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

Abstract

The DSM-5 presents two competing diagnostic frameworks for personality disorders: the standard categorical model and the Alternative Model of Personality Disorders (AMPD). The AMPD was initially criticized for being too complex and theory laden for clinical implementation. Though inter-rater reliability studies have contested initial claims of the model’s complexity, little attention has been paid to how clinicians experience the usability and learnability of either model. We interviewed twenty Norwegian clinicians about their experiences with either the SCID-II/5-PD (n = 9), SCID-5-AMPD-I (n = 8), or both (n = 3). Separate thematic analyses were conducted for SCID-II/5-PD and SCID-5-AMPD-I groups, and group themes were compared. We identified four themes for each group, relating to required skills, training, challenges and information gained through the interview. We found that training and clinical experience were considered to be important for both interviews. Moreover, the SCID-5-AMPD-I was considered to rely more explicitly on theory specific to the development and content of the AMPD model in general and the LPFS specifically We also identified shared and unique challenges and shortcomings of each interview. We comment on how our findings relate to the debate surrounding the AMPD, and recommend development of clear training guidelines for both interviews.

The fifth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-5) presents two competing diagnostic frameworks for personality disorders (PDs): 1) the standard categorical model; and 2) the Alternative Model of Personality Disorders (American Psychiatric Association, 2013). The standard categorical model delineates specific PD diagnostic categories defined by the presence of a specified number of symptom criteria, whereas the AMPD is a hybrid model including both dimensional aspects and categorical diagnoses. The dimensional nature of personality pathology is captured by the A and B criteria of the AMPD. The A criterion is the Levels of Personality Functioning Scale (LPFS), a global scale of personality functioning ranging from 0 (little or no impairment) to 4 (extreme impairment). The B criterion is a collection of five pathological personality trait domains, organized into twenty-five trait facets rated on a scale of 0 to 3, with 0 indicating “very little or not at all descriptive,” and 3 indicating “very descriptive.” A PD diagnosis requires a moderate level of impairment of personality functioning (level 2), as well as the presence of at least one pathological personality trait.

The AMPD was designed to replace the standard categorical model and to solve commonly cited issues with the categorical model (for a review, see Clark, 2007) by providing a dimensional framework for classifying personality disorders. Due to a number of concerns, such as the model’s supposed complexity, lack of empirical support, and insufficient continuity with the standard categorical model, the AMPD was placed in Section III for further research and development (see Zachar et al., (2016) for a review of the AMPD development process).

Since the publication of the DSM-5, studies on inter-rater reliability among untrained raters have contested initial claims of the AMPD’s complexity (e.g., Few et al., 2013; Garcia et al., 2018; Zimmermann et al., 2014). Garcia et al. (2018) found that relatively untrained raters could provide ratings similar to more experienced expert raters and that agreement between these novices and experts increased with training. Morey (2018) further found that the concepts of the first module of the DSM-5, the Levels of Personality Functioning Scale (LPFS), could be reliably applied by untrained students who had not previously been introduced to the AMPD. Although these studies presented important findings about the learnability of the AMPD, few studies have directly compared the learnability and clinical utility of the AMPD to the standard categorical model. Furthermore, few studies considering the clinical utility of these models have been carried out in a naturalistic clinical setting.

Morey et al. (2014) had mental health professionals (both psychiatrists and non-psychiatrists) rate the clinical utility of both models. They found that, although clinicians generally considered criterion A and B of the AMPD as having either equal or greater clinical utility in several respects (such as informing treatment and communicating with patients), criterion A was deemed more difficult to use than the standard categorical model. This illustrates that experienced clinicians can still find it challenging to apply this criterion to their patients, as critics predicted (e.g., Gunderson, 2013; Pilkonis et al., 2011; Verheul, 2012).

Standardized clinical interviews have the potential to mitigate such challenges, as they have been shown to improve reliability and to greatly enhance the identification of personality disorders in clinical settings (Rogers, 2003). The American Psychiatric Association (APA) has published such interviews for both the standard categorical model and the AMPD. The Structured Clinical Interview for DSM-IV Axis II disorders (SCID-II) and its successor, the Structured Clinical interview for DSM-5 Personality Disorders (SCID-5-PD), were developed for the standard categorical model (First & Gibbon, 1997; First et al., 2015), whereas the Structured Clinical Interview for DSM-5 Alternative Model of Personality Disorders (SCID-5-AMPD) was developed for the AMPD (Bender et al., 2018). The SCID-5-AMPD includes three modules: the first module (SCID-5-AMPD-I) assesses the LFPS; the second module focuses on the 25 personality traits; and the third module can be used to make a specific PD diagnosis within the AMPD framework. The first module is pivotal in deciding whether a person has a PD diagnosis. It can therefore be expected that this module will be used most frequently in clinical practice (Buer Christensen et al., 2018). The SCID-5-AMPD-I is also the most innovative and complex. Exploration of the usability and learnability of this module is therefore warranted.

The SCID-II and SCID-5-PD are highly similar in both content and structure; hereafter, we therefore refer to them together as SCID-II/5-PD. The SCID-II/5-PD and the SCID-5-AMPD-I, however, differ substantially in both content and administration rules. The SCID-II/5-PD encompasses ninety items covering the ten PD diagnostic categories of the DSM-IV/5. In these interviews, questions are asked consecutively and are scored on a three-point scale indicating “absent,” “sub-threshold,” and “present.” The SCID-5-AMPD-I assesses the Levels of Personality Functioning Scale on twelve subdomains, three for each of the four over-arching domains of “identity,” “self-directedness,” “empathy,” and “intimacy.” Screening questions are used to estimate functioning under each subdomain. Then, specific questions are selected to explore increasing levels of the LPFS, until the patient no longer qualifies for the level being explored. Although domain scores can be calculated by averaging sub-domain scores, only the total score (derived from averaging these domain scores) has a strong theoretical basis. A recent study conducted by our research group found strong support for the essential unidimensionality of the SCID-5-AMPD-I (Hummelen et al., 2021).

Although being called structured clinical interviews, both the SCID-II/5-PD and the SCID-5-AMPD-I are in fact semi-structured, since administrators are encouraged to follow up on patients’ responses and ask for concrete examples to substantiate them. The clinician makes a clinical judgment about whether the answers given by the patient fulfill the scoring criteria. This requires clinicians to use their clinical knowledge and experience more actively during administration than they would when using other instruments, such as self-report and fully structured interviews. It can therefore be challenging to administer these interviews, if raters have not received adequate training.

The administration of the SCID-II/5-PD and SCID-5-AMPD-I is said to require both training and clinical experience (Columbia University, 2020). However, the matter about what kind of training and how much training is required is largely left unspecified.

We argue direct comparisons of these two kinds of interviews on the dimensions of usability and learnability is needed to further inform the current debate surrounding the standard categorical model and the AMPD, as well as for understanding what a transition to a more dimensional model will require from practicing clinicians.

In this study, we therefore conduct qualitative interviews with clinicians to assess their experiences of learning and using SCIDs developed for both the standard categorical model and the LPFS, in naturalistic clinical settings. Our aim is to inform the current debate surrounding the supposed complexity of the AMPD, and to provide insight into the training requirements and potential challenges associated with the SCID-II/5-PD and SCID-5-AMPD-I. The research questions are as follows:

1. How do clinicians experience the usability of each interview;

2. What challenges do clinicians experience when learning to use these interviews;

3. In what way do these two interviews pose similar and/or different challenges to clinicians who are learning to use them for the first time?

To answer these research questions, we apply what we have dubbed a “sequential multigroup approach.” This approach incorporates the use of qualitative comparison groups (Lindsay, 2019), and adds a sequential element often seen in sequential designs in mixed methods research (see Bishop, 2015 for an overview). The use of qualitative comparison groups is thought to facilitate both the representations of experiences from differing groups of participants (RQ 1 and 2), and a comparison of experiences between these groups (RQ3) (Lindsay, 2019). We further add a sequential element wherein preliminary data analysis is used to inform further data collection. In doing so, we seek to further inform between-group comparison and incorporate a form of member checking (see for instance Birt et al., 2016) whereby members of the intended target populations of our research are asked to comment on, correct, and elaborate on our preliminary results.

Method

A brief overview of sequential multigroup approach

Our sequential multigroup approach informed all stages of this study. First, two separate qualitative comparison groups were identified and recruited. These groups differed in terms of which instruments (SCID-II/5-PD or SCID-5-AMPD-I) participants were experienced with and were interviewed about. The same interview guide was used for both groups, and separate thematic analyses (Braun & Clarke, 2006) were carried out for each group, in order to identify themes relating to ease of use and learnability of each instrument. Preliminary results of these separate thematic analyses and an initial comparison of themes and thematic content between groups were used to develop a second interview guide, which was administered to a third group of participants. Participants in this third interview group had been introduced to both the SCID-II/5-PD and the SCID-5-AMPD-I at approximately the same time, at an early stage in their careers. These participants were asked to a) conduct their own comparison of the two instruments, and b) comment on observed differences and similarities between the themes and thematic content of the two qualitative comparison groups. Coded interviews from this third group of participants were explored for confirmations, elaborations and contradictions to our observed differences and similarities between the first two interview groups and to incorporate a form of member checking to reduce bias in analysis.

Participants

A total of twenty clinicians participated in this study. The twenty clinicians were recruited from both rural and central areas of Norway and worked in general outpatient settings, or in specialist clinical settings (where clinicians reported PD patients as the main patient group served by their unit). Their ages ranged from twenty-nine to fifty-eight. Participants were recruited to one of three groups, based on the type(s) of interview with which they had prior experience. The S-II/5 group consisted of participants who had experience with either the SCID-II, the SCID-5-PD, or both. The S-LPFS group consisted of experienced clinicians who had used the SCID-5-AMPD-I, after having already used the SCID-II/5-PD interview for years. Lastly, the mixed group consisted of clinicians who started using both the SCID-II/5-PD and the SCID-5-AMPD-I, either during their studies or directly after having obtained their university degree, and who therefore had a comparable level of experience with both interviews. Table 1 gives an overview of participant characteristics and group affiliation.

Table 1. Overview of participants.

Group	Interviewed about	Male/Female	Generalist/Specialist	Central/Rural	Mean age (SD)
S-II/5	SCID-II/5-PD	5/4	4/5	5/4	39.56 (8.21)
S-LPFS	SCID-5-AMPD-I	5/3	7/1	3/5	42.63 (7.66)
Mixed	SCID-II/5-PD and SCID-5-AMPD-I	0/3	1/2	3/0	32.67 (2.62)

The S-II/5 group participants represented seven outpatient clinics from three out of the four major health districts in Norway. Two of these participants were initially recruited to test the interview guide developed for the S-II/5 and the S-LPFS group (see materials). As no changes were made to the interview guide after pilot testing, the pilot interviews were included in the final dataset. Participants in this group were invited to participate in this study through administrators of outpatient clinics. Administrators were asked to forward invitation letters to clinical staff, or to identify prospective participants we could contact directly via email or telephone.

Participants in the S-LPFS and mixed group represented four outpatient clinics and were recruited from two of the four major health districts in Norway. As the SCID-5-AMPD-I was not widely adopted and used in Norway when this study was conducted, participants in these groups were recruited from lists of clinicians who had contributed to data collection for at least one of two inter-rater reliability studies of the SCID-5-AMPD-I, carried out by the Norwegian research group The Norwegian Multicenter Study of the AMPD. One of these studies has already been published (NorAMPD; Buer Christensen et al., 2018). The Norwegian translation of the SCID-5-AMPD-I used in these studies was developed by the NorAMPD research group and the translation process is partially outlined in Buer Christensen et al. (2018). Contributors to the NorAMPD interrater reliability studies were contacted via email and telephone and were informed about the study and asked to participate.

Decisions about when to terminate participant recruitment were informed by the model for evaluating information power introduced by Malterud et al. (2016). This approach argues that appropriate sample size can be determined based on the amount of study-relevant information found in the data produced by the sample. In line with this method, we decided when to terminate participant recruitment, based on the diversity of experiences represented in our samples (as indicated by diversity in age, geographical location, workplace-setting and years of work experience among participants), and the richness and quality of our data in relation to our research questions (as evaluated after initial analysis of three interviews from each sample, and continuously throughout the data collection process).

Materials

In keeping with our sequential multi-group approach, two interview guides were developed for this study: one for the S-II/5 and S-LPFS group and one for the mixed group. We first developed the interview guide for the S-II/5 and S-LPFS group. After completing all interviews with these groups, a preliminary analysis of three S-II/5 and three S-LPFS group interviews was carried out to identify similarities and differences between clinicians’ experience with the SCID-II/5-PD and the SCID-5-AMPD-I. The mixed-group interview guide was then developed in order to explore these differences and similarities further through mixed-group interviews and allow these participants to comment on our initial findings.

The interview guide for the S-II/5 and S-LPFS group covered four main topics: 1) general experiences with the use of the instrument (SCID-II/5-PD or SCID-5-AMPD-I); 2) experiences regarding the process of learning to use the instrument; 3) changes experienced over time as clinicians became more experienced in using the instrument; 4) advice to novices who are about to begin using the instrument in clinical practice. This interview guide was pilot-tested with two participants who fit the inclusion criteria for the S-II/5 group. Since no further adaptions were made after this pilot, the interviews of these participants were included in the analysis.

The mixed-group interview guide covered four main topics: 1) general experiences of using and learning to use each of the two instruments (SCID-II/5-PD and SCID-5-AMPD-I); 2) challenges encountered in learning to use each of the two instruments; 3) experienced differences and similarities between the two instruments; and 4) thoughts and comments on our preliminary results from the S-II/5 and S-LPFS group interviews. This interview guide was not pilot tested due to its overall similarity to the interview guide for the S-II/5 and S-LPFS group, and a lack of prospective participants who fit the inclusion criteria for the mixed group.

Ethics

This study did not need approval from the Regional Committee for Medical and Health Research Ethics, according to Norwegian regulations. All procedures were in accordance with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Data-handling procedures were reviewed and approved by the Data Protection Officer at Oslo University Hospital.

Procedure

All interviews for the S-II/5 group and three interviews for the S-LPFS group were carried out face-to-face between December 2019 and March 2020. Due to restrictions related to the Covid-19 pandemic, the remaining interviews were carried out either via Skype for business or via telephone from March to April 2020. Prior to the interviews, participants received a short description of the research project’s aim, and a list of the four topics covered in the interview. They were told that they could, but did not have to, prepare for the interview by reviewing these topics in advance.

After obtaining informed consent, an encrypted laptop was used to make an audio recording of the interviews. Recordings were stored on a secure research server at the Oslo University Hospital. Interviews were transcribed verbatim, either by the first author or by a trained research assistant, under supervision of the first author. Interview transcripts were also stored on the secure research server. The interviews lasted between twenty minutes and two hours. The participants received a 400 NOK gift card, usable at an online retailer, as compensation for their time and participation.

Analysis

Our analysis consisted of two separate thematic analyses – one for the S-II/5 group and one for the S-LPFS group. Carrying out separate thematic analyses for qualitative comparison groups is one of several suggested approaches outlined by Lindsay (2019). The two thematic analyses carried out in this study followed the six steps outlined by Braun and Clarke (2006), and are explained in more detail below. The purpose of these separate thematic analyses was, first, to answer research questions 1 and 2, and, second, to produce a set of themes associated with each instrument (SCID-II/5-PD and SCID-5-AMPD-I) which could serve as a basis for comparison to answer research question 3. Following our sequential multi-group approach, we also conducted a comparison of themes found for the S-II/5 and S-LPFS group. An initial comparison of themes and thematic content from the S-II/5 and S-LPFS group informed the development of an interview guide for the mixed group, and results of these interviews were used to further inform the between-groups comparison. This second stage of analysis followed two main steps: 1) themes from the S-II/5 and S-LPFS group were compared by the research team to formulate initial ideas of observed differences and similarities between the groups; 2) our observed differences and similarities were compared with the results of the mixed-group interviews.

For the development of the themes we applied the six steps by Braun and Clarke (2006). The first step consisted of familiarization with the data. The first author transcribed half of the interviews and read all interviews. Next, emergent descriptive codes were developed through open inductive coding. All initial coding was carried out by the first author. After coding three interviews from each interview group, a codebook containing codes and sub-codes, their definitions, and examples of coded content was evaluated and discussed in collaboration with the second and last author. This codebook was updated and expanded throughout coding of subsequent interviews. The completed codebook was then re-applied to all coded data, to ensure consistent coding. To check coder agreement, a research assistant was asked to label text segments within three interviews with an appropriate code, based on descriptions in the hierarchical codebook. She was provided with the hierarchical codebook outlining the definitions of all codes from all three interview groups and was asked to prepare for the task by reading this codebook thoroughly. She was then provided with three interviews (one from each interview group) with marked text segments representing all coded text segments in each interview and was asked to apply an appropriate code to each labeled text segment. Proportion of inter-rater agreement ($\frac{\mathit{Number}\mathrm{ }of\mathrm{ }\mathit{agreements}}{\mathit{Number}\mathrm{ }of\mathrm{ }\mathit{agreements}\mathrm{ }+\mathrm{ }\mathit{Disagreements}\mathrm{ }}$) was calculated for each of the three interviews and reached .69, .78 and .88 at the lowest sub-code level for the S-II/5, S-LPFS and mixed-group interviews, respectively. Disagreements were discussed until consensus was reached. Only minor revisions in the coding of the data (re-coding or splitting text excerpts between codes) were performed. The codebook remained unchanged.

In the next steps, codes were grouped according to thematic similarity, consistency of content within prospective themes, and external uniqueness between themes. Subthemes and themes were named based on the meanings of coded content of grouped codes. Lastly, themes were summarized into a cohesive narrative, and excerpts were selected to substantiate and illustrate their content. During all steps of the thematic analysis, the first, second and last author reviewed the process by considering potential overlap across themes, their relationship to each other and to the research questions.

Results

Table 2 summarizes the four main themes and subthemes developed for the S-II/5 and S-LPFS group and their relationship to research question 1 and 2. Table 3 summarizes results of the between-group comparison, and highlights observed differences and similarities between the S-II/5 and S-LPFS group under each thematic topic, alongside mixed-group elaborations and additions. In this section, we first describe the themes from the S-II/5 and S-LPFS group. We then present the results of the between-groups comparison. Note that some subthemes in Table 2 are italicized. These are subthemes we consider to be related to PD assessment in general and to the specific contexts of our participants, or otherwise to not directly be related to the SCID-II/-5PD and SCID-5-AMPD-I. These subthemes will only be briefly summarized in this section.

Table 2. Overview of themes, subthemes and thematic topics for the standard and alternative interview groups.

Research Questions	Thematic topics	S-II/5-PD group		S-LPFS group
		Themes	Subthemes	Themes	Subthemes
RQ 1 - How do clinicians experience the ease of use of each interview?	Skills and competencies	Applying both general and specific skillsets	Theoretical knowledge and experiential familiarity General therapeutic skills Developing technical skills and growing confidence ^a	The SCID-5-AMPD-I requires practical experience as well as theoretical knowledge
	Training received and/or required	Access to training is varied and novices often rely on their colleagues for guidance and support	Formal courses Collegial training and support Own reading and experiential training	Formal training is essential and should cover both theory and practice
	Outcome and insights gained	Clinicians do more than just score the SCID-II/5-PD	Diagnostic test Source of insight	The SCID-5-AMPD-I is more in line with clinicians’ interests and needs	General theoretical framework Focus on important topics which are of interest to clinicians
RQ2 - What challenges do clinicians experience when learning to use each interview?	Experienced challenges	Challenges arise from both within and outside the SCID-II/5-PD	Shortcomings of the interview Use of compensatory measures Navigating the complexities of personality disorders ^a	Challenges arise from the context, novelty of the SCID-5-AMPD-I and current shortcomings	Challenges in learning and using the interview Initial feelings of being overwhelmed Shortcomings of the interview Administering the interview in a research context ^a

^a = Italicized items represent subthemes related to the assessment of PD in general, the specific contexts of participants or are otherwise not directly tied to the SCID-II/5-PD and SCID-5-AMPD-I.

Table 3. Between group comparison of themes.

Thematic Topics	Observed Similarities	Observed Differences	Mixed-group Elaborations
Skills and competencies	Theoretical knowledge Experience with patient group	General PD theory v/s interview-specific theory General therapeutic skills as a separate theme in S-II/5-PD group interviews	Theory is more explicit in SCID-AMPD-I SCID-II/5-PD is more intuitive and familiar to novices
Experienced challenges	Complex language/poor translations Shortcomings leading to scoring difficulties	Over-inclusive vs double-barreled criteria. Layout, novelty and feeling overwhelmed v/s Difficulty getting and evaluating information Navigating the complexities of personality disorders as a separate sub-theme for S-II/5-PD group interviews.	Confirmed difference in how shortcomings affect scoring Double-barreled criteria in SCID-5-AMPD-I could be due to remnants of categorical thinking Novelty aspect of SCID-AMPD-I may lessen over time
Training received and/or required	Training is important	Receiving v/s not receiving formal training Collegial training and support as separate sub-theme in S-II/5-PD group interviews.	Confirmed differences in training received SCID-5-AMPD-I may require more theoretical clarification during training
Outcome and insights gained	Want insight into patients individual functioning This insight is achievable	Getting it directly v/s going beyond interview questions	SCID-AMPD-I may be more in line with what clinicians want and need Going beyond questions in SCID-II/5-PD grants insight

S-II/5 group themes

Theme S1: Applying both general and specific skillsets

Within this theme we identified three subthemes, relating to general and specific skills, as well as growing competence.

Theoretical knowledge and experiential familiarity

Participants emphasized the importance of having theoretical knowledge about personality disorders when using the SCID-II/5-PD. To one participant, this meant having: “fundamental knowledge about personality, development[al] [psychology] and personality pathology” (P8). Similarly, many also stressed the usefulness of experiential familiarity with the patient group: “One basically has to meet a lot of patients and learn to understand what personality pathology is, through meeting them and seeing what the challenges in their lives are, and well, ‘what does a personality disorder look like?’” (P9). Participants described relying on theory and prior experience when scoring answers of new patients: “I have the criteria, have all these cases in the back of my mind, when I meet new patients and assess or diagnose [them]. So, I think that it becomes easier to recognize” (P4).

General therapeutic skills

In addition to theory and experience, participants reported using general therapeutic skills to foster a safe environment. This was important to them, because they considered the interview to be a challenging experience for the patient. Participants described taking the time to get acquainted with the patient before administering the interview, and communicating openly with them about both the purpose and outcome of the interview:

[I administer the SCID-II/5-PD] after a few appointments…when I feel I know something about this person. (P3)

Typically, this is to […] Get people on board […] with what we are actually doing and why we are doing it. (P8)

Some clinicians also mentioned attempts to illustrate their conclusions with examples provided by the patient, in order to make the outcome of the interview more relatable and acceptable: “[I gather and use] everyday examples, for every individual patient. Which makes it easier for them to take a potential diagnosis or a description of weaknesses to heart.” (P4).

Developing technical skills and growing confidence

As may be expected when using any standardized assessment instrument, our participants described growing more confident and skillful in their use of the interview over time. Greater familiarity with the interview content enabled them to conduct the interview by means of more free-form conversation and adapt the interview to their informational needs. Furthermore, they reported growing more comfortable with clinical decision making based on information elicited from the interview.

Summary and interpretation

These subthemes illustrate two central points: 1) theoretical knowledge and therapeutic skills are required for valid administration and scoring of the SCID-II/5-PD; 2) technical skills and confidence in one’s role facilitate flexibility in interview administration and trust in conclusions, pointing to the central role of training and experience.

Theme S2: Access to training is varied and novices often rely on their colleagues for guidance and support

Only five out of the nine participants in the S-II/5 group had attended a formal SCID-II/5-PD course. Of these five, four had received formal training after switching jobs or by participating in data collection for research projects. Emphasis on collegial and experiential learning stood out in the interviews. Three subthemes delineate different modes of learning.

Formal courses

Formal courses were described as combining lectures, videotaped interview rating and plenum discussions, and as focusing on understanding diagnostic criteria and the purpose of interview questions. One participant stated that courses promoted confidence and reliability through discussions and plenum scoring of videotaped interviews:

[…] because there was some calibration going on when we watched the video, and "what would you have scored," right? That way you realize that, yes indeed, you are sort of on the same page with [the others] […] and that you have interpreted the manual correctly. (P14)

Participants also advised novice clinicians to seek out formal courses, rather than trying to learn the interview on their own: “So, take a course. [….] There is a lot to gain from learning it [using the SCID-II/5-PD] properly.” (P3).

Collegial training and support

Both novices and experienced clinicians described discussing difficult cases with colleagues. They described this as being useful both for reaching a diagnostic conclusion and for their own sense of security: “You can write it down in the casefile that ‘I have now discussed it [the diagnosis] with others and we agree on this.’ That lifts some of that burden of being the one who makes the call” (P11).

Several participants reported having learned to use the interview primarily through informal collegial training, involving lectures, rating the interviews together, watching a supervisor perform the interview, or watching videotapes of colleagues conducting the interview: “In addition, I watched while my supervisor was conducting two interviews. […] I just watched and assigned scores. […] Afterwards we compared our scores. And then I started using it [the SCID-II/5-PD] on my own” (P10).

Own reading and experiential training

Most participants, at some point, mentioned the benefits of learning through practical experience. Some also mentioned the benefits of reading the manual for the interview, or of reading about PDs in general. Learning “on your own” was described, either as a supplement to more formal training or as a substitute when lacking it. Some reported having a lack of formal training as daunting, while others did not.

Summary and interpretation

These subthemes highlight three main points: 1) formal courses can cover relevant theory and practice needed for the SCID-II/5-PD interview; 2) colleagues are an important resource, and their input can compensate for a lack of formal training; 3) learning by doing and reading relevant literature is useful, but sub-optimal, when compared to learning from or with others.

Theme S3: Clinicians do more than score the SCID-II/5-PD

Participants spoke of the SCID-II/5-PD as both a diagnostic test and a source of insight into patients’ functioning. Two subthemes delineate this distinction.

Diagnostic test

Participants described the SCID-II/5-PD as a means of arriving at a diagnosis, which should be communicated to the patient and used as a selection criterion for treatment: “It [the SCID-II/5-PD] is the pillar in our assessment process, it is part of what… determines whether someone gets the green light with respect to treatment at our clinic. So, it is a very important part of what we do, you know” (P3).

Source of insight – Participants also described the interview more broadly as a source of insight into the patient’s experiences and functioning. Many reported actively seeking out detailed, individual information, beyond what interview questions asked for, in order to inform scoring and understanding. Gaining insight and understanding was generally described as at least equally important as arriving at a diagnosis:

[…] I don't feel that counting the number of criteria met is the most important part of the SCID[-II/5-PD] interview, but that you can use it to foster a useful clinical dialogue about how…or what are the potentially problematic aspects of this person's personality. (P11)

Summary and interpretation

These subthemes illustrate that participants are competent and actively engaged in using SCID-II/5-PD for more than just reaching a diagnostic conclusion. This speaks to their informational needs as well as the potential clinical utility of the SCID-II/5-PD.

Theme S4: Challenges arise from both within and outside the SCID-II/5-PD

Participants outlined challenges relating to both specific shortcomings of the interview and the complexity of assessing personality disorders more generally. We delineated three subthemes.

Shortcomings of the interview

Participants noted that interview questions for narcissistic personality disorder were phrased pejoratively, leading patients to refuse to identify with these characteristics, even if they satisfied symptom criteria: “Apart from narcissistic personality disorder, the SCID[II/5-PD] is useless in that respect. […] You don't have to be particularly bright to understand that you just don't answer those questions affirmatively” (P6). Participants also pointed out that the complex language and poor translations forced them to explain and clarify interview questions to their patients:

[…] Even though I sometimes think that it is obvious that it [the SCID-II/5-PD] has been translated from American English […] So at times, I just have to tell the patient that “I know it sounds a bit awkward, and some of the formulations are rather old-fashioned” and so on. (P2)

Participants considered some interview questions too broad, leading patients to confirm diagnostic criteria they did not qualify for, creating a risk of misdiagnosing patients and producing false positives. These issues were viewed by some participants as being related to specific diagnostic categories, with one participant stating:

In any case, I have the impression that there is…that there is a risk for overdiagnosis for some of the disorders that the interview [the SCID-II/5-PD] targets, and perhaps underdiagnosis for others. […] I have the impression that there is some risk for overdiagnosis for these cluster C-diagnoses; dependent, avoidant, obsessive-compulsive, mainly because [] those questions, they capture so many. (P9)

Use of compensatory measures

To overcome certain shortcomings of the interview, participants spoke of re-phrasing or skipping interview questions and providing supplementary examples. One participant even mentioned presenting patients with a prepared set of follow-up questions in the beginning of the interview to make them understand that the interview questions are meant to assess pervasive and long-lasting tendencies and not rare or occasional experiences:

So, I end up spending a lot of time sort of [asking] "Are you usually like this? Always been like this?" […] After a while, they sort of know that that is what I am after. […] "Right, you want to know how I am doing most of the time, not just how I have been feeling at times." (P6)

Navigating the complexities of personality disorders

In addition to the interview-related challenges and compensatory measures, participants also emphasized the complexity of assessing personality disorders more generally. Here, they pointed out the challenges they faced when diagnosing patients with symptoms of several different PDs, and having to decide whether symptoms were severe enough to satisfy diagnostic criteria, and controlling for external factors which can impact patients’ responses. External factors included patients’ motivations, indecisiveness, and capacity or ability to understand and answer interview questions.

Summary and interpretation

These subthemes illustrate an important distinction between challenges in assessing personality disorders in general and challenges in assessing personality disorders with the SCID-II/5-PD. Shortcomings of the interview can potentially be addressed in future iterations, while more general challenges may be more closely related to limitations of the standard categorical model.

S-LPFS group themes

Theme A1: the SICD-5-AMPD-I requires practical experience as well as theoretical knowledge

Participants reported needing both theoretical competence and experience with the patient group when conducting the SCID-5-AMPD-I:

If you have prior experience with personality disorders, I think you can do it very quickly. If you do not have such experience, and you have not acquainted yourself with the relevant literature, then I think it is a world apart from your usual way of thinking. (P5)

Several participants especially highlighted interview-specific theoretical knowledge, underlining this as being a distinct difference from the SCID-II/5-PD:

So if you were to sort of give a recommendation regarding its use then it would be helpful to have access to information describing how this came about… how they decided which facets, why they think these facets and domains specifically, how they kind of relate to the human psyche. (P7)

In principle, one could very well use the SCID-II without knowing an awful lot about personality functioning, because it basically consists of a list of questions you have to get answers to and assign scores to. [] My point is that this [SCID-AMPD-I] is an instrument where one cannot just go “chop, chop, chop” and then “Tada, here we have the score!,” or something. Here one needs a somewhat deeper level of understanding. (P16)

Clinical experience, educational background and theoretical orientation were listed as a source of such interview-specific theoretical knowledge: “But I think when it comes to the professional degree in psychology, regardless of where you study, similar topics are covered. Psychodynamic theory is covered […] So I think that psychologists are well equipped to handle this” (P5). Participants did not, however, say that practitioners with other educational backgrounds could not use the interview. As one participant stated: “You don’t need to be a medical doctor or psychologist or something to use this interview. But you need certain background knowledge. About what is relevant and what is not, you know” (P17).

Summary and interpretation

These answers indicate that the administration of SCID-5-AMPD-I may require additional theoretical knowledge as compared to the SCID-II/5-PD. Furthermore, they highlight that potential users will likely vary regarding their familiarity with relevant theory, depending on their educational background and clinical experience, which should be considered when developing training programs for this interview.

Theme A2: Formal training is essential and should cover both theory and practice

Formal training was described as being important by participants, as it enabled them to conduct this interview properly. Participants expressed being satisfied with having received theoretical and conceptual clarifications, alongside practical training with case examples and the rating of videotaped interviews: “It [SCID-AMPD-I training] was thorough, first theory, then some practical [exercises]… and there were follow-up meetings” (P16). The opportunity to calibrate scoring through group discussions and feedback was also described as useful:

Yes, it [SCID-AMPD-I training] was very helpful. And I think it is a necessity, I really do. Because if we had gone and studied by ourselves, then I think it would have resulted in… different levels of misuse […] because our scoring and administration strategies would have been so very different. (P15)

One participant pointed out however, that the training they had received may have required a certain level of theoretical background knowledge, highlighting the novelty of some interview-related concepts:

In other words, this was a training program targeted at those with a special interest [in personality disorders] …in part at specialists in the field one could say. […] And they [the instructors] could have taken into account that some of the people who will be using it [SCID-AMPD-I] probably are not very familiar with the concepts covered in the interview. (P13)

Summary and interpretation

These subthemes substantiate concerns about the complexity of the AMPD. The last quote (participant 13) further highlights how clinicians vary regarding their familiarity with the concepts of this interview and how this may impact training needs. This should be considered when developing training guidelines, to avoid the assumption that all clinicians will be equally prepared to use this interview in their practice.

Theme A3: the SCID-5-AMPD-I is more in line with clinicians’ interests and needs

Participants reported liking the SCID-5-AMPD-I, pointing out that it provided individualized as well as dimensional assessments, and that it focused on important topics, not often assessed elsewhere, which contributed to patients feeling seen and understood. We delineated two subthemes illustrating these points.

General theoretical framework

Participants expressed that they favored a dimensional interview which focused on the individual features of a patient’s psychological make-up, rather than assigning patients to a diagnostic category:

In other words, the SCID-II basically says little about…and it is…it is just categorical in a way: "personality disorder yes or no." It says nothing about what, what you should focus on, other than…what type of personality disorder it concerns. But the LPFS provides one with a more qualitative…description of how this disorder manifests itself for the given patient. (P17)

Focus on important topics which are of interest to clinicians

Participants reported being interested in the topics which the interview explored (namely, the four domains). These were described as being in line with their own theoretical viewpoint that PD is made up of issues in intrapersonal and interpersonal functioning: “So, you would ask questions you otherwise would not […] we focus much more on symptoms normally, but here one looks beyond the symptoms, and rather focuses on the cause of things” (P16). The interview’s focus on these topics was further considered to help clinicians with both planning treatment and communicating results of the interview:

Yes, in other words, one can achieve a higher degree of personalization than for example with the SCID[-II/5-PD]. It aligns well with individual experiences and one can relate these to each other. This way, it is easier for patients and their loved ones to understand. That is my experience, anyway. (P18)

Summary and interpretation

These subthemes can be viewed as indicating that the SCID-5-AMPD-I may address clinicians’ informational needs better than other interviews. Several participants said that the SCID-5-AMPD-I compared favorably to the SCID-II/5-PD and other assessment tools, further substantiating this assumption.

Theme A4: Challenges arise from the context, novelty of the SCID-5-AMPD-I, and current shortcomings

Participants pointed out challenges relating to administering the interview in a research context, gaining mastery of a novel interview, and overcoming key shortcomings of the interview. Four subthemes delineate these challenges.

Challenges in learning and using the interview

Several participants expressed having difficulty in getting used to the layout of the interview guide. Some concepts and terms in the interview were also said to be confusing and difficult to understand: “The thing was that the layout was terribly confusing and […] I wasn't always clear on what I was actually asking [the patient]. ‘What is this question actually tapping into?’” (P5). In addition, some participants expressed some difficulty scoring the interview. This was partially ascribed to the complexity of using interview questions to evaluate the fit of prototypical descriptions at each level of personality functioning: “And so you end up scoring a ‘2’ because the description of level 2 basically … even though the patient answered ‘no, no,’ the description basically fits. [] Yes. But that is where I think things can get difficult.” (P13).

Initial feelings of being overwhelmed

The challenges associated with novelty reported under the previous subtheme seem to have made participants feel overwhelmed when first presented with the interview. These feelings were sometimes described as being unfounded, easily overcome, or expected when learning something new:

No, I remember thinking, when this was presented to me, thinking “O god, how, how…WHAT?!” It was like “How in the world will I be able to understand this thing?” […] Because it was new, and it was unfamiliar, and it was a different way of [thinking]. (P15)

Shortcomings of the interview

Participants also underlined central shortcomings of the interview. They reported difficulty in scoring certain disorders such as narcissistic and avoidant PD:

Perhaps the reason is that the questions were not able to capture [empathy], or perhaps we were not alert enough…when it comes to [identifying] challenges related to avoidant [PD]. […] As a result, it slipped through the cracks a bit […]. (P13)

This was, by some, ascribed to the “double barreled” nature of the SCID-5-AMPD-I criteria. For instance, when interview questions asked about both narcissistic and avoidant characteristics: “You are trying to identify what we call avoidant pathology and narcissistic pathology in one go. […] and the thing is, that patients with a bit of avoidant… don't identify with it [the topics addressed in the questions]” (P5).

Some also expressed concerns that interview questions could be perceived as confrontational and difficult to understand: “Some of the questions… I ended up skipping or using infrequently. I only used them when I had to, to get the scoring to [add up]. Basically, because the formulations were overly complex” (P19).

Administering the interview in a research context

In addition to these interview-specific challenges, participants also pointed out some contextual challenges. The design of the aforementioned NorAMPD test-retest studies precluded prior familiarity with patients before administering the interview. Our participants explained that this prevented them from developing a satisfactory working alliance with the patient before administering the interview. This relative lack of familiarity with the patient made it difficult to evaluate whether it was safe to push for more information or confront patients about inconsistencies in their statements. Furthermore, it left clinicians without a frame of reference to compare patients’ responses in order to establish their truthfulness.

Summary and interpretation

Together, these subthemes illustrate that challenges associated with the SCID-5-AMPD-I relate to both its novelty and some central shortcomings which should be addressed. The emphasis on novelty highlights the importance of adequate training during initial implementation. The effect of novelty may also be expected to lessen over time, as the administrator becomes more familiar with the interview and its theoretical framework.

Between group comparison of themes

Table 3 gives an overview of the most central observed differences and similarities, as well as the elaborations offered by the mixed-group interviews. Our presentation of these results will follow the four thematic topics outlined in Tables 2 and 3.

Skills and competencies

Observed similarities and differences

The S-II/5 and S-LPFS groups both emphasized the importance of theoretical knowledge and experience with PD patient groups. The two groups differed, however, in that the S-II/5 group highlighted general PD theory, whereas the S-LPFS group highlighted theories specific to the development and content of the AMPD model in general and the LPFS specifically. This was taken to indicate that the SCID-5-AMPD-I might require more specific theoretical clarification than the SCID-II/5-PD interview.

We also observed that the S-II/5 group subtheme of general therapeutic skills did not emerge for the S-LPFS group. We interpreted this in terms of differences in the administration of these interviews in clinical versus research contexts. This interpretation was supported by the subtheme of administering the interview in a research context, which emerged for the S-LPFS group (see Table 2).

Mixed-group elaborations

Mixed-group participants also indicated that the SCID-5-AMPD-I might require a deeper understanding of relevant theory than the SCID-II/5-PD. They elaborated on this, explaining that the SCID-5-AMPD-I may emphasize its theoretical framework more explicitly in the interview guide, making it difficult to use this interview without having the necessary familiarity with that framework:

Because it is clear that…the way things look in the DSM-IV and DSM-5 is informed by theory. […] Maybe it just comes to the forefront more clearly in the LPFS manual, or in the interview… than it does in the SCID-II. […] You can apply the SCID-II in a much more straightforward manner… ask the questions as they are written. If you don't have a lot of prior knowledge. But that will not be an easy feat when using the LPFS interview […]. (P20)

Experienced challenges

Observed similarities and differences

Both the S-II/5 group and the S-LPFS group stated that the complex language and poor translations posed an impediment to patients and clinicians alike to understanding the interview questions. Both groups also pointed out specific shortcomings of their respective interviews, albeit that the shortcomings mentioned by the two groups differed from each other in nature: the S-II/5 group described over-inclusive criteria as posing a potential risk of overdiagnozing certain patient groups, whereas the S-LPFS group described double-barreled items as posing a potential risk of underdiagnosis of certain disorders.

We also observed that the S-LPFS group’s reports of difficulty getting used to the layout of the SCID-5-AMPD-I and feeling overwhelmed did not emerge in the S-II/5 group. This observed difference was seen as being the result of the novelty of the SCID-II/5-AMPD interview, as well as an indication that the layout of this interview may be less intuitive than that of the SCID-II/5-PD.

Mixed-group elaborations

Mixed-group participants did not directly confirm the issue of over-inclusive questions in the SCID-II/5-PD. They did however cite challenges in deciding whether enough information was obtained to score SCID-II/5-PD criteria, with one participant stating: “It was difficult to decide, sort of, yes: ‘Did I gather enough information?’ ‘Do I assign a subthreshold score to this trait? Is the threshold met for this trait?’ And I still…think that can be challenging sometimes” (P21). We considered that this may have been a result of patients frequently confirming interview questions without qualifying for symptom criteria, and viewed this as lending support to our initial observation regarding over-inclusive interview questions in the SCID-II/5-PD.

Mixed-group participants also confirmed the observation regarding double-barreled interview questions for the SICD-5-AMPD-I: “[…] and then there were indeed some questions I remember. Yes, when reading them, in retrospect I thought that some of the questions - among other things - actually ask several questions in one [question] and they are quite lengthy […]” (P22). They also elaborated on observed challenges for this interview, explaining that some of the categorical thinking found in the SCID-II/5-PD seemed to be present in the SCID-5-AMPD-I, leading to inconsistences between interview questions and the various levels of personality functioning in each domain: “The levels are described in a somewhat prototypical manner. Level 1 is accompanied by descriptions typical of obsessive-compulsive PD, level 2 a bit more narcissistic, and level 3 emotionally unstable and so on” (P22).

Mixed-group participants also argued that the layout of the SCID-5-AMPD-I might be viewed as being more complicated than that of the SCID-II/5-PD, suggesting the reason for that could be that the SCID-II/5-PD is in line with the way in which clinicians are generally trained in making a diagnosis, and that it therefore appears more familiar to novices:

[Starting with] the SCID-II; […] I think I only… well, I don't think we learned to use it during my studies. I am not entirely sure. But the layout was, well, familiar in a way. We did learn about the SCID[-I], the regular SCID interview. And besides, it [the SCID-II/5-PD] is quite straightforward when reading the protocol. It looks quite orderly. (P20)

Furthermore, they pointed out that the novelty of the SCID-5-AMPD-I will likely lessen over time as the AMPD model becomes more familiar to both clinicians and educators:

But now the ICD-11 has arrived of course. I don't know how it would be now… that students receive training at university for example. […] If you study both of them equally, a less intensive training would be required for the LPFS, for example. (P20)

Training received and/or required

Observed similarities and differences

Considering that the S-LPFS group participants had emphasized interview-specific theory as a prerequisite for the SCID-5-AMPD-I, we assumed that this interview might require more training than the SCID-II/5-PD interviews. Though the two groups did not differ in the emphasis they placed on training, they did differ in the training they had received, with S-LPFS group participants describing having received more comprehensive formal training than the S-II/5 group participants. This difference indicated a potential difference in the amount of training offered in research versus clinical contexts.

Mixed-group elaborations

Mixed-group participants partly confirmed our assumption that the SCID-5-AMPD-I might require more training than the SCID-II/5-PD, suggesting that this interview may require more theoretical clarification: “Yes. Yes. Training is important. Hmm. […] I think, to understand the SCID AMPD, one is required to study the theoretical underpinnings” (P21).

Mixed-group participants also expressed having received more training for the SCID-5-AMPD-I than for the SCID-II/5-PD:

When it comes to the LPFS, we received a more concrete explanation about the model itself and its underpinnings. […] while with the SCID[-II/5-PD] it is more something I read up on as I went along, more learning by doing, on my own. (P22)

Outcome and insight gained

Observed similarities and differences

Both S-II/5 and S-LPFS group participants emphasized that they wanted and needed individualized insight into patients’ functioning, in order to better inform diagnostic decisions and to plan treatment, and effectively communicate results to patients. Both groups also reported achieving such insight through their respective interviews. It appeared however, that S-II/5 group participants gained this insight by following up on patients’ answers and asking additional questions when administering the SCID-II/5-PD. We also observed that S-LPFS group participants found that the SCID-5-AMPD-I compared favorably to the SCID-II/5-PD in terms of its descriptive capabilities; it seemed to be better equipped to assess and describe the unique problems of individual patients.

Mixed-group elaborations

Mixed-group interviews confirmed our assumption that clinicians may need to go beyond interview questions of the SCID-II/5-PD, by asking additional questions and following up on patients answers to achieve insight into patients’ functioning:

Eh… Also, when using the SCID[-II/5-PD], you know, when conducting a thorough assessment… and probe deeply you know. I mean, when you are sincerely interested in the patient, when you are open to asking yourself the question “What could this be?” […] and you want to paint a complete picture… then you end up digging into some of the topics covered by the LPFS interview… possibly. […] But it won't be in a very structured manner (P20)

Furthermore, we noted again the tendency to spontaneously compare the SCID-5-AMPD-I favorably to the SCID-II/5-PD. Mixed-group participants also explained that how one interprets the output generated from each interview could depend on one’s frame of reference: “I believe that those who have only ever used the SCID-II, they don't know any better. Like the AMPD, exactly, that now there is a better alternative. If I am allowed to say that” (P21).

Discussion

In this study, we explored clinicians’ experiences with the use of SCID-II/5-PD and SCID-5-AMPD-I. We identified challenges in each interview and compared these to identify challenges faced by clinicians in learning and in using the interviews. We found that our participants considered theoretical knowledge and clinical experience as requirements for both interviews. The SCID-5-AMPD-I however, was deemed to require more interview-specific theory, because it more explicitly highlights its theoretical framework and therefore appears less intuitively familiar and straightforward to clinicians. Both interviews were also described as requiring comprehensive training for clinicians in order to be able to administer them properly. The SCID-5-AMPD-I appeared to require more theoretical clarification during training than the SCID-II/5-PD, however. Both interviews were reported to contain complex language, which participants partly attributed to translation issues, posing a challenge to administration. Furthermore, specific shortcomings and challenges were identified for each interview. The SCID-5-AMPD-I was described as being initially overwhelming and difficult to master due to its complexity; however, these challenges were considered worthwhile overcoming, since this interview was deemed more capable of describing patients’ problems than the SCID-II/5-PD.

Our participants described correct administration of both the SCID-II/5-PD and SCID-5-AMPD-I as being dependent on fully understanding the meaning of the interview questions. This understanding relied on both general PD theory and, for the SCID-5-AMPD-I, familiarity with interview-specific theories and concepts. These findings could be explained by differences in form and content between the two interviews. The purpose of the SCID-II/5-PD is to determine the presence or absence of PD. This requires assessing whether the patient satisfies DSM PD criteria, as well as evaluating whether these criteria are pervasive, persistent, and problematic (American Psychiatric Association, 2013; First & Gibbon, 2004). Administration of the SCID-II/5-PD therefore requires clinical experience and knowledge about general PD theory, which was also emphasized by our S-II/5 participants. The SCID-5-AMPD-I was developed to assess personality functioning, which has a rich and complex theoretical background (Bender et al., 2011). This may explain why our S-LPFS and mixed-group participants stressed that administrations of SCID-5-AMPD-I required familiarity with the specific theoretical background of the LPFS. This difference in background knowledge required for the administration of the interviews is important, and has clear implications for what training programs should emphasize for each of these interviews. Furthermore, we would argue that novice clinicians are likely to develop a familiarity with PD criteria over time through interactions with PD patients or through discussions with more experienced colleagues, as our S-II/5 participants also described. It is unclear however, whether novice clinicians will develop an understanding of the interview-specific theory of the SCID-5-AMPD-I, simply by interacting with patients or having discussions with colleagues who are equally unfamiliar with the LPFS.

At the time of writing, no specific guidelines had been issued concerning the kind and amount of training required for the administration of the SCID-II/5-PD and SCID-5-AMPD-I, beyond what has already been written on the official website for these interviews (Columbia University, 2020). Here, the developers state that they can be administered by mental health professionals or clinically trained researchers, and that clinical experience is the most important qualification. They offer no specific training requirements, but state that if clinical experience is lacking, more training will be required. Our study suggests that adequate training is highly important for both interviews, and that receiving clarification about interview-specific theories is especially important for novice users of the SCID-5-AMPD-I. We therefore suggest that separate training guidelines be developed for each interview. We recommend that these emphasize teaching of specific and general PD diagnostic criteria for the SCID-II/5-PD and that interview-specific theories and concepts be emphasized for the SCID-5-AMPD-I. We further recommend that practical exercises incorporating group discussions and scoring calibration be included in the training guidelines accompanying the instruments.

Our study also identified both shared and unique challenges associated with each interview. Complex language was described as a central shortcoming of both the SCID-II/5-PD and SCID-5-AMPD-I in this study, and could present a potential threat to both reliability and diagnostic accuracy. However, Norwegian translations of the SCID-II and SCID-5-AMPD-I have been shown to have acceptable inter-rater reliability (Arnevik et al., 2009; Buer Christensen et al., 2018), indicating a limited impact on reliability. The impact on usability has not yet been explored, however. Some participants in this study suggested complex language could be the result of poor translation. It may be, however, that the original English versions of these interviews contain complex language, as well, which was carried over into the Norwegian translations. We therefore recommend that this be further explored by carrying out similar studies with the original English version and other translations of these interviews.

Our participants also identified shortcomings of each interview in terms of over-inclusive questions in the SCID-II/5-PD, and double-barreled questions in the SCID-5-AMPD-I. Participants in both groups reported that such questions were distributed throughout each interview, complicating scoring of PD criteria or impairment of personality functioning. We therefore recommend, that in future iterations of these interviews, interview questions be closely examined by test developers and discussed with both patient and clinician representatives, to ensure they are clearly stated, specific and unambiguous.

Some remaining challenges identified for each interview may be related to shortcomings of the diagnostic models on which they are based. For the SCID-II/5-PD, we found that participants frequently encountered patients who presented with mixed-symptom profiles and/or who qualified for more than one personality disorder. Furthermore, they struggled to decide whether symptoms were severe enough to qualify for a diagnosis. These findings are in line with the criticism that the standard categorical model presents poorly delimited diagnostic categories with arbitrary diagnostic thresholds (see Clark, 2007 for review). For the SCID-5-AMPD-I, we found that participants felt initially overwhelmed by the complex layout and content of the interview. They pointed out, furthermore, that this interview relied more explicitly on unfamiliar theories. These findings support the criticism that the AMPD presents a complex diagnostic framework based on theories with which clinicians may not be familiar (Gunderson, 2013; Pilkonis et al., 2011; Shedler et al., 2010; Verheul, 2012).

Some critics have expressed concerns that the AMPD’s complexity and reliance on unfamiliar theory would constitute a discouragement to adopt the model (Gunderson, 2013; Pilkonis et al., 2011). Our study does not find support for these concerns, however. Participants in our S-LPFS and mixed group all expressed a preference for the SCID-5-AMPD-I over the SCID-II/5-PD. They ascribed this preference to their experience that the SCID-5-AMPD-I provided better descriptions of patients’ individual problems and every-day functioning than the SCID-II/5-PD. Similar findings have been reported by Morey et al. (2014). These findings may be explained by the fact that the scoring of the SCID-5-AMPD-I involves the evaluation of descriptive statements about personality functioning associated with the various levels in each subdomain. Furthermore, they indicate that participants in our study and the study of Morey et al. (2014) may consider the added complexity of the AMPD worthwhile because it provides valuable insight into core experiences of their patients.

It is worth noting however, that the final product of the SCID-5-AMPD assessment is a total score ranging from 0 to 4, alongside one of five accompanying prototypical patient descriptions. The SCID-5-AMPD-I has furthermore been found to be unidimensional (Hummelen et al., 2021), which further supports interpretation of the total score as a single construct. This total score does not reflect the detailed information our participants describe having obtained through the interview. The second module of the SCID-5-AMPD however, is intended to record individual characteristics of PD through assessing the presence of pathological PD traits. It is therefore important to be aware that, though our participants describe having obtained detailed insight through this initial assessment stage and describe this insight as being useful for communicating with their patients and informing therapy, formal recording of this insight is not carried out until later stages, and the remaining SCID-5-AMPD modules are therefore needed for complete standardized assessment of AMPD personality disorders.

One additional and particularly interesting finding of this study is the indication that training offered within a clinical setting may be less comprehensive than training offered in a research context. Previous studies have shown that personality disorders are diagnosed more frequently, accurately, and reliably when using structured clinical interviews as opposed to unstructured clinical interviews (Rogers, 2003; Zimmerman et al., 2008). If clinicians receive less comprehensive training than raters in these research studies, one may rightfully question whether the benefits of structured clinical interviews replicate in a natural clinical setting. For this reason, we recommend surveying what kind of training is offered for these and other interviews in clinical practice, and examining whether the use of structured clinical interviews for DSM personality disorders are equally effective in a naturalistic clinical context.

Strengths and limitations

A major strength of this study is that our sample provides a broad coverage of relevant populations. In our S-II/5 group, participants represented three out of four major healthcare districts in Norway, and consisted of clinicians from both central and rural areas who worked in both generalist and specialist clinical settings and who had varying levels of professional experience. The S-LPFS group included most of the Norwegian clinicians who had used the SCID-5-AMPD-I, and our mixed group included all Norwegian clinicians who had been introduced to both the SCID-II/5-PD and the SCID-5-AMPD-I at approximately the same time, at the beginning of their careers. Including mixed-group interviews in the between group comparison of themes, enabled us to both further explore and elaborate on observed differences and similarities between the S-II/5 and S-LPFS groups and inform interpretations of our findings. We therefore consider our sample to be highly suitable for identifying potential challenges clinicians may experience when learning to administer the Norwegian versions of these interviews, and for identifying potential differences and similarities in the challenges presented by each interview. Whether similar findings would be obtained for the original English versions or other translations of these interviews is unknown, however. We therefore recommend that similar studies be carried out with samples from other countries that have used other translations, as well as the original English version of these interviews.

Another strength of this study is our use of a sequential multi-group qualitative approach. Although this is a novel approach, it incorporates the use of qualitative comparison groups (Lindsay, 2019), and a form of member checking, both of which are existing methodological elements thought to reduce bias by enhancing the rigor of our analysis and the credibility of our findings (Birt et al., 2016; Lindsay, 2019).

A potential limitation of our study was that participants in the S-LPFS group were all experienced clinicians who used the SCID-5-AMPD-I in a relatively uniform research setting. This may have limited our insight into how novice clinicians may experience the SCID-5-AMPD-I, and how the interview may be experienced, if used in a more general clinical setting. Mixed group participants were however novices when first presented with the SCID-5-AMPD-I and did not offer any additional challenges with this interview, not seen in the S-LPFS group.

Conclusion

This study aimed to document clinicians’ experiences with the SCID-II/5-PD and the SCID-5-AMPD-I, and identifies both unique and shared challenges and shortcomings of SCIDs developed for the two competing diagnostic frameworks for personality disorders presented in the DSM-5. We consider our most significant finding to be the importance of adequate training in the administration of both instruments and recommend the development of specific training guidelines for each interview. We recommend that these guidelines emphasize general and specific DSM PD criteria for the SCID-II/5-PD and interview-specific LPFS theory and concepts for the SCID-5-AMPD-I. We also recommend further exploration into how interview questions for each interview are interpreted by both clinicians and patients. In closing, we want to highlight that although the SCID-AMPD-I appears to require more theoretical knowledge and training than the SCID-II/5-PD and may be initially unfamiliar to clinicians, the topics covered by this interview were deemed by our participants to be important, interesting, and clinically useful. It is worth noting, however, that the rich information they extract throughout the interview is not formally recorded and is not reflected in the final score, and that the implementation of later modules of the SCID-5-AMPD is required to make it possible to conduct a complete, standardized AMPD personality disorder assessment.

Acknowledgments

We would like to thank all clinicians who participated in this study. We would also like to thank The Norwegian Multicenter Study of the AMPD for their help in recruiting participants and the National Competency Center for Personality Psychiatry for their help in establishing contact with mental health clinics across Norway.

Disclosure statement

No potential conflict of interest was reported by the authors.

Data availability statement

Due to the nature of this research, participants of this study did not agree for their data to be shared publicly, so supporting data is not available.

Additional information

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by a FRIPRO Young Research Talent grant for the last author (Grant no. NFR 286893), awarded by the Research Council of Norway.