The heritability of cluster A personality disorders assessed by both personal interview and questionnaire.
ABSTRACT Personality disorders (PDs) as assessed by questionnaires and personal interviews are heritable. However, we know neither how much unreliability of measurement impacts on heritability estimates nor whether the genetic and environmental risk factors assessed by these two methods are the same. We wish to know whether the same set of PD vulnerability factors are assessed by these two methods.
A total of 3334 young adult twin pairs from the Norwegian Institute of Public Health Twin Panel (NIPHTP) completed a questionnaire containing 91 PD items. One to 6 years later, 1386 of these pairs were interviewed with the Structured Interview for DSM-IV Personality (SIDP-IV). Self-report items predicting interview results were selected by regression. Measurement models were fitted using Mx.
In the best-fit models, the latent liabilities to paranoid personality disorder (PPD), schizoid personality disorder (SPD) and schizotypal personality disorder (STPD) were all highly heritable with no evidence of shared environmental effects. For PPD and STPD, only unique environmental effects were specific to the interview measure whereas both environmental and genetic effects were found to be specific to the questionnaire assessment. For SPD, the best-fit model contained genetic and environmental effects specific to both forms of assessment.
The latent liabilities to the cluster A PDs are highly heritable but are assessed by current methods with only moderate reliability. The personal interviews assessed the genetic risk for the latent trait with excellent specificity for PPD and STPD and good specificity for SPD. However, for all three PDs, the questionnaires were less specific, also indexing an independent set of genetic risk factors.
- [Show abstract] [Hide abstract]
ABSTRACT: There is scant knowledge on the presentation of paranoid personality disorder in clinical psychiatric settings. In this study, the charts of 15 consecutive patients diagnosed with paranoid personality disorder were retrospectively analyzed. Information was gathered concerning descriptive behavioral and psychopathological characteristics including occurrence of delusional psychosis. With respect to ICD-10 research criteria, ‘excessive sensitivity’ and ‘self-reference’ were most consistently present. Conversely, ‘suspiciousness’ and ‘jealousy’ were only recorded in half of the individuals. Seven individuals had episodes of delusional psychosis and four others were for periods of time suspected of delusion development. Occurrence of delusions was associated with a prolonged psychiatric course. All individuals had positive depression ratings. Implications for conceptualization of paranoid personality disorder are discussed.Current psychology (New Brunswick, N.J.) 06/2014; 33(2):219-228. · 0.45 Impact Factor
- [Show abstract] [Hide abstract]
ABSTRACT: The study of schizotypal personality disorder (SPD) is important clinically, as it is understudied, challenging to treat, often under-recognized or misdiagnosed, and associated with significant functional impairment. SPD also represents an intermediate schizophrenia-spectrum phenotype, and therefore, can provide a better understanding of the genetics, pathogenesis, and treatment of related psychotic illnesses. In this review we discuss recent findings of SPD related to epidemiology and functional impairment, heritability and genetics, working memory and cognitive impairments, social-affective disturbances, and neurobiology. Additionally, we examine the challenges associated with treating patients with SPD, as well as clinical recommendations. Finally, we address future directions and areas in need of further exploration.Current Psychiatry Reports 07/2014; 16(7):452. · 3.05 Impact Factor
- [Show abstract] [Hide abstract]
ABSTRACT: Personality disorders (PDs) reduce global functioning, are associated with high levels of work disability, and are thus also likely to influence long-term sick leave (LTSL). Previous research has indicated significant genetic influence on both DSM-IV PDs and LTSL. To what degree genes contributing to PDs also influence LTSL has not been investigated. The aims of the current study were to investigate which PDs were significantly associated with LTSL, to what extent the genetic contributions to these PDs account for the heritability of LTSL, and to explore the hypothesis of a causal association between PDs and LTSL. The sample consisted of 2,771 young, adult Norwegian twins, born 1967-1979. PDs were assessed using the Structured Interview for DSM-IV Personality (SIDP-IV). The age range for the interview was 20-32. The data were subsequently linked to public records of LTSL (sick leave >16 days) up to 11 years later. The odds ratio for being in the highest LTSL category (>15% sick leave) when fulfilling the DSM-IV criteria for any PD diagnosis was 2.6 (1.8-3.8, 95% CI). Dimensional representations of schizotypal, paranoid, and borderline PD were independently and significantly associated with LTSL. The heritability of LTSL was 0.50. Genetic factors shared with the PDs accounted for 20% of this. The association between PDs and LTSL was due to shared genetic and not environmental influences, and was mainly explained by one common genetic factor. The hypothesis of a causal association was not supported, indicating that the association is explained by overlapping genetic liability between PDs and LTSL.Twin Research and Human Genetics 01/2014; 17(01):1-9. · 1.92 Impact Factor
The heritability of cluster A personality disorders
assessed by both personal interview and questionnaire
KENNETH S. KENDLER1,2*, JOHN MYERS1, SVENN TORGERSEN3,
MICHAEL C. NEALE1,2AND TED REICHBORN-KJENNERUD4
University, Richmond, VA, USA;3Institute of Psychology, University of Oslo, Center for Child and Adolescent
Mental Health Eastern and Southern Norway and Nic Waal’s Institute, Norway;4Division of Mental Health,
Norwegian Institute of Public Health and the Institute of Psychiatry, University of Oslo, Norway
1Psychiatry and2Human Genetics, Medical College of Virginia of Virginia Commonwealth
Background. Personality disorders (PDs) as assessed by questionnaires and personal interviews are
heritable. However, we know neither how much unreliability of measurement impacts on herita-
bility estimates nor whether the genetic and environmental risk factors assessed by these two
methods are the same. We wish to know whether the same set of PD vulnerability factors are
assessed by these two methods.
Method. A total of 3334 young adult twin pairs from the Norwegian Institute of Public Health
Twin Panel (NIPHTP) completed a questionnaire containing 91 PD items. One to 6 years later,
1386 of these pairs were interviewed with the Structured Interview for DSM-IV Personality (SIDP-
IV). Self-report items predicting interview results were selected by regression. Measurement models
were fitted using Mx.
Results. In the best-fit models, the latent liabilities to paranoid personality disorder (PPD), schizoid
personality disorder (SPD) and schizotypal personality disorder (STPD) were all highly heritable
with no evidence of shared environmental effects. For PPD and STPD, only unique environmental
effects were specific to the interview measure whereas both environmental and genetic effects were
found to be specific to the questionnaire assessment. For SPD, the best-fit model contained genetic
and environmental effects specific to both forms of assessment.
Conclusions. The latent liabilities to the cluster A PDs are highly heritable but are assessed by
current methods with only moderate reliability. The personal interviews assessed the genetic risk for
the latent trait with excellent specificity for PPD and STPD and good specificity for SPD. However,
for all three PDs, the questionnaires were less specific, also indexing an independent set of genetic
Two different approaches have been developed
towards the measurement of personality dis-
orders (PDs): structured interviews (SIs) (e.g.
Loranger, 1988; First et al. 1995; Pfohl et al.
1995) and self-report questionnaires (SRQs)
(e.g. Livesley et al. 1992; Clark, 1993). To date,
all twin studies of PDs have used one these
assessment methods (Livesley et al. 1993; Jang
et al. 1996; Torgersen et al. 2000; Kendler et al.
2006; Reichborn-Kjennerud et al. 2006). A gen-
etically informative study that included the
symptoms of PDs as assessed by both ap-
proaches could clarify the degree to which these
two assessment methods index the same genetic
and environmental risk factors. That is, such a
study could answer the question: ‘Do SIs and
* Address for correspondence: Kenneth S. Kendler, M.D.,
Virginia Institute for Psychiatric and Behavioral Genetics, Virginia
Commonwealth University Medical School, Box 980126, 800 E.
Leigh Street, Room 1-123, Richmond, VA 23298-0126, USA.
Psychological Medicine, 2007, 37, 655–665.
f 2007 Cambridge University Press
First published online 16 January 2007Printed in the United Kingdom
SRQs assess the same or different PD-related
In twin or adoption studies, the calculation of
heritability from a single assessment confounds
the roles of individual-specific environment
and measurement error. For example, differ-
ences in the scores of a pair of monozygotic
(MZ) twins on a particular PD could result
from differences in their prior environmental
experiences or from assessment error. Error in
the assessment of PDs is a concern because PD
symptoms might not be as temporally stable
as once thought (McGlashan et al. 2005) and
agreement on the level of PD symptomatology
across assessment methods is typically modest
(Zimmerman, 1994; Widiger & Coker, 2002).
Furthermore, the heritabilities obtained re-
cently from a population-based sample for
cluster A (Kendler et al. 2006) and cluster C PDs
(Reichborn-Kjennerud et al. 2006) assessed
by SI were lower than those typically seen for
normative personality traits (e.g. Loehlin, 1992;
Lake et al. 2000). Such low heritabilities could
arise either because PDs as assessed by SIs
are reliably measured and have modest heri-
tabilities or because they have heritabilities
similar to standard personality traits but are
assessed with lower reliability.
In the Norwegian Institute of Public Health
Twin Panel (NIPHTP), interviews including the
Structured Interview for DSM-IV Personality
(SIDP-IV; Pfohl et al. 1995) were conducted
between 1999 and 2004. The results of these in-
terviews form the basis of our prior twin analy-
ses of PDs in this sample (Kendler et al. 2006;
Reichborn-Kjennerud et al. 2006). However, in
1998, the NIPHTP completed the Dysfunctional
Personality Questionnaire, a 91-item SRQ that
assessed dimensional representations of a wide
range of PDs.
In this report, we incorporate into twin
models assessments by both SI and SRQ of di-
mensional representations of the three cluster
A PDs: paranoid personality disorder (PPD),
schizoid personality disorder (SPD) and schizo-
typal personality disorder (STPD). With these
models, we addressed two questions. First, ac-
counting for unreliability of measurement by
using two measures differing in both time and
mode of assessment, how heritable is the liab-
ility to PPD, SPD and STPD? Second, what is
the relative sensitivity and specificity with which
SI and SRQ assessments index the underlying
liability to these individual PDs?
Sample and assessment methods
Twins in the NIPHTP (described in detail else-
where; Harris et al. 2002) were identified
through the Norwegian National Medical Birth
Registry, established 1 January 1967, which re-
ceives mandatory notification of all live births.
The current panel began with 15370 like- and
unlike-sexed twins born 1967–1979. Two ques-
tionnaire studies have been conducted in 1992
(twins born 1967–1974) and in 1998 (twins born
1967–1979). Altogether, 12700 twins received
the second questionnaire, and 8045 responded
after one reminder (response rate 63%). The
sample included 3334 pairs and 1377 single re-
sponders. The second questionnaire contained
the Dysfunctional Personality Questionnaire,
which contained 91 items selected by one of us
(S.T.)to assess DSM PD traits. These items were
instruments (Foulds, 1965; Lazare et al. 1966;
Conte et al. 1980) or developed and sub-
The Dysfunctional Personality Questionnaire
was also administered in a prior large-scale
(Torgersen et al. 2001).
In the personal interview phase of this study,
twins were not approached for interview until
preliminary consent had been obtained from
both members of the pair. Participants were re-
cruited among 3153 complete pairs who, in the
second questionnaire, agreed to participate in
the interview study, and 68 pairs who were
drawn directly from the NIPHTP. Of these 3221
eligible pairs, 0.8% were unwilling or unable to
participate, and in 16.2% of pairs only one twin
agreed to the interview. After two contacts re-
questing participation, 38.2% did not respond.
Where only one twin agreed to the interview, the
uncooperative twin either did not respond to
our contacts (96.0%), had an unknown address
(2.9%) or refused (1.1%). Altogether, 2794
twins (44% of those eligible) were interviewed
for the assessment of PDs. Approval was re-
ceived from the Norwegian Data Inspectorate
and the Regional Ethical Committee, and writ-
ten informed consent was obtained from all
656K. S. Kendler et al.
participants after complete description of the
PDs were assessed by a Norwegian version of
the SIDP-IV (Pfohl et al. 1995). Both DSM-III-
R and DSM-IV versions have been used pre-
viously in major Norwegian studies (Torgersen
et al. 2001; Helgeland et al. 2005). The SIDP
is a comprehensive semi-structured diagnostic
interview for the assessment of all DSM-IV
axis II diagnoses. The instrument includes non-
pejorative questions organized into topical sec-
tions to produce a natural flow in the interview.
The SIDP uses the ‘5-year rule’, meaning that
behaviors, cognitions and feelings that have
been predominant for most of the past 5 years
are considered to be representative of the in-
dividual’s long-term personality functioning.
This rule is supported by empirical evidence
of high stability of normative personality traits
DSM-IV criterion is scored as 0=absent, 1=
Interviewers (mostly experienced psychology
students and psychiatric nurses) were trained by
one psychiatrist and two psychologists with
previous extensive experience with the instru-
ment. The interviews, largely conducted face-to-
face, were carried out between June 1999 and
May 2004. For practical reasons, 231 interviews
(8.3%) were conducted over the telephone.
Members of a pair were assessed by different
Inter-rater reliability was assessed by two
raters scoring 70 audiotaped interviews. They
obtained high intra-class (and polychoric) cor-
relations for the number of endorsed criteria at
the subthreshold level: PPD +0.92 (+0.94),
SPD +0.81 (+0.86) and STPD+0.86 (+0.90).
Zygosity was determined by standard ques-
tionnaire items used in a discriminant analysis
with results of 24 microsatellite markers avail-
able on 676 of the like-sex pairs in the sample.
From these data, we estimated that, in our entire
1% (Neale, 2003).
Subjects missing scores on 10% or more of their
SRQ items (210 out of 8030, or 2.6%) were ex-
cluded from the analysis, producing a total
sample of 7820. For those missing less than
10% (1782 twins, or 22.8% of the sample, of
whom 1001 were missing a single item and
343 two items), the missing values were imputed
using IVEware (Raghunathan et al. 2000).
Separately for each of the three cluster A PDs,
we then conducted step-wise ordinal logistic re-
gression analyses using PROC LOGISTIC in SAS
(SAS Institute, 2006) in twin 1 from each
twin pair (n=1363), attempting to predict, from
responses to the Dysfunctional Personality
Questionnaire, the number of criteria endorsed
with a score of 1 or higher. The significance level
for entry and exit into this regression analysis
was 0.20. We then took these resulting items and
repeated the analyses in the second twin in each
pair (n=1368), this time using an entry and exit
criteria of 0.05.
We use a liability-threshold model to estimate
the genetic and environmental contributions to
twin resemblance for PDs. For SRQs, this liab-
ility is indexed by the number of items re-
sponded to in the positive direction. For the SI,
liability is indexed by the number of DSM-IV
criteria endorsed at the subthreshold level. In
this paper, we refer, for convenience, to PDs but
we are in fact assessing a dimensional represen-
tation of these PDs, operationalized as the
number of endorsed criteria. We showed pre-
viously, using the multiple threshold model, that
the number of criteria for the cluster A PDs
could be regarded as differences of severity on
a single normally distributed continuum of
liability (Kendler et al. 2006).
The model-fitting used here divides the vari-
ation in liability to PDs into three classes: (i)
additive genetic (A), which contributes twice
as much to the correlation in MZ twins as di-
zygotic (DZ) twins, (ii) family or ‘common’ en-
vironment (C), which contributes equally to the
correlation in MZ and DZ twins, and (iii) indi-
vidual specific environment (E), which reflects
environmental experiences not shared by both
members of a twin pair and therefore contribute
to differences between them in their liability
Our model for PDs, previously referred to as
a measurement model (Foley et al. 1998), uses
simultaneously both our SRQ and SI data from
our twin sample. As illustrated in Fig. 1, the
model assumes that there is a true latent liability
to each PD. The latent liability to the PD is
Heritability of cluster A personality disorders 657
indexed by both items from the SRQ and DSM-
IV criteria assessed by SI. The magnitude of this
relationship is reflected in the paths lSand lI,
where S and I refer to self-report and inter-
viewed-based assessments respectively. Genetic
(A), shared environmental (C) and individual-
specific environmental effects (E) are included in
the model for the latent liability to the PD (in-
dicated by the subscript L), specific to the self-
report questionnaire (indicated by the subscript
S), and specific to the self-report structured in-
terview (indicated by the subscript I).
In our measurement model (Fig. 1), if there
are no shared environmental effects, the two l
paths are unconstrained, and genetic effects
ment; models that assign the unique genetic
effect to one or other time of measurement will
typically have identical fits. To avoid this con-
found, we added the constraint to our models:
lIolS. Given that the SI was specifically
designed to operationalize the DSM PD criteria,
we have made the plausible assumption that the
latent liability to the individual PDs would be
indexed at least as well by the SI as by the SRQ.
To maximize power, we fitted models, in the
software program Mx (Neale et al. 2003), to the
raw data from all twins including those without
a co-twin and twins who had completed the
SRQ but not the SI. Alternative models are
evaluated by comparing the difference in their
x2relative to the difference in their degrees of
freedom (df), according to the principle of par-
simony – models with fewer parameters are
preferable if they do not provide a significantly
worse fit. We operationalized this balance be-
tween explanatory power and parsimony by the
use of Akaike’s Information Criterion (AIC;
is calculated as x2x2df, where df equals the
difference in the number of degrees of freedom
between the two models being compared. The
lower (or more negative) the value of the AIC,
the better is the balance between explanatory
power and parsimony.
Questionnaire item selection by regression
NineSRQ items from
Personality Questionnaire were selected from
our two-stage regression analyses to maximally
predict the number of endorsed PPD criteria at
interview. As shown in Table 1, these items had
varied content. Two of them (items 126 and 127)
directly reflect aspects of suspiciousness. Two
items reflect general emotionality (items 99 and
116) and two reflect problems in self-concept
(items 157 and 165). Items 137 and 149 reflect
problems in inter-personal relationships, being
either unpredictable or aggressive. The poly-
choric correlation between the sum of these
SRQ items and the number of endorsed PPD
criteria at SI (n=2731) was +0.34.
Only six SRQ items were selected from the
two-stage multiple regression analysis to predict
the number of endorsed SPD criteria at inter-
view. As shown in Table 1, four of them (items
92, 107 135 and 148) directly reflect schizoid/
introverted traits. Of note, item 126, also seen in
PPD, was scored negatively, so not endorsing
this item predicts meeting SPD criteria. The
correlation between the sum of these items and
liability to personality disorder (PD) is indexed by both items from
the self-report questionnaire (SRQ) and DSM-IV criteria as assessed
the structured interview. The magnitude of this relationship is re-
flected in the paths lSand lIrespectively, where S and I refer to self-
report questionnaire and interviewed-based assessment. Genetic (A),
shared environmental (C) and individual-specific environmental
effects (E) are included in the model for the latent liability to PD
(indicated by subscript L), specific to the self-report questionnaire
(indicated by subscript S), and specific to the self-report structured
interview (indicated by subscript I).
The measurement model used in this report. The latent
658 K. S. Kendler et al.
the number of endorsed SPD criteria at SI
Seven SRQ items from the Dysfunctional
Personality Questionnaire were selected from
our two-stage multiple regression analysis to
predict the number of endorsed STPD criteria at
interview. As shown in Table 1, the content of
these items was variable and reflected traits of
suspiciousness (item 127), social ill-ease (items
128 and 158), soft psychotic-like symptoms
(item 138) and lack of even-temperedness (item
99) and groundedness (item 110). The poly-
choric correlation between the sum of these
items and the number of endorsed STPD criteria
at SI was +0.32.
Paranoid personality disorder (PPD)
We begin by describing in detail the results
in model fitting for PPD (Table 2). Model I,
the full model, allowed for qualitative and
quantitative sex effects and genetic shared and
individual-specific environmental effects for
the latent index to PPD as well as the PPD
measurements obtained by SRQ and SI. In
models II and III, we omitted the qualitative
and quantitative sex effects respectively, both
times producing an improvement in the fit as
indexed by the AIC. Working from model III,
we then dropped all shared environmental and
all genetic effects respectively in models IV
but deteriorated substantially with model V.
Working from model IV, we then dropped one
at a time, in models VI through VIII, additive
genetic effects for the latent liability to PPD and
then for the PPD measurements obtained by
SRQ and SI. Of these three models, by far the
best fit was obtained by model VIII, which
omitted the genetic effects for the personal inter-
view. We tried to simplify the model further by
Table 1.Items from the self-report questionnaire selected to assess liability to paranoid, schizoid
and schizotypal personality disorder
Item numberItem content
Paranoid personality disorder
x99On the average, I am calm and even-tempered
116My mood will easily change in accordance with the environment
126I easily get hurt if someone ridicules me or makes derogatory remarks
127It is difficult for me to trust people, since they very often turn their
backs on me or let me down
137I behave in a way that people consider as unexpected or changing
149Some of the people that know me think that I am rather aggressive
153 People treat me as if I were an ‘object’
157 I have been involved in relationships where I was unable to identify whether
thoughts and emotions belonged to me or the other person
165I wonder who I really am
Schizoid personality disorder
92 Among other people, I prefer to stand back
107I decided long ago that it was best to have little to do with others
115If things turn up suddenly and unexpectedly, I feel completely confused
x126I easily get hurt if someone ridicules me or makes derogatory remarks
135I am afraid of close relationships
x148Many people regard me as a lively person
Schizotypal personality disorder
95I act the way I feel
x99 On the average, I am calm and even-tempered
x110I keep both feet on the ground. I stick to what is tangible rather
than be lost in reverie
127It is difficult for me to trust people, since they very often turn their backs
on me or let me down
128Somehow I feel it is hard for me to know how I should behave
among other people
138 I have seen or heard things that have no logical explanation
158 In situations where I ought to speak out, I sometimes become speechless
and unable to say a single word
A minus sign (x) before the item indicates that it was reverse coded.
Heritability of cluster A personality disorders659
dropping genetic effects for the SRQ (model IX)
and the latent liability (model X) but the fit de-
teriorated substantially in both cases, indicating
that model VIII was our best fit.
Parameter estimates (and 95% confidence
intervals) for this best-fit model for PPD are
shown in Fig. 2. Five results are noteworthy.
First, as indicated in the model fitting, shared
environment plays no apparent role in the
etiology of PPD. Second, the latent liability to
PPD, as indexed by SRQ and SI, is fairly heri-
table, with an estimated heritability of 66%.
Third, the latent liability to PPD is indexed
equally well by the SRQ (lS=+0.58) and the
SI (lI=+0.58). Fourth, the only factors that
impact specifically on SI are individual-specific
environmental in nature. That is, the model sug-
gests that a set of factors unique to each member
of a twin pair, such as measurement error and
environmental experiences, acts specifically on
the SI assessment of PPD. Fifth, by contrast, we
found evidence for substantial genetic effects
specific to the SRQ assessment of PPD. Using
the rules of path analysis, we can calculate that
the total heritability of our SRQ PPD-related
items is 43%. Of that total, 51% comes from
genes that index the latent liability to PPD (and
are thus shared with the SI assessed criteria) and
49% are unique to the SRQ measures. These
results indicate that the SRQ and SI assessments
of PPD traits are not genetically equivalent. The
SRQ measures index a broad array of genetic
risk factors, only some of which are shared by
the SI measures.
Schizoid personality disorder (SPD)
The model fitting results for SPD, depicted in
Table 2, differed in one important way from
Table 2. Model fitting results for paranoid, schizoid and schizotypal personality disorder
Qual, Qualitative sex effects; Quant, quantitative sex effects; df, degrees of freedom; AIC, Akaike’s Information Criterion (Akaike, 1987).
* Best-fit model.
660K. S. Kendler et al.
those obtained for PPD. Model IV, which con-
tains specific genetic effects for both the SI and
SRQ, fit slightly better than model VIII, the
best-fit model for PPD. Because the difference in
AIC values between these two models (0.58
units) is too small to permit us to choose be-
tween them with confidence, we present par-
ameter estimates for both in Figs 3a and 3b
respectively. Model IV predicts a heritability of
the latent liability to SPD of 55%. This liability
is equally well indexed by the SRQ (lS=+0.55)
and the SI (lI=+0.55). Genes specific to each
method of measurement contribute a substan-
tially greater proportion of variance to the SRQ
(0.442=19%) than to the SI (0.302=9%).
Expressed in another way, the proportion of
total genetic effects on the SRQ and SI measures
of SPD that derive from factors unique to that
form of measurement equals 53% and 35% re-
spectively. The parameter estimates from model
VIII (Fig. 3b) are similar. The latent liability to
SPD is estimated to be slightly more heritable
(59%) and genetic specific effects are seen only
for the SRQ. The proportion of genetic effects
on the SRQ measures of SPD due to genetic ef-
fects specific to that form of assessment is 50%.
Schizotypal personality disorder (STPD)
Model fitting results for STPD were identical to
those seen for PPD producing model VIII as the
in parentheses from the best-fit model IV (see Table 2) for paranoid
personality disorder (PPD). Path estimates must be squared to equal
the proportion of variance in the dependent variable accounted for
by the independent variable. A stands for additive genetic and E for
individual-specific environmental effects.
Parameter estimates in bold and 95% confidence intervals
in parentheses from (a) the best-fit model IV and (b) the second-best-
fit model VIII (see Table 2) for schizoid personality disorder (SPD).
Path estimates must be squared to equal the proportion of variance
in the dependent variable accounted for by the independent variable.
A stands for additive genetic and E for individual-specific environ-
Parameter estimates in bold and 95% confidence intervals
Heritability of cluster A personality disorders661
best-fit model. The parameter estimates of this
model for STPD are shown in Fig. 4. As seen
with PPD, the best-fit model contained no evi-
dence for shared environmental effects for
STPD. The heritability of the latent liability to
STPD is estimated at 72%, modestly higher
than that seen for PPD and SPD. As with the
other two PDs, the latent liability to STPD was
equally well indexed by the SRQ (lS=+0.57)
and the SI (lI=+0.57). As with PPD, but not
SPD, in the best-fit model for STPD, the only
factors that impact specifically on the SI are in-
dividual-specific environmental in nature, but
substantial genetic effects specific to the SRQ
assessment were seen. Of the total heritability of
our SRQ STPD-related items, 61% comes from
genes that index the latent liability to STPD and
39% are unique to the SRQ measures.
In our previous analyses of the cluster A PDs
as assessed at SI, we were surprised at the
(Kendler et al. 2006), which were substantially
lower than commonly seen for normative per-
sonality (e.g. Loehlin, 1992; Lake et al. 2000).
Were these results an indication that the cluster
A constructs as operationalized by SI were less
genetic than more standard personality traits or
could they have arisen because of problems of
measurement? We obtained a clear answer to
this crucial question in the present analyses.
When examined using a measurement model
that corrects for unreliability of assessment,
the heritability of PPD, SPD and STPD was
fairly high, ranging from 55% to 72%. Taking
into account imperfections in measurement, the
etiologic role of genetic factors appears to be at
least as strong in these pathologic dimensions of
personality function as they are in the better
studied normative traits.
These analyses also addressed a subtler but
equally important question: what is the relative
sensitivity and specificity with which SI and
SRQ assessments index the underlying liability
to the individual PDs? For all three PDs, the
best-fit model estimated the two lambda paths
(lSand lI) to be equal, meaning that the SI and
SRQ assessments were equally sensitive at de-
tecting the latent liability to PPD, SPD and
STPD. However, they were not equally specific.
For two of the three cluster A PDs, the best-fit
model contained no genetic risk factors unique
to the SI assessment. For the third PD (SPD),
the impact of these unique genetic factors on the
SI was modest, and in a second model that
fit nearly as well, was absent altogether. By
contrast, for all three of these PDs, the best-fit
model contained a unique set of genetic risk
factors, which impacted substantially on the
self-report questionnaire data, that were not re-
flected in the interview-based measures. Thus,
in our results, the SI had greater specificity in
indexing the genetic risk for the latent liability to
the individual PDs than did the SRQ.
There are two plausible explanations for
these findings. First, the items in our SRQ did
not map in any one-to-one manner with the
DSM criteria for the cluster A PDs. Therefore,
the questionnaire-specific genetic effects might
result from item content unique to the SRQ.
If this is the case, then these results are prob-
ably specific to the Dysfunctional Personality
Questionnaire and would probably not general-
ize to other SRQ measurements of PD. Second,
the genetic effects specific to the SRQ might
in parentheses from the best-fit model VIII (see Table 2) for schizo-
typal personality disorder (STPD). Path estimates must be squared
to equal the proportion of variance in the dependent variable ac-
counted for by the independent variable. A stands for additive gen-
etic and E for individual-specific environmental effects.
Parameter estimates in bold and 95% confidence intervals
662K. S. Kendler et al.
arise because of something more fundamental to
the assessment method; that is, SRQ responses
might tap a set of genetically influenced traits
distinct from those that impact on SI assess-
We do not have data to directly address the
relative plausibility of these two explanations.
We can, however, indirectly evaluate them in
in these analyses were selected for their ability to
predict results of the SIDP interview, perhaps
these relationships reflect only idiosyncratic fea-
tures of our own data. However, the Dysfunc-
tional Personality Questionnaire was also used
along within the SIDP interview in an earlier
epidemiological study in Norway (Torgersen
et al. 2001). In 2053 adults living in and around
Oslo, the polychoric correlations between the
SRQ items depicted in Table 1 and the count
of DSM criteria assigned on personal interview
with the SID-P were +0.35, +0.39 and +0.33
for PPD, SPD and STPD respectively. These
correlations were only slightly higher than those
observed in our twin sample and this difference
could easily have arisen because these two
measures were completed within a week of one
another in the epidemiological study and at least
a year apart in our twin sample.
Second, the modest correlations between our
SI and SRQ assessments for the three cluster A
PDs, averaging +0.32, might suggest that these
two measures were indeed tapping different
traits. However, these results are congruent with
the prior evidence for modest levels of agree-
ment between different assessment approaches
to PDs when the item content is similar
(Zimmerman, 1994). Widiger & Coker (2002)
reviewed 18 studies reporting correlations in
dimensional PD ratings between SRQ and SI
assessments that, for the cluster A disorders,
averaged +0.33. Five of these studies used the
1994), which contained items written to rep-
resent each of the DSM PD diagnostic criteria.
These studies obtained correlations in the di-
mensional scores between the SRQ and SI
measures of PPD, SPD and STPD, which aver-
aged +0.35, hardly greater than that found in
our sample given that these studies typically
obtained SRQ and SI measurements at the same
time while these measures were separated by a
year or more in our sample.
varied had we used different SRQ items – pick-
ing PPD as an example because its DSM-IV
criteria set are relatively homogeneous in con-
tent. K.S.K. reviewed all SRQ items and picked
seven items typifying the features of PPD. Three
of these were among those selected by the re-
gression analyses. Using these items to form a
new SRQ assessment of PPD, model VIII was
again the best-fit model and it produced par-
ameter estimates very similar to those seen in
Fig. 2. For example, using both the ‘clinician-
selected’ and regression-derived items for PPD,
49% of the total genetic variance for the SRQ
measures come from sources unique to that
mode of assessment. The specific genetic factors
obtained for the SRQ assessments, at least for
PPD, are not simply a result of the apparently
variable content of the selected items.
The genetic effects specific to the SRQ that we
found for all three cluster A PDs in this sample
certainly could reflect idiosyncratic features of
our SRQ items and/or our sample. However,
our results suggest that the second hypothesized
explanation, that these unique genetic factors
might reflect something fundamental about
how PDs are assessed by SRQ versus SI, de-
serves serious consideration. We hope that fu-
ture research will further clarify this important
These results should be interpreted in the
context of four potentially significant methodo-
logical limitations. First, we examined dimen-
sional representations of PPD, SPD and STPD
rather than diagnoses. This made sense given
that our goal was to compare the performance
of SRQ and SI evaluations of PDs. This ap-
proach was also partially dictated by the rarity
of the fully syndromal diagnoses in our sample
and meant that much of the information in our
analyses about the SI assessment of the cluster
A PDs came from individuals who endorsed one
or a few criteria. However, many researchers
have argued that the PDs are best conceptu-
alized as dimensional constructs rather than
dichotomous diagnostic entities (Oldham &
Skodol, 2000; Skodol et al. 2005; Widiger &
Samuel, 2005; Widiger & Simonsen, 2005;
Cramer et al. 2006). Second, our sample only
contained one assessment of PDs by SI and one
Heritability of cluster A personality disorders 663
by SRQ separated by at least a year. Therefore,
we could not determine the degree to which the
correlation between our two assessments were
the result of method variance, underlying un-
reliability of assessment and/or true change in
PD symptoms over the time separating the two
assessments. Third, our sample included only
young adult Norwegians. Our results cannot be
assumed to extrapolate to other cultural and
ethnic groups. Fourth, considerable attrition
was observed in this sample from the original
birth registry through three waves of contact.
Detailed analyses of the predictors of non-
response across waves suggest that psychopath-
ology was unrelated to cooperation although
retention in the sample was strongly predicted
by female sex, monozygosity, older age and
higher educational status (Harris et al. unpub-
lished observations). In addition, as outlined
previously (Kendler et al. 2006), cooperation at
the stage of personal interview was not pre-
dicted by scores for any cluster A FPD dimen-
sion as assessed by SRQ. While we cannot be
certain that our sample was representative with
respect to cluster A psychopathology, these find-
ings suggest that a significant bias is unlikely.
This work was supported in part by grant MH-
068643 from the National Institutes of Health,
and by grants from The Norwegian Research
Council, The Norwegian Foundation for Health
and Rehabilitation, The Norwegian Council for
Mental Health, and The European Commission
under the program ‘Quality of Life and
Management of the Living Resources’ of the
5th Framework Program (no. QLG2-CT-2002-
DECLARATION OF INTEREST
Akaike, H. (1987). Factor analysis and AIC. Psychometrika 52,
Clark, L. A. (1993). Schedule for Nonadaptive and Adaptive
Personality (SNAP): Manual for Administration, Scoring, and
Interpretation. University of Minnesota Press: Minneapolis.
Conte, R. C., Plutchik, R., Karasu, T. B. & Jerrett, I. (1980). A self-
report borderline scale. Discriminative validity and preliminary
norms. Journal of Nervous and Mental Disease 168, 428–435.
Cramer, V., Torgersen, S. & Kringlen, E. (2006). Personality dis-
orders and quality of life. A population study. Comprehensive
Psychiatry 47, 178–184.
First, M. B., Spitzer, R. L., Gibbon, M. & Williams, J. B. W. (1995).
Structured Clinical Interview for DSM-IV Axis II Personality
Disorders – Patient (SCID-I/P) (Version 2.0). New York State
Psychiatric Institute: New York.
Foley, D. L., Neale, M. C. & Kendler, K. S. (1998). Reliability of a
lifetime history of major depression: implications for heritability
and co-morbidity. Psychological Medicine 28, 857–870.
Foulds, G. A. (1965). Personality and Personal Illness. Tavistock:
Harris, J. R., Magnus, P. & Tambs, K. (2002). The Norwegian
Institute of Public Health Twin Panel: a description of the sample
and program of research. Twin Research 5, 415–423.
Helgeland, M. I., Kjelsberg, E. & Torgersen, S. (2005). Continuities
between emotional and disruptive behavior disorders in ado-
lescence and personality disorders in adulthood. American Journal
of Psychiatry 162, 1941–1947.
Hyler, S. E. (1994). Personality Diagnostic Questionnaire-4 (PDQ-4)
Unpublished Test. New York State Psychiatric Institute: New
Jang, K. L., Livesley, W. J., Vernon, P. A. & Jackson, D. N. (1996).
Heritability of personality disorder traits: a twin study. Acta
Psychiatrica Scandinavica 94, 438–444.
Kendler, K. S., Czajkowski, N., Tambs, K., Torgersen, S., Aggen,
S. H.,Neale, M. C.&Reichborn-Kjennerud,
Dimensional representation of DSM-IV Cluster A personality
disorders in a population-based sample of Norwegian twins:
Lake, R. I., Eaves, L. J., Maes, H. H., Heath, A. C. & Martin, N. G.
(2000). Further evidence against the environmental transmission
of individual differences in neuroticism from a collaborative study
of 45,850 twins and relatives on two continents. Behavior Genetics
Lazare, A., Klerman, G. L. & Armor, D. J. (1966). Oral, obsessive,
and hysterical personality patterns. An investigation of psycho-
analytic concepts by means of factor analysis. Archives of General
Psychiatry 14, 624–630.
Livesley, W. J., Jackson, D. N. & Schroeder, M. L. (1992). Factorial
structure of traits delineating personality disorders in clinical and
general population samples. Journal of Abnormal Psychology 101,
Livesley, W. J., Jang, K. L., Jackson, D. N. & Vernon, P. A. (1993).
Genetic and environmental contributions to dimensions of per-
sonality disorder. American Journal of Psychiatry 150, 1826–1831.
Loehlin, J. C. (1992). Genes and Environment in Personality
Development. Sage Publications: Newbury Park, CA.
Loranger, A. W. (1988). Personality Disorder Examination (PSE).
DV Communications: Yonkers, NY.
McCrae, R. R. & Costa Jr., P. T. (1990). Personality in Adulthood.
Guilford Press: New York, NY.
McGlashan, T. H., Grilo, C. M., Sanislow, C. A., Ralevski, E.,
Morey, L. C., Gunderson, J. G., Skodol, A. E., Shea, M. T.,
Zanarini, M. C., Bender, D., Stout, R. L., Yen, S. & Pagano, M.
(2005). Two-year prevalence and stability of individual DSM-IV
criteria for schizotypal, borderline, avoidant, and obsessive-com-
pulsive personality disorders: toward a hybrid model of axis II
disorders. American Journal of Psychiatry 162, 883–889.
Neale, M. C. (2003). A finite mixture distribution model for data
collected from twins. Twin Research 6, 235–239.
Neale, M. C., Boker, S. M., Xie, G. & Maes, H. H. (2003). Mx:
Statistical Modeling. Department of Psychiatry, Virginia Com-
monwealth University Medical School: Box 980126, Richmond,
Oldham, J. M. & Skodol, A. E. (2000). Charting the future of axis II.
Journal of Personality Disorders 14, 17–29.
Pfohl, B., Blum, N. & Zimmerman, M. (1995). Structured Interview
for DSM-IV Personality (SIDP-IV). University of Iowa, Depart-
ment of Psychiatry: Iowa City.
664 K. S. Kendler et al.
Raghunathan, T. E., Solenberger, P. & Van Hoewyk, J. (2000).
IVEware: Imputation and Variance Estimation Software: In-
stallation Instructions and User Guide. Survey Research Center,
Institute for Social Research, University of Michigan: Ann
Reichborn-Kjennerud, T., Czajkowski, N., Neale, M. C., Orstavik,
R. E., Torgersen, S., Tambs, K., Roysamb, E., Harris, J. R. &
Kendler, K. S. (2006). Genetic and environmental influences
on dimensional representations of DSM-IV Cluster C person-
ality disorders: a population-based multivariate twin study.
Psychological Medicine. Published online: 30 November 2006.
SAS Institute (2006). SAS OnlineDoc 9.1.3. SAS Institute Inc.: Cary,
Skodol, A. E., Oldham, J. M., Bender, D. S., Dyck, I. R., Stout, R. L.,
Morey, L. C., Shea, M. T., Zanarini, M. C., Sanislow, C. A., Grilo,
C. M., McGlashan, T. H. & Gunderson, J. G. (2005). Dimensional
representations of DSM-IV personality disorders: relationships
to functional impairment. American Journal of Psychiatry 162,
Torgersen, S. (1980a). Hereditary-environmental differentiation
of general neurotic, obsessive, and impulsive hysterical person-
ality traits. Acta Geneticae Medicae et Gemellologiae 29, 193–207.
Torgersen, S. (1980b). The oral, obsessive, and hysterical personality
syndromes. A study of hereditary and environmental factors by
means of the twin method. Archives of General Psychiatry 37,
Torgersen, S., Kringlen, E. & Cramer, V. (2001). The prevalence of
personality disorders in a community sample. Archives of General
Psychiatry 58, 590–596.
Torgersen, S., Lygren, S., Oien, P. A., Skre, I., Onstad, S., Edvardsen,
J., Tambs, K. & Kringlen, E. (2000). A twin study of personality
disorders. Comprehensive Psychiatry 41, 416–425.
Widiger, T. A. & Coker, L. A. (2002). Assessing personality dis-
orders. In Clinical Personality Assessment: Practical Approaches
(ed. J. N. Butcher), pp. 407–434. Oxford University Press:
Widiger, T. A. & Samuel, D. B. (2005). Diagnostic categories or
dimensions: a question for the Diagnostic and Statistical Manual
of Mental Disorders – fifth edition. Journal of Abnormal Psy-
chology 114, 494–504.
Widiger, T. A. & Simonsen, E. (2005). Alternative dimensional
models of personality disorder: finding a common ground. Journal
of Personality Disorders 19, 110–130.
Williams, L. & Holahan, P. (1994). Parsimony-based fit indices for
multiple-indicator models: do they work? Structural Equation
Modeling 1, 161–189.
Zimmerman, M. (1994). Diagnosing personality disorders. A review
of issues and research methods. Archives of General Psychiatry 51,
Heritability of cluster A personality disorders 665