More is more: Evidence for the incremental value of the SCID-II/SCID-5-PD specific factors over and above a general personality disorder factor

(1)

University of Groningen

More is more

Paap, Muirne C. S.; Heltne, Aleksander; Pedersen, Geir; Germans Selvik, Sara; Frans, Niek;

Wilberg, Theresa; Hummelen, Benjamin

Published in:

Personality Disorders: Theory, Research, and Treatment

DOI:

10.1037/per0000426

IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from

it. Please check the document version below.

Document Version

Publisher's PDF, also known as Version of record

Publication date:

2021

Link to publication in University of Groningen/UMCG research database

Citation for published version (APA):

Paap, M. C. S., Heltne, A., Pedersen, G., Germans Selvik, S., Frans, N., Wilberg, T., & Hummelen, B.

(2021). More is more: Evidence for the incremental value of the SCID-II/SCID-5-PD specific factors over

and above a general personality disorder factor. Personality Disorders: Theory, Research, and Treatment.

https://doi.org/10.1037/per0000426

Copyright

Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).

Take-down policy

If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.

Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.

(2)

Personality Disorders: Theory, Research, and

Treatment

More Is More: Evidence for the Incremental Value of the SCID-II/SCID-5-PD

Specific Factors Over and Above a General Personality Disorder Factor

Muirne C. S. Paap, Aleksander Heltne, Geir Pedersen, Sara Germans Selvik, Niek Frans, Theresa Wilberg, and Benjamin Hummelen

Online First Publication, February 18, 2021. http://dx.doi.org/10.1037/per0000426

CITATION

Paap, M. C. S., Heltne, A., Pedersen, G., Germans Selvik, S., Frans, N., Wilberg, T., & Hummelen, B. (2021, February 18). More Is More: Evidence for the Incremental Value of the SCID-II/SCID-5-PD Specific Factors Over and Above a General Personality Disorder Factor. Personality Disorders: Theory, Research, and Treatment. Advance online publication. http://dx.doi.org/10.1037/per0000426

(3)

More Is More: Evidence for the Incremental Value of the

SCID-II/SCID-5-PD Specific Factors Over and Above a General Personality Disorder Factor

Muirne C. S. Paap

1, 2

, Aleksander Heltne

1, 3

, Geir Pedersen

4, 5

, Sara Germans Selvik

6, 7

, Niek Frans

1, 2

,

Theresa Wilberg

1, 3

, and Benjamin Hummelen

1

Department of Research and Innovation, Division of Mental Health and Addiction, Oslo University Hospital, Oslo, Norway 2

Faculty of Behavioural and Social Sciences, The Nieuwenhuis Institute for Educational Research, University of Groningen 3

Institute of Clinical Medicine, University of Oslo 4

Department of Personality Psychiatry, Division of Mental Health and Addiction, Oslo University Hospital, Oslo, Norway 5

The Norwegian Centre of Mental Disorders Research (NORMENT), Institute of Clinical Medicine, University of Oslo 6

Department of Psychiatry, Helse Nord-Trønderlag, Namsos Hospital, Namsos, Norway 7

Department of Mental Health, Norwegian University of Science and Technology (NTNU)

Currently, 3 competing conceptualizations of personality dysfunction can be distinguished: the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM–5) categorical model delineating 10 dis-tinct types of personality disorders (PDs); the alternative model for PDs (DSM–5 Section III), which assesses personality functioning and traits separately; and the International Classification of Diseases, 11th_{Version conceptualization, which provides 1 single code for the presence of a PD (which is based on} problems in functioning) as well as codes that specify the level of the disorder (mild/moderate/severe), and prominent trait domains or patterns (5 domains and 1 pattern). The current study aims to assess the incremental value of the DSM–5 PDs over and above a global personality dysfunction factor, using expert ratings obtained with the Structured Clinical Interview for DSM–IV PDs and the Structured Clinical Interview for DSM–5 PDs interview in a large sample of clinical patients (N = 3,851). All estimated bifac-tor models provided adequatefit to the data. We found a surprisingly low explained common variance for the g-factor (,40%), indicating that ignoring the specific PD factors would lead to a substantial loss of in-formation. The strongest specific PDs in terms of explained common variance were the avoidant, schizoty-pal, and schizoid PD factors and the conduct disorder criteria set if included. Correlations between our factors and external variables were relatively low, except for the Severity Indices of Personality Problems, which aims to measure personality functioning. Ourfindings suggest that specific PDs still have an impor-tant role to play in the assessment of personality pathology.

Keywords: personality pathology, bifactor model, personality disorders, ICD-11, DSM–5 Supplemental materials:https://doi.org/10.1037/per0000426.supp

Over the past two decades, there has been a strong ongoing debate in the field of personality disorders (PDs) regarding the labeling and definitions used in the field (Huprich, 2015).

Although it may have seemed that a majority of those dedicated to studying personality disorders (PDs) agreed that the classification system was in need of a thorough revision, no consensus on the issue and the best way forward had yet been reached at the time of publication of the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM–5;American Psychiatric Associa-tion, 2013), resulting in the retainment of the classification system used in the previous edition. Meanwhile, alternatives to the tradi-tional classification model have been proposed. This has led to the inclusion of the alternative model of PD (AMPD) in Section III of the DSM–5, and the inclusion of the recent radical revision of the classification system used in the International Classification of Diseases, 10th version (10), with the publication of the ICD-11 (World Health Organization, 2018). In sum, three competing conceptualizations of personality dysfunction can currently be dis-tinguished: the“traditional” model delineating 10 distinct types of PDs (DSM–5); the AMPD (DSM–5 Section III), which assesses personality functioning and traits separately, but still allows for diagnosing six distinct types of PDs; and the ICD-11

Muirne C. S. Paap https://orcid.org/0000-0002-1173-7070

Aleksander Heltne https://orcid.org/0000-0002-3524-203X

Niek Frans https://orcid.org/0000-0001-6684-0684

Benjamin Hummelen https://orcid.org/0000-0002-8717-8076

The author(s) disclosed receipt of the followingﬁnancial support for the research, authorship, and/or publication of this article: This research was supported by a FRIPRO Young Research Talent Grant to Muirne C. S. Paap (Grant NFR 286893), awarded by the Research Council of Norway. The authors declare there are no conﬂicts of interest.

Correspondence concerning this article should be addressed to Muirne C. S. Paap, Department of Research and Innovation, Division of Mental Health and Addiction, Oslo University Hospital, P.O. Box 4956 Nydalen, 0424 Oslo, Norway. Email:m.c.s.paap@rug.nl

1

Personality Disorders: Theory, Research, and Treatment

ISSN: 1949-2715 https://doi.org/10.1037/per0000426 Thi s d o cumen t is copy righte d b y the Amer ican Ps ycholo gic al Asso ciatio n o r o n e of its all ied publi she rs. Con tent may be shar ed at no co st, but any reque sts to reu se thi s co ntent in par t o r whol e m u st g o thro ugh the Amer ican Psyc holog ical Asso ciatio n.

(4)

conceptualization, which provides one single code for the presence of a PD (which is based on problems in functioning) as well as codes that specify the level of the disorder (mild/moderate/severe), and prominent trait domains or patterns (ﬁve domains and one pattern).

In the DSM–5, the following PDs are distinguished: paranoid, schizoid, schizotypal, antisocial, borderline, histrionic, narcissistic, avoidant, dependent, obsessive–compulsive, and PD not otherwise speciﬁed. The disorders are rated as either present or absent, and diagnoses are made by assessing how many symptoms (behaviors, characteristics) are present. The number of symptoms needed for a diagnosis differs across the PDs (from three toﬁve).

Many experts have called for a shift to a continuous (often referred to as dimensional) approach in the DSM–5 (to what degree are symptoms present), to allow for a more fine-grained and ultimately more fitting diagnosis (Crawford, et al., 2011; Hyman, 2010;Skodol et al. 2005;Tyrer, 2005;Widiger & Clark, 2000;Widiger & Trull, 2007). In addition, a number of PD experts vocalized additional concerns specific to the diagnosis of PDs. Of-ten repeated points of critique include within-diagnosis heteroge-neity (Widiger & Trull, 2007) and substantial covariation or comorbidity among PD diagnoses (Clark, 2007; Widiger & Samuel, 2005;Widiger & Trull, 2007).

Covariations in PD diagnoses across individuals have been well documented in previous editions of the DSM (Clark, 2007; Wat-son & Sinha, 1998). This covariation has been regarded by many as a major challenge to the integrity and clinical usefulness of cat-egorical PD diagnoses (Clark, 2007), as it could be taken to indi-cate that existing PD indi-categories do not capture the natural underlying structure of personality pathology. Although some have argued that the issue of PD covariation and its implications has been somewhat overstated (Zimmerman, 2012), others have called for an exploration of shared features that cut across PD diagnostic categories, alongside speciﬁc factors that make up sty-listic differences in PD expression (Tyrer, 2005).

The AMPD resolves the latter issue by distinguishing an overall level of personality functioning (Criterion A) from stylistic traits (Criterion B), using separate steps in the diagnostic process. Crite-rion A is supposed to capture both the presence and severity of PD, which can be realized by administering theﬁrst module of the Structured Clinical Interview for the DSM–5 AMPD, that is, the SCID-5-AMPD-I (Bender, et al., 2018), assessing the level of per-sonality functioning. The AMPD has retained six of the traditional PD diagnoses, which are assessed by a combination of criterion A and B, using the third module of the SCID-5-AMPD. The ICD-11 goes a step further, discarding all PD categories except the border-line PD (which is referred to as “borderline pattern”). As in the AMPD, personality pathology is assessed by a combination of per-sonality functioning and stylistic traits. Several self-report instru-ments have been developed to assess personality pathology according to the ICD-11 (Bagby & Widiger, 2020). However, most of said instruments focus on the trait model, and structured diagnostic interviews for the ICD-11 are not yet available. It has been suggested by some that the SCID-5-AMPD-I may be used to assess severity as operationalized by the ICD-11, for the time being (Bach & First, 2018). Thus, there seems to be general con-sensus about the validity of distinguishing between a general se-verity factor and speciﬁc traits. However, only a handful of

clinical studies using structured diagnostic interviews have focused on this issue.

Given these recent developments, the future of the PD catego-ries recognized in the DSM seems uncertain. Has the time come for drastic change, as proposed in the ICD-11? Are we ready to abandon the PD categories? A number of recent studies have explored the latter question by factor analyzing the“traditional” PD criteria (Conway, et al., 2016;Ringwald et al. 2019;Sharp et al. 2015;Williams et al. 2018). Bifactor modeling has been espe-cially popular, due to its ability to disentangle the total variance (interpersonal differences) into variance attributable to a general factor (something all items have in common) and specific factors (unique aspects measured by items). This seems to align well with recent developments which seek to distinguish personality func-tioning from specific specifiers. Although aforementioned studies make important contributions to the dialogue surrounding the mer-its of hierarchical models in the PD field, it is unfortunate that most of said studies focus primarily on the modelfit when discus-sing the utility of the found model. Bifactor models are known to have superiorfit to other models simply due to the nature of the method. In other words, it is highly likely that factor models will show superiorfit to competing models in most cases—which could be regarded as a method effect rather than a clinically interesting result (Bonifay et al. 2017;Reise et al. 2016).

We suggest that it may be much more relevant to consider other statistics, such as the explained common variance (ECV) associ-ated with each factor in the model. The ECV of the general factor has been suggested as an indicator for essential unidimensionality (with cutoff values of 60% or 70%;O'Connor Quinn, 2014;Reise et al. 2013). Exploring the incremental contribution of the speciﬁc factors over and above the general factor using traditional diagnos-tic instruments such as the Structured Clinical Interview for DSM–5 PDs (SCID-5-PD), may yield insights that can be used to further inform the debate surrounding the optimal model for PDs. How dominant is the general factor measured by the SCID-5-PD? Can we distinguish general personality functioning from more sty-listic elements associated with the PD categories currently still being used in the DSM–5? If the 10 speciﬁc PD diagnoses explain a considerable amount of variance, this would indicate that we would stand to lose a great deal of information if they were to be dismissed.

In this study, we will apply conﬁrmatory bifactor analyses to the Section II DSM–5 PD criteria to explore the relative contribu-tion of PDs currently listed in the DSM–5, as assessed using struc-tured clinical interviews for DSM disorders (SCID-II and SCID-5-PD). Furthermore, we will explore the relationship of the estimated factors using measures of psychosocial functioning, symptom dis-tress and interpersonal functioning.

Method

Participants and Diagnostic Process

In this multisite, naturalistic, and explorative study, the material comprised data from 16 outpatient units within the Norwegian Network for Personality Disorders (Karterud, et al., 1998). All units are specialized in treatment of PDs and personality related difﬁculties. The participants were 3,851 psychiatric outpatients

Thi s d o cumen t is copy righte d b y the Amer ican Ps ycholo gic al Asso ciatio n o r o n e of its all ied publi she rs. Con tent may be shar ed at no co st, but any reque sts to reu se thi s co ntent in par t o r whol e m u st g o thro ugh the Amer ican Psyc holog ical Asso ciatio n.

(5)

treated in the period January 2009 to March 2019, whose mean age was 32 years (SD = 9) and of whom 75% were women. The different treatment units collected patient data, which were regis-tered in an anonymous central database, administrated by the Department for Personality Psychiatry, Oslo University Hospital, Oslo, Norway. The State Data Inspectorate and the Regional Com-mittee for Medical Research and Ethics have approved these procedures.

In all, 69% of the sample was diagnosed with one or more PDs; 51% patients had one PD diagnosis; 13% had two PD diagnoses; and 4% had three or more PD diagnoses. Avoidant PD was the most common PD (32%), followed by borderline PD (25%), and PD not otherwise speciﬁed (14%). Other common PD diagnoses were paranoid PD (8%), obsessive–compulsive PD (6%), and de-pendent PD (4%). Schizoid PD, schizotypal PD, narcissistic PD, and histrionic PD all had a prevalence of less than 1.0%. In all, 93% of the patients had one or several symptom disorders. Mood disorders were most common (64%) including dysthymia (8%), bipolar disorder Type I (3%) and bipolar disorder Type II (2%). Social phobia was the most common anxiety disorder (19%), fol-lowed by panic disorder (14%), posttraumatic stress disorder (10%), and generalized anxiety disorder (9%). Moreover, 8% of patients had a substance use disorder. Symptom disorder diagnoses were made using the Mini International Neuropsychiatric Inter-view (Sheehan & Lecrubier, 1994). Either of two instruments was used to assess PDs; 80% of the patients were assessed using the Structured Clinical Interview for DSM–IV Axis II PDs (SCID-II; First et al. 1997), and the remaining 20% of the patients were assessed using the SCID-5-PD (First, et al., 2015). Diagnostic reli-ability was not investigated. However, diagnostic assessments were performed in each unit by clinical staff who had received systematic training in diagnostic interviews and principles of the LEAD procedure (Longitudinal, Expert, All-Data;Pedersen et al. 2013;Spitzer, 1983). This means that diagnoses were based on all available information, including referral letters, self-reported his-tory and complaints, overall clinical impression together with the two diagnostic interviews.

Measures

SCID-II and SCID-5-PD

The SCID-II and SCID-5-PD are semistructured clinical inter-views consisting of 90 items in total, measuring the DSM PD cate-gories avoidant, dependent, obsessive–compulsive, paranoid, schizotypal, schizoid, histrionic, narcissistic, borderline, and anti-social PDs as well as the criteria for conduct disorder. Both the SCID-II and the SCID-5-PD have been shown to have good inter-rater reliability in European samples (Lobbestael, Leurgans, & Arntz, 2011;Somma et al. 2017). All items are scored on a 3-point-scale: not present, subthreshold, and present. The differences between the two versions are minor. The three response categories for each criterion were adjusted from 1–2–3 in the SCID-II to 0–1–2 in the SCID-5-PD (0 = criterion absent; 1 = subthreshold; and 2 = threshold or true). In this study, we used the new scoring procedure and recoded the SCID-II scores by subtracting 1 point from all SCID-II item scores.

Work and Social Adjustment Scale

Psychosocial functioning was assessed by the Work and Social Adjustment Scale (WSAS;Mundt et al. 2002); afive-item self-report scale of functional impairment that measures the level of impairment on a scale from 0 to 8, with 0 indicating no impair-ment and 8 indicating severe impairimpair-ment. The scores on thefive different items are summed to create a total score (0–40). Mundt et al. reported sufficient levels of reliability in patients being treated for depression (Cronbach’sa ..80), and patients being treated for obsessive–compulsive disorder (Cronbach’s a ..78, test-retest = .73).Pedersen et al. (2017) reported similar levels of reliability (Cronbach’sa = .77–.83 depending on subsample) and measure-ment invariance for gender in a subsample of the data used in the current study.

Patient Health Questionnaire

Self-reported level of depression was measured by the Patient Health Questionnaire (PHQ-9;Kroenke et al. 2001). Patients are asked how much each symptom has bothered them over the past 2 weeks, with four response options: 0 (not at all),“1 (several days), 2 (more than half the days), and 3 (nearly every day). PHQ-9 scores were computed as the sum score of all nine items, thus ranging from 0 to 27. Sufﬁcient to high levels of reliability (..80) as well as high levels of sensitivity and speciﬁcity (both 88% for a cutoff value of 10) have been reported for primary care patients (Kroenke, et al., 2001).

Generalized Anxiety Disorder Scale

Self-reported level of anxiety was measured by the seven-item Generalized Anxiety Disorder Scale (GAD-7;Spitzer et al. 2006). The GAD-7 uses the same response format as the PHQ-9. GAD-7 scores were computed by summing the scores on all nine items, resulting in a total score ranging between 0 and 21. High levels of reliability (..90) and sufﬁcient levels of sensitivity and speciﬁcity (both.80% for a cut-off value of 10) have been reported for pri-mary care patients (Spitzer, et al., 2006).

Circumplex of Interpersonal Problems

Interpersonal problems were measured using a 48-item short version of the Inventory of Interpersonal Problems–Circumplex (Alden, Wiggins, & Pincus, 1990) called the Circumplex of Inter-personal Problems (CIP;Pedersen, 2002). CIP measures the same eight interpersonal problem areas as the Inventory of Interpersonal Problems–Circumplex, using the same 5-point response format (not at all [0] to extremely [4]). In the current study, the total score was used by averaging all 48 items. As part of a validity study of the CIP (Pedersen, et al., 2011), a high test–retest stability of the total score was found, using a 3- to 4-day interval in a clinical sample (n = 53): intraclass coefﬁcient = .957 (95% CI [.925, .975]; unpublished data).

Severity Indices of Personality Problems

The Severity Indices of Personality Problems (SIPP-118) is a self-report questionnaire developed byAndrea et al. (2007) that aims to measure core components of (mal)adaptive personality functioning that can change over time. The instrument contains 118 items that cover 16 facets. The response categories range from THE ADDED VALUE OF THE SCID-II/SCID-5-PD SPECIFIC FACTORS

₃

(6)

1 to 4 (fully disagree to fully agree), and higher total scores indi-cate more adaptive functioning. A recall period of 3 months is used. The facets were developed using an approach that was con-tent-driven: experts identiﬁed concepts, generated items, and these were in turn evaluated by patients. In a recent study, using both a community and two clinical samples, Cronbach’s a estimates ranged between .63 and .85 (lowest value for the facet respect, highest for aggression regulation and self-respect), with most val-ues exceeding .70 (Pedersen, et al., 2019).

Analyses

Latent Variable Modeling

In this study, we used conﬁrmatory bifactor analysis (Cai, 2010; Reise, 2012) to evaluate the latent structure of the SCID-II and SCID-5-PD. In a bifactor model, all items load directly on their re-spective subscale as well as on the general factor. This sets it apart from a correlated-trait model, where items only load on their own respective factors, but these factors are allowed to correlate1

. Using the results from the bifactor analysis, one can calculate the ECV. For a conﬁrmatory bifactor analysis where all factors are or-thogonal, the ECV for a factor equals the sum of all squared load-ings for that factor divided by the sum of all squared loadload-ings across all factors. This index can be seen as an indicator of relative factor strength (Reise, 2012).Reise et al. (2013)tentatively pro-posed that when the ECV for the general factor in a bifactor model is larger than 60%, the factor loading estimates for a unidimen-sional model are close to the true loadings on the general factor in the bifactor model, and can be interpreted as essentially unidimen-sional. More recently, O'ConnorQuinn (2014)proposed a cutoff of 70%. As the authors of both papers state, these numbers should not be used as strict cutoff values, but instead as general guidelines.

Because two test versions of the SCID for PDs (SCID-II and SCID-5-PD) were used in this study, we evaluated whether they are comparable in psychometric terms before proceeding with the analyses. To test this, two sets of models were estimated: models in which loadings and intercepts were allowed to vary across the test versions (the unconstrained model) and models in which the loadings were constrained to be equal across the test versions (the constrained model). For a bifactor model, this translates to the loadings on the speciﬁc factors being equal across the test ver-sions, but loadings on the general factor can still vary. If invari-ance was found, the constrained model was taken as the basis for interpretation, and if invariance was not supported, the uncon-strained model was used. Each set consisted of two models:

1. A model based on all diagnostic criteria;

2. A model based on all diagnostic criteria except for the conduct disorder criteria.

For each of these models, the assignment of items to speciﬁc factors corresponded with the assignment of items to PDs as per the DSM–5. All items in each model thus loaded on one of the spe-ciﬁc factors as well as on the general factor. Model 2 was esti-mated to evaluate whether including the conduct disorder criteria had an impact on the outcomes.

Unconstrained and constrained models were compared using the Bayesian Information Criterion (Schwarz, 1978) as well as visual inspection of the size of the differences. Furthermore, we used the following indices and rules-of-thumb to evaluate whether a model showed adequatefit: the comparative fit index (CFI), good fit if CFI $ .95, and acceptable fit if CFI was between .90 and .95; the Tucker–Lewis Index, good fit if TLI $ .90; and the root mean square error of approximation (RMSEA), goodfit if RMSEA # .06, acceptable fit if RMSEA was between .06 and .08 (Cook, et al., 2009;Hu & Bentler, 1998).

Interpretation of the Factors: Loading Patterns and

Correlations With Other Measures

To help us understand how the general and specific factors could be interpreted, the factor loadings were examined. Fur-thermore, Pearson correlations coefficients were calculated to assess the strength of the linear relationship between the SCID-II/SCID-5-PD factors estimated with the bifactor mod-els described in the previous section, and measures of psy-chological and/or psychosocial functioning: the WSAS, the PHQ-9, the GAD-7, the CIP, and the g-factor of the SIPP. This g-factor was obtained using a bifactor model, and taken as an indication of personality dysfunction (this approach is comparable to the one used by Bastiaansen, et al., 2016). Given the large sample size, we chose to focus on effect size of correlations rather than statistical significance tests. Fol-lowing Ringwald et al. (2019), correlation coefficients and loadings equal to or greater thanj.30j were considered clini-cally relevant effects.

Software

All statistical analyses were performed in the open-source soft-ware program R Version 3.6.1 (R Core Team, 2017). The bifactor model was estimated using a full information maximum likelihood approach in the R package mirt Version 1.31 (Chalmers, 2012), which follows the analytic strategy outlined byCai (2010).

Results

Descriptive Statistics

The descriptive statistics for the seven included external varia-bles measuring psychological and/or psychosocial functioning are reported inTable 1.

Constrained Versus Unconstrained Models

The CFI, TLI and RMSEA estimates for the constrained and unconstrained models were highly comparable (all in the ac-ceptable to good range), and the Bayesian Information Criterion values were lower (indicating betterﬁt) for all the constrained models (Table 2). These ﬁndings indicate that the factor

1

We refer the interested reader to the online supplemental material accompanying a paper byPaap et al. (2015)for a more detailed comparison of bifactor analysis to other commonly used techniques for assessing dimensionality. Thi s d o cumen t is copy righte d b y the Amer ican Ps ycholo gic al Asso ciatio n o r o n e of its all ied publi she rs. Con tent may be shar ed at no co st, but any reque sts to reu se thi s co ntent in par t o r whol e m u st g o thro ugh the Amer ican Psyc holog ical Asso ciatio n.

(7)

structures found were invariant/comparable across test ver-sions. Therefore, the ECV results were studied for the con-strained models only.

Relative Strength of the g- and s-Factors

The Model Including All PD Diagnoses

For the model incorporating all SCID-II/SCID-5-PD diagnoses, the g-factor was found to be relatively weak (,60%) in terms of ECV. The speciﬁc factors, on the other hand, could be considered strong in terms of ECV, with the strongest being avoidant, schizo-typal, and schizoid PD. These results were found regardless of whether or not the conduct disorder criteria were incorporated. When the conduct disorder criteria were included, these criteria formed the strongest speciﬁc factor by far. ECV patterns were inspected to compare models with different numbers of criteria (Table 3). Values were very similar across the two models. In both cases, the ECV attributable to the g-factor was smaller than 60%; and the weakest sub factors were the same as well, with the excep-tion of antisocial PD. The latter factor was substantially stronger in terms of ECV in the model where the conduct disorder criteria were omitted.

Factor Loadings

For the model incorporating all SCID-II/SCID-5-PD criteria (Table 4) factor loadings were fairly high for all speciﬁc factors.

This was especially true for avoidant PD (factor loadings ranged from .57 to .85, with a mean loading of .73), schizoid PD (factor loadings ranged from .63 to .81, with a mean loading of .72), and schizotypal PD (factor loadings ranged from .43 to .86, with a mean loading of .59). For the remaining specific factors, factor loadings ranged from .13 to .73. Factor loadings on both the spe-cific factors and the general factor did not change substantially when the conduct disorder criteria where dropped from the model (see the online supplemental material). In general, PD criteria belonging to a PD diagnosis represented by a strong specific factor tended to load consistently on that factor, whereas PD criteria with high loadings on the general factor belonged to a wide range of different PD diagnoses. Factor loadings on the g-factor ranged from.17 to .71, with a mean factor loading of .40.

The PD criteria showing the highest loadings on the g-factor were narcissistic, borderline, histrionic and antisocial PD. The par-anoid PD criteria also had high loadings on the g-factor. Further-more, several schizotypal PD criteria exhibited high loadings on the g-factor. The schizotypal PD criteria that loaded highest on the g-factor represented odd beliefs and experiences as well as suspi-ciousness and ideas of reference. Factor loadings on the g-factor for PD criteria from the remaining diagnostic categories (avoidant, dependent, obsessive–compulsive and schizoid PD) were all fairly low, with mean factor loadings falling below .30. For avoidant and schizoid PD criteria, factor loadings on the g-factor were particu-larly low with none higher than .26 and most falling below .20.

Correlations With Other Measures

Prior to interpreting the correlations, scatterplots were visually checked to see whether the relationships deviated from linearity. None of them did. The correlations among the SCID-II/SCID-5-PD g-factors and the included external variables were weak (.01) to strong (.56) in size (Table 5, upper part). The strongest correla-tions were found for the SIPP g-factor: .55 and .56 for the total SCID-II and SCID-5-PD g-factor, respectively. For the remaining correlations, the highest values were found for the GAD-7 Table 3

Explained Common Variance for the General (g) and Specific (s) Factors, Based on Two Constrained Models

Total with CD Total without CD

g 38.75 37.40 AVPD 7.45 9.11 DPD 4.78 5.64 OCPD 4.97 5.94 PPD 3.58 4.34 STPD 6.73 8.18 SZPD 7.30 8.98 HPD 4.46 5.20 NPD 4.45 5.66 BPD 3.33 3.95 CD 10.50 NA ASPD 3.69 5.61

Note. AVPD = avoidant personality disorder; DPD = dependent person-ality disorder; OCPD = obsessive-compulsive personperson-ality disorder; PPD = paranoid personality disorder; STPD = schizotypal personality disorder; SZPD = schizoid personality disorder; HPD = histrionic personality disor-der; NPD = narcissistic personality disordisor-der; BPD = borderline personality disorder; CD = conduct disorder; ASPD = antisocial personality disorder.

Table 2

Fit Statistics for All Estimated Bifactor Models

BIC CFI TLI RMSEA

Total with CD Constrained 383434 .916 .914 .024 Unconstrained 384673 .917 .913 .024 Total without CD Constrained 351006 .906 .904 .026 Unconstrained 352011 .908 .903 .027

Note. Constrained = model in which the loadings were constrained to be equal across test versions; Unconstrained = model in which loadings and intercepts were allowed to vary over the test versions; CD = conduct dis-order; BIC = Bayesian Information Criterion; CFI = comparative fit index; TLI = Tucker–Lewis Index; RMSEA = root mean square error of approximation.

Table 1

Descriptive Statistics for the Seven Included External Variables

n M SD Min Max GAF-F 3,834 52.96 7.67 20.00 83.00 GAF-S 3,835 52.31 6.41 21.00 97.00 WSAS 3,738 22.83 8.63 0.00 40.00 PHQ-9 2,699 17.25 5.85 0.00 27.00 GAD-7 698 13.21 4.78 0.00 21.00 CIP 3,466 1.73 0.47 0.08 3.29 SIPP-g 1,601 0.01 0.95 4.43 3.64

Note. GAF-F = functioning scale of Global Assessment of Functioning; GAF-S = symptoms scale of Global Assessment of Functioning; WSAS = Work and Social Adjustment Scale; PHQ-9 = nine-item Patient Health Questionnaire; GAD-7 = seven-item Generalized Anxiety Disorder scale; CIP = Circumplex of Interpersonal Problems scale; SIPP-g = g-factor of the Severity Indices of Personality Problems.

THE ADDED VALUE OF THE SCID-II/SCID-5-PD SPECIFIC FACTORS

₅

(8)

Table 4

Factor Loadings for Constrained Model With CD

Item Description g AVPD DPD OCPD PPD STPD SPD HPD NPD BPD CD ASPD

AVPD 1 Avoids social work .01 .75

AVPD 2 Must be liked .05 .85

AVPD 3 Resistant to intimacy .08 .57 AVPD 4 Preoccupied with rejection .07 .77

AVPD 5 Socially inhibited .17 .80

AVPD 6 Views self as socially inept .00 .64 AVPD 7 Avoids risks and new activities .06 .70

DPD 1 Reassurance seeking .17 .62

DPD 2 Avoids personal responsibility .28 .63 DPD 3 Fears losing support/approval .12 .52 DPD 4 Lacks confidence in abilities .16 .62 DPD 5 Volunteers for unpleasant jobs .19 .38 DPD 6 Fear of inability for self-care .37 .59 DPD 7 Seeks new relationships urgently .46 .41 DPD 8 Fear of being left to care for self .37 .56

OCPD 1 Orderly .20 .69

OCPD 2 Perfectionism .19 .64

OCPD 3 Workaholic .16 .67

OCPD 4 Moral inflexibility .18 .56

OCPD 5 Hoarding .27 .41

OCPD 6 Reluctant to delegate .40 .54

OCPD 7 Miserly .08 .54

OCPD 8 Rigidity .50 .32

PPD 1 Suspects exploitations .46 .63

PPD 2 Doubts loyalty .39 .73

PPD 3 Reluctance to confide .23 .71

PPD 4 Hostile attribution bias .47 .46

PPD 5 Holds persistent grudges .46 .35

PPD 6 Sensitive to defamation .66 .13 PPD 7 Suspects infidelity .53 .17 STPD 1 Ideas of reference .39 .45 STPD 2 Odd beliefs .52 .43 STPD 3 Odd experiences .48 .48 STPD 4 Odd thinking/speech .29 .86 STPD 5 Suspicious .58 .36 STPD 6 Constricted affect .26 .77 STPD 7 Odd behavior/appearance .31 .83

STPD 8 Lacks close friends .05 .54

STPD 9 Social anxiety .18 .57

SZPD 1 Uninterested in close relationships .22 .81

SZPD 2 Chooses solitary activities .13 .72

SZPD 3 Asexual .06 .63

SZPD 4 Anhedonia .15 .74

SZPD 5 Lacks close friends .11 .63

SZPD 6 Indifference to praise/criticism .26 .74

SZPD 7 Flattened affect .03 .79

HPD 1 Attention-seeking .61 .52

HPD 2 Sexually seductive .65 .50

HPD 3 Rapidly shifting emotions .55 .64

HPD 4 Dresses provocatively .59 .53 HPD 5 Superficial speech .57 .63 HPD 6 Dramatism .63 .54 HPD 7 Suggestibility .21 .32 HPD 8 Overestimates intimacy .50 .50 NPD 1 Grandiose .57 .58

NPD 2 Preoccupied with themselves .59 .54

NPD 3 Believes s/he is special .71 .45

NPD 4 Needs admiration .64 .46 NPD 5 Entitlement .71 .52 NPD 6 Exploitative .66 .48 NPD 7 Lacks empathy .58 .51 NPD 8 Envious .56 .39 NPD 9 Arrogant .60 .54 BPD 1 Avoids abandonment .60 .41 BPD 2 Interpersonal instability .64 .52 (table continues) Thi s d o cumen t is copy righte d b y the Amer ican Ps ycholo gic al Asso ciatio n o r o n e of its all ied publi she rs. Con tent may be shar ed at no co st, but any reque sts to reu se thi s co ntent in par t o r whol e m u st g o thro ugh the Amer ican Psyc holog ical Asso ciatio n.

(9)

(anxiety) with the total SCID-II/SCID-5-PD g-factor (with or without the conduct disorder criteria).

The correlations among the SCID-II/SCID-5-PD speciﬁc fac-tors and the included external variables were nonexistent (0) to moderate (.37). Only one correlation exceeded .30: borderline PD with SIPP-g. One correlation came close (.29): avoidant PD with CIP. Of all speciﬁc factors, the avoidant PD factor showed the highest correlations with the external variables (.10.29).

Discussion

In this study, we investigated the relative contribution of specific PD factors over and above a general factor of PD, based on SCID-II and SCID-5-PD ratings in a large clinical sample from the Norwegian Network for Personality Disorders. Notably, all criteria were included in the analyses and the crite-ria were scored on a 3-point scale by experienced clinicians. We conducted a series of confirmatory bifactor analyses of the PD criteria, where the criteria were allowed to load on both a general factor and their own specific PD factor. Because the factor structures of the SCID-II and the SCID-5-PD were found to be equivalent, only the results of the constrained models were considered.

All estimated bifactor models provided adequateﬁt to the data. We found a surprisingly low ECV for the g-factor (,40%),

indicating that ignoring the speciﬁc PD factors would lead to a substantial loss of information. The strongest speciﬁc PDs in terms of ECV were the avoidant, schizotypal, and schizoid PD factors; and the conduct disorder criteria set if included. Of note, the avoi-dant PD criteria did not load on the g-factor at all. Correlations with the g-factor scores and the other clinical variables were gen-erally modest, except for the g-factor scores of the SIPP-118 g-fac-tor (self-reported impairment of personality functioning).

Because the g-factor explained less than 40% of the variance in our bifactor analyses, personality pathology as assessed by the SCID-II/SCID-5-PD seems to be better conceptualized as a multidimensional construct than as a unidimensional construct. Elaborating on this, thefinding that the specific factors explained a substantial amount of the variance warrants further inspection of these factors. In terms of ECV, the avoidant PD criteria constituted the largest specific factor in our analyses after CD, and did not load on the g-factor at all; supporting the distinctiveness of the avoidant PD factor. When it comes to the correlations found for avoidant PD with external variables, these may look low atfirst sight, but it is important to note that these correlations are not calculated based on the total score for the avoidant PD subscale; but for the specific factor—which reflects the variance that remains after one has cor-rected for the variance explained by the general factor. Therefore, the correlations between the avoidant PD specific factor and exter-nal variables are lower than they would have been if subscale scores had been used. With this in mind, we would argue that the Table 4 (Continued)

Item Description g AVPD DPD OCPD PPD STPD SPD HPD NPD BPD CD ASPD

BPD 3 Identity disturbance .60 .47 BPD 4 Self-harming impulsivity .64 .36 BPD 5 Suicidality .44 .45 BPD 6 Affective instability .62 .58 BPD 7 Feeling empty .34 .33 BPD 8 Intense anger .66 .37 BPD 9 Transient dissociation .47 .30 CD 1 Threatened/scared others .50 .63 CD 2 Started fights .51 .62 CD 3 Used weapons .51 .68

CD 4 Physically hurt others .51 .63

CD 5 Physically hurt animals .51 .45

CD 6 Mugged .54 .65

CD 7 Forced sexual contact .66 .48

CD 8 Started fires .46 .63

CD 9 Destroyed property .48 .63

CD 10 Breaking and entering .44 .67

CD 11 Lied .56 .52

CD 12 Stole valuables .46 .61

CD 13 Out all night before 15 .48 .55

CD 14 Out all night before 13 .52 .59

CD 15 Skipped school before 13 .39 .51

ASPD 1 Failure to conform .56 .61

ASPD 2 Deceitfulness .70 .36

ASPD 3 Impulsivity .71 .40

ASPD 4 Irritable/aggressive .59 .48

ASPD 5 Disregard for safety .42 .63

ASPD 6 Irresponsible .54 .50

ASPD 7 Lacks remorse .61 .56

Note. Loadings,j.30j are printed in gray. g = g-factor; AVPD = avoidant personality disorder; DPD = dependent personality disorder; OCPD = obses-sive-compulsive personality disorder; PPD = paranoid personality disorder; STPD = schizotypal personality disorder; SZPD = schizoid personality disor-der; HPD = histrionic personality disordisor-der; NPD = narcissistic personality disordisor-der; BPD = borderline personality disordisor-der; CD = conduct disordisor-der; ASPD = antisocial personality disorder.

₇

(10)

correlations found for the avoidant PD speciﬁc factor (ranging between .10 and .29) are still high enough to be of clinical interest.

Avoidant PD is a comprehensive construct representing features from different trait domains; notably, negative affectivity and detachment (Bach & First, 2018). Avoidant PD is one of the most common PDs encountered in clinical practice ( Zimmer-man et al. 2005). Moreover, avoidant PD has proven its clinical utility in a number of studies, delineating a disorder associated with significant functional impairment and high risk of a poor course (Lampe & Malhi, 2018). It is increasingly recognized that avoidant PD represents particular treatment challenges and requires specially adapted treatments (Simonsen, et al., 2019). By dismissing this category, one not only risks the loss of a considerable amount of information, but also the loss of a diag-nostic category with useful clinical applications. Thus, it is of the utmost importance that patients who meet the criteria for avoidant PD in the traditional systems are identified adequately, if the new diagnostic systems (i.e., AMPD or ICD-11) are used. In the AMPD, personality functioning is measured first, after which traits are considered if the diagnostic threshold for the Level of Personality Functioning Scale is met. If this procedure were to be retained, we argue that the Level of Personality Functioning Scale should be carefully scrutinized to see if it adequately captures personality functioning elements relevant to patients with avoidant PD. Alternatively, one could discard the multistage process and administer both measures of

personality functioning and traits, regardless of whether or not the diagnostic threshold for personality functioning is met; one may expect that in this way it should be possible to capture essential elements of avoidant PD, since negative affectivity and detachment would be covered by the trait domains. The lat-ter strategy could also be applied, if the ICD-11 classiﬁcation is used, where trait domain qualiﬁers are then optional.

As far as we know, of all bifactor studies involving PD criteria, only the research ofRingwald et al. (2019)reported the ECV for the g-factor, which was much larger than ours (86%). However, the study of Ringwald et al. is also the only previous bifactor study involving PD criteria in which the avoidant PD criteria did not load on the g-factor. In the other bifactor studies, the avoidant cri-teria loaded on both the g-factor and a specific factor (Conway, et al., 2016;Jahng et al. 2011;Sharp et al. 2015;Williams et al. 2018). When comparing the correlations between the estimated g-factor and external variables in our study to the values reported by previous bifactor studies, we observe that some of the correlations between the estimated g-factor and external variables were lower than could have been expected based on previous research. For example,Williams et al. (2018)found substantially stronger corre-lations between the g-factor they estimated and anxiety and depression, which they measured with the Mini-International Neu-ropsychiatric Interview (Sheehan et al. 1998), than we did. Fur-thermore, Conway et al. (2016) found standardized regression coefficients when regressing their g-PD factor on several measures of functioning that were somewhat higher than the correlations we found. Discrepancies between previous bifactor studies and our study could be explained by a number of methodological differen-ces, that is, sampling differences (pure clinical samples vs. com-bined clinical/community samples, younger vs. older participants), differences in PD assessments (use of lay persons vs. use of clini-cians; assessment of all PD criteria vs. a selection of criteria; systematic use of structured clinical interviews vs. clinical assess-ment), and differences in statistical techniques (confirmatory bifactor analyses vs. exploratory bifactor analysis; the number of specific factors included in the analyses). In this study, we used a large dataset containing ratings made by experienced clinicians using standardized instruments to assess all current DSM PD crite-ria. We chose to use a confirmatory analysis, as this type of analy-sis allows for a careful investigation of PDs as currently defined in the DSM system. We do acknowledge, however, that most of the external variables in our study were measured with relatively short instruments, and all of them were based on self-report; which may have impacted the correlations. More research on the associations between the g-PD factor and relevant external criteria is needed to clarify this issue further. Furthermore, in order to comply with the General Data Protection Regulation of the EU, we were not per-mitted to register treatment units in the database. As a conse-quence, we were not able to formally investigate the potential influence of the dependence structure (patients clustered in outpa-tient units) on ourfindings. This being said, we do not expect that the clustering resulted in systematic bias, since all involved thera-pists followed the same diagnostic strategy and participated in the same diagnostic assessment training, which resulted in a similar level of supervision and experience.

It should also be noted that all previous studies using bifactor analyses to investigate covariance among PD criteria were U.S.-based. Although there is some indication that cultural differences Table 5

Correlations Between Factors Based on Constrained Models and External Variables

WSAS PHQ-9 GAD-7 CIP SIPP-g

g-factors Total .15 .24 .32 .22 .56 Total with CD .15 .23 .30 .21 .55 Specific factors AVPDa _.24 _.18 _.10 _.29 _.15 DPDa _.17 _.15 _.13 _.20 _.11 OCPDa _.15 _.10 _.06 _.14 _.07 PPDa _.12 _.12 _.10 _.18 _.02 STPDa _.19 _.08 _.04 _.14 _.01 SZPDa _.15 _.04 _.14 _.04 _.02 HPDa _.07 _.01 _.00 _.01 _.04 NPDa _.02 _.10 _.12 _.01 _.00 BPDa _.08 _.19 _.21 _.08 _.37 CDa _.02 _.09 _.10 _.04 _.05 ASPDb _.00 _.01 _.02 _.01 _.04

Note. Underlined, bolded correlations.j.30 j. AVPD = avoidant person-ality disorder; DPD = dependent personperson-ality disorder; OCPD = obsessive-compulsive personality disorder; PPD = paranoid personality disorder; STPD = schizotypal personality disorder; SZPD = schizoid personality disorder; HPD = histrionic personality disorder; NPD = narcissistic per-sonality disorder; BPD = borderline perper-sonality disorder; CD = conduct disorder; ASPD = antisocial personality disorder; WSAS = Work and Social Adjustment Scale; PHQ-9 = nine-item Patient Health Questionnaire; GAD-7 = seven-item Generalized Anxiety Disorder scale; CIP = Circumplex of Interpersonal Problems scale; SIPP-g = g-factor of the Severity Indices of Personality Problems.

a_{Factor scores are based on the constrained model including all} PDs. b_{Factor scores are based on the constrained model including all} PDs except the CD criteria. Pairwise deletion was used; seeTable 1for effective sample sizes.

(11)

may not have a large impact on the structure of personality dimen-sions (Allik, 2005), it would be of clinical and scientiﬁc interest to conduct more bifactor studies outside the US.

Schizotypal PD is another diagnosis that is retained in the AMPD. In the empirical literature, it is often emphasized that schizotypal PD is composed of two main components, often la-beled as the “perceptual dysregulation factor” (“ideas of refer-ence,” “magical thinking,” and “unusual perceptual experiences”) and“oddness factor” or “disorganized factor” (“odd thinking and speech,” “constricted affect,” and “odd appearance or behavior”; Hummelen, et al., 2012). Our study found that the criteria belong-ing to the oddness factor did not load on the g-factor, which is in accordance with two other studies that reported separate factor loadings for the PD criteria (Sharp, et al., 2015;Williams et al. 2018). A plausible explanation for thisﬁnding is that these criteria belong to the schizophrenia spectrum, whereas the criteria of the perceptual dysregulation factor are in a stronger degree related to the general factor of personality pathology. In a former twin study, Torgersen et al. (2002)found that schizotypal patients with pre-dominant cognitive-perceptual features had elevated levels of bor-derline symptoms without having an increased prevalence of schizophrenia among their relatives, whereas the criteria of the oddness factor like inadequate rapport and odd communication, appeared to be the genetic core of schizotypy, as it is related to schizophrenia.

In line with most previous studies using bifactor modeling to investigate the structure of personality pathology, we found that the borderline PD criteria loaded relatively strongly on the g-factor (Conway et al., 2016;Ringwald et al. 2019; Sharp et al. 2015; Williams et al. 2018). A closer inspection of the factor loadings revealed that these criteria seem to be highly representative for self-regulation difficulties, especially within interpersonal con-texts; for example, “aggressive reactions to imagined slander” (paranoid PD), “theatrical expressions of emotion” (histrionic PD), “entitlement” (narcissistic PD), “interpersonal instability” (borderline PD), and“impulsivity” (antisocial PD). Thus, one way to approach the g-factor in our study is to consider it as represent-ing self-regulation problems. This position is supported by the finding that our g-factor was strongly correlated to the g-factor of the SIPP-118, which was labeled as self-regulation in a former study on the SIPP-118 which included over 10,000 patients (Paap et al., 2020). Another interpretation posits that our g-factor relates to the well-known externalizing factor as conceptualized by the Hierarchical Taxonomy of Psychopathology (HiTOP) consortium (Kotov et al., 2017). However, the g-factor proposed by the HiTOP consortium (Kotov et al., 2017) appears to be a broader construct compared to the g-factor found in the current study. Because many criteria did not load on the g-factor at all but rather constitute strong specific factors, our results do not lend strong support for the HiTOP approach.

In sum, ourfindings indicate that having specific PDs may still be pivotal in the assessment of personality pathology. Because the specific factors together explained more than 60% of the common variance, while the g-factor explained less than 40% of the com-mon variance, the existence of such a g-factor could not be used as an argument to dismiss all specific PD categories. This finding should not come as a surprise; the PD criteria were never intended to capture a g-factor. This being said, we do not mean to suggest that a g-PD factor could not be of clinical importance. Rather, one

may need instruments speciﬁcally developed to capture broad underlying constructs as personality functioning. Indeed, in a recent study, we found support for a strong common factor under-lying the Level of Personality Functioning Scale as measured by the SCID-5-AMPD-I (Hummelen, et al., 2020), which has a strong focus on self- and interpersonal functioning as core concepts of personality functioning. Our results do suggest that if such a mea-sure is used to identify presence of a PD, it is highly important that aspects pertaining to avoidant PD are covered as well, either by including them in the measure of personality functioning or by measuring them using trait-based instruments. Our results may be taken as tentative support for the current approach to PD assess-ment offered by the AMPD, which allows for separate instruassess-ments being used for the assessment of overall personality functioning, traits, and diagnosis of speciﬁc PDs.

References

Alden, L. E., Wiggins, J. S., & Pincus, A. L. (1990). Construction of Cir-cumplex Scales for the Inventory of Interpersonal Problems. Journal of Personality Assessment, 55(3–4), 521–536. https://doi.org/10.1080/ 00223891.1990.9674088

Allik, J. (2005). Personality dimensions across cultures. Journal of Person-ality Disorders, 19(3), 212–232.https://doi.org/10.1521/pedi.2005.19.3 .212

American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders (DSM-5®_{). Author.}

Andrea, H., Verheul, R., Berghout, C. C., Dolan, C., van der Kroft, P. J. A., Bateman, A. W., Fonagy, P., & Busschbach, J. J. (2007). Meas-uring the Core Components of Maladaptive Personality: Severity Indi-ces of Personality Problems (SIPP-118) - The first technical report. Viersprong Institute for Studies on Personality Disorders (VISPD) Med-ical Psychology and Psychotherapy.

Bach, B., & First, M. B. (2018). Application of the ICD-11 classification of personality disorders. BMC Psychiatry, 18(1), 351.https://doi.org/10 .1186/s12888-018-1908-3

Bagby, R. M., & Widiger, T. A. (2020). Assessment of the ICD-11 dimen-sional trait model: An introduction to the special section. Psychological Assessment, 32(1), 1–7.https://doi.org/10.1037/pas0000785

Bastiaansen, L., Hopwood, C. J., Van den Broeck, J., Rossi, G., Schotte, C., & De Fruyt, F. (2016). The twofold diagnosis of personality disor-der: How do personality dysfunction and pathological traits increment each other at successive levels of the trait hierarchy? Personality Disor-ders: Theory, Research, and Treatment, 7(3), 280–292.https://doi.org/ 10.1037/per0000149

Bender, D. S., Skodol, A. E., First, M. B., & Oldham, J. M. (2018). Struc-tured Clinical Interview for the DSM-5 Alternative Model for Personal-ity Disorders (SCID-5-AMPD); Module I: Level of PersonalPersonal-ity Functioning Scale. American Psychiatric Association Publishing. Bonifay, W., Lane, S. P., & Reise, S. P. (2017). Three concerns with

applying a bifactor model as a structure of psychopathology. Clinical Psychological Science, 5(1), 184–186.https://doi.org/10.1177/2167702 616657069

Cai, L. (2010). A two-tier full-information item factor analysis model with applications. Psychometrika, 75(4), 581–612. https://doi.org/10.1007/ s11336-010-9178-0

Chalmers, R. P. (2012). mirt: A multidimensional item response theory package for the R environment. Journal of Statistical Software, 48(6), 1–29.https://doi.org/10.18637/jss.v048.i06

Clark, L. A. (2007). Assessment and diagnosis of personality disorder: Pe-rennial issues and an emerging reconceptualization. Annual Review of

₉

(12)

Psychology, 58(1), 227–257. https://doi.org/10.1146/annurev.psych.57 .102 904.190200

Conway, C. C., Hammen, C., & Brennan, P. A. (2016). Optimizing predic-tion of psychosocial and clinical outcomes with a transdiagnostic model of personality disorder. Journal of Personality Disorders, 30(4), 545–566.https://doi.org/10.1521/pedi_2015_29_218

Cook, K. F., Kallen, M. A., & Amtmann, D. (2009). Having a fit: Impact of number of items and distribution of data on traditional criteria for assessing IRT’s unidimensionality assumption. Quality of Life Research, 18(4), 447–460.https://doi.org/10.1007/s11136-009-9464-4

Crawford, M. J., Koldobsky, N., Mulder, R., & Tyrer, P. (2011). Classify-ing personality disorder accordClassify-ing to severity. Journal of Personality Disorders, 25(3), 321–330.https://doi.org/10.1521/pedi.2011.25.3.321

First, M. B., Gibbon, M., Spitzer, R. L., Williams, J. B. W., & Benjamin, L. S. (1997). The Structured Clinical Interview for DSM–IV Axis II Per-sonality Disorders (SCID-II). American Psychiatric Press.

First, M. B., Williams, J. B. W., Karg, R. S., & Spitzer, R. L. (2015). User’s Guide for the SCID-5-PD (Structured Clinical Interview for DSM-5 Personality Disorder). American Psychiatric Association. Hu, L., & Bentler, P. M. (1998). Fit indices in covariance structure

model-ing: Sensitivity to unparameterized model misspecification. Psychologi-cal Methods, 3(4), 424–453.https://doi.org/10.1037/1082-989X.3.4.424

Hummelen, B., Braeken, J., Buer Christensen, T., Nysaeter, T.-E., Germans Selvik, S., Walther, K., Pedersen, G., Eikenaes, I., & Paap, M. C. S. (2020). A Psychometric Analysis of the Structured Clinical Interview for the DSM-5 Alternative Model for Personality Disorders Module I; Level of Personality Functioning Scale (SCID-5-AMPD-I). Assessment. Advance online publication.https://doi.org/10.1177/107319 1120967972

Hummelen, B., Pedersen, G., & Karterud, S. (2012). Some Suggestions for the DSM-5 Schizotypal personality disorder construct. Comprehensive Psychiatry, 53(4), 341–349. https://doi.org/10.1016/j.comppsych.2011 .05.009

Huprich, S. K. (2015). Introduction: Personality disorders into the 21st century. In S. K. Huprich (Ed.), Personality disorders: Toward theoreti-cal and empiritheoreti-cal integration in diagnosis and assessment (pp. 3–19). American Psychological Association.https://doi.org/10.1037/14549-001

Hyman, S. E. (2010). The diagnosis of mental disorders: The problem of reification. Annual Review of Clinical Psychology, 6(1), 155–179.

https://doi.org/10.1146/annurev.clinpsy.3.022806.091532

Jahng, S., Trull, T. J., Wood, P. K., Tragesser, S. L., Tomko, R., Grant, J. D., Bucholz, K. K., & Sher, K. J. (2011). Distinguishing general and specific personality disorder features and implications for substance de-pendence comorbidity. Journal of Abnormal Psychology, 120(3), 656–669.https://doi.org/10.1037/a0023539

Karterud, S., Pedersen, G., Friis, S., Urnes, Ø., Irion, T., Brabrand, J., Falkum, L. R., & Leirvåg, H. (1998). The Norwegian network of psy-chotherapeutic day hospitals. Therapeutic Communities, 19(1), 15–28. Kotov, R., Krueger, R. F., Watson, D., Achenbach, T. M., Althoff, R. R.,

Bagby, R. M., Brown, T. A., Carpenter, W. T., Caspi, A., Clark, L. A., Eaton, N. R., Forbes, M. K., Forbush, K. T., Goldberg, D., Hasin, D., Hyman, S. E., Ivanova, M. Y., Lynam, D. R., Markon, K., . . . Zimmerman, M. (2017). The hierarchical taxonomy of psychopathology (HiTOP): A dimensional alternative to traditional nosologies. Journal of Abnormal Psychology, 126(4), 454–477. https://doi.org/10.1037/abn 0000258

Kroenke, K., Spitzer, R. L., & Williams, J. B. W. (2001). The PHQ-9: Va-lidity of a brief depression severity measure. Journal of General Internal Medicine, 16(9), 606–613. https://doi.org/10.1046/j.1525-1497.2001 .016009606.x

Lampe, L., & Malhi, G. S. (2018). Avoidant personality disorder: Current insights. Psychology Research and Behavior Management, 11, 55–66.

https://doi.org/10.2147/PRBM.S121073

Lobbestael, J., Leurgans, M., & Arntz, A. (2011). Inter-rater reliability of the Structured Clinical Interview for DSM–IV Axis I Disorders (SCID I) and Axis II Disorders (SCID II). Clinical Psychology and Psychother-apy, 18(1), 75–79.https://doi.org/10.1002/cpp.693

Mundt, J. C., Marks, I. M., Shear, M. K., & Greist, J. H. (2002). The Work and Social Adjustment Scale: A simple measure of impairment in func-tioning. The British Journal of Psychiatry, 180(5), 461–464.https://doi .org/10.1192/bjp.180.5.461

O’Connor Quinn, H. (2014). Bifactor models, explained common variance (ECV), and the usefulness of scores from unidimensional item response theory analyses. University of North Carolina at Chapel Hill.https://cdr .lib.unc.edu/concern/dissertations/w95051780

Paap, M. C. S., Brouwer, D., Glas, C. A. W., Monninkhof, E. M., Forstreuter, B., Pieterse, M. E., & van der Palen, J. (2015). The St George’s Respiratory Questionnaire revisited: A psychometric evalua-tion. Quality of Life Research, 24(1), 67–79. https://doi.org/10.1007/ s11136-013-0570-y

Paap, M. C. S., Hummelen, B., Braeken, J., Arnevik, E. A., Walderhaug, E., Wilberg, T., Berghuis, H., Hutsebaut, J., & Pedersen, G. (2020). A multi-center bi-factor analysis of The Severity Indices of Personality Problems 118 (SIPP-118): Do we really need all those facets? Quality of Life Research. Advance online publication. https://doi.org/10.1007/ s11136-020-02654-8

Pedersen, G. (2002). Norsk revidert versjon av Inventory of Interpersonal Problems—Circumplex (IIP-C) [Revised Norwegian version of the In-ventory of Interpersonal Problems—Circumplex (IIP-C)]. Tidsskrift for Norsk Psykologforening, 39(1), 25–34.

Pedersen, G., Arnevik, E. A., Hummelen, B., Walderhaug, E., & Wilberg, T. (2019). Psychometric properties of the Severity Indices of Personality Problems (SIPP) in two samples: A Norwegian community sample and clinical samples of patients with and without personality disorders. Eu-ropean Journal of Psychological Assessment, 35(5), 698–711.https:// doi.org/10.1027/1015-5759/a000436

Pedersen, G., Hagtvet, K. A., & Karterud, S. (2011). Interpersonal prob-lems: Self-therapist agreement and therapist consensus. Journal of Clini-cal Psychology, 67(3), 308–317.https://doi.org/10.1002/jclp.20762

Pedersen, G., Karterud, S., Hummelen, B., & Wilberg, T. (2013). The impact of extended longitudinal observation on the assessment of per-sonality disorders. Perper-sonality and Mental Health, 7(4), 277–287.

https://doi.org/10.1002/pmh.1234

Pedersen, G., Kvarstein, E. H., & Wilberg, T. (2017). The Work and Social Adjustment Scale: Psychometric properties and validity among males and females, and outpatients with and without personality disorders. Personality and Mental Health, 11(4), 215–228.https://doi.org/10.1002/ pmh.1382

R Core Team. (2017). R: A language and environment for statistical computing. R Foundation for Statistical Computing. https://www.R-project.org/.

Reise, S. P. (2012). The Rediscovery of Bifactor Measurement Models. Multivariate Behavioral Research, 47(5), 667–696. https://doi.org/10 .1080/00273171.2012.715555

Reise, S. P., Kim, D. S., Mansolf, M., & Widaman, K. F. (2016). Is the bifactor model a better model or is it just better at modeling implausible responses? Application of Iteratively Reweighted Least Squares to the Rosenberg Self-Esteem Scale. Multivariate Behavioral Research, 51(6), 818–838.https://doi.org/10.1080/00273171.2016.1243461

Reise, S. P., Scheines, R., Widaman, K. F., & Haviland, M. G. (2013). Multidimensionality and structural coefficient bias in structural equation modeling: A bifactor perspective. Educational and Psychological Mea-surement, 73(1), 5–26.https://doi.org/10.1177/0013164412449831

Ringwald, W. R., Beeney, J. E., Pilkonis, P. A., & Wright, A. G. C. (2019). Comparing hierarchical models of personality pathology. Jour-nal of Research in PersoJour-nality, 81, 98–107.https://doi.org/10.1016/j.jrp .2019.05.011 Thi s d o cumen t is copy righte d b y the Amer ican Ps ycholo gic al Asso ciatio n o r o n e of its all ied publi she rs. Con tent may be shar ed at no co st, but any reque sts to reu se thi s co ntent in par t o r whol e m u st g o thro ugh the Amer ican Psyc holog ical Asso ciatio n.

(13)

Schwarz, G. (1978). Estimating the Dimension of a Model. Annals of Sta-tistics, 6(2), 461–464.https://doi.org/10.1214/aos/1176344136

Sharp, C., Wright, A. G., Fowler, J. C., Frueh, B. C., Allen, J. G., Oldham, J., & Clark, L. A. (2015). The structure of personality pathology: Both general (‘g’) and specific (‘s’) factors? Journal of Abnormal Psychology, 124(2), 387–398.https://doi.org/10.1037/abn0000033

Sheehan, D. V., & Lecrubier, Y. (1994). Mini International Neuropsychi-atric Interview (M.I.N.I.). University of South Florida Institute fore Research in Psychiatry/INSERM-Hôpital de la Salpétrière.

Sheehan, D. V., Lecrubier, Y., Sheehan, K. H., Amorim, P., Janavs, J., Weiller, E., Hergueta, T., Baker, R., & Dunbar, G. C. (1998). The Mini-International Neuropsychiatric Interview (M.I.N.I.): the development and validation of a structured diagnostic psychiatric interview for DSM–IV and ICD-10. The Journal of Clinical Psychiatry, 59(Suppl 20), 22–57.

Simonsen, S., Eikenæs, I. U.-M., Nørgaard, N. L., Normann-Eide, E., Juul, S., & Wilberg, T. (2019). Specialized treatment for patients with severe avoidant personality disorder: Experiences from Scandinavia. Journal of Contemporary Psychotherapy, 49(1), 27–38. https://doi.org/10.1007/ s10879-018-9395-x

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 rep-resentations of DSM–IV personality disorders: Relationships to func-tional impairment. The American Journal of Psychiatry, 162(10), 1919–1925.https://doi.org/10.1176/appi.ajp.162.10.1919

Somma, A., Borroni, S., Maffei, C., Besson, E., Garbini, A., Granozio, S., Limuti, B., Perego, G., Pietrobon, A., Rugi, C., Turano, E., & Fossati, A. (2017). Inter-rater reliability of the Italian Translation of the Struc-tured Clinical Interview for DSM-5 Personality Disorders (SCID-5-PD): A study on consecutively admitted clinical adult participants. Journal of Psychopathology, 23(3), 105–111.

Spitzer, R. L. (1983). Psychiatric diagnoses: Are clinicians still necessary? Comprehensive Psychiatry, 24(5), 399–411. https://doi.org/10.1016/ 0010-440X(83)90032-9

Spitzer, R. L., Kroenke, K., Williams, J. B., & Lowe, B. (2006). A brief measure for assessing generalized anxiety disorder: The GAD-7.

Archives of Internal Medicine, 166(10), 1092–1097.https://doi.org/10 .1001/archinte.166.10.1092

Torgersen, S., Edvardsen, J., Øien, P. A., Onstad, S., Skre, I., Lygren, S., & Kringlen, E. (2002). Schizotypal personality disorder inside and out-side the schizophrenic spectrum. Schizophrenia Research, 54(1-2), 33–38.https://doi.org/10.1016/S0920-9964(01)00349-8

Tyrer, P. (2005). The problem of severity in the classification of personal-ity disorder. Journal of Personalpersonal-ity Disorders, 19(3), 309–314.https:// doi.org/10.1521/pedi.2005.19.3.309

Watson, D. C., & Sinha, B. K. (1998). Comorbidity of DSM–IV personality disorders in a nonclinical sample. Journal of Clinical Psychology, 54(6), 773–780.

Widiger, T. A., & Clark, L. A. (2000). Toward DSM-V and the classifica-tion of psychopathology. Psychological Bulletin, 126(6), 946–963.

https://doi.org/10.1037/0033-2909.126.6.946

Widiger, T. A., & Samuel, D. B. (2005). Diagnostic categories or dimen-sions? A question for the diagnostic and statistical manual of mental dis-orders–fifth ed Journal of Abnormal Psychology, 114(4), 494–504.

https://doi.org/10.1037/0021-843X.114.4.494

Widiger, T. A., & Trull, T. J. (2007). Plate tectonics in the classification of personality disorder: Shifting to a dimensional model. American Psy-chologist, 62(2), 71–83.https://doi.org/10.1037/0003-066X.62.2.71

Williams, T. F., Scalco, M. D., & Simms, L. J. (2018). The construct valid-ity of general and specific dimensions of personalvalid-ity pathology. Psycho-logical Medicine, 48(5), 834–848.https://doi.org/10.1017/S003329171 7002227

World Health Organization. (2018). ICD-11 clinical descriptions and diagnostic guidelines for mental and behavioural disorders. Author. Zimmerman, M. (2012). Is there adequate empirical justification for

radi-cally revising the personality disorders section for DSM-5? Personality Disorders: Theory, Research, and Treatment, 3(4), 444–457.https://doi .org/10.1037/a0022108

Zimmerman, M., Rothschild, L., & Chelminski, I. (2005). The prevalence of DSM–IV personality disorders in psychiatric outpatients. The Ameri-can Journal of Psychiatry, 162(10), 1911–1918.https://doi.org/10.1176/ appi.ajp.162.10.1911