Skip to main content
NCBI home page
Wiley Open Access Collection logoLink to Wiley Open Access Collection
. 2026 Aug 3;20(3):e70093. doi: 10.1002/pmh.70093

Comparison of Self‐Report and Structured Clinical Interview Scores for the Assessment of the Level of Personality Functioning Scale

Muirne C S Paap 1,✉, Sara Germans Selvik 2,3, Tore Buer Christensen 4, Tor Erik Nysæter 4, Ingeborg Eikenaes 5, Benjamin Hummelen 5
PMCID: PMC13430580  PMID: 42544898

ABSTRACT

This study addresses the need for an examination of the criterion validity of a highly popular self‐report instrument which is used to gain a quick first impression of personality dysfunction: the Level of Personality Functioning Scale‐Brief Form 2.0 (LPFS‐BF 2.0). To this end, the relationship between LPFS‐BF 2.0 scores and SCID‐5‐AMPD‐I scores rated by trained clinicians was investigated in a clinical Norwegian sample (N = 155). Moreover, we investigated the relationship between the Norwegian version of the LPFS‐BF 2.0 and other self‐report instruments (the WSAS, GAD‐7, and PHQ‐9). Correlations, a linear regression model, as well as sensitivity and specificity for cut‐off values proposed in the literature, were estimated. The results of the linear regression analysis with SCID‐5‐AMPD‐I scores as the dependent variable, showed a strong and significant relationship between LPFS‐BF 2.0 scores and SCID‐5‐AMPD‐I scores, also when controlling for other variables. The LPFS‐BF 2.0 showed moderate to high correlations with all other instruments. The sensitivity values for cut‐off scores suggested in the literature for the LPFS‐BF 2.0 were too low. In sum, we found support for the concurrent and criterion validity of the LPFS‐BF 2.0. Additionally, we tentatively suggest that the cut‐off values currently in use may need to be lowered.

Keywords: alternative model for personality disorders, AMPD, DSM‐5, LPFS, personality disorders, psychometrics, SCID‐5‐AMPD‐I, self‐report

1. Introduction

Personality dysfunction that is severe enough to meet the criteria of a personality disorder (PD) is associated with considerable suffering and/or lasting impairment of social adaptiveness. More specifically, it has been shown that patients with severe personality dysfunction experience a disease burden that is comparable to that of severe somatic illnesses (Soeteman et al. 2008). Patients diagnosed with PDs often experience psychosocial impairment, have a decreased capacity for work, inadequate skills for establishing lasting interpersonal relationships, and have substantially lower life expectancy (Chesney et al. 2014; Shadid et al. 2025; Tyrer et al. 2015; Tyrer et al. 2021). Since PDs are notoriously difficult to treat, correctly identifying severe personality dysfunction is of paramount importance: using appropriate treatment designed especially for PDs has been shown to help patients improve (Gunderson et al. 2000; Zanarini et al. 2008).

To increase the odds of accurately identifying clinical levels of personality dysfunction, self‐report screening instruments can be employed as a first step (see Zimmerman and Holst 2018). To maximize the benefits of using a screening instrument, it is critical that it correctly identifies people with dysfunction, while keeping the costs associated with unnecessary full diagnostic evaluations low. Therefore, it is crucial to evaluate the psychometric properties, and in particular the criterion validity, of such screening instruments across various populations. In this article, we will focus on the criterion validity of a highly popular self‐report instrument which is used to gain a quick first impression of personality dysfunction: the Personality Functioning‐Brief Form 2.0 (LPFS‐BF 2.0).

The LPFS‐BF 2.0 (Weekers et al. 2019) is a self‐report inventory that was specifically developed to tap into Criterion A of the Alternative Model for Personality Disorders (AMPD), which was introduced in section III of the DSM‐5 (American Psychiatric Association 2013). Contrary to the traditional categorical approach, where the focus is on classification into distinct PD categories based on features specific to those categories, the first step when applying the AMPD is on assessing PD severity along a continuum; this scale is known as the Level of Personality Functioning Scale (LPFS; criterion A). As a second step, the nature of the dysfunction is explored by scoring pathological personality traits (criterion B). Currently, the gold standard approach to assessing Criterion A is to administer a semi‐structured interview by a trained clinician, such as the SCID‐5‐AMPD Module I: LPFS (SCID‐5‐AMPD‐I; Bender et al. 2018) or the Semi‐Structured Interview for Personality Functioning DSM–5 (STiP 5.1; Hutsebaut et al. 2017).

After the AMPD was introduced, self‐report inventories specifically developed to assess Criterion A/the LPFS quickly became available, of which the LPFS‐BF 2.0 is perhaps the most well‐known. It can be used free of charge, is currently available in 25 languages, and was included in the International Consortium for Health Outcomes Measurement (ICHOM) set of recommended instruments for PDs a few years ago (Prevolnik Rupel et al. 2021). Importantly, many different types of psychometric properties of this instrument have been assessed across a wide range of populations. Several studies found support for the use of the total scores in both general and clinical populations (Paap et al. 2024; Weekers et al. 2023), and recently our research group also found support for measurement invariance across time (Paap et al. 2025). Some support for measurement invariance across countries (Natoli et al. 2022), and for sensitivity to change has also been reported (Weekers et al. 2019). With respect to reliability, research findings are mixed. Several studies that studied global reliability reported adequate values (in the range of 0.74–0.90; Paap et al. 2024; Weekers et al. 2019; Weekers et al. 2023); however, in a recent study our group showed that local reliability (estimated with item response theory models) only exceeded 0.8 for a small range of the latent trait in a large clinical Norwegian sample (Paap et al. 2025).

A crucial psychometric property of the LPFS‐BF 2.0 that has not yet received a lot of attention is the criterion validity, when comparing LPFS‐BF 2.0 scores against clinician‐rated LPFS scores. We would argue that out of all psychometric properties, this is perhaps the most important one yet, since it concerns a questionnaire that is used as a screening instrument, and clinician ratings are still considered the gold standard for Criterion A (and for good reason). If the criterion validity of the LPFS‐BF 2.0 were high, this would solidify its place in screening procedures for PDs. Yet, the number of studies that have examined the alignment of self‐report screening instruments for PDs designed for the dimensional approach with diagnostic interviews is still limited.

Using the Italian translations of the SCID‐5‐AMPD‐I and LPFS‐BF in a clinical sample, Somma et al. (2020) found that the LPFS‐BF correlated 0.53 with the clinician‐rated LPFS scores. Somewhat higher correlations were reported between the STIP‐5.1 and LPFS‐BF 2.0: Spearman's rho = 0.77 for the Estonian versions in a mixed sample (general population sample and clinical sample; Oitsalu et al. 2022), and r = 0.74 in a college sample in the United States (Ran et al. 2025).

In this study, we take a much‐needed step by examining the correlations between the SCID‐5‐AMPD‐I and the LPFS‐BF 2.0 in a clinical Norwegian sample, as well as sensitivity and specificity values for tentative cut‐off scores suggested in the literature. To evaluate whether the LPFS‐BF 2.0 captures unique variance in personality functioning beyond general distress, we also performed a linear regression analysis, including measures of depression (PHQ‐9) and anxiety (GAD‐7) as control variables. Given that Tyrer et al. (2021) found strong evidence supporting the hypothesis that social functioning is a key manifestation of personality dysfunction, we treated the Work and Social Adjustment Scale (WSAS) as part of the nomological network of Criterion A and examined its correlations accordingly.

2. Methods

2.1. Participants and Procedure

In the current study, we used data from Wave 2 of the Norwegian multicenter study of the AMPD (NorAMP). Data for Wave 1 of the NorAMP study were collected between March 2015 and March 2017, and data collection for Wave 2 took place between January 2023 and August 2025 (Buer Christensen et al. 2020; Buer Christensen et al. 2018). Patients eligible for participation in the NorAMP study were recruited via their therapists. We used the data from the second wave only, since the LPFS‐BF 2.0 was not included in the first wave. For Wave 2, most interviews were administered online, using video conferencing platforms approved by the local Data Protection Officer. A detailed description of the sample obtained for the first wave can be found in the relevant study by Buer Christensen et al. (2018).

All patients (N = 282) who took part in the Wave 1 study were contacted again 8 years later, except for eight participants who had passed away during the follow‐up period and seven participants who at baseline had indicated that they did not wish to be contacted again. Participants were primarily contacted by telephone by a research collaborator. If a current phone number was unavailable or out of service, they received an invitation to participate by mail, instead. Out of the 275 persons who were approached, 155 agreed to participate.

We checked to what degree the 155 patients who participated in Wave 2 differed from the total group that participated in Wave 1. Patients who did not participate did not differ significantly from those who did participate 8 years ago in terms of gender, level of LPFS as measured using the SCID‐AMPD‐I in Wave 1, number of PD criteria, number of borderline PD criteria, number of avoidant PD criteria, number of symptom disorders, level of self‐reported psychosocial impairment, number of episodes with self‐injury or number of suicide attempts. However, patients who participated in Wave 1 but not in Wave 2 were significantly older than patients who participated in wave 2 (34.7 vs. 30.6 years in Wave 1) and had a significantly higher number of antisocial PD criteria. Patients with ASPD were also significantly older than those without ASPD (35.7 vs. 32.0 years), which may partly explain the age difference between participants and non‐participants in Wave 2. In addition, patients with ASPD were more frequently lost to follow‐up due to death: four out of 29 patients with ASPD were no longer alive, compared to five out of 239 patients without ASPD.

The NorAMP study was set up in such a way that the whole severity spectrum of personality pathology would be covered, which was facilitated by including patients from a wide range of recruitment sites, including general mental health inpatient and outpatient departments, group psychotherapy outpatient and day treatment units, and a substance abuse outpatient unit that also served a prison (Buer Christensen et al. 2020; Buer Christensen et al. 2018). The group psychotherapy units were all part of the Norwegian Network for Personality Disorders (Pedersen et al. 2023); this is a large collaborative network of clinical units that specialize in PD assessment and PD treatment. The exclusion criteria were: schizophrenia spectrum disorder (except schizotypal PD), sequelae after brain injury, pervasive developmental disorders (i.e., autism spectrum disorders), intellectual disability, severe ongoing substance abuse, and lack of understanding of Norwegian.

2.2. Ethics

Both the original and follow‐up NorAMP studies were approved by the Regional Committee for Medical and Health Research Ethics in Norway. All Wave 1 participants had provided written consent to be recontacted for one or more follow‐up interviews. Before participating in Wave 2, participants provided new written informed consent.

2.3. Measurement Instruments

In the current study, we report on scores from four self‐report instruments (LPFS‐BF 2.0, WSAS, GAD‐7, and PHQ‐9), as well as LPFS scores rated by a clinician who used the SCID‐AMPD. The instruments which were used will be described below.

2.3.1. The Level of Personality Functioning Scale Brief Form 2.0 (LPFS‐BF 2.0)

The LPFS‐BF 2.0 is a 12‐item instrument in which each item corresponds to one of the 12 indicators of the LPFS as outlined in the AMPD. Items 1–6 were developed to measure self‐functioning (Identity and Self‐direction) and items 7–12 relate to Interpersonal functioning (Empathy and Intimacy). The LPFS‐BF was translated into Norwegian by a group of four clinical researchers, two of whom are fluent in both Dutch and Norwegian, following the guidelines outlined by Hambleton et al. (2004) see also Paap et al. (2024).

The LPFS‐BF 2.0 is scored using a 4‐point ordered response format. Note that the first version of the LPFS‐BF used a 0–3 scoring format, which was adopted in the Norwegian Network. However, studies on the second version typically use a 1–4 scoring format. In the current study, we chose to use the 1–4 scoring format, in order to facilitate comparison with results reported in other studies. The answering categories (ordered from lowest to highest score) are as follows: “Very False or Often False,” “Sometimes or Somewhat False,” “Sometimes or Somewhat True,” and “Very True or Often True.” In this study, we will report mean scores, since this is most common. Some studies refer to sum scores, however. The mean scores reported here can be easily transformed into sum scores by multiplying them by 12. The global reliability of the LPFS‐BF 2.0, as estimated with Guttman's Lambda‐2 (Guttman 1945), was 0.88 in the current sample.

Three cut‐off values have been suggested by Weekers et al. (2023): one for moderate, one for severe, and one for very severe impairment. These were based on a sample of the Danish general population, taking the classification rules of average score in the sample + 1.5 SD, average score + 2 SDs, and average score + 2.5 SDs. In the paper by Weekers et al., the cut‐off values are reported on the total score scale. Here, we will convert them to the mean score scale: 2.58 (moderate), 3.00 (severe), and 3.38 (very severe).

2.3.2. The Work and Social Adjustment Scale (WSAS)

The WSAS is a self‐report instrument that was developed to measure impairment in the domains of work and social functioning, and is thought to be useful in treatment outcomes studies (Mundt et al. 2002). The WSAS consists of five items using a 9‐point response format (0–8), from “Not at all” to “Very Severely.” The scores on the five items are added up to obtain a total score in the range of 0–40. Scale reliability of the WSAS items, estimated using Guttman's Lambda‐2, was 0.90 in the current sample.

2.3.3. The Generalized Anxiety Disorder 7‐Item (GAD‐7)

The GAD‐7 is a self‐report instrument developed to screen for the DSM‐IV‐based diagnosis Generalized Anxiety Disorder (GAD; Spitzer et al. 2006). The GAD‐7 consists of 7 items that are scored using a 4‐point response format (0–3) from “Not at all” to “Nearly every day.” It inquires about symptoms in the past 2 weeks. GAD‐7 scores were computed as the sum score of all seven items, ranging from 0 to 21. Scale reliability of the GAD‐7 items, as estimated with Guttman's Lambda‐2, was 0.91 in the current sample.

2.3.4. The Patient Health Questionnaire‐9 (PHQ‐9)

The PHQ‐9 is based on the depression module of the PRIME‐MD instrument used to diagnose common mental disorders, which was designed to score the 9 DSM‐IV depression criteria (Kroenke et al. 2001). The PHQ‐9 is a self‐report questionnaire consisting of 9 items that are scored using a 4‐point response format (0–3) from “Not at all” to “Nearly every day.” To obtain the total scores, item scores are summed, resulting in a total score that ranges from 0 to 27. Scale reliability of the PHQ‐9, as estimated with Guttman's Lambda‐2, was 0.87 in the current sample.

2.3.5. The SCID‐5‐AMPD Module I: LPFS (SCID‐5‐AMPD‐I)

The SCID‐5‐AMPD‐I aims to measure the severity of personality pathology across the twelve subdomains and four domains of the LPFS (Identity, Self‐Direction, Empathy, and Intimacy; Bender et al. 2018; First et al. 2018). Each subdomain is scored on a scale ranging from 0 (little to no impairment) to 4 (extreme impairment). These scores are then averaged yielding four domain scores and a total LPFS score. Based on earlier studies by our research group, an average LPFS score of 1.5 was used as a cut‐off value for the presence of a PD diagnosis (Buer Christensen et al. 2020). This implies that an average score of 1.5 or higher signifies personality dysfunction that warrants clinical attention.

To get a basic sense of the interviewee's level of personality functioning, the SCID‐5‐AMPD‐I starts with eight preliminary questions about view of self and quality of interpersonal relationships. Subsequently, each subdomain is assessed by using a combination of screening questions and level of determination questions. Thus, the interview follows a two‐step administration procedure in which information gathered during the first step (the preliminary questions) can be used in the second step to decide which screening questions to administer for each subdomain, which levels to assess, and which of the level‐specific questions to use. In other words, the interview is conducted in an adaptive manner, which allows for a more tailored assessment. This also means not all questions need to be administered. Studies by our research group have shown that the instrument is well liked by clinicians, and that they use the funnel structure effectively to tailor their assessment based on relevant patient characteristics (Heltne et al. 2022, 2023).

The interviews were conducted by experienced psychiatrists and clinical psychologists. Prior to data collection, all interviewers participated in a two‐day workshop designed to standardize procedures and enhance reliability. The training emphasized consensus‐building techniques, ensuring that clinicians applied uniform criteria and minimized subjective variability during assessments. The same workshop‐based training approach was implemented in both Wave 1 and Wave 2 to maintain consistency across data collection phases and strengthen inter‐rater agreement.

2.4. Statistical Analysis Plan

Correlations were estimated to explore the relationship among the included variables, with a specific focus on the LPFS‐BF 2.0 and the SCID‐5‐AMPD‐I. Subsequently, multiple linear regression models were estimated, with clinician‐rated LPFS scores as a dependent variable. The following independent variables were added in three blocks: 1) age and gender; 2) the total scores on the GAD‐7, and PHQ‐9; and 3) self‐reported LPFS scores. The goal of estimating this model was to explore the relationship between self‐reported and clinician‐rated LPFS scores, while controlling for the effect of age, gender, anxiety, and depression. Assumptions underlying linear regression modeling were checked for the final model, and additional analyses were carried out in case assumptions were not met. Results were considered statistically significant when p‐value ≤ 0.05. Furthermore, sensitivity and specificity values were calculated for cut‐off scores suggested in the literature for both instruments. In case sensitivity and specificity values were inadequate, additional ROC analyses were performed to identify more appropriate cut‐off values. The statistical analyses were conducted in the freely available software R (R Core Team 2021).

3. Results

3.1. Descriptive Statistics

The median age in the sample was 28 years (IQR = 24–36), and 69% (n = 108) were female. The mean LPFS‐BF 2.0 total scores were on the low side for a clinical sample: 1.92 (SD = 0.60). The mean SCID‐5‐AMPD‐I score equaled 1.46 (SD = 0.92). For patients who met the 1.5 threshold on the SCID‐5‐AMPD‐I, the mean LPFS‐BF 2.0 score was 2.23 (SD = 0.56), and for patients who did not meet that threshold, the mean LPFS‐BF 2.0 was 1.64 (SD = 0.47). The mean scores for the WSAS, GAD‐7 and PHQ‐9 equaled 3.25 (SD = 1.96), 1.35 (SD = 0.81) and 1.33 (SD = 0.71), respectively. Figure 1 shows the distribution of LPFS‐BF 2.0 scores for participants who were rated below and above the SCID‐5‐AMPD‐I threshold of 1.5.

FIGURE 1.

FIGURE 1

Open in a new tab

Distributions of LPFS‐BF 2.0 scores for persons scoring positive (solid line) and negative (dashed line) on the SCID‐5‐AMPD‐I using the 1.5 threshold.

3.2. Correlations

All continuous variables showed moderate‐strong correlations with each other (Table 1), including the self‐reported LPFS and clinician‐rated LPFS scores (r = 0.576). The highest correlations were found for the PHQ‐9 and the WSAS, and for the PHQ‐9 and the GAD.

TABLE 1.

Correlation matrix for the continuous variables to be included in the regression model.

Variable 1. 2. 3. 4. 5.
1. LPFS clinician‐rated 1
2. WSAS 0.502 1
3. GAD‐7 0.403 0.632 1
4. PHQ‐9 0.440 0.768 0.784 1
5. LPFS self‐report 0.576 0.608 0.629 0.725 1
Open in a new tab

3.3. Model Building With Linear Regression Analysis

Two of the three steps resulted in a significant change in R 2: from steps 1 to 2 and from steps 2 to 3 (Table 2), with the largest change from steps 1 to 2. This implies that even when controlling for the other variables, the self‐reported LPFS score still contributed to a large degree to predicting clinician‐rated LPFS scores.

TABLE 2.

Results of hierarchical multiple linear regression at each of the four steps with clinician‐rated LPFS scores as the dependent variable.

Step Predictors B SE B* 95% CI B t p Adj. R 2
LL UL
1. Intercept 1.653 0.302 — 1.056 2.250 5.471 < 0.001 0.007
Age −0.011 0.009 −0.099 −0.028 0.007 −1.226 0.222
Gender 0.200 0.164 0.099 −0.124 0.523 1.220 0.224
2.** Intercept 0.652 0.323 — 0.015 1.290 2.022 0.045 0.186
Age −0.001 0.008 −0.009 −0.017 0.015 −0.114 0.910
Gender 0.072 0.150 0.036 −0.224 0.368 0.482 0.631
GAD‐7 0.175 0.137 0.152 −0.095 0.445 1.279 0.203
PHQ‐9 0.416 0.155 0.318 0.110 0.723 2.686 0.008
3.** Intercept 0.535 0.297 — −0.052 1.122 1.801 0.074 0.313
Age 0.001 0.008 0.010 −0.014 0.016 0.142 0.887
Gender 0.017 0.138 0.008 −0.256 0.289 0.121 0.904
GAD‐7 0.082 0.127 0.071 −0.168 0.333 0.648 0.518
PHQ‐9 0.005 0.162 0.004 −0.315 0.325 0.030 0.976
LPFS self‐report 0.830 0.156 0.530 0.522 1.138 5.326 < 0.001
Open in a new tab

Note: **The F change statistic was statistically significant.

Abbreviations: B = unstandardized beta; B* = standardized beta; CI = confidence interval; LL = lower limit; UL = upper limit.

When looking at Table 2 more closely, a few things stand out. The PHQ‐9 was a significant predictor for clinician‐rated LPFS in step 2 but not in step 3. In step 3, the only significant effect is found for self‐reported LPFS scores. The beta coefficient is quite large: 0.830 with a confidence interval ranging from 0.522 to 1.138. This indicates that an increase in LPFS‐BF 2.0 scores of 1 point is expected to lead to a substantial and clinically relevant increase of 0.83 points in clinician‐rated LPFS scores.

3.4. Assumption Checks

Assumptions were checked for the complete model (the model estimated in step 3). A visual check of a Q‐Q plot of the residuals for the estimated model supported the assumption of normally distributed residuals. The assumption of homoscedasticity was checked both visually and with the Breusch‐Pagan test by using the ncvTest function for the car package in R. These checks did not reveal a breach of the assumption.

Different cut‐off values have been suggested for flagging variables for collinearity using VIF values. Widely referenced standard texts mention cut‐off values of 4, 5, or 10 (James et al. 2013; Menard 2002; Miles and Shevlin 2001; Vittinghoff et al. 2005). In our case, the highest VIF value found equaled 3.4 for the PHQ‐9 in the step 3 model. Since this value does not exceed any of the proposed cut‐offs, we suspect collinearity was not a problem for our models.

3.5. Sensitivity and Specificity for Various Cut‐Off Values

Using the cut‐off score for moderate impairment suggested by Weekers et al. (2.58 on the scale of mean scores where answer options are scored from 1–4), only 12.8% met that threshold, and 4.5% met the suggested threshold (3.00) for severe impairment. In contrast, based on the LPFS scores rated by clinicians, 72.4% of the sample met the 1.5 threshold (and were thus labeled as positive cases). Using 2.58 as a cut‐off value resulted in a specificity of 0.94, but a sensitivity as low as 0.20. For a cut‐off value of 3.00, the sensitivity as estimated in our sample dropped to almost 0.

As it has been suggested in the literature that sensitivity deserves to be emphasized (over specificity) for a screening instrument (see Zimmerman and Holst 2018), we decided to run a ROC analysis to explore which cut‐off value on the LPFS‐BF 2.0 would result in high (ideally ≥ 0.90) sensitivity values. In this analysis, the LPFS‐BF 2.0 was used as a predictor, and the SCID‐5‐AMPD‐I dichotomization based on the 1.5 threshold as the outcome. The results showed that achieving the desired sensitivity of 0.90 as suggested by Zimmerman and Holst (2018) was not feasible, since it resulted in a specificity of only 0.46. Lowering the sensitivity to 0.80 seemed more reasonable; this resulted in a specificity of 0.65. The associated cut‐off value equals 1.79. This cut‐off score yielded 58 true positives (37.9%), 16 false negatives (10.4%), 27 false positives (17.6%), and 52 true negatives (34.0%).

4. Discussion

In this study, we examined aspects of criterion validity of the LPFS‐BF 2.0, with a specific focus on concurrent validity in relation to LPFS scores rated by clinicians who used the SCID‐5‐AMPD‐I. We found that the relationship between self‐reported LPFS scores and clinician‐rated LPFS was moderately strong, and this effect remained rather stable when controlling for scores on other relevant self‐report instruments (GAD‐7 and PHQ‐9). This shows convincing support for the criterion validity of the LPFS‐BF 2.0. Finally, we also looked at sensitivity and specificity levels, by using cut‐off values for both instruments that have previously been suggested in the literature. For the cut‐off values suggested by Weekers et al. (2023), the sensitivity of the LPFS‐BF 2.0 was unacceptably low in our sample. An exploratory follow‐up analysis indicated that it might be worth considering lowering the cut‐off value used for the LPFS‐BF 2.0 to a value close to 1.79. In our sample, using this value resulted in a sensitivity of 0.80 and a specificity of 0.65.

The correlation we found (r = 0.57) between LPFS‐BF 2.0 scores and clinician‐rated SCID‐5‐AMPD‐I scores was highly similar to the correlation reported by Somma et al. (2020) for the Italian versions of the instruments (r = 0.53). What our regression analyses indicate is that the LPFS‐BF 2.0 is a strong predictor of SCID‐5‐AMPD‐I scores, even when controlling for other variables. This is an interesting finding, as it shows that the LPFS‐BF 2.0 captures unique variance in personality dysfunction beyond general distress. We also found high correlations between the LPFS‐BF 2.0 and the other included self‐report measures that were used, including the WSAS. This strong relationship is in line with findings by Tyrer et al. (2021), who tested whether social functioning could serve as a proxy measure for PD severity and found solid evidence supporting this relationship, strongly indicating that social dysfunction is intrinsic to the construct of personality pathology.

Studies focusing on other important psychometric properties of the LPFS‐BF 2.0 have shown support for content validity (Waugh et al. 2020), construct validity (at least, of the total score; Paap et al. 2024; Weekers et al. 2023), as well as measurement invariance across time (Paap et al. 2025); all of which are crucial psychometric properties. Taking those findings together with the findings of the current study, one could argue that the evidence in favor of the psychometric quality of LPFS‐BF 2.0 may be considered good enough to warrant using it as a screening instrument for personality dysfunction. However, since results of various studies have been mixed (support for several scales versus one; adequate global reliability but low local reliability values for certain samples), we do argue that more research is needed. Important next steps are: 1) studying the sensitivity and specificity of the LPFS‐BF 2.0 in various populations for multiple cut‐off values; 2) using a think‐aloud procedure with patients and clinicians to clarify their thought process when completing the LPFS‐BF 2.0 and the SCID‐5‐AMPD‐I, and using qualitative data analysis to investigate whether these two instruments tap into a common construct; and 3) exploring whether other self‐report instruments that were designed to measure a construct related to personality functioning, such as the IPO, SIPP‐118, PAI‐BOR, DSM‐5 Level of Personality Functioning Questionnaire (DLOPFQ; Huprich et al. 2018), show stronger criterion validity as compared to the LPFS‐BF 2.0.

We need to acknowledge a few limitations of the current study. One of the main limitations is the restricted representativeness of the sample. Although all participants were recruited from clinical units in Wave 1 (baseline), conducting a screening assessment 8 years later is not common practice in an ordinary clinical setting, and therefore has limited clinical applicability and usefulness with respect to the identification of personality pathology in routine care. The representativeness of the sample is further limited by the inclusion of many patients recruited from specialized treatment units for PDs, resulting in a relatively high prevalence of patients with PD in Wave 1. Moreover, the LPFS interviews were conducted without any prior knowledge about the participants, which may have reduced the validity of the findings. In routine clinical practice, structured diagnostic interviews for PDs are typically administered after an initial intake session and/or assessment of symptom disorders, and are informed by a broad range of clinical information, such as referral letters, medical records, and previous assessments. At this point, it is therefore difficult to determine the generalizability of our findings. This underlines the importance of carrying out a similar study with different samples (e.g., using clinical and general population samples obtained in other countries, or samples from clinical populations slightly different from the one under study here). Furthermore, it is common knowledge that sensitivity and specificity estimates are directly tied to the prevalence of diagnostic cases in the population under study (e.g., Murad et al. 2023; Vijfschagt et al. 2025), which is another important reason to examine the properties of the suggested cut‐off value in other samples. That being said, this is one of the first studies explicitly comparing self‐reported LPFS scores to LPFS scores rated by a trained clinician using the SCID‐5‐AMPD‐I. One of the core strengths of this study is that it used a very clinically relevant sample of adequate size, which is hard to obtain in practice, especially in regard to data obtained through clinical interviews. Therefore, we are still confident that our findings provide a highly relevant contribution to the existing literature.

It has been pointed out in numerous studies that self‐reported scores cannot be viewed as interchangeable with scores rated by a clinician using a semi‐structured interview, including studies specifically focusing on PDs (e.g., Müller et al. 2025). Both types of scores are important in their own right, and have their own strengths and weaknesses. Reported strengths of self‐reported scores include that 1) no‐one else observes the respondents in as many different situations as the respondents themselves, and they have access to a great quantity and breadth of information that they can draw on when filling out a questionnaire; and 2) it is highly practical to use self‐report instruments (Paulhus and Vazire 2007). Risks or weaknesses, especially in a psychiatric setting, include that 1) they respondent may not interpret the question as intended, and 2) the respondent may lack the self‐awareness or self‐insight to answer the question adequately (Kessler et al. 2000). It is partly because of the latter two reasons that semi‐structured interviews remain the gold standard in psychiatry. A risk that is often reported when it comes to clinicians rating a patient, is that their bias may influence their judgment (e.g., Kendell 1968; Lewis and Williams 1989). We argue that this is one of the reasons that proper training is a necessity (and not just a recommendation) before a clinician starts using a certain semi‐structured questionnaire. Interviews conducted with clinicians by our research team support this notion (Heltne et al. 2022). Other strategies to diminish the effect of bias include triage with colleagues, combining multiple sources of information to obtain an integrated impression of the patient, and asking the patient for input and feedback during the diagnostic process. Ultimately, self‐report and clinician ratings are two highly relevant pieces of a puzzle: complimentary rather than interchangeable. When using a self‐report instrument as a screener, however, it is still important to establish high levels of both concurrent and discriminant validity. Which psychometric properties are most important in a given situation, always depends on for which goal the instruments are used (e.g., Sireci and Sukin 2013).

One of the choices that impact the suitability of an instrument for screening purposes is the cut‐off score used. Ran et al. (2025) found support for a cut‐off score very close to the moderate cut‐off score suggested by Weekers et al. (2023) in a mixed sample of college students and persons with self‐identified Borderline Personality Disorder. Our findings, however, indicate that the cut‐off scores recently suggested for the LPFS‐BF 2.0 based on a Danish general population sample by Weekers et al. (2023) may need adjustment, as they resulted in very low sensitivity estimates, which is undesirable for a screening instrument. In light of these findings, we deem it advisable to slightly lower the cut‐off value. However, as the number of studies currently available is insufficient to inform a definite recommendation, we argue that more research is needed across various populations before a specific value can be decided upon. In our study, the cut‐off value of 1.79 was associated with moderate sensitivity and acceptable specificity.

4.1. Conclusion

Our study found convincing support for the criterion validity of the LPFS‐BF 2.0 scores when compared to clinician‐rated LPFS scores in a Norwegian clinical sample. Based on our findings, we tentatively suggest considering lowering the recently suggested cut‐off values that were based on the aforementioned Danish general population sample.

Funding

The authors have nothing to report.

Ethics Statement

Anonymized data were collected from each treatment unit and subsequently transferred to a common research database. The Data Protection Officer (DPO) at each contributing unit approved the data collection procedures, while the DPO at Oslo University Hospital, Oslo, Norway approved the security procedure for the research database. The Regional Committee for Medical and Health Research Ethics has approved the project.

Consent

All participating patients from all treatment units provided written consent for anonymous use of their clinical data for research purposes.

Conflicts of Interest

The authors declare no conflicts of interest.

Acknowledgements

We would like to thank the patients who participated in this study for their time, being so open, and being willing to contribute to research. We would also like to express our gratitude to the therapists who conducted the interviews, and to the research assistants and secretaries who provided invaluable practical administrative support.

No AI/LLMs were used in developing any portion of this manuscript.

Data Availability Statement

Due to restrictions imposed by the Regional Medical Ethics Committee regarding patient confidentiality, data are only available upon reasonable request. Requests for data may be sent to the Privacy and Data Protection Officer at Oslo University Hospital, Oslo, Norway: personvern@oslo-universitetssykehus.no.

References

  1. American Psychiatric Association . 2013. Diagnostic and Statistical Manual of Mental Disorders. 5th ed. American Psychiatric Publishing. 10.1176/appi.books.9780890425596. [DOI] [Google Scholar]
  2. Bender, D. S. , Skodol A. E., First M. B., and Oldham J. M.. 2018. “Module I: Structured Clinical Interview for the Level of Personality Functioning Scale.” In Structured Clinical Interview for the DSM–5 Alternative Model for Personality Disorders (SCID–AMPD), edited by First M. B., Skodol A. E., Bender D. S., and Oldham J. M.. American Psychiatric Association Publishing. [Google Scholar]
  3. Buer Christensen, T. , Hummelen B., Paap M. C. S., et al. 2020. “Evaluation of Diagnostic Thresholds for Criterion A in the Alternative DSM‐5 Model for Personality Disorders.” Journal of Personality Disorders 34, no. Supplement C: 40–61. 10.1521/pedi_2019_33_455. [DOI] [PubMed] [Google Scholar]
  4. Buer Christensen, T. , Paap M. C. S., Arnesen M., et al. 2018. “Interrater Reliability of the Structured Clinical Interview for the DSM‐5 Alternative Model of Personality Disorders Module i: Level of Personality Functioning Scale.” Journal of Personality Assessment 100, no. 6: 630–641. 10.1080/00223891.2018.1483377. [DOI] [PubMed] [Google Scholar]
  5. Chesney, E. , Goodwin G. M., and Fazel S.. 2014. “Risks of All‐Cause and Suicide Mortality in Mental Disorders: A meta‐Review.” World Psychiatry 13, no. 2: 153–160. 10.1002/wps.20128. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. First, M. B. , Skodol A. E., Bender D. S., and Oldham J. M.. 2018. User's Guide for the SCID‐5‐AMPD: Structured Clinical Interview for the DSM‐5 Alternative Model for Personality Disorders. American Psychiatric Association Publishing. [Google Scholar]
  7. Gunderson, J. G. , Shea M. T., Skodol A. E., et al. 2000. “The Collaborative Longitudinal Personality Disorders Study: Development, Aims, Design, and Sample Characteristics.” Journal of Personality Disorders 14, no. 4: 300–315. [DOI] [PubMed] [Google Scholar]
  8. Guttman, L. 1945. “A Basis for Analyzing Test‐Retest Reliability [Journal Article].” Psychometrika 10, no. 4: 255–282. 10.1007/bf02288892. [DOI] [PubMed] [Google Scholar]
  9. Hambleton, R. K. , Merenda P. F., and Spielberger C. D.. 2004. “Issues, Designs, and Technical Guidelines for Adapting Tests Into Multiple Languages and Cultures.” In Adapting Educational and Psychological Tests for Cross‐Cultural Assessment, edited by Hambleton R. K., Merenda P. F., and Spielberger C. D., 15–50. Psychology Press. [Google Scholar]
  10. Heltne, A. , Bode C., Hummelen B., Falkum E., Selvik S. G., and Paap M. C. S.. 2022. “Norwegian Clinicians' Experiences of Learnability and Usability of SCID‐II, SCID‐5‐PD and SCID‐5‐AMPD‐I Interviews: A Sequential Multi‐Group Qualitative Approach.” Journal of Personality Assessment 104, no. 5: 599–612. 10.1080/00223891.2021.1975726. [DOI] [PubMed] [Google Scholar]
  11. Heltne, A. , Braeken J., Hummelen B., Germans Selvik S., Buer Christensen T., and Paap M. C. S.. 2023. “Do Flexible Administration Procedures Promote Individualized Clinical Assessments? An Explorative Analysis of How Clinicians Utilize the Funnel Structure of the SCID‐5‐AMPD Module I: LPFS.” Journal of Personality Assessment 105, no. 5: 636–646. 10.1080/00223891.2022.2152344. [DOI] [PubMed] [Google Scholar]
  12. Huprich, S. K. , Nelson S. M., Meehan K. B., et al. 2018. “Introduction of the DSM‐5 Levels of Personality Functioning Questionnaire.” Personality Disorders, Theory, Research, and Treatment 9, no. 6: 553–563. 10.1037/per0000264. [DOI] [PubMed] [Google Scholar]
  13. Hutsebaut, J. , Kamphuis J. H., Feenstra D. J., Weekers L. C., and De Saeger H.. 2017. “Assessing DSM–5‐Oriented Level of Personality Functioning: Development and Psychometric Evaluation of the Semi‐Structured Interview for Personality Functioning DSM–5 (STiP‐5.1).” Personality Disorders, Theory, Research, and Treatment 8, no. 1: 94–101. 10.1037/per0000197. [DOI] [PubMed] [Google Scholar]
  14. James, G. , Witten D., Hastie T., and Tibshirani R.. 2013. An Introduction to Statistical Learning: With Applications in R. 1st ed. Springer. [Google Scholar]
  15. Kendell, R. E. 1968. “An Important Source of Bias Affecting Ratings Made by Psychiatrists.” Journal of Psychiatric Research 6, no. 2: 135–141. 10.1016/0022-3956(68)90035-6. [DOI] [PubMed] [Google Scholar]
  16. Kessler, R. C. , Wittchen H.‐U., Abelson J., and Zhao S.. 2000. “Methodological Issues in Assessing Psychiatric Disorders With Self‐Reports.” In The Science of Self‐Report: Implications for Research and Practice, 229–255. Lawrence Erlbaum Associates Publishers. [Google Scholar]
  17. Kroenke, K. , Spitzer R. L., and Williams J. B.. 2001. “The PHQ‐9: Validity of a Brief Depression Severity Measure.” Journal of General Internal Medicine 16, no. 9: 606–613. 10.1046/j.1525-1497.2001.016009606.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Lewis, G. , and Williams P.. 1989. “Clinical Judgement and the Standardized Interview in Psychiatry.” Psychological Medicine 19, no. 4: 971–979. 10.1017/S0033291700005699. [DOI] [PubMed] [Google Scholar]
  19. Menard, S. 2002. Applied Logistic Regression Analysis. SAGE Publications, Inc. 10.4135/9781412983433. [DOI] [Google Scholar]
  20. Miles, J. , and Shevlin M.. 2001. Applying Regression & Correlation: A Guide for Students and Researchers. SAGE Publications Inc. [Google Scholar]
  21. Müller, S. , Schroeders U., Bachrach N., et al. 2025. “Revisiting the Structure of Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition, Section II Personality Disorder Criteria Using Individual Participant Data Meta‐Analysis.” Personality Disorders, Theory, Research, and Treatment No Pagination Specified‐No Pagination Specified 17, no. 1: 1–16. 10.1037/per0000736. [DOI] [PubMed] [Google Scholar]
  22. Mundt, J. C. , Marks I. M., Shear M. K., and Greist J. M.. 2002. “The Work and Social Adjustment Scale: A Simple Measure of Impairment in Functioning.” British Journal of Psychiatry 180, no. 5: 461–464. 10.1192/bjp.180.5.461. [DOI] [PubMed] [Google Scholar]
  23. Murad, M. H. , Lin L., Chu H., et al. 2023. “The Association of Sensitivity and Specificity With Disease Prevalence: Analysis of 6909 Studies of Diagnostic Test Accuracy.” CMAJ 195, no. 27: E925–e931. 10.1503/cmaj.221802. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Natoli, A. P. , Bach B., Behn A., et al. 2022. “Multinational Evaluation of the Measurement Invariance of the Level of Personality Functioning Scale‐Brief Form 2.0: Comparison of Student and Community Samples Across Seven Countries.” Psychological Assessment 34, no. 12: 1112–1125. 10.1037/pas0001176. [DOI] [PubMed] [Google Scholar]
  25. Oitsalu, M.‐L. , Kreegipuu M., and Hutsebaut J.. 2022. “Psychometric Evaluation of the Estonian Version of the Semi‐Structured Interview for Personality Functioning DSM‐5 (STiP‐5.1).” Borderline Personality Disorder and Emotion Dysregulation 9, no. 1: 28. 10.1186/s40479-022-00197-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Paap, M. C. S. , Lortz S. A. J., Hummelen B., Fjermestad‐Noll J., Kvarstein E. H., and Pedersen G.. 2025. “An Investigation of Longitudinal Measurement Invariance and Item Response Theory Analysis of the LPFS‐BF 2.0 in a Large Clinical Norwegian Sample.” Personality and Mental Health 19, no. 3: e70032. 10.1002/pmh.70032. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Paap, M. C. S. , Pedersen G., Kvarstein E., and Hummelen B.. 2024. “Evaluating the Construct Validity of the Norwegian Version of the Level of Personality Functioning Scale ‐ Brief Form 2.0 in a Large Clinical Sample.” Journal of Personality Assessment 106, no. 1: 49–59. 10.1080/00223891.2023.2182694. [DOI] [PubMed] [Google Scholar]
  28. Paulhus, D. L. , and Vazire S.. 2007. “The Self‐Report Method.” In Handbook of Research Methods in Personality Psychology, 224–239. Guilford Press. [Google Scholar]
  29. Pedersen, G. , Wilberg T., Hummelen B., and Hartveit Kvarstein E.. 2023. “The Norwegian Network for Personality Disorders – Development, Contributions and Challenges Through 30 Years.” Nordic Journal of Psychiatry 77, no. 5: 512–520. 10.1080/08039488.2022.2147995. [DOI] [PubMed] [Google Scholar]
  30. Prevolnik Rupel, V. , Jagger B., Fialho L. S., et al. 2021. “Standard Set of Patient‐Reported Outcomes for Personality Disorder.” Quality of Life Research 30, no. 12: 3485–3500. 10.1007/s11136-021-02870-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. R Core Team . 2021. “R: A Language and Environment for Statistical Computing R Foundation for Statistical Computing.” https://www.R‐project.org/.
  32. Ran, T. , Balzen K. M., Kerr S., Hutsebaut J., and Sharp C.. 2025. “Level of Personality Functioning Scale‐Brief Form 2.0: Factor Structure and Clinical Cutoffs.” Journal of Personality Disorders 39, no. 3: 185–205. 10.1521/pedi.2025.39.3.185. [DOI] [PubMed] [Google Scholar]
  33. Shadid, J. , Ferrari A. J., Bach B., et al. 2025. “The Global Epidemiology of Personality Disorder: A Systematic Review and meta‐Regression.” Lancet Psychiatry 12, no. 12: 932–946. 10.1016/S2215-0366(25)00299-8. [DOI] [PubMed] [Google Scholar]
  34. Sireci, S. G. , and Sukin T.. 2013. “Test validity.” In APA Handbook of Testing and Assessment in Psychology, Vol. 1: Test Theory and Testing and Assessment in Industrial and Organizational Psychology, 61–84. American Psychological Association. 10.1037/14047-004. [DOI] [Google Scholar]
  35. Soeteman, D. I. , Verheul R., and Busschbach J. J. V.. 2008. “The Burden of Disease in Personality Disorders: Diagnosis‐Specific Quality of Life.” Journal of Personality Disorders 22, no. 3: 259–268. 10.1521/pedi.2008.22.3.259. [DOI] [PubMed] [Google Scholar]
  36. Somma, A. , Borroni S., Gialdi G., et al. 2020. “The Inter‐Rater Reliability and Validity of the Italian Translation of the Structured Clinical Interview for DSM‐5 Alternative Model for Personality Disorders Module I and Module II: A Preliminary Report on Consecutively Admitted Psychotherapy Outpatients.” Journal of Personality Disorders 34, no. Suppl C: 95–123. 10.1521/pedi_2020_34_511. [DOI] [PubMed] [Google Scholar]
  37. Spitzer, R. L. , Kroenke K., Williams J. B. W., and Löwe B.. 2006. “A Brief Measure for Assessing Generalized Anxiety Disorder: The GAD‐7.” Archives of Internal Medicine 166, no. 10: 1092–1097. 10.1001/archinte.166.10.1092. [DOI] [PubMed] [Google Scholar]
  38. Tyrer, P. , Reed G. M., and Crawford M. J.. 2015. “Classification, Assessment, Prevalence, and Effect of Personality Disorder.” Lancet 385, no. 9969: 717–726. 10.1016/S0140-6736(14)61995-4. [DOI] [PubMed] [Google Scholar]
  39. Tyrer, P. , Yang M., Tyrer H., and Crawford M.. 2021. “Is Social Function a Good Proxy Measure of Personality Disorder?” Personality and Mental Health 15, no. 4: 261–272. 10.1002/pmh.1513. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Vijfschagt, N. D. , Burger H., Berger M. Y., et al. 2025. “Variation in Sensitivity and Specificity of Diverse Diagnostic Tests Across Health‐Care Settings: A Meta‐Epidemiological Study.” Journal of Clinical Epidemiology 184: 111816. 10.1016/j.jclinepi.2025.111816. [DOI] [PubMed] [Google Scholar]
  41. Vittinghoff, E. , Shiboski S. C., Glidden D. V., and McCulloch C. E.. 2005. Regression Methods in Biostatistics: Linear, Logistic, Survival, and Repeated Measures Models. Springer. [Google Scholar]
  42. Waugh, M. H. , McClain C. M., Mariotti E. C., et al. 2020. “Comparative Content Analysis of Self‐Report Scales for Level of Personality Functioning.” Journal of Personality Assessment 103: 161–173. [DOI] [PubMed] [Google Scholar]
  43. Weekers, L. C. , Hutsebaut J., and Kamphuis J. H.. 2019. “The Level of Personality Functioning Scale‐Brief Form 2.0: Update of a Brief Instrument for Assessing Level of Personality Functioning.” Personality and Mental Health 13, no. 1: 3–14. 10.1002/pmh.1434. [DOI] [PubMed] [Google Scholar]
  44. Weekers, L. C. , Sellbom M., Hutsebaut J., Simonsen S., and Bach B.. 2023. “Normative Data for the LPFS‐BF 2.0 Derived From the Danish General Population and Relationship With Psychosocial Impairment.” Personality and Mental Health 17, no. 2: 157–164. 10.1002/pmh.1570. [DOI] [PubMed] [Google Scholar]
  45. Zanarini, M. C. , Frankenburg F. R., Reich D. B., Fitzmaurice G., Weinberg I., and Gunderson J. G.. 2008. “The 10‐Year Course of Physically Self‐Destructive Acts Reported by Borderline Patients and axis II Comparison Subjects.” Acta Psychiatrica Scandinavica 117, no. 3: 177–184. 10.1111/j.1600-0447.2008.01155.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Zimmerman, M. , and Holst C. G.. 2018. “Screening for Psychiatric Disorders With Self‐Administered Questionnaires.” Psychiatry Research 270: 1068–1073. 10.1016/j.psychres.2018.05.022. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Availability Statement

Due to restrictions imposed by the Regional Medical Ethics Committee regarding patient confidentiality, data are only available upon reasonable request. Requests for data may be sent to the Privacy and Data Protection Officer at Oslo University Hospital, Oslo, Norway: personvern@oslo-universitetssykehus.no.

Articles from Personality and Mental Health are provided here courtesy of Wiley

RESOURCES