96(21.99)

60.90(22.06)

Compulsive

39.18(18.02)

40.88(21.85)

39.66(19.16)

Passive-aggressive

82.67 (23.36)

81.38(25.30)

82.30(23.89)

Schizotypal

65.60(11.16)

66.48(14.14)

65.85(12.07)

Borderline*

73.57(12.19)

80.12(16.40)

75.45(13.81)

Paranoid

68.11 (12.96)

71.40(16.31)

69.05(14.05)

'P<Q.O\.

Vol 44, No 6

borderlineour sample had the highest scores and, in most

cases, they differed significantly from those of other samples.
Our sample had the lowest scores for the narcissistic and compulsive scales. On 3 scaleshistrionic, antisocial, and paranoidour sample was close to the midpoint of the
distribution.
The mean BR scores and standard deviations of our sample
were also compared with those of 3 other Quebec clinical
samples. ANOVA and a Student-Newman-Keuls test were
used. Our sample obtained higher scores on several scales
(Table 4). These results also indicate that these 4 samples
present different Axis II profiles.
Discussion
The MCMI is not a diagnostic instrument (24). It measures
whether certain features have attained clinical significance.
Our own test-retest reliability study (23) has shown that the
intensity of symptoms decreases after a few weeks of treatment. All authors except Millon have reported similar results
(25). Toxic effects of substances probably decrease after a
few weeks of treatment (26).

Table 4. Comparison of Millon Clinical Multiaxial Inventory scale score between 4 samples
Alcohol and drug
abuse

Sexual
dysfunction

Rape

Pedophilia

Total

(n = 255)

(n = 173)

(n = 34)

(n = 79)

(n = 541)

Schizoid

70.61

45.95-

61.50

61.77-

60.86

Avoidant

81.05

53.01-

70.50-

72.62-

70.19

Dependent

75.85

57.40-

67.15

76.38

69.48

Histrionic

58.63

55.08-

56.15

52.65

56.46

57.53

59.12

Axis II

Narcissistic

56.58

63.35+

60.41

Antisocial

60.90

66.55+

67.38

59.56

62.92

Compulsive

39.66

60.14+

51.29+

51.63+

48.69

Passive-aggressive

82.30

51.65-

63.94-

63.76-

68.64

Schizotypal

65.85

58.21-

62.09

65.29

63.09

Borderline

75.45

62.6&-

67.12-

69.96-

Paranoid

69.05

69.46

67.32

+This sample mean is significantly higher than the drug alcohol abuse sample.
-This sample mean is significantly lower than the drug alcohol abuse sample.

subjects scoring more than 84 on 1 scale. For the cutoff score

of 74, the mean nimiber of scales was 4.48, with 95% of subjects scoring at least 1 scale. The Mann-Whitney U nonparametric test was used to make comparisons by gender, since
this variable is not continuous. On average, women scored
84+ on 3.08 scales, compared with 2.52 scales for men (t/ =
5614.5, P = 0.05). There was no statistically significant difference for the 74 cutoff point.
Our results were compared with those of 8 samples mentioned in Table 1. ?-Tests showed that for 6 scalesschizoid,
avoidant, dependent, passive-aggressive, schizotypal, and

Our results reveal relatively severe

personality disorganization in the
upper range, compared with other
studies with addicted clients.
These results confirm that no single PD is typical of all addicts (27).
However, these comparisons are
subject to limitations. The length
of time between admission to treatment and test completion may
have varied from one study to another. Particular psychosocial
characteristics of the different
samples may have influenced the
results.

The comparison with 3 Quebec

clinical samples confirms the more
69.68
69.16
severe Axis II features of our sample. These findings are consistent
with the results of a previous study
(28) in 3 rehabilitation centres using a validated French version of the Addiction Severity Index (29), which found greater deterioration of the
psychological, family, and legal spheres in Quebec samples
than in those from outside Quebec. The consistency confirms
the validity ofthe 2 instruments and also hints about the links
between Axis I and II disorders.
70.04

Results concerning the passive-aggressive PS ofthe MCMI-I

should be interpreted with caution. Several clinicians noted,
as we did, that its high scores suggest an overlap with SD. In
the DSM-IV, the passive-aggressive PD has been relegated to

August 1999

Personality Disorders and Addiction

an annex because of doubts as to its specificity. Millon excluded this scale from the MCMI-III.

Clinical Implications

Results according to gender show that PDs are more prevalent among women, no matter how the results are analyzed.
Numerous studies have demonstrated similar results (30).
Women, however, have as good a prognosis as men and, in
some cases, a better one (31). The specific effect of PD on
treatment outcome in reference to gender, however, is unknown.

The prevalence of Axis II disorders is high among clients with

an addiction.
Axis II disorders should be given as much attention as Axis I
disorders.
Those in charge of treatment programs should examine how
best to adapt their services to deal with Axis II comorbidity.

Limitations

SDs and Axis II disorders must be treated concurrently

(10,32,33). Those in charge of treatment programs should examine how best to adapt their services to deal with Axis II
comorbidity.

The Millon Clinical Multiaxial Inventory (MCMI) is not a diagnostic instrument.

The intensity of Axis II disorders decreases after a few weeks
of treatment for clients with substance disorders.
Results conceming the passive-aggressive personality scale
on the MCMC-I should be interpreted with caution.

Acknowledgements
This research was supported by a grant from the Conseil quebecois
de la recherche sociale to the team Recherche et Intervention sur les
Substances psychoactives - Quebec (RISQ).

17. Nace EP, Davis CW. Treatment outcome in substance-abusing patients with a
18.

References
1.
2.

3.

4.
5.
6.
7.

8.

9.
10.
11.
12.
13.

14.

15.
16.

Millon T. Millon Clinical Multiaxial Inventory. Minneapolis (MN): National

Computer Systems; 1983.
Calsyn DA, Saxon AJ. Personality disorder subtypes among cocaine and opioid
addicts using the Millon Clinical Multiaxial Inventory. International Joumal of
the Addictions 1990;25:1037-49.
Craig RJ. A psychometric study ofthe prevalence of DSM-III personality disorders among treated opiate addicts. International Journal of the Addictions
I988;23:l 15-24.
Craig RJ, Olson RE. MCMI comparisons of cocaine abusers and heroin addicts. J
Clin Psychol 1990;46;23(>-7.
Dougherty RJ, Lesswing NJ. Inpatient cocaine abusers: an analysis of psychological and demographic variables. J Subst Abuse Treat 1989;6:45-7.
Khantzian EJ, Treece C. DSM-III psychiatric diagnosis of narcotic addicts. Arch
Gen Psychiatry 1985;42:1067-71.
Koenigsberg HW, Kaplan RD, Gilmore MM, Cooper AM. The relationship between syndrome and personality disorder in DSM-III: experience with 2,462 patients. Am J Psychiatry 1985;142:207-12.
Marsch DT, Stile SA, Stoughton NL, Trout-Landen BL. Psychopathology of opiate addiction: comparative data from the MMPI and MCMI. Am J Drug Alcohol
Abuse I988;14:17-27.
Weiss RD, Mirin SM, Michael JL, Sollogub AC. Psychopathology in chronic cocaine abusers. Am J Drug Alcohol Abuse 1986; 121:17-29.
Matano RA, Locke KD, Schwartz K. MCMI personality subtypes for male and female alcoholics. J Pers Assess 1994;63:250-64.
DeJong CA, Van Den Brink W, Harteveld FM, Van Der Wielen EG. Personality
disorders in alcoholics and drug addicts. Compr Psychiatry 1993;34:87-94.
Sheenan MF. Dual diagnosis. Special issue: contemporary topics in drug dependence and alcoholism. Psychiatr Q 1993;64:107-34.
Clopton JR, Weddige RL, Contreras SA, Fliszar GM, Arredondo R. Treatment
outcome for substance misuse patients with personality disorder. International
Joumal ofthe Addictions 1993; 28:1147-1153.
Boudreault L, Duhamel D, Maltais K, Marceau JP. Les troubles de la personnalit^
dans le contexte de l'intervention en toxicomanie. In: Brisson P, editor. L'usage
des drogues et la toxicomanie. Volume II. Montreal: GaStan Morin; 1994. p
437-57.
Reich JH, Green AI. Effect of personality disorders on outcome of treatment. J
Nerv Ment Dis 1991;179:74-82.
Reich JH, Vasile EG. Effect of personality disorders on the treatment of Axis I
conditions: an update. J Nerv Ment Dis 1993; 181:475-84.

595

19.
20.

21.

22.
23.

24.
25.
26.
27.
28.

29.
30.

31.

32.
33.

personality disorder. Am J Addict 1993;2:26-33.

Penick EC, Nickel EJ, Cantrell PF, Powell BJ, Read MR, Thomas MM. The
emerging concept of dual diagnosis: an overview and implications. [Special issue:
Managing the dually diagnosed patient: current issues and clinical approaches.]
Joumal of Chemical Dependency Treatment 1990;3:l-27.
Brown HP. Substance abuse and the disorders ofthe self: examining the relationship. J Psychoactive Drugs 1992;21:139-44.
Ness R, Handelsman L, Aronson MJ, Hershkowitz A, Kanof PD. The acute effects of rapid medical detoxification upon dysphoria and other psychopathology
experienced by heroin abusers. J Nerv Mem Dis 1994; 182:353-9.
Craig RJ, Verinis JS, Vexler S, Personality characteristics of drug addicts and alcoholics on the Millon Clinical Multiaxial Inventory. J Pers Assess
1985;49:156-60.
D'Elia A, LagierPM. Inventaire clinique multiaxial de Millon. Montreal: Bureau
d'intervendon psychosociale; 1986.
Landry M, Nadeau L, Racine S. La prevalence des troubles de la personnalit^ dans
huit centres de readaptation du Quebec. Cahier de recherche. Montreal: Recherche et Intervention sur les Substances psychoacdves/Quibec; 1996.
Perry JC. Problems and considerations in the valid assessment of personality disorders. Am J Psychiatry I992;149:1645-53.
ZimmermanM. Diagnosing personality disorders: a review of issues and research
methods. Arch Gen Psychiatry 1994;51:225^5.
Schuckit MA, Monteiro MS. Alcoholism, anxiety and depression. British Joumal
of Addiction 1983;83:1373-80.
Craig RJ. The role of personality in understanding substance abuse. Alcoholism
Treatment Quarterly 1995;13:17-27.
Guyon L, Landry M. L'abus de substances psychoactives, un problteie parmi
d'autres? Portrait d'une population en traitement. Psychotropes, revue intemationale des toxicomanies 1996;1:61-81.
McLellan AT, Luborsky L, O'Brien CP. An improved evaluation instrument for
substance abuse patients. J Nerv Mem Disorder 1981;168:26-33
Nadeau L. Les femmes et l'alcool: itat de la question. In: Brisson P, editor.
L'usage des drogues et la toxicomanie. Volume II. Montreal: Gatan Morin;
1994. p 231-54.
Lightfoot L, Adrian M, Leigh G, Thompson J. Women's use of and views on substance abuse treatment. In: Adrian M, Lundy C, Eliany M, editors. Women's use
of alcohol, tobacco and other drugs in Canada/Les Canadiennes et l'usage d'alcool, de tabac et d'autres drogues. Toronto: Addiction Research Foundation;
1996. p 228^6.
Klein RH, Orieans JF, Soule CR. The Axis II group: treating severely characterologically disturbed patients. Int J Group Psychother 1991;41:97-115.
Southwick SM, Satel SL. Exploring the meanings ofsubstance abuse: an important dimension of early work with borderline patients. Am J Psychother
1990;44:61-7.

596

The Canadian Joumai of Psychiatry

Vol 44, No 6

Resume
Cette etude a determine la prevalence des troubles de personnalite parmi les clients en traitement pour toxicomanie ; cette prevalence a ete comparee a celles constatees dans des etudes semblables et dans des echantillons
cliniques de personnes souffrant d 'autres troubles de I 'axe I. Notre echantillon se composait de 255 sujets. La premiere edition de I 'Inventaire clinique multiaxial de Millon (MCMI) a ete utilisee. Seulement 11,8% des sujets n 'ont
pas obtenu de resultat superieur a 84 a aucune des II echelles de I'axe II. Le nombre moyen d'echelles dans la
categorie des 84 etplus etait 2,68. Les comparaisons indiquent que cet echantillon etait plus gravement malade
dans la plupart des cas.