Maladaptive Personality Traits in Primary Care: Validation of the German Modified Brief Personality Inventory for DSM-5 (PID5BF + M)

Abstract

This study evaluated the psychometric properties of the modified brief Personality Inventory for DSM-5 (PID5BF + M) in primary care (PC) using data from n = 1,030 German patients. Furthermore, differences in maladaptive personality traits between PC patients and the general population were explored. Confirmatory factor analysis supported factorial validity (CFI = 0.949, TLI = 0.942, RMSEA = 0.044, and SRMR = 0.058). Reliability was adequate across domain scales (ω_H: 0.75–0.85). PID5BF + M total and domain scale scores, particularly negative affectivity, correlated significantly positively with depression (PHQ-9), anxiety (GAD-7), and somatic symptoms (PHQ-15), indicating convergent validity. Regression analyses showed PC to be associated with lower levels of maladaptive personality traits, compared with a representative German population sample (n = 4,172). These findings highlight the PID5BF + M as a valid and reliable tool for assessing personality pathology and maladaptive traits in PC, enabling general practitioners to screen for transdiagnostic indicators of mental health conditions.

Introduction

Mental Health Conditions in Primary Care: Limitations of Categorical Diagnostic Taxonomies

The two leading diagnostic taxonomies currently in use, the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) section II (American Psychiatric Association, 2013) and the International Classification of Diseases, 10th Revision (ICD-10; World Health Organisation, 1993), classify mental disorders as distinct entities based on the binary occurrence (i.e., present/not present) of a certain set of symptoms. This approach has several shortcomings: First, it has been suggested that established diagnostic thresholds are largely arbitrary in nature, following that strict cut-offs defining when a specific diagnosis is given may be questionable (Clark et al., 2017). Second, great inter-individual heterogeneity exists within diagnoses (Newson et al., 2021). For instance, with only five out of nine possible symptoms being required for a borderline personality disorder diagnosis, two affected individuals could match on only one symptom (Woods et al., 2020). The lack of empirical equivalence between single symptoms within an associated mental disorder is also of concern (Boschloo et al., 2015). Third, assigning a clear diagnosis to an individual can be difficult due to the frequent overlap of symptoms between diagnoses, indicating low validity of diagnostic constructs (Rodriguez-Seijas et al., 2015). For example, core symptoms of major depressive disorder are shared with 35 other diagnoses (Forbes et al., 2023). This heterogeneity is further reflected by individuals’ tendency to shift between mental disorders across their lifespan (Caspi et al., 2020; Menzies et al., 2024), implying an overly rigid categorization of symptoms into specific disorder clusters (Dalgleish et al., 2020).

Particularly in primary care (PC), the diagnosis of mental illness according to standard classification systems may be inadequate (Hanel et al., 2009; World Health Organisation, 2008). Usually being the first point of contact for individuals with symptoms, PC serves as the interface between the general population and more specialized healthcare settings (Green et al., 2001). Compared to the general population, PC patients can be distinguished by a higher disease burden (Mokraoui et al., 2016), with the early identification of mental health conditions as a key function of PC (Alexander et al., 2010; Gaebel et al., 2013; World Health Organisation, 2018). However, assessing a large number of specific mental disorder criteria has been deemed unfeasible for general practitioners (GPs; Brown & Wissow, 2012; Hanel et al., 2009). This approach seems not only impractical in terms of the time available per patient (Irving et al., 2017; Talen et al., 2013) but also of limited clinical utility. As a gateway to psychiatric care, PC is characterized by patients who frequently present with mild to moderate, nonspecific psychological distress. Symptoms often exist along a continuum and may overlap with multiple mental disorders (Gili et al., 2010; Linden, 2004). Given these distinct characteristics of PC, screening tools are needed that support rather than burden GPs in their preliminary assessment of psychologically distressed patients (Neulinger et al., 2024; Talen et al., 2013).

Personality Disorders and the Shift to a Dimensional Approach to Psychopathology

Instead of assessing psychopathology through distinct, disorder-specific categories, a continuous understanding has been proposed (Dalgleish et al., 2020; Haslam et al., 2020; Kotov et al., 2017; Widiger & Samuel, 2005). Given the aforementioned non-specificity of psychological symptoms in PC (Brown & Wissow, 2012; Hanel et al., 2009; Linden, 2004), this approach may be promising for the diagnostic practice of GPs. A prime example of a shift toward dimensional models is represented by personality disorders. In response to accumulating evidence on the dimensional nature of personality (Hopwood et al., 2018), an alternative model of personality disorders (AMPD) was implemented in section III of the DSM-5 (American Psychiatric Association, 2013), and a similar model was adopted in the ICD-11 (World Health Organisation, 2019/2021). In both models, personality pathology is defined in terms of a general level of personality dysfunction and specific personality domains. For the latter, the DSM-5 and ICD-11 each define a total of five maladaptive personality traits. While four traits are highly similar (i.e., negative affectivity, detachment, antagonism/dissociality, and disinhibition), both models differ regarding the fifth domain, which refers to psychoticism in the DSM-5 and anankastia in the ICD-11.

Transdiagnostic Mental Health Screening in PC: The Potential of the PID5BF + M

Two questionnaires that are widely used to measure the personality traits specified by each classification system are the Personality Inventory for DSM-5 (PID-5) (Krueger et al., 2012) and the Personality Inventory for ICD-11 (PiCD; Oltmanns & Widiger, 2018). Recently, a brief modified version of the PID-5 was developed. Originally comprising 34 items (PID5BF+) (Kerber et al., 2022), it has been further refined with 36 items (PID5BF + M) (Bach et al., 2020). Due to its extension by the domain of anankastia, the PID5BF + M allows the maladaptive personality traits of both classification systems to be assessed using trait-specific sum scores. Furthermore, the total score can be used to indicate the overall severity of personality pathology (Pires et al., 2023; Zimmermann et al., 2020).

The PID-5 in its brief form has been proposed as a valuable screener for PC (Porcerelli et al., 2019). In addition to assessing personality pathology –25% of PC patients show signs of a personality disorder [Tyrer et al., 2015]) – the PID5BF + M may be particularly useful in PC, as psychologically distressed patients without a personality disorder still present with elevated levels of maladaptive personality traits (Pires et al., 2021). Although defined separately, maladaptive personality traits and psychological symptoms show some functional relationship (Abdi & Pak, 2019; Hong & Tan, 2021; Perkins et al., 2019; Sauer-Zavala et al., 2022; Stanton et al., 2016). Accordingly, maladaptive personality traits were suggested as higher-order constructs, operating as transdiagnostic indicators for a wide range of lower-order symptoms of mental illness (Fowler et al., 2022; Kerber et al., 2022; Kotov et al., 2010; Porcerelli et al., 2019; Rek et al., 2021; Subica et al., 2016; Wendt et al., 2024; Wright et al., 2015). Negative affectivity (i.e., neuroticism) appears to play a particular role (Kotov et al., 2010; Rek et al., 2021; Stanton et al., 2016) as it is considered to be an important marker of “emotion-based disorders” (Barlow et al., 2021; Tyrer et al., 2016). Characterized by a heightened experience of (negative) emotions and subsequent avoidance-based coping mechanisms (Bullis et al., 2019), common mental disorders in PC, such as depression, anxiety, and somatic symptom disorders (Hanel et al., 2009; Roca et al., 2009), can be subsumed under this umbrella term. Routinely screening for maladaptive personality traits – especially increased neuroticism/negative affectivity – may enable GPs to provisionally identify patients who are at risk of or affected by mental illness (Böhnke et al., 2014; Lahey, 2009; Neulinger et al., 2024; Rek et al., 2021; Widiger & Oltmanns, 2017). The PID5BF + M could therefore be a more feasible, but no less informative, addition to the diagnostic process of GPs than disorder-specific screening (Böhnke et al., 2014; Porcerelli et al., 2019).

Research Gap: The Psychometric Properties and Clinical Utility of the PID5BF + M in PC

A recent systematic review on screening questionnaires in PC recommended evaluating personality questionnaires as tools for the transdiagnostic assessment of mental illness (Neulinger et al., 2024). While research on the psychometric properties of the PID5BF + M has been conducted in inpatient and community samples (Bach et al., 2020; Riegel et al., 2021; Zinchuk et al., 2023), its evaluation in PC remains limited. To date, only Porcerelli et al. (2019) have examined the discriminant and convergent validity of the PID5BF – another abbreviated version of the PID-5 – using a small sample of 100 patients. The authors explicitly called for replication studies and stressed the importance of evaluating the utility of personality screening in relation to commonly used depression- and anxiety-focused screeners. As a result, key gaps remain in understanding the psychometric properties and clinical utility of the PID5BF + M in PC. To address these gaps, this study aimed to (1) assess the factorial validity and reliability of the PID5BF + M in a German PC sample; (2) evaluate its potential as an early indicator of mental illness in PC by examining convergent validity with established measures of depressive, anxiety, and somatic symptom severity; and (3) analyze the association between the PC setting and maladaptive personality traits, using a representative sample of the German general population for comparison.

Methods

Contextual Background on the German Healthcare System

The German healthcare system operates on a dual model, comprising statutory health insurance and private health insurance. The statutory health insurance is based on the principle of solidarity and is primarily funded through wage-related contributions of about 14% of income, with employees and employers each paying half. This structure ensures universal health coverage for all residents, as health insurance is mandatory in Germany (Busse et al., 2017). Primary healthcare is predominantly delivered by GPs. Particularly for patients with mental health conditions, GPs are often the sole treatment providers (Häfner & Petzold, 2007). Notably, patients have the autonomy to choose and change their GP of choice at any time and to consult specialists directly without a GP referral (Blümel et al., 2020).

Procedure and Participants

Cross-sectional data collection took place between December 2022 and May 2023 in 10 PC practices in Munich, Germany. PC patients were informed about the study by members of the research team or medical assistants in the waiting room of their PC practice. If willing to participate, and after giving oral informed consent, they completed a self-report questionnaire assessing sociodemographic and health-related data.

Considering only PC patients with complete data (n = 1,030), 59.6% identified as women, the mean age was 46.5 years (SD = 16.61), and 90.6% stated to have a German citizenship. 66.8% participants were in a relationship or married (single: 20.7%; annulled: 7.8%; widowed: 4.7%). 46.5% had a high-school diploma (no school-leaving certificate: 0.9%; lower secondary school diploma: 24.1%; secondary school diploma: 28.4%). A vocational mastership, technical college, or university degree was held by 35.7% (no vocational training: 6.0%; vocational training: 58.3%). The employment rate was 70.1%. Among participants not working, 11.6% were housewives/housemen, 63.9% were retired, 17.7% were students, 6.8% were unemployed (see Appendix 1 for a comparison of sociodemographic and mental health symptom data between included and excluded PC patients).

The reference sample of the German general population was derived from the standard edition of the GESIS panel (Leibniz Institute for the Social Sciences, 2023). Funded by federal and state governments and administered by the Leibniz Institute for the Social Sciences, the GESIS panel holds data of a representative sample (n = approximately 5,200 panelists) of the adult German-speaking population (18–70 years of age). Initiated in 2013, today, a total of four cohorts are part of the GESIS panel who complete questionnaires four times per year (i.e., in 4 waves) on different topics, among others, also concerning maladaptive personality traits (Bosnjak et al., 2017; Leibniz Institute for the Social Sciences, 2023). Specifically, for this study, the wave “Ge” was selected (Rek et al., 2021). The identifier “Ge” refers to data collection conducted between October and November 2019, including participants of the first three cohorts who completed the PID5BF + (difference to the PID5BF + M is described below). Later GESIS data were not included to avoid the influence of the COVID-19 pandemic (Wendt et al., 2024). The extracted data comprised 4,172 participants (50.4% female) with a mean age of 58.9 years (SD = 14.5). 65.1% of participants were in a relationship or married (single: 19.8%; annulled: 9.4%; widowed: 5.8%). A high-school diploma was held by 49.1% (no school-leaving certificate: 0.9%; lower secondary school diploma: 16.9%; secondary school diploma: 33.2%), while 40.0% had a vocational mastership, technical college, or university degree (no vocational training: 8.4%; vocational training: 51.6%).

Measures

Modified Personality Inventory for DSM-5 Brief Form

The PID5BF + M (Bach et al., 2020) assesses the five DSM-5 personality traits of psychoticism, disinhibition, negative affectivity, detachment, and antagonism, and is supplemented by the ICD-11 personality domain of anankastia. With a total of 36 items, each domain (e.g., negative affectivity) is divided into three facets (e.g., emotional lability, anxiousness, and separation insecurity) and two items per facet rated on a 4-point Likert scale (0: “does not apply at all”–3: “applies exactly”). A total score and domain-specific scores can be formed, indicating the overall severity of personality pathology and the levels of trait-specific manifestations, respectively. Higher values suggest greater impairment.

In the GESIS panel, the PID5BF+ (Kerber et al., 2022) was employed. Compared to the PID5BF + M, it covers the five DSM-5 personality traits with the same six items each. However, the ICD-11 trait anankastia is measured with only four items, two of which are also included in the PID5BF + M, while the other two differ in content.

For copyright reasons, the exact wording of the PID5BF + /M items was not provided, but is available on the website of the American Psychiatric Association (2025).

Patient Health Questionnaire-9

Depressive symptoms were examined using the Patient Health Questionnaire-9 (PHQ-9; Kroenke et al., 2001), a nine-item measure evaluating symptom severity over the past 2 weeks on a 4-point Likert scale (0: “not at all present”–3: “present nearly every day”). The total score can range from 0 (= no depression) to 27. An acceptable internal consistency was observed in the PC sample (ω = 0.88).

Generalized Anxiety Disorder Screener-7

General anxiety symptoms were measured using the Generalized Anxiety Disorder Screener-7 (GAD-7; Löwe et al., 2008). With seven items rated on a 4-point Likert scale (0: “not at all present”–3: “present nearly every day”), symptom severity over the past 2 weeks is assessed. The sum score can vary between 0 (=no anxiety) and 21. McDonald’s Omega (ω = 0.90) suggested an acceptable internal consistency in the PC sample.

Patient Health Questionnaire-15

Somatic symptom severity was evaluated using the Patient Health Questionnaire-15 (PHQ-15; Kroenke et al., 2002), with 15 items each assessing the experience of a specific somatic symptom in the past 4 weeks on a 3-point Likert scale (0: “not bothered at all”–2: “bothered a lot”). The total score can lie between 0 (=no somatic symptoms) and 30. An acceptable internal consistency was found in the PC sample (ω = 0.84).

Data Analysis

Statistical analysis was conducted with R (R Core Team, 2021), using the following packages: “readxl” (Wickham & Bryan, 2023), “lavaan” (Rosseel et al., 2022), “semTools” (Jorgensen et al., 2022), “psych” (Revelle, 2024), “dplyr” (Wickham et al., 2023), “correlation” (Makowski et al., 2022), “car“ (Fox & Weisberg, 2019), “officer” (Gohel et al., 2025), “flextable” (Gohel & Skintzos, 2024), and “polycor” (Fox, 2022).

For all PID5BF + M items and domain scales, descriptive statistics were calculated, including the mean, standard deviation (SD), skewness, and kurtosis. The reliability of the PHQ-9, GAD-7, and PHQ-15 was estimated using categorical McDonald’s Omega total (ω), whereas the reliability of the PID5BF + M domain scales was examined with McDonald’s Omega hierarchical (ω_H). For the former, ω > 0.65 and for the latter, ω_H > 0.80 were assumed to imply acceptable reliability (Kalkbrenner, 2023).

Factorial validity of the PID5BF + M in the PC sample was assessed using ordinal confirmatory factor analysis (CFA) based on robust weighted least squares means and variances (WLSMV) adjusted estimation, which has been shown to yield more accurate factor loadings for ordinal data than robust maximum likelihood (MLR; Li, 2016). Consistent with previous research (Kerber et al., 2022; Rek et al., 2021), a PID5BF + M higher-order model was specified with two related items (e.g., item 01 and item 19) loading on a corresponding first-order facet (e.g., emotional lability) and three related facets (e.g., emotional lability, anxiousness, and separation insecurity) in turn loading on a corresponding second-order domain (e.g., negative affectivity). As goodness of fit indices, the χ²-likelihood ratio statistic, comparative fit index (CFI), Tucker–Lewis index (TLI), root mean square error of approximation (RMSEA), and standardized root mean square residual (SRMR) were considered. A CFI and TLI value > 0.95, as well as an RMSEA value < 0.06 and a SRMR value < 0.08, were applied as approximate cut-offs indicating adequate model fit (Hu & Bentler, 1999). Convergent validity of PID5BF + M was tested by applying Pearson correlations between the factor scores of the domain scales and the factor scores of the PHQ-9, GAD-7, and PHQ-15.

To examine differences in maladaptive personality traits between PC patients and the general population, regression analyses were conducted. Each model included one of the five maladaptive personality domains (negative affectivity, detachment, antagonism, disinhibition, and psychoticism) as the dependent variable. The independent variables comprised sample setting (PC vs. general population) and sociodemographic factors, which were added to control for potential confounding. Sociodemographic factors included age, gender (female and male), marital status (single, relationship/married, annulled, and widowed), highest level of education (no school-leaving certificate, lower secondary school diploma, secondary school diploma, and high school diploma), and vocational/professional training (no vocational training, vocational training, and mastership/technical college/university degree). Anankastia was excluded from regression analyses due to its different operationalization in the PID5BF+ (completed by the general population sample) than in the PID5BF + M (completed by the PC sample; see Appendix 2 for a figure of the regression model). To ensure comparability of PID5BF+/M data between PC patients and the general population, measurement invariance was tested across both samples. Due to the occurrence of a Heywood case in the higher-order PID5BF + M model when applying CFA in the PC sample, an alternative model was implemented. Following prior studies (Bach et al., 2020; Riegel et al., 2021; Zinchuk et al., 2023), two related items per facet were aggregated to form facet scores, which were then used as indicators in a multi-group CFA (e.g., facet scores for emotional lability, anxiousness, and separation insecurity were defined to load on a corresponding domain factor representing negative affectivity). As the facet scores were derived from the sum of two items each and were therefore not considered ordinal, MLR was used instead of WLSMV. Measurement invariance was assessed sequentially at three levels: (1) evaluating the factor structure (configural invariance), (2) constraining factor loadings (metric invariance), and (3) restricting item intercepts (scalar invariance), with the latter being necessary for meaningful group comparisons (Putnick & Bornstein, 2016). Consistent with previous research addressing measurement invariance of the PID5BF+/M (Kerber et al., 2022; Riegel et al., 2021), a change of less than |0.01| in CFI and RMSEA between levels was used to indicate equivalence.

Results

Descriptive Statistics

In the PC sample, the mean values of the PID5BF + M items ranged from 0.20 (SD: 0.48) to 1.12 (SD: 0.94), with the majority of means < 1. Concerning domain scales, antagonism showed the lowest (mean = 2.42 [SD: 2.69]) and anankastia the highest (mean = 4.95 [SD: 3.82]) mean score (Table 1). The mean total PID5BF + M score was 22 (SD: 14.55).

Table 1.

Descriptive Statistics, McDonald’s ω_H, and CFA Results of a 6-Factor PID5BF+M Higher-Order Model.

Items	M	SD	Skew	Kurtosis	Factor loadings
Negative affectivity (ωH: .80)	4.83	3.72	0.69	−0.02	6-factor higher-order model
Item 01	1.12	0.94	0.27	−0.99	0.81	Emotional Lability: 0.79
Item 19	1.00	0.94	0.51	−0.78	0.94
Item 07	1.08	0.95	0.42	−0.84	0.90	Anxiousness: 0.85
Item 25	0.72	0.85	0.99	0.16	0.91
Item 13	0.70	0.87	1.05	0.17	0.79	Separation Insecurity: 0.84
Item 31	0.21	0.55	2.93	9.03	0.76
Detachment (ωH: .79)	3.35	3.15	1.19	1.49
Item 04	0.88	0.85	0.60	−0.53	0.72	Withdrawal: 0.85
Item 22	0.42	0.68	1.63	2.19	0.83
Item 10	0.46	0.75	1.63	2.05	0.84	Anhedonia: 0.84
Item 28	0.63	0.80	1.02	0.08	0.81
Item 16	0.63	0.84	1.14	0.37	0.71	Intimacy Avoidance: 0.85
Item 34	0.33	0.66	2.10	4.06	0.83
Antagonism (ωH: .81)	2.42	2.69	1.46	2.45
Item 02	0.52	0.77	1.39	1.20	0.71	Manipulativeness: 0.73
Item 20	0.22	0.55	2.87	8.53	0.96
Item 08	0.62	0.79	1.05	0.26	0.73	Deceitfulness: 1.02
Item 26	0.42	0.66	1.46	1.45	0.75
Item 14	0.38	0.67	1.83	3.00	0.69	Grandiosity: 0.96
Item 32	0.26	0.57	2.42	6.04	0.74
Disinhibition (ωH: .75)	3.62	3.08	0.92	0.66
Item 03	0.36	0.64	1.78	2.83	0.75	Irresponsibility: 0.72
Item 21	0.33	0.68	2.20	4.30	0.66
Item 09	0.68	0.80	0.92	0.00	0.77	Impulsivity: 0.96
Item 27	0.70	0.78	0.86	−0.01	0.77
Item 15	0.69	0.82	0.99	0.23	0.87	Distractibility: 0.76
Item 33	0.86	0.87	0.69	−0.38	0.78
Psychoticism (ωH: .81)	2.82	2.94	1.29	1.49
Item 05	0.94	0.96	0.50	−1.01	0.65	Unusual beliefs & experiences: 0.99
Item 23	0.44	0.80	1.83	2.45	0.67
Item 11	0.41	0.74	1.74	2.24	0.80	Eccentricity: 0.90
Item 29	0.53	0.76	1.27	0.73	0.78
Item 17	0.30	0.61	2.14	4.32	0.87	Perceptual dysregulation: 0.80
Item 35	0.20	0.48	2.68	7.58	0.84
Anankastia (ωH: .85)	4.95	3.82	0.68	0.07
Item 06	1.09	0.93	0.37	−0.86	0.80	Perfectionism: 0.92
Item 18	0.90	0.93	0.66	−0.65	0.82
Item 12	1.10	0.93	0.26	−1.02	0.81	Rigidity: 0.94
Item 24	0.85	0.84	0.55	−0.71	0.71
Item 30	0.48	0.79	1.58	1.57	0.80	Orderliness: 0.92
Item 36	0.53	0.77	1.30	0.89	0.79
					Fit indices
					χ2; df; p-value	1,690.535; 561; p < .001
					CFI	0.949
					TLI	0.942
					RMSEA (90% CI)	0.044 [0.042, 0.047]
					SRMR	0.058

Note. M = mean; SD = standard deviation; ω_H = McDonald’s Omega hierarchical; bold = negative latent variance; χ² = chi-squared; df = degrees of freedom; CFI = comparative fit index; TLI = Tucker–Lewis index; RMSEA = root mean square error of approximation; CI = confidence interval; SRMR = standardized root mean square residual; The column “6-factor higher-order model” is divided into a left-hand column with the standardized factor loadings for the items and a right-hand column with the standardized factor loadings for the higher-order facets, each corresponding to two items (e.g., facet Emotional Lability: items 01 and 19).

Factorial Validity and Reliability

Standardized factor loadings of the 6-factor PID5BF + M higher-order model ranged from 0.65 to 0.96 for the items and from 0.72 to 1.02 for the facets. Model fit indices were acceptable: CFI = 0.949, TLI = 0.942, RMSEA = 0.044, SRMR = 0.058. Notably, the model estimation resulted in a Heywood case (i.e., a negative variance for the facet deceitfulness, see Table 1). In a next step, following Kerber et al. (2022), unstandardized facet factor loadings of items were constrained to 1 to avoid potential facet factor underidentification associated with only two items per facet. However, the negative variance for the facet deceitfulness remained (see link to statistical output in the data availability statement). To test whether this may be related to distinct characteristics of the PC sample, another CFA was run – without constraining unstandardized loadings to 1 – including data from both the PC and general population samples. Due to operationalization differences in the domain anankastia between the PID5BF+ (completed by the general population sample) and the PID5BF + M (completed by the PC sample), it was excluded from the analysis. No negative variance occurred, while the overall model fit was similar (factor loadings items: 0.48–0.96; factor loadings facets: 0.66–0.97; CFI: 0.931; TLI: 0.921; RMSEA: 0.058, CI [0.057, 0.059]; SRMR: 0.054; Appendix 3).

McDonald’s ω_H values between 0.75 and 0.85 were observed, indicating an acceptable reliability of PID5BF + M domain scales (Table 1).

Convergent Validity

Factor scores of all PID5BF + M domain scales and total scores correlated significantly positively with the factor scores of the PHQ-9, GAD-7, and PHQ-15. Correlation estimates were highest for the PID5BF + M total score and the domain scale score of negative affectivity. The overall lowest correlation estimates were found for the domain scale score of antagonism (Table 2).

Table 2.

Correlation Estimates of PID5BF + M Domain Scale and Total Factor Scores With PHQ-9, GAD-7, and PHQ-15 Factor Scores.

		Correlation estimates
Questionnaires	Negative affectivity	Detachment	Antagonism	Disinhibition	Psychoticism	Anankastia	Total score
PHQ-9	0.68	0.67	0.38	0.64	0.55	0.37	0.72
GAD-7	0.73	0.54	0.39	0.60	0.53	0.44	0.72
PHQ-15	0.57	0.49	0.24	0.48	0.46	0.37	0.58

Note. All correlation estimates were significant at p < .001; PHQ-9 = Patient Health Questionnaire-9; GAD-7 = Generalized Anxiety Disorder Screener-7; PHQ-15 = Patient Health Questionnaire-15.

Association Between Setting and Maladaptive Personality Traits

Measurement invariance testing showed changes in CFI and RMSEA remaining within the accepted threshold (<|0.01|) from configural to metric invariance (△CFI = −0.005; △RMSEA = −0.000) and from metric to scalar invariance (△CFI = −0.008; △RMSEA = 0.001), supporting group comparisons.

Regression analysis indicated that the PC setting was incrementally associated with significantly lower factor scores of all maladaptive personality traits (excluding anankastia) compared to the general population. Concerning the sociodemographic variables, significantly lower levels of maladaptive personality traits were mainly found for female gender (except for negative affectivity), increasing age, being or having been in a relationship/married, as well as for participants with a higher level of education and vocational/professional training. However, the associations with education and training appeared smaller and less consistent across traits. The unadjusted coefficients of determination (R²) ranged from 0.05 to 0.13 across the five personality domains, indicating that the sample setting and socio-demographic variables explained a limited share of total variance in maladaptive personality traits (Table 3). For details on variance inflation factor (VIF) values, see the link to the statistical output in the data availability statement. Across all regression models, VIF values were <2, falling within the recommended threshold (VIF < 5; James et al., 2013).

Table 3.

Regression Analysis of the Association Between Setting (PC, General Population) and PID5BF+/M Domain Scale Scores, Controlled for Gender, Age, Marital Status, Highest Level of Education, and Vocational/Professional Training.

Sociodemographic factors	Standardized regression estimates
	Negative affectivity	Detachment	Antagonism	Disinhibition	Psychoticism
R 2	0.081	0.081	0.132	0.050	0.096
Setting
General population
PC	−0.178***	−0.140***	−0.190***	−0.128***	−0.257***
Gender
Male
Female	0.170***	−0.126***	−0.236***	−0.015	−0.081***
Age (increasing age)	−0.095***	0.035	−0.221***	−0.154***	−0.073**
Marital status
Single
Relationship/married	−0.096***	−0.249***	−0.089***	−0.103***	−0.138***
Annulled	−0.049*	−0.108***	−0.033	−0.024	−0.035
Widowed	−0.030	−0.082***	−0.006	−0.004	−0.029
Highest level of education
No school-leaving certificate	0.011	0.027	0.005	0.018	0.025
Lower secondary school diploma	0.069**	0.024	−0.028	0.040	0.055**
Secondary school diploma	0.057**	0.019	−0.013	0.042	0.024
High school diploma
Vocational/ professional training
No vocational training	0.052**	0.010	0.017	0.041*	0.037*
Vocational training	0.048*	0.032	0.027	0.052*	0.068**
Mastership/technical college/university degree

Note.*p < .05. **p < .01. ***p < .001.

R ² = coefficient of determination, unadjusted; italic = reference category; PC = primary care.

Regression coefficients indicate the estimated difference in maladaptive personality trait scores for each category compared to its respective reference category, controlling for all other variables in the model. Positive values reflect higher trait scores relative to the reference category, while negative values reflect lower scores.

As recent data on vocational/professional training of the general population were only available for the second and third cohort of the GESIS wave “Ge” (data on vocational/professional training from 2016 and 2018, respectively), another regression analysis was performed, excluding the participants of the first cohort (data on vocational/professional training from 2013). Results on the association between the PC setting and maladaptive personality traits remained unchanged (Appendix 4).

Discussion

Assessing personality pathology and maladaptive traits as transdiagnostic indicators of mental illness (Fowler et al., 2022; Kerber et al., 2022; Kotov et al., 2010; Rek et al., 2021) may assist the diagnostic procedure of GPs (Neulinger et al., 2024; Porcerelli et al., 2019). This study aimed to examine the factorial validity, reliability, and convergent validity of the PID5BF + M in German PC patients. In addition, the association between the PC setting and maladaptive personality traits was investigated in comparison to the general population.

Model fit of the PID5BF + M

The overall acceptable model fit aligns with the initial validation study of the PID5BF + M by Bach et al. (2020) and subsequent validation studies in Czech (Riegel et al., 2021) and Russian (Zinchuk et al., 2023) samples. These studies used exploratory structural equation modeling (ESEM) instead of CFA. ESEM has been suggested to better accommodate the complexity of personality due to its less restrictive modeling assumptions (Hopwood & Donnellan, 2010). However, the three studies examined a less complex model of the PID5BF + M, estimating model fit only at the facet level; that is, loading three related facets onto their respective personality domain, without accounting for individual items. Considering model complexity similar to the current study, Rek et al. (2021) and Kerber et al. (2022) both specified a higher-order model to examine the PID5BF+ model fit. Results indicated an overall acceptable model fit (CFI: 0.925/0.941; RMSEA: 0.060/0.055; SRMR: 0.057/0.059, respectively). Notably, both studies reported scaled rather than robust model fit indices. To ensure comparability, the present study took the same approach. Thus, findings should be viewed in light of potential overestimation of model fit (Savalei, 2018, 2021), as robust indices would have indicated a poorer model fit in the PC sample (robust CFI: 0.860; robust TLI: 0.843; robust RMSEA: 0.074). In general, inadequate CFA results for established personality questionnaires are common and may stem from the complexity of operationalizing personality (Hopwood & Donnellan, 2010). Model complexity rather than model misspecification may have also contributed to the negative variance obtained for the deceitfulness facet, which was also reported by Kerber et al. (2022). A smaller sample size and fewer indicators per latent variable have been associated with an increased likelihood of a Heywood case (Cooperman & Waller, 2022), particularly in hierarchical CFA (Marsh, 1987). Thus, using a larger combined sample (i.e., PC and general population) likely improved model estimation, allowing for a complex higher-order structure with only two items per facet while avoiding a Heywood case and achieving a comparable model fit.

Reliability of the PID5BF + M

McDonald’s ω_H values for the PID5BF + M domain scales were higher than those reported by Rek et al. (2021; ω_H = 0.66–0.71). It should be noted that a fixed cut-off indicating acceptable reliability for McDonald’s ω_H is still under debate (Garcia-Garzon et al., 2021; Reise et al., 2013). Internal consistency estimates ≥ 0.70 were suggested to be adequate for research settings (Nunnally, 1978). Using categorical McDonald’s Omega, both the Czech (Riegel et al., 2021) and Russian (Zinchuk et al., 2023) studies obtained acceptable internal consistency (ω = 0.65–0.75; ω = 0.83–0.90, respectively) for all domain scales, except for detachment (ω = 0.59) in the Czech-speaking sample. The high McDonald’s ω_H coefficient found for anankastia in the current study further supports the superiority of defining anankastia with the three facets of perfectionism, rigidity, and orderliness, rather than the two facets of rigid perfectionism and perseveration, as is done in the PID5BF+ (Bach et al., 2020; Kerber et al., 2022; Riegel et al., 2021).

Association Between Maladaptive Personality Traits and the PC Setting

Personality disorders are likely to be more prevalent in individuals seeking healthcare services (Tyrer et al., 2015). Therefore, the significantly lower manifestations of maladaptive personality traits in the PC sample relative to the general population sample were unexpected. Other studies assessing PID-5 personality traits have similarly reported higher values in community samples (Bach et al., 2020; Rek et al., 2021) and among PC patients (Porcerelli et al., 2019). One explanation may be socioeconomic status, demonstrated to be negatively associated with personality pathology (Luo et al., 2024; Rek et al., 2021) and mental health (Fryers et al., 2005; Kivimäki et al., 2020; Lampert et al., 2018; Pinto-Meza et al., 2013). As the PC sample was mainly collected in the urban and suburban regions of Munich – one of the most affluent cities in Germany (Fink et al., 2019; City of Munich, 2022, p. 61) – a higher overall socioeconomic status of the PC sample relative to the general population sample can be assumed. In addition, given the role of PC as the first point of contact for individuals with mental health concerns (Alexander et al., 2010), the setting likely selects for individuals from the general population experiencing heightened mental distress. As a result, the variable “setting” may capture some of the variance in mental distress between PC patients and the general population. However, this study was unable to explicitly account for this factor, as no corresponding measure of mental distress was available for the general population sample. Thus, future research should aim to include measures of both socioeconomic status and mental distress to examine additional explanatory factors. The coefficients of determination are also noteworthy: only a limited proportion of the total variance in maladaptive personality traits (5–13%) was accounted for by the sociodemographic variables and sample setting. Within the context of simple linear regression, however, these R² values correspond to correlation estimates between 0.22 and 0.36, which are generally regarded as moderate to large effect sizes (Funder & Ozer, 2019).

The observed associations between maladaptive personality traits and the sociodemographic variables age, gender, marital status, and level of education have already been discussed by Rek et al. (2021), as the majority of data for the regression analyses were based on the same German population sample (i.e., same data of GESIS).

Convergent Validity of the PID5BF + M

The correlation estimates found between personality pathology and maladaptive personality traits with depressive, anxiety, and somatic symptom severity support the convergent validity of the PID5BF + M. Fowler et al. (2022) reported a similar correlation pattern between the PID-5 domain scale scores and the PHQ-9, GAD-7, and PHQ-15 scores. Severity of personality pathology and negative affectivity in particular, as well as detachment, were most commonly associated with psychopathology, linking both to the presence of mental health conditions and their change over time (Böhnke et al., 2014; Fowler et al., 2022; Hong & Tan, 2021; Kerber et al., 2022; Rek et al., 2021; Wendt et al., 2024; Wright et al., 2015; Zinchuk et al., 2023). Furthermore, psychoticism was shown to be a differentiating factor between mental health conditions of varying severity (Kerber et al., 2022). For disinhibition, a longitudinal link with changes in psychological functioning was indicated (Wright et al., 2015). Conversely, in a mixed German sample of students and inpatients, no incremental association was found between disinhibition and the presence of depressive, anxiety, or somatic symptoms (Zimmermann et al., 2014). Concerning anankastia, a link with obsessive-compulsive personality disorder was suggested (Bach et al., 2020), symptoms of which co-occur with depression, anxiety, and somatic symptom disorder (Bienvenu et al., 2012; Sharma et al., 2021; Torres et al., 2016). In addition, perfectionism, as a facet of anankastia, was identified as an underlying factor of internalizing mental disorders (Bienvenu et al., 2012; Sharma et al., 2021; Torres et al., 2016). The comparably lower correlations of antagonism with the PHQ-9, GAD-7, and PHQ-15 scores may reflect the stronger attribution of this domain to externalizing than internalizing psychopathology (Lynam & Miller, 2019).

Implications for Assessment and Treatment

The findings support the notion that maladaptive personality traits function as higher-order transdiagnostic factors for lower-order mental health conditions (Böhnke et al., 2014; Fowler et al., 2022; Hong & Tan, 2021; Kerber et al., 2022; Kotov et al., 2010; Rek et al., 2021; Stanton et al., 2016; Subica et al., 2016; Wendt et al., 2024; Wright et al., 2015). In agreement with literature (Böhnke et al., 2014; Kerber et al., 2022; Neulinger et al., 2024; Porcerelli et al., 2019; Rek et al., 2021; Widiger & Oltmanns, 2017), especially overall personality pathology and negative affectivity – while not disregarding other personality traits (Kotov et al., 2010) – may assist GPs in screening patients for mental health conditions, particularly depressive, anxiety, and somatoform symptom severity. Given the time constraints of PC consultations (Irving et al., 2017), administering the full 36-item PID5BF + M for routine screening may not be feasible. A pragmatic alternative could be a two-step screening approach (Neulinger et al., 2024; Rogers et al., 2021) in which the negative affectivity subscale is first used to flag potential psychological distress, followed by the full PID5BF + M for a more comprehensive assessment if indicated. To inform subsequent treatment, further research is needed to identify bridging factors that mediate the relationship between maladaptive personality traits and mental health symptoms (Rodriguez-Seijas et al., 2015; Sauer-Zavala et al., 2022). One promising example is the Unified Protocol for Transdiagnostic Treatment of Emotional Disorders (Barlow et al., 2017), which targets transdiagnostic factors such as mindfulness, cognitive flexibility as well as emotion-based avoidance – underlying mechanisms of neuroticism – and has been shown to effectively reduce both mental health symptoms (Carlucci et al., 2021) and neuroticism itself (Sauer-Zavala et al., 2021). Ultimately, a shift toward dimensional classification models that reflect the hierarchical structure of psychopathology (see e.g., HiTOP; Kotov et al., 2017) would allow for identification of broad vulnerability factors at higher levels, followed by targeted assessment of specific lower-order mechanisms, facilitating both diagnostic precision and treatment planning in clinical practice.

Strengths and Limitations

This study is the first to provide a comprehensive evaluation of the psychometric properties of the PID5BF + M in the PC setting, assessing its confirmatory factorial validity, reliability, and convergent validity. Other strengths of the study are the large sample size, the use of the PID5BF + M as a standalone measure instead of extracting data from longer PID-5 versions (Bach et al., 2020), and complementing analyses with a representative sample of the German population.

The generalization of results to the entire PC setting is limited by the restriction of data collection to the area of Munich, Germany. Another limitation is that measurement invariance of the PID5BF + M between the PC and general population samples was assessed only at the facet level, warranting reevaluation at the item level. However, previous research has already supported measurement invariance of the PID-5 across clinical and non-clinical samples (Bach et al., 2018). Also noteworthy is the exclusion of items on anankastia from the measurement invariance and regression analyses due to the different operationalization of this domain in the PID5BF+ and PID5BF + M. Although supported by research (Pires et al., 2023; Zimmermann et al., 2020), it should further be considered that the total score of the PID5BF + M was used as a surrogate measure of the severity of personality pathology. Lastly, the PID5BF + M is a very brief version of the original Personality Inventory for DSM-5. Thus, any presented results should be understood as provisional rather than confirmatory.

Conclusion

The PID5BF + M demonstrated adequate factorial validity and reliability in the PC sample. Correlations between maladaptive personality traits and pathology with depressive, anxiety, and somatic symptom severity supported its convergent validity. GPs may consider the PID5BF + M as a valuable tool to screen for maladaptive personality traits and pathology in their patients, providing transdiagnostic insights into lower-order symptoms of mental illness. Future research should reexamine the psychometric properties of the PID5BF + M higher-order model by applying robust model fit indices and less stringent model specifications (e.g., allowing for correlated residuals) to better capture the complexity of personality assessment. In addition, the association between PC and the manifestation of maladaptive personality traits and pathology warrants further investigation. Finally, future studies could explore the mechanisms linking maladaptive personality traits to mental health symptoms to refine screening approaches and inform targeted treatment strategies in clinical practice.

Appendix

Appendix 1.

Comparison of Characteristics Between Included and Excluded Primary Care Patients.

	Group
	Included primary care patients (n = 1,030)	Excluded primary care patients (n = 280)	Mann–Whitney U
Variable	M (SD)	M (SD)	U	P
Age	46.53 (16.61)	48.89 (18.96)	98,558.0	.124
PHQ-9 score	5.57 (4.78)	5.71 (5.23)	142,953.5	.824
GAD-7 score	4.46 (4.27)	4.34 (4.52)	138,504.5	.307
PHQ-15 score	6.99 (4.93)	7.26 (6.01)	141,630.5	.646
	%	%
Gender
Female	60	61
Male	40	39
Nationality
German	90	81
Other	10	19
Marital status
Annulled	8	10
Relationship/married	67	60
Single	21	26
Widowed	5	5
Vocational/professional training
No vocational training	6	11
Vocational training	58	62
Mastership/technical college/university degree	36	27
Level of education
No school-leaving certificate	1	3
Lower secondary school diploma	24	42
Secondary school diploma	28	23
High school diploma	47	32
Occupation
Employed	70	62
Unemployed	30	38

Note. M = mean, SD = standard deviation, U = Mann–Whitney-U, p = p-value, n = sample size, PHQ-9 = Patient Health Questionnaire-9, GAD-7 = Generalized Anxiety Disorder-7, PHQ-15 = Patient Health Questionnaire-15.

Appendix 2.

Regression model figure.

Note. For each maladaptive personality trait, one regression model was specified, including setting and sociodemographic factors as independent variables.

Appendix 3.

CFA Results for the Combined Primary Care/General Population Sample With a 5-Factor PID5BF+/M Higher-Order Model.

Items	Factor loadings
Negative affectivity	5-factor higher-order model
PID5BF+/M: item 01	0.71	Emotional Lability: 0.66
PID5BF+/M: item 19	0.94
PID5BF+/M: item 07	0.84	Anxiousness: 0.76
PID5BF+/M: item 25	0.96
PID5BF+/M: item 13	0.71	Separation Insecurity: 0.93
PID5BF + M: item 31/PID5BF+: item 30	0.75
Detachment
PID5BF+/M: item 04	0.67	Withdrawal: 0.81
PID5BF+/M: item 22	0.82
PID5BF+/M: item 10	0.77	Anhedonia: 0.89
PID5BF+/M: item 28	0.73
PID5BF+/M: item 16	0.62	Intimacy Avoidance: 0.90
PID5BF + M: item 34/PID5BF+: item 33	0.77
Antagonism
PID5BF+/M: item 02	0.68	Manipulativeness: 0.72
PID5BF+/M: item 20	0.96
PID5BF+/M: item 08	0.64	Deceitfulness: 0.95
PID5BF+/M: item 26	0.80
PID5BF+/M: item 14	0.68	Grandiosity: 0.97
vPID5BF + M: item 32/PID5BF+: item 31	0.71
Disinhibition
PID5BF+/M: item 03	0.67	Irresponsibility: 0.79
PID5BF+/M: item 21	0.63
PID5BF+/M: item 09	0.68	Impulsivity: 0.91
PID5BF+/M: item 27	0.75
PID5BF+/M: item 15	0.83	Distractibility: 0.79
PID5BF + M: item 33/PID5BF+: item 32	0.78
Psychoticism
PID5BF+/M: item 05	0.48	Unusual beliefs & experiences: 0.92
PID5BF+/M: item 23	0.75
PID5BF+/M: item 11	0.79	Eccentricity: 0.91
PID5BF+/M: item 29	0.78
PID5BF+/M: item 17	0.84	Perceptual Dysregulation: 0.84
PID5BF + M: item 35/PID5BF+: item 34	0.84
Fit indices
χ2	7,031.073
df; p-value	380; p < .001
CFI	0.931
TLI	0.921
RMSEA	0.058
90% confidence interval	[0.057, 0.059]
SRMR	0.054

Note. CFA = confirmatory factor analysis; χ² = chi-squared, df = degrees of freedom, CFI = comparative fit index, TLI = Tucker–Lewis index, RMSEA = root mean square error of approximation, SRMR = standardized root mean square residual.

The column “5-factor higher-order model” is divided into a left-hand column with the standardized factor loadings for the items and a right-hand column with the standardized factor loadings for the higher-order facets, each corresponding to two items (e.g., facet Emotional Lability: items 01 and 19).

Appendix 4.

Regression Analysis of the Association Between Setting (Primary Care, General Population) and PID5BF + M Domain Scale Scores, Controlled for Gender, Age, Marital Status, Highest Level of Education, and Vocational/Professional Training (only second and third cohorts of GESIS wave “Ge” included).

	Standardized regression estimates
	Negative affectivity	Detachment	Antagonism	Disinhibition	Psychoticism
R 2	0.076	0.096	0.132	0.063	0.111
Setting
General population
Primary care	−0.194***	−0.159***	−0.205***	−0.150***	−0.308***
Gender
Male
Female	0.175***	−0.110***	−0.211***	0.004	−0.052*
Age (increasing age)	−0.105**	0.022	−0.228***	−0.155***	−0.082**
Marital status
Single
Relationship/married	−0.090**	−0.264***	−0.113***	−0.105**	−0.144***
Annulled	−0.022	−0.078**	−0.004	−0.002	−0.034
Widowed	−0.023	−0.071**	0.024	0.027	−0.023
Highest level of education
No school-leaving certificate	0.006	−0.008	−0.003	−0.006	0.009
Lower secondary school diploma	0.055	−0.028	−0.065*	−0.024	0.013
Secondary school diploma	0.046	0.006	−0.029	0.036	0.025
High school diploma
Vocational/professional training
No vocational training	0.014	0.021	−0.004	0.059*	0.046*
Vocational training	0.030	0.068*	0.027	0.071*	0.069**
Mastership/technical college/university degree

Note.*p < .05. **p < .01; ***p < .001.

R ² = coefficient of determination, unadjusted; italic = reference category.

Regression coefficients indicate the estimated difference in maladaptive personality trait scores for each category compared to its respective reference category, controlling for all other variables in the model. Positive values reflect higher trait scores relative to the reference category, while negative values reflect lower scores.

Appendix 5.

Variance–covariance Matrices of the PID5BF+/M Facets and Heterogeneous Correlation Matrices of the Sociodemographic Variables for Both the Primary Care and the Combined Primary Care/General Population Datasets.1. Primary care dataset.1.1. Variance-covariance matrix of PID5BF + M facets.

Variable	Emotional lability	Anxiousness	Separation insecurity	Withdrawal	Anhedonia	Intimacy avoidance	Manipulativeness	Deceitfulness	Grandiosity
Emotional lability	2.939	1.629	0.708	0.660	0.619	0.434	0.288	0.539	0.440
Anxiousness	1.629	2.777	1.007	0.796	0.835	0.565	0.248	0.712	0.444
Separation insecurity	0.708	1.007	1.462	0.366	0.565	0.371	0.288	0.482	0.423
Withdrawal	0.660	0.796	0.366	1.736	0.896	0.706	0.286	0.408	0.277
Anhedonia	0.619	0.835	0.565	0.896	1.869	0.728	0.335	0.547	0.307
Intimacy avoidance	0.434	0.565	0.371	0.706	0.728	1.638	0.312	0.455	0.326
Manipulativeness	0.288	0.248	0.288	0.286	0.335	0.312	1.316	0.741	0.365
Deceitfulness	0.539	0.712	0.482	0.408	0.547	0.455	0.741	1.500	0.595
Grandiosity	0.440	0.444	0.423	0.277	0.307	0.326	0.365	0.595	1.043
Irresponsibility	0.419	0.401	0.298	0.412	0.457	0.332	0.464	0.513	0.316
Impulsivity	1.209	0.984	0.592	0.551	0.640	0.601	0.524	0.836	0.520
Distractibility	0.896	0.972	0.597	0.678	0.844	0.650	0.402	0.690	0.391
Unusual beliefs	1.017	0.895	0.506	0.704	0.480	0.631	0.479	0.650	0.533
Eccentricity	0.761	0.774	0.501	0.594	0.623	0.565	0.504	0.657	0.533
Perceptual dysregulation	0.369	0.445	0.328	0.308	0.333	0.381	0.291	0.453	0.337
Perfectionism	1.248	1.105	0.582	0.584	0.401	0.458	0.337	0.581	0.470
Rigidity	0.976	1.015	0.501	0.560	0.426	0.454	0.237	0.536	0.488
Orderliness	0.682	0.833	0.499	0.393	0.379	0.337	0.251	0.438	0.458

1.1. Variance–covariance matrix of PID5BF + M facets (continued).

Variables	Irresponsibility	Impulsivity	Distractibility	Unusual beliefs	Eccentricity	Perceptual dysregulation	Perfectionism	Rigidity	Orderliness
Emotional lability	0.419	1.209	0.896	1.017	0.761	0.369	1.248	0.976	0.682
Anxiousness	0.401	0.984	0.972	0.895	0.774	0.445	1.105	1.015	0.833
Separation insecurity	0.298	0.592	0.597	0.506	0.501	0.328	0.582	0.501	0.499
Withdrawal	0.412	0.551	0.678	0.704	0.594	0.308	0.584	0.560	0.393
Anhedonia	0.457	0.640	0.844	0.480	0.623	0.333	0.401	0.426	0.379
Intimacy avoidance	0.332	0.601	0.650	0.631	0.565	0.381	0.458	0.454	0.337
Manipulativeness	0.464	0.524	0.402	0.479	0.504	0.291	0.337	0.237	0.251
Deceitfulness	0.513	0.836	0.690	0.650	0.657	0.453	0.581	0.536	0.438
Grandiosity	0.316	0.520	0.391	0.533	0.533	0.337	0.470	0.488	0.458
Irresponsibility	1.190	0.542	0.607	0.491	0.398	0.279	0.175	0.127	0.091
Impulsivity	0.542	1.883	0.939	0.935	0.905	0.494	0.912	0.798	0.640
Distractibility	0.607	0.939	2.242	0.736	0.739	0.496	0.556	0.583	0.430
Unusual beliefs	0.491	0.935	0.736	2.053	0.949	0.547	0.952	0.757	0.582
Eccentricity	0.398	0.905	0.739	0.949	1.680	0.492	0.744	0.712	0.695
Perceptual dysregulation	0.279	0.494	0.496	0.547	0.492	0.919	0.489	0.406	0.420
Perfectionism	0.175	0.912	0.556	0.952	0.744	0.489	2.737	1.490	1.230
Rigidity	0.127	0.798	0.583	0.757	0.712	0.406	1.490	2.318	1.153
Orderliness	0.091	0.640	0.430	0.582	0.695	0.420	1.230	1.153	1.828

1.2. Heterogeneous correlation matrix of sociodemographic variables.

Variable	Age	Gender	Marital status	Highest level of education	Vocational/professional training
Age	1.000	−0.119	0.483	−0.214	0.040
Gender	−0.119	1.000	0.108	−0.076	−0.092
Marital status	0.483	0.108	1.000	−0.062	0.054
Highest level of education	−0.214	−0.076	−0.062	1.000	0.710
Vocational/professional training	0.040	−0.092	0.054	0.710	1.000

2. Combined primary care/general population dataset.

2.1. Variance–covariance matrix of PID5BF+/M facets.

Variable	Emotional lability	Anxiousness	Separation insecurity	Withdrawal	Anhedonia	Intimacy avoidance	Manipulativeness	Deceitfulness	Grandiosity
Emotional lability	2.104	1.029	0.496	0.294	0.272	0.163	0.118	0.315	0.320
Anxiousness	1.029	2.506	0.832	0.577	0.614	0.451	0.149	0.553	0.434
Separation insecurity	0.496	0.832	1.385	0.305	0.442	0.375	0.230	0.415	0.482
Withdrawal	0.294	0.577	0.305	1.538	0.681	0.590	0.284	0.423	0.326
Anhedonia	0.272	0.614	0.442	0.681	1.386	0.555	0.291	0.444	0.360
Intimacy avoidance	0.163	0.451	0.375	0.590	0.555	1.363	0.238	0.388	0.398
Manipulativeness	0.118	0.149	0.230	0.284	0.291	0.238	1.057	0.592	0.375
Deceitfulness	0.315	0.553	0.415	0.423	0.444	0.388	0.592	1.444	0.559
Grandiosity	0.320	0.434	0.482	0.326	0.360	0.398	0.375	0.559	1.135
Irresponsibility	0.235	0.217	0.243	0.284	0.297	0.258	0.338	0.402	0.296
Impulsivity	0.841	0.741	0.466	0.358	0.426	0.378	0.321	0.593	0.406
Distractibility	0.608	0.773	0.517	0.478	0.573	0.491	0.316	0.598	0.420
Unusual beliefs	0.565	0.669	0.397	0.488	0.327	0.441	0.339	0.497	0.497
Eccentricity	0.486	0.652	0.501	0.606	0.565	0.513	0.453	0.618	0.539
Perceptual dysregulation	0.369	0.545	0.479	0.327	0.379	0.458	0.285	0.474	0.486

2.1. Variance–covariance matrix of PID5BF+/M facets (continued).

Variable	Irresponsibility	Impulsivity	Distractibility	Unusual beliefs	Eccentricity	Perceptual dysregulation
Emotional lability	0.235	0.841	0.608	0.565	0.486	0.369
Anxiousness	0.217	0.741	0.773	0.669	0.652	0.545
Separation insecurity	0.243	0.466	0.517	0.397	0.501	0.479
Withdrawal	0.284	0.358	0.478	0.488	0.606	0.327
Anhedonia	0.297	0.426	0.573	0.327	0.565	0.379
Intimacy avoidance	0.258	0.378	0.491	0.441	0.513	0.458
Manipulativeness	0.338	0.321	0.316	0.339	0.453	0.285
Deceitfulness	0.402	0.593	0.598	0.497	0.618	0.474
Grandiosity	0.296	0.406	0.420	0.497	0.539	0.486
Irresponsibility	0.892	0.353	0.429	0.278	0.368	0.297
Impulsivity	0.353	1.433	0.703	0.579	0.670	0.500
Distractibility	0.429	0.703	1.723	0.473	0.687	0.549
Unusual beliefs	0.278	0.579	0.473	1.684	0.770	0.609
Eccentricity	0.368	0.670	0.687	0.770	1.591	0.610
Perceptual dysregulation	0.297	0.500	0.549	0.609	0.610	1.222

Note. Facets associated with the personality domain of anankastia were excluded.

2.2. Heterogeneous correlation matrix of sociodemographic variables.

Variable	Age	Gender	Marital status	Highest level of education	Vocational/professional training
Age	1.000	−0.095	0.510	−0.308	0.135
Gender	−0.095	1.000	0.151	−0.037	−0.178
Marital status	0.510	0.151	1.000	−0.229	0.102
Highest level of education	−0.308	−0.037	−0.229	1.000	0.543
Vocational/professional training	0.135	−0.178	0.102	0.543	1.000