The DSM-IV and ICD-10 personality questionnaire (DIP-Q): Construction and preliminary validation

Abstract

This paper describes the construction and preliminary validation of a new selport inventory for personality disorders — DSM-IV and ICD-10 Personality Disorder Questionnaire (DIP-Q). In a consensus process the criteria sets of DSM-IV and ICD-10 were scrutinized. Twenty-seven criteria were judged completely identical in both systems. In addition, 20 criteria were close to identical. The total number of different criteria could thus be reduced from 161 to 114. Since 24 criteria could not be covered by a single statement, the final version of the DIP-Q includes 135 statements reflecting the criteria and additionally 5 statements reflecting the general criteria. The questionnaire is generally completed within 20 min. In the preliminary validation among 33 psychiatric patients Cronbach's alpha coefficients within each personality disorder were acceptable for most personality disorders and somewhat higher for the DSM-IV personality disorders than for those in the ICD-10. The criteria set of Dissocial disorder in the ICD-10 showed a negative alpha coefficient. When analysed dimensionally, the Pearson correlation between pairs of disorders in the ICD-10 and the DSM-IV varied from 0.77 to 0.99. Kappa coefficients between pairs from each system varied from 0.47 to 0.69. In conclusion, the ICD-10 and the DSM-IV are similar enough to enable the construction of a brief and comprehensive questionnaire evaluating personality disorders from both systems. There are, however, significant differences between systems which must be further analysed in future full-scale validation studies.

Full text

Original article
Eur Psychiatry 1998 ; 13 : 246-53
0 Elsevier, Paris
DSM-IV and ED-10 personality disorders:
a comparison of a self-report questionnaire (DIP-Q)
with a structured interview
H Ottosson’, 0 Bodlundl, L Ekselius2, M Grannl, L von Knorring2,
G Kullgrenl*, E Lindstrijm2, S Sijderbergl
1 Department of Psychiatry, University of UmeH, S-901 85 UmeB:
2 Department of Psychiatry, University Hospital, Uppsala University, Uppsala, Sweden
(Received 17 June 1997; accepted 20 February 1998)
Summary - Objective: Diagnosing personality disorders according to structured expert interviews is time-consuming and costly. For
epidemiological studies, self-report instruments have several advantages. The DSM-IV and ICD-10 personality questionnaire (DIP-Q) is a self-
report questionnaire constructed to identify personality disorder according to DSM-IV and ICD- 10.
Method: The DIP-Q is validated vs a structured expert interview in a clinical sample of 138 individuals. In addition, prevalence rates yielded by
DIP-Q among 136 healthy volunteers are assessed and compared to expected prevalence.
Results: For DSM-IV the agreement for any personality disorder as measured by Cohen’s Kappa was 0.61 and 0.56 for ICD-10. Overall
sensitivity for any personality disorder was for DSM-IV 0.84 and for ICD-10 0.85. However, specificity was lower: 0.77 and 0.70, respectively.
When dimensional scores between self-report and interview for each personality disorder were compared, the intraclass correlation for the DSM-
IV entities was 0.37-0.87 and for the ICD-10 entities 0.33-0.73. Among healthy volunteers the base rate of personality disorders was found to
be 14%.
Conclusion: DIP-Q can be used as a screening instrument for personality disorders according to DSM-IV and ICD-10. Self-report
questionnaires such as DIP-Q will probably play an increasingly important role in future epidemiological studies. 0 1998 Elsevier, Paris
personality disorder / DSM-IV / ICD*lO / self-report I DIP-Q / structured interview
INTRODUCTION
The third edition of the American Psychiatric Associ-
ation’s DSM classification, DSM-III [l], was the first
international system to specify diagnostic criteria for
personality disorders. The DSM-III was revised in
1987 (DSM-III-R) [2] and in 1994 (DSM-IV) [3]. The
World Health Organisation’s 9th revision of the Inter-
national Classification of Diseases [25] provided only
brief clinical descriptions for its diagnostic categories
of personality disorders, but the 10th revision, ICD-10
[26], provides a set of criteria to be evaluated for each
specified personality disorder. Also, diagnostic criteria
for research have been published, the Draft Criteria
for Research, DCR [27].
The DCR-10 (in this paper further referred to as
ICD-10) and DSM-IV are highly concordant with
regard to personality disorders. All eight diagnostic
categories in ICD-10 are also found in DSM-IV.
Analysis of the criteria sets for the categories, shows
that 70% of the ICD-10 criteria are covered by
DSM-IV and 60% of the DSM criteria are covered
by ICD [ 191. Both the DSM-IV and the ICD-10 use a
polythetic system in which a specified number of the
criteria must be fulfilled for each personality disor-
der diagnosis (with the exception of Emotionally
*Correspondence and reprints.

Validity of DIP-Q
247
unstable personality disorder in ICD-10). In addition
to diagnostic criteria, both DSM-IV and ICD-10
equivalently specify a set of general diagnostic crite-
ria, which must be met to warrant a personality dis-
order. In short, the general criteria state that only
when personality traits are inflexible and maladap-
tive and cause significant functional impairment or
subjective distress, they constitute a personality dis-
order.
The process of testing and refining diagnostic
instruments of psychiatric diagnoses forms a vital
basis for reliable communication, research and treat-
ment planning. Three different types of instruments
for diagnosing personality disorders are available;
structured interviews and self-report questionnaires,
both being the interest of this paper, and lastly infor-
mant-based interviews and questionnaires. Struc-
tured interviews have improved interrater reliability
as well as test-retest reliability, compared to unstan-
dardised clinical assessments [22]. Many structured
and semistructured interviews have been developed.
Of current interest is the International Personality
Disorder Examination (IPDE), a structured inter-
view, for the assessment of both the DSM-III-R and
the ICD-10 personality disorders. In a “silent
observer” interrater design with 141 psychiatric
patients included, the overall Cohen’s Kappa was
0.57 for DSM-III-R and 0.65 for ICD-10 [16]. In a
review of Zimmerman’s [28], 15 joint-interview
studies reported similar findings with nearly 80% of
the Kappa values above 0.69. Different base rates in
the different studies, however, hamper comparisons
WI.
Comparisons of interviews and self-report ques-
tionnaires are frequent in the literature. Hyler et al
[ 141 administered the Personality Diagnostic Ques-
tionnaire Revised (PDQ-R), a self-report question-
naire for the assessment of the DSM-III-R personality
disorders, to 87 applicants for inpatient treatment for
severe personality psychopatology. Blind to these
results, diagnoses were made according to the Struc-
tured Clinical Interview for DSM-III-R Personality
Disorders (SCID-II) and the Personality Disorder
Examination (PDE). The overall agreement, as meas-
ured by Cohen’s Kappa between PDQ-R and SCID-II
was 0.41 and between PDQ-R and PDE 0.36. SCID-II
is accompanied by a self-report instrument, the
SCID-screen [12]. In a study on a Swedish version of
the SCID-screen, 69 psychiatric patients completed
the SCID screen and were subsequently interviewed
with the SCID-II [I 11. Without adjustments, the
SCID-screen was overinclusive but when cut-off lev-
els were adjusted by requiring one more criterion for
diagnosis of each of the personality disorders, the
overall kappa was as high as 0.78, quite up to the
standard of agreement reported for joint-interview
studies.
Accordingly, overall agreement of personality disor-
ders present between interviews and questionnaires has
been shown to be acceptable. However, agreement for
certain personality disorder diagnoses is less accept-
able. In general, self-report instruments tend to overdi-
agnose personality disorders, but may still be useful
for screening purposes. The constructors of both the
SCID-screen and the PDQ-R recommend positive
diagnoses according to the self-report instruments to
be verified by clinician-administered interviews. This
two-step process is still time-saving in comparison to a
complete personality disorder evaluation.
Self-administered questionnaires will probably
increase in importance in the future. They have
good screening properties, they are easily adminis-
trated, timesaving and free from systematic biases
of interviewers [28]. Furthermore questionnaires
present means of recording dimensional scores, in
addition to categorical personality disorder diag-
noses, as a more reliable and valid way of reporting
personality pathology [lo, 181. By means of com-
puter processing, dimensional scores and subthres-
hold pathology are easily presented in a graphically
accessible way.
In addition to good screening properties, self-report
assessments may provide data qualitatively disparate
from interviews. In a 5-year follow-up study of trans-
sexuals in the process of changing sex, one of the
major risk factors for negative outcome was a categor-
ical personality disorder diagnosis according to SCID-
screen, while a personality disorder only identified
from clinical interview was of lower predictive value
[7]. In addition, self-rating scales may have an advan-
tage to expert ratings when evaluating course and
effect of treatment [ 151.
Based on our previous experiences from the self-
report instrument SCID-screen [7, 101 and analyses of
ICD-10 and DSM-IV diagnostic criteria [19], a new
self-report instrument was developed, the DSM-IV
and ICD-10 Personality Questionnaire (DIP-Q). Apart
from adjusting the self-report items according to
revised and new criteria, an impairment/distress scale
was added, corresponding to the general diagnostic
criterion. Thus, in contrast to previous instruments
DIP-Q provides diagnoses from both systems and the
general criterion is assessed through the impair-
ment/distress scale.
The aim of the present study was to validate this
new DIP-Q instrument vs a fully structured interview
in a clinical sample. In addition, observed prevalence
in a healthy-volunteer sample of the specific personal-
ity disorders according to DIP-Q is compared to
expected prevalence according to the literature.

248
H Ottosson et al
METHOD
Subjects
The clinical sample consisted of 138 patients (58% women
and 42% men) with a mean age of 37.2 (range 1843, stan-
dard deviation [SD] 12.0), and was selected to cover a wide
range of clinical settings; psychiatric out patients (22%),
general psychiatric inpatients (1 l%), patients hospitalised
after suicide attempt (38%) and patients recruited from an
inpatient assessment unit for sleep disorders and pain syn-
dromes (29%). All included patients were judged to have an
intellectual and verbal capacity sufficient to allow them to
understand and respond to a self-rating questionnaire. All
patients gave informed consent to participate in the study,
which was approved by the Ethics Committees of Uppsala
and Umei Universities.
A sample of healthy volunteers, 136 individuals (69%
women and 31% men) with a mean age of 28.0 (range
18-55, SD 8.7) were recruited from a college for medical
care.
The DSM-IV and ICD-10 Personality Questionnaire
(DIP-Q)
DIP-Q is a 140 item true/false self-report questionnaire
designed to measure all ten DSM-IV and all eight ICD-10
personality disorders. Also included is the ICD- 10 schizo-
typal disorder. It requires approximately 20 min to complete
and the scoring is best done by computer. The construction
and preliminary validation has been further described by
Ottosson and co-workers [19].
Out of the 140 self-report items in the questionnaire,
135 items reflect the diagnostic criteria of the DSM-IV and
ICD-10 personality disorders. Each item constitutes a brief
statement reflecting the major aspect of the corresponding
criterion, and the respondent is asked to score the statement
as true or false. For example: the DSM-IV borderline per-
sonality disorder criterion A [3]: “identity disturbance:
markedly and persistently unstable self image or sense of
self ‘, corresponds to the DIP-Q item “I feel vev losf inside-
I don’t really know who I am”. In DIP-Q, in contrast to
SCID-screen, criteria related to observed behaviour occur-
ring in schizoid, schizotypal, histrionic and narcissistic per-
sonality disorder, are also included. Twenty-one items are
reversed in the sense that “false” as a response indicates ful-
filment of the corresponding criterion.
Five items constitute the impairment/distress scale (ID-
scale), which is based on the scale included in the Personal-
ity Diagnostic Questionnaire [13]. Complementary to the
ID-scale, a self-report version of the Global Assessment of
Functioning (GAF) Scale is included. This scale consists of
the original, O-100 point scale [3], but with fewer defining
characteristics than in the original version. The GAF self-
report version has been further described elsewhere [6].
Different cut-off levels as concerns the ID-scale and the
GAF self report scale can be applicable. In this study the
cut-off score on the ID-scale is set to two or higher. On the
GAF scale a scoring at the anchoring point of 70 indicates
some degree of psychological distress and/or some diffi-
culty in social or occupational functioning, therefore a
score of 70 or less was chosen as cut-off level. Accord-
ingly, a categorical diagnosis requires firstly that the num-
ber of criteria for the specific personality disorder reaches
the threshold specified by the DSM-IV and ICD-10 manu-
als, and secondly a scoring of two or higher on the ID-scale
or a scoring of 70 or less on the GAF self-report scale.
Dimensional scores are calculated as the number of positive
criteria for each personality disorder diagnosis, regardless
of ID or GAF scores.
The original DIP-Q is in Swedish, but it is available in
English, French, Spanish, Finnish, Danish, Icelandic, and
Norwegian on request from the author.
The DSM-IV and ICD-10 Personality Interview
(The DIP-I)
The DSM-IV and ICD-10 Personality Interview (DIP-I), is a
structured interview, developed by our group, for the assess-
ment of the DSM-IV and ICD-10 personality disorders, plus
schizotypal disorder in ICD-10. The construction is in all
essentials similar to the SCID-II [12]. In a recent study by
Maffei and co-workers [ 171 interrater reliability was shown
to be quite adequate (Cohen’s Kappa 0.48-0.98) and we
have no reason to believe that DIP-Q differs in this respect.
The interview starts with a set of overview questions and
then systematically covers each criterion of the personality
disorders in turn, making it easy for the interviewer to assess
one disorder at a time. The interviewer is encouraged to ask
questions referring to the general diagnostic criteria. Based
on the knowledge of the patients history and answers to the
questions, the interviewer makes a clinical rating of each
personality criterion.
Procedure
After informed consent was obtained, the interview started
with a brief overview, focusing on the individual’s intellec-
tual capacity and self-reflection. As co-occurring Axis-I dis-
order may confound assessment of personality disorders, an
unstructured Axis I interview was also conducted. The clini-
cal sample completed both the DIP-I and the DIP-Q, and
statistical agreement measures were computed for both cate-
gorical diagnosis and personality dimensions (number of
criteria fulfilled for each personality disorder). The majority
of the subjects (83%) completed the DIP-Q from 1 day up to
a week before the interview was made, 14% completed the
DIP-Q the same day before the interview was made and 3%
2 or more days after the interview. The interviews were

Validity of DIP-Q
249
Table I. Prevalence of DSM-IV Axis I principal diagnoses in the
clinical sample.
NO (‘36) ofpatients {n = 138) Female Male
Substance-related disorder
Schizophrenia
Other psychotic disorder
Depressive disorder
Bipolar disorder
Anxiety disorder
Eating disorder
Sleep disorder
Pain syndromes
Adjustment disorder
No Axis I disorder present
1 (1.3) 0 -
4 (5) 2 (3.4)
3 (3.8) 0 -
27 (33.8) 17 (29.3)
4 (5) 2 (3.4)
12 (15) 8 (13.8)
3 (3.8) 0 -
9 (11.3) 14 (24.1)
3 (3.8) 2 (3.4)
2 (2.5) 2 (3.4)
12 (15.1) 11 (18.9)
performed by three psychiatrists with extensive prior experi-
ence with the SCID-II, and one psychologist provided with
initial training by passive joint-interviewing. The interview-
ers were blind to the DIP-Q results.
To roughly estimate the properties of DIP-Q in assessing
DSM-IV personality pathology among non-patients, a sam-
ple of healthy volunteers completed the DIP-Q, and preva-
lence of personality disorder was compared with expected
prevalence based on epidemiological data [24]. All healthy
volunteers completed the DIP-Q at home and mailed it back
anonymously.
Statistics
Categorical personality disorder diagnosis were estimated
from DIP-Q scores in two ways: a) with adjustments by the
ID-scale and the GAF-self-report scale, referred to as DIP-Q-
adjusted, and b) without any adjustment, referred to as
DIP-Q-raw.
Agreement between categorical DIP-Q diagnoses and
interview diagnoses, respectively, was calculated using
Cohen’s Kappa [9], a chance corrected measure of congru-
ence for binary ratings. The value of Kappa ranges from
-1 .O to 1 .O with higher values representing higher levels of
agreement.
Dimensional agreement, ie, the number of criteria ful-
filled for each personality disorder according to DIP-Q
and interview, respectively, was calculated using the
intraclass correlation. The so-called Case 2 formula was
used [21].
Specificity was calculated as proportion of correctly clas-
sified negative personality disorder cases and sensitivity as
the proportion correctly classified positive cases.
RESULTS
Table I shows the distribution of the diagnostic groups
according to DSM-IV.
Table II. Prevalence rates of personality disorders in the clinical
sample (n = 138) according to DIP-interview and DIP-Q-adjusted.
The prevalence rates according to DIP-Q without adjustments by
the ID-scale and GAF self-report scale is referred to as DIP-Q-raw.
Personality disorder DIP-I, DIP-Q, DIP-Q adjusted raw
DSM-IV
Paranoid
Schizoid
Schizotypal
Antisocial
Borderline
Histrionic
Narcissistic
Avoidant
Dependent
Obsessive-compulsive
Any PD
21 34.1 39.9
4.3 13.8 15.2
8 26.8 28.3
10.1 10.1 10.9
33.3 40.6 45.7
5.8 4.3 4.3
5.1 6.5 7.2
37 39.1 44.9
9.4 18.1 21
28.3 37.7 49.3
65.9 58.7 77.5
[CD-IO
Paranoid
Schizoid
Schizotypal disorder
Dissocial
Emotionally unstable PD
- Borderline type
- Impulsive type
Histrionic
Anxious
Dependent
Anankastic
Any PD
21 37.7 42
16.7 32.6 36.2
13 41.3 46.4
7.2 13.8 15.2
25.7 30.4 34.8
18.8 18.1 21
6.5 5.8 5.8
33.3 38.4 42
13 25.4 29.7
29.7 45.7
63 Cl 76.1
In the clinical sample, the overall prevalence of any
personality disorder according to the interview was
66% for DSM-IV and 63% for ICD-10. The preva-
lence for a specific DSM-IV diagnosis varied from 4%
for schizoid personality disorder to 37% for avoidant
personality disorder. For a specific ICD-10 diagnosis,
the prevalence varied from 7% for histrionic personal-
ity disorder to 33% for anxious personality disorder.
The highest discrepancy between DSM and ICD was
noted for schizoid personality disorder. With the
exception of histrionic personality disorder, both DIP-
Q adjusted and DIP-Q raw yielded more personality
disorder diagnoses than the interview. As expected the
prevalence rates of DIP-Q raw was throughout higher
than for DIP-Q-adjusted (table ZZ).
The mean number of DSM-IV diagnoses for those
with at least one personality disorder was for the inter-
view 2.5 and for DIP-Q-adjusted 3.9. The mean num-
ber of ICD-10 diagnoses was for the interview 3.0 and
for DIP-Q-adjusted 4.7.
The chance-corrected agreement (Kappa) for cate-
gorical diagnoses, based on interview and DIP-Q with
adjustments, showed a high degree of variance, from
0.08 for narcissistic personality disorder to 0.59 for

250
H Ottosson et al
Table III.
Cohen’s Kappa Coefficient of agreement between cate-
gorical DIP-I personality disorder diagnosis and categorical DIP-Q-
adjusted personality disorder diagnosis in the clinical sample
(n = 138). The Kappas between DIP-I and DIP-Q without adjust-
ments by the ID-scale and GAF self-report scale are presented as
DIP-Q-raw.
Personality disorder DIP-Q
adjusted
DIP-Q
raw
Paranoid
Schizoid
Schizotypal
Antisocial
Borderline
Histrionic
Narcissistic
Avoidant
Dependent
Obsessive-compulsive
Any cluster A
Any cluster B
Any cluster C
Any personality disorder
Paranoid
Schizoid
Schizotypal disorder
Dissocial
Emotionally unstable PD
- Borderline type
- Impulsive type
Histrionic
Anxious
Dependent
Anankastic
Any personality disorder
DSM-IV
0.41 0.39
0.15 0.25
0.33 0.30
0.51 0.42
0.52 0.55
0.17 0.25
0.08 0.18
0.52 0.54
0.50 0.44
0.52 0.39
0.45 0.36
ICD-IO
0.48 0.52
0.63 0.48
0.61 0.47
0.48 0.39
0.33 0.33
0.21 0.19
0.30 0.23
0.59
0.25
0.11
0.59
0.49
0.50
0.56
0.48
0.21
0.06
0.58
0.43
0.39
0.43
anxious and borderline type of emotionally unstable
personality disorder. Kappa coefficient for “any per-
sonality disorder” was 0.61 for the DSM diagnoses
and 0.56 according to the ICD. Highest agreement,
0.63, was calculated for “any cluster C personality
disorder diagnosis” (table ZZZ).
The sensitivity of DIP-Q-raw was in general higher
than for DIP-Q-adjusted. For the occurrence of any
personality disorder, the sensitivity was 0.92 for DSM
and 0.91 for ICD. The specificity for DIP-Q-raw was
in the range of 0.58-0.90 (table IV).
The dimensional agreement, ie, the correlation
between the number of fulfilled criteria, based on
interview and DIP-Q for the specific personality dis-
orders, varied from 0.37 to 0.87 for DSM and from
0.33 to 0.73 for ICD. Highest correlation was found
for conduct disorder and avoidant/anxious personality
disorders (table V).
In
the healthy-volunteer sample the prevalence
was calculated only for the DSM-IV personality dis-
orders, since comparative data for ICD-10 was not
available. The overall prevalence of “any personal-
ity disorder” was reduced considerably when
ID-adjustments were done, from 38% to 14%, a
reduction roughly to the expected level of present
epidemiological data as presented by Weissman [24]
(table VZ).
DISCUSSION
A common feature in both the DSM-IV and the ICD- 10
chapter on personality disorders is the two levels of
definition, firstly, the general diagnostic criteria, and
secondly, the specific diagnostic criteria. The general
definition is required to separate personality disorders
from non-pathological personality traits. DSM-IV
states: “only when personality traits are inflexible and
maladaptive and cause significant functional impair-
ment or subjective distress do they constitute a per-
sonality disorder” [3, p 6301. The complex question of
how to evaluate this crucial criterion in self-report
questionnaires is of great importance, but often disre-
garded. Shortcomings in this respect might explain the
tendency for self-report instruments to overdiagnose
personality disorders, in relation to structured inter-
views [ 11, 141.
Modifications of self-report instruments in order to
reduce this problem can be made in several ways. One
way is to raise the threshold for the number of criteria
required for a categorical diagnosis. This can be justi-
fied by the fact that higher cut-off points probably
reflects a higher degree of social impairment. In a pre-
vious study using SCID-screen adjustment by cut-off,
ie, one more criterion for each personality disorder
requested, agreement improved between self-report
and interview diagnosis [ 111. Another option is to
assess impairment or distress of the individual as
reported by Bodlund et al [5]. In this study, diagnoses
based on SCID-screen, combined with a self-reported
GAF of 70 or less, were in good agreement with clini-
cal diagnosis.
DIP-Q allows for adjustment by the impairment/dis-
tress scale and the GAF self report scale, which repre-
sent different aspects of the general criterion and
could be regarded as complementary. As personality
traits are often egosyntonic, underreporting might be
expected from the ID-scale since it requires that the
respondent has acknowledged difficulties connected
to maladaptive personality traits. The GAF scale, on
the other hand, permits the respondent to rate himself
as malfunctioning without referring to personality
pathology.
Agreement between the structured interview and the
self rating questionnaire was acceptable with regard to

Validity of DIP-Q
251
Table IV. Sensitivity and specificity of DIP-Q-adjusted and DIP-Q-raw (without adjustments) in relation to the diagnostic interview (DIP-I).
Personality disorder Sensitivity Sensitivity Specificity Specificity
DIP-Q-adjusted DIP-Q-raw DIP-Q-adjusted DIP-Q-raw
DSM-IV
Paranoid 0.83 0.80 0.17 0.71
Schizoid 0.50 0.67 0.88 0.87
Schizotypal 0.90 0.83 0.78 0.77
Antisocial 0.53 0.50 0.95 0.90
Borderline 0.81 0.85 0.75 0.75
Histrionic 0.17 0.25 0.97 0.97
Narcissistic 0.17 0.29 0.93 0.94
Avoidant 0.75 0.81 0.79 0.76
Dependent 0.82 0.85 0.89 0.85
Obsessive-compulsive 0.79 0.85 0.78 0.64
Any Cluster A 0.80 0.81 0.72 0.65
Any Cluster B 0.75 0.81 0.74 0.73
Any Cluster C 0.84 0.89 0.78 0.58
Any personality disorder 0.84 0.92 0.77 0.51
ED-10
Paranoid
Schizoid
Schizotypal disorder
Dissocial
Emotionally unstable PD
- Borderline type
- Impulsive type
Histrionic
Anankastic
Anxious
Dependent
Any personality disorder
0.88 0.83 0.74 0.69
0.75 0.78 0.73 0.72
0.80 0.79 0.63 0.59
0.50 0.45 0.89 0.88
0.77 0.72 0.85 0.80
0.42 0.38 0.85 0.83
0.16 0.11 0.95 0.95
0.77 0.78 0.77 0.67
0.84 0.83 0.79 0.78
0.87 0.89 0.83 0.79
0.85 0.91 0.70 0.49
the distinction of personality disorder versus no per-
sonality disorder on overall and cluster level, but low
for some of the specific personality disorders. Agree-
ment was modest in particular for, schizoid, histrionic
and narcissistic personality disorder. Obviously some
of the self-report items are inadequate in assessing the
diagnostic criteria. Items referring to histrionic or nar-
cissistic personality disorder criteria can particularly
be apprehended as challenging or offending. Agree-
ment between diagnostic criteria assessed on inter-
view and corresponding self-report items, related to
observed behaviour, was low, indicating that criteria
related to observed behaviour are not easily adapted
for self reporting [4].
Both on a cluster level and on a global level the
sensitivity of the DIP-Q without adjustments is
high. The sensitivity is also high for the specific
Cluster C personality disorders, and for paranoid,
schizotypal and borderline personality disorder.
Thus, clinicians can generally be confident that sig-
nificant Axis II pathology is uncommon in patients
who do not reach threshold on any of these disorders.
However, for schizoid, antisocial, histrionic and
narcissistic personality disorder, the sensitivity is
0.5 or lower, making the DIP-Q less useful as a
screening instrument.
When regarded from a dimensional perspective,
agreement between the DIP-Q and the interview is
high. Highest correlation is noted for the Conduct Dis-
order, probably explained by the fact that all criteria
are defined by clearly defined childhood behaviours.
When prevalence of any personality disorder in
the healthy volunteer sample was examined accord-
ing to DIP-Q, with and without adjustment, preva-
lence dropped from 37.5% to 14.0%, emphasising
the significant impact of the general criteria. This is,
of course, most pronounced in general population
studies whereas in clinical samples impairment or
distress is almost by definition to be expected. In a
previous general population study, Reich et al [20]
found a very similar drop in prevalence, from 28%
to 11% when the general criteria were applied in
addition to fulfilment of diagnostic criteria. The
overall prevalence of 14% in our sample is somewhat

252
H Ottosson et al
Table V. Intraclass correlation (ICC,,,) estimates between number
of criteria fulfilled for each personality disorder according to DIP-Q
and interview in the clinical sample (n = 138). All correlations are
significant at the P < 0.001 level.
Personality disorder ICC Estimate
DSM-IV
Paranoid
Schizoid
Schizotypal
Antisocial
- Adult criteria (A criteria)
- Conduct Disorder (C criteria)
Borderline
Histrionic
Narcissistic
Avoidant
Dependent
Obsessive-compulsive
Cluster A criteria
Cluster B criteria
Cluster C criteria
0.67
0.47
0.44
0.53
0.87
0.68
0.58
0.37
0.78
0.57
0.46
0.60
0.78
0.66
ICD-IO
Paranoid
Schizoid
Schizotypal disorder
Dissocial
Emotionally unstable PD
- Borderline type
- Impulsive type
Histrionic
Anxious
Dependent
Anankastic
0.59
0.48
0.33
0.43
0.39
0.55
0.50
0.73
0.54
0.51
high but within the expected range suggested by
several studies [8, 241. However, for some diag-
noses, ie, paranoid personality disorder, DIP-Q
seems to be overinclusive. There are however, sev-
eral drawbacks when comparing our observed prev-
alence among students with studies performed in
samples using other instruments and caution is
therefore recommended when interpreting these
results.
CONCLUSION
Categorical assessment of personality pathology with
the self-report questionnaire DIP-Q can be made with
acceptable reliability, when validated vs a structured
interview. Sensitivity and specificity is in general high
and DIP-Q shows good screening properties. The
advantage of DIP-Q in relation to previous instru-
ments is, among others, its duality in assessing the
general criterion.
Table VI. Prevalence rates of personality disorders based on
epidemiological data (expected) and among a sample of healthy
volunteers (n = 136) according to DIP-Q-adjusted. The prevalence
rates according to DIP-Q without adjustments by the ID-scale is
referred to as DIP-Q raw.
DSM-IV
Personality disorder
diagnosis
Expected DIP-Q DIP-Q
[241 adjusted raw
Paranoid 0.4-0.9 3.1 8.1
Schizoid 0.4-0.9 1.6 2.2
Schizotypal 3-5.6 2.3 4.4
Antisocial 2-3 - 0.7
Borderline 4.6 6.2 8.1
Histrionic 1.3-3 1.6 1.5
Narcissistic 0.4 1.6 4.4
Avoidant 0.4-l .3 3.9 4.4
Dependent 1.64 -
Obsessive-compulsive 1.6-6 8.5 25
Any 10-13.5 14 37.5
When used for dimensionally assessing of degree of
personality pathology for the specific disorders, agree-
ment is strong between interview and self-report
diagnoses. This is particular promising, since a dimen-
sional approach is closer to the true nature of person-
ality pathology.
ACKNOWLEDGEMENT
The study was supported by grants from Stiftelsen
SGderstriim-Konigska, Gaudelius Minnesfond and the
Swedish Medical Research Council.
REFERENCES
American Psychiatric Association. Diagnostic and Statistical
Manual
of
Mental Disorders - 3rd ed. Washington, DC: APA;
1980
American Psychiatric Association. Diagnostic and Statistical
Manual of Mental Disorders - 3rd ed revised. Washington, DC:
APA; 1987
American Psychiatric Association. Diagnostic and Statistical
Manual
of
Mental Disorders - 4th ed. Washington, DC: APA;
1994
Amtz A, van Beijsterveldt B, Hoekstra R, Hofman A, Eussen M,
Sallaerts S. The interrater reliability of a Dutch version of the
Structured Clinical Interview for DSM-III-R Personality Disor-
ders. Acta Psychiatr Stand 1992 ; 85 : 394-400
Bodlund 0. Transsexualism and Personality: Methodological
and Clinical Studies on Gender Identity Disorders. Doctoral
dissertation. University of Umea: Department of Psychiatry;
1994
Bodlund 0, Kullgren G, Ekselius L, Lindstriim E, von Knorring L.
Axis V - global assessment of functioning: Evaluation of a self-
report version. Acta Psychiatr Stand 1994 ; 90 : 342-7
Bodlund 0, Kullgren G. Transsexualism: General outcome and
prognostic factors. Arch Sex Behaviour 1996 ; 25 : 303-16
Bodlund 0, Grann M, Ottosson H, Svanborg C. Validation of
the self-report questionnaire DIP-Q in diagnosing DSM-IV per-

Validity of DIP-Q 253
sonality disorders: a comparison of three psychiatric samples.
Acra Psychiatr Stand 1988 ; 97 : 433-9
9 Cohen J. A coefficient of agreement for nominal scales. Educ
Psycho1 Meas 1960 ; 20 : 37-46
10 Ekselius L, Lindstrom E, von Knorring L, Bodlund 0, Kullgren G.
Personality disorders in DSM-III-R as categorical or dimen-
sional. Acta Psychiatr &and 1993 ; 88 : 183-7
11 Ekselius L, Lindstrlim E, von Knorring L, Bodlund 0, Kullgren G.
SCID interviews and the SCID screen questionnaire as diagnos-
tic tool for personality disorders in DSM-III-R. Acra Psychiatr
Stand 1994 ; 90 : 120-3
12 First MB, Spitzer RL, Gibbon M, Williams JBW. The struc-
tured clinical interview for DSM-III-R personality disorders
(SCID-II). Part I: description. J Pers Disoid 1995 ; 9: 83-91
13 Hyler SH, Rieder RO, Williams JBW, Snitzer RL. Hendler J.
Lyons M. The Personality Diagnostic Questionnaire: Develop-
ment and preliminary results. J Pers Disord 1988 ; 2 : 229-37
14 Hyler SE, Skodol AE, Kellman HD, Oldham JM, Rosnick L.
Validity of the Personality Diagnostic Questionnaire-Revised:
Comparison with two structured interviews. Am J Psychiatry
1990;147: 1043-8
15 Kellner R, Rada R, Andersen T, Pathak D. The effects of chlor-
diazepoxide on self-rated depression, anxiety an well-being.
Psychopharmacology 1979 ; 65 : 185-91
16 Loranger AW, Sartorius N, Andreoli A, Berger P, Buchheim P,
Channabasavanna SM et al. The International Personality Dis-
order Examination. Arch Gen Psychiatry 1994 ; 51 : 215-24
17 Maffei C, Fossati A, Agostoni I, Barraco A, Bagnato M, Debo-
rah D et al. Interrater reliability and internal consistency of the
structured clinical interview for DSM-IV Axis-II personality dis-
orders @CID-II), version 2.0. J Per-s Disord 1997 ; 11 : 279-84
18
19
20
21
22
23
24
25
26
27
28
McDavid JD, Pilkonis PA. The stability of personality disorder
Diagnosis. / Pers Disord 1996 ; 10 : l-15
Ottosson H, Bodlund 0, Ekselius L, Lindstrom E, von
Knorring L, Kullgren G, Soderberg S. The DSM-IV and
ICD-10 Personality Questionnaire (DIP-Q): Construction
and preliminary validation. Nord J Psychiatry 1995 ; 49 :
285-91
Reich JH, Yates W, Nduaguba M. Prevalence of DSM-III per-
sonality disorders in the community. Sot Prychialr Epidem
1989;24: 12-6
Shrout PE, Fleiss JL. Intraclass correlations: Uses in assessing
rater reliability. Psycho1 Bull 1979 ; 86 : 420-8
Skodol AE, Spizer RL. The development of reliable diagnostic
criteria in psychiatry. Ann Rev Med 1982 ; 33 : 317-26
Uebersax JS. Diversity of decision - making models and the
measurement of interrater agreement. Psycho1 Bull 1987 ; 101 :
140-6
Weissman MM. The epidemiology of personality disorders: A
1990 Update. J Pers Disord 1993 ; 7 : 44-62
World Health Organization. Internafional Statistical Classifica-
tion of Diseases and Related Health Problems, 9th revision.
Geneva: WHO; 1978
World Health Organization. International Statistical Classifica-
tion of Diseases and Related Health Problems, 10th revision.
Geneva: WHO; 1992
World Health Organization. The ICD-IO Classification of Men-
tal and Behavioural Disorders. Diagnostic criteria for research.
Geneva: WHO; 1993
Zimmerman M. Diagnosing personality disorders - a review of
issues and research methods. Arch Gen Psychiatry 1994 ; 51 :
225-45