Challenges assessing personality disorders with the SCID-5-PD in psychiatric patients

Background: The SCID-5-PD is frequently used to diagnose personality disorders. The aim of this study is to compare the diagnostic outcomes of the SCID-5-PD with expert clinical assessment in an ICD-10 setting. Methods: A random sample of a total of 30 psychiatric in- and outpatients (mean age = 34 ± 16, 17 males and 13 females) went through a comprehensive clinical assessment conducted by experts. Subsequently, the patients were assessed with the SCID-5-PD by specifically trained novice raters. Results: 55% (n=11) of patients with clinical diagnosis within the schizophrenia spectrum were allocated one or more diagnoses of personality disorder according to the SCID-5-PD, primarily borderline personality disorder (n=6). In contrast, of all patients with a clinical diagnosis outside the schizophrenia spectrum, only one patient qualified for a diagnosis of personality disorder with the SCID-5-PD. Meanwhile, 70% (n=7) of patients with a clinical diagnosis of ICD-10 schizotypal disorder did not meet the criteria for this disorder when assessed with the SCID-5-PD. Conclusions: When considering a differential diagnosis within the schizophrenia spectrum, outcomes from the SCID-5-PD should be interpreted cautiously.


Background
Structured interviews are increasingly becoming the gold standard in psychiatric assessment.
Diagnostic information is obtained based on the patient's responses and the clinician's observations. These interviews attempt to identify symptoms and syndromes, which meet specific diagnostic criteria(1), e.g. the Structured Clinical Interview for DSM-5 Personality Disorders (SCID-5-PD).
There is no consensus on a formal definition of the term structured interview. While the SCID-5-PD considers itself a semi-structured interview(2), we consider the interview to be fully structured as it consists of a set of predetermined questions, presented in a definite order (1). This type of interview is popular in research as well as in everyday clinical settings to diagnose a wide variety of psychiatric conditions including personality disorders (3).
These disorders are believed to be of high prevalence in western countries (around 12% in the general population(4)) and even more so among psychiatric inpatients, where the prevalence of borderline personality disorder (BPD) alone is estimated to be around 20% (5).
The Structured Clinical Interview for DSM-5 Personality Disorders (SCID-5-PD) is frequently used for assessing these disorders.
This interview schedule is widespread also in countries using the ICD-10. When used clinically, the DSM-5 diagnoses are usually converted to their equivalent in the ICD-10. However, there are important differences between the two diagnostic systems: The ICD-10 is a hierarchical system, in which lower order diagnoses such as personality disorders are generally overruled or included by higher order diagnoses such as schizophrenia, unless they are clearly independent conditions. Patients with schizophrenia are usually not given an additional diagnosis of BPD, because schizophrenia spectrum disorders (SSD) rank higher than personality disorders and disturbances in personality are inherent to schizophrenia. The ICD-10 hierarchy is based on the continental European tradition in psychopathology as described by Jaspers (6). The DSM-5 allows for a higher degree of comorbidity, as it does not have the same hierarchical order. Furthermore, schizotypal disorder is considered a personality disorder in the DSM-5, whereas ICD-10 considers it a higher-order disorder within the schizophrenia spectrum.
The SCID-5-PD suggests a diagnosis of personality disorder when a certain number of diagnostic criteria are met [1] . However, none of the symptoms or diagnostic criteria for personality disorders are specific to personality disorder alone. Similar phenomena can be seen in various psychiatric disorders (7), which makes the process of differential diagnosis paramount. This was the subject of a recent study by Zandersen and Parnas, in which 30 patients diagnosed with BPD underwent a rigorous clinical examination conducted by experts (8). According to this assessment, two thirds of the patients met the criteria for a SSD conforming to DSM-5 and three fourths for ICD-10.
A general challenge in psychiatric diagnosing is the lack of a gold standard for the assessment of psychiatric disorders. Robert Spitzer, one of the creators of the DSM-III and the SCID-5-PD proposed that the validity of structured interviews could be tested against a Longitudinal Expert assessment of All Data (interviews with the patient, relatives, etc. as well as all available clinical records; abbr. LEAD) (9). Following Spitzer's proposal, we decided to investigate how outcomes from the SCID-5-PD correlated with those from expert clinical assessment.

Aim
The aim of the study is to compare SCID-5-PD assessments of psychiatric patients with expert assessment using all available information.

Participants
A total of 30 hospitalized and outpatient participants were recruited from Mental Health Center Glostrup in the capital region of Denmark. Outpatients were generally in a more stable clinical condition than hospitalized patients. We included both groups in order to study patients across different severities of psychiatric illness.
Intellectual disability. These criteria were selected to study psychiatric patients ecologically, without the confounding factors developmental disorders and organic disease. The recruitment of patients took place from 11/4-2019 to 27/10-2019. Hospitalized patients were included randomly until 15 were obtained. Patients from the outpatient clinic were invited consecutively, based on the date of appointment. Of 37 eligible outpatients, 15 were recruited. Two patients were excluded after inclusion: One due to not being able to complete the interview, the other due to an uncertainty in the clinical diagnosis. Two other patients were included instead.

Assessment
Clinical diagnoses were assessed according to regular procedure and allocated according to ICD-10.
Hospitalized patients were assessed during hospitalization by an attending senior psychiatrist. Results Table 1 shows demographic data for hospitalized patients and outpatients. Table 2 shows the clinical ICD-10 diagnoses and the DSM-5 diagnoses from the SCID-5-PD.
A total of 12 patients qualified for at least one SCID-5-PD diagnosis. 11 of them had a clinical diagnosis within the ICD-10 schizophrenia spectrum. This amounts to 55% of patients with clinical schizophrenia spectrum diagnoses also qualifying for at least one SCID-5-PD diagnosis (median number of diagnoses 1, ranging from 1 to 4). The most frequent SCID-5-PD diagnosis for these patients was BPD. 3 of the 10 patients with a clinical diagnosis of schizotypal disorder were diagnosed with schizotypal personality disorder (SPD) with the SCID-5-PD.   Examination Questionnaire (IPDEQ) (14) and outcomes were compared to those of healthy controls.
The study found that patients suffering from schizophrenia were 8 times more likely to screen positive for a personality disorder than a healthy control. Patients with schizophrenia were also more likely to screen positive for personality disorders across clusters than healthy controls. While these results are similar to ours, they are interpreted as co-morbidity and linked to specific prognostic outcomes. There is little discussion regarding the possible connection between the altered personality and the underlying condition of schizophrenia. Our findings too could be understood as a reflection of comorbidity, i.e. the personality disorders are co-morbid to schizophrenia. However, co-morbidity in ICD-10 is only relevant for disorders that are not aspects of the same clinical entity. SSD patients often exhibit disturbances that superficially mimic personality disorders but are, in fact, expressions of a more profound psychopathology (15,16).
It is not always clear what kind of phenomena the diagnostic criteria for a personality disorder cover.
An example of this is the BPD criteria of "identity disturbance" and "chronic feelings of emptiness" that can be interpreted at different levels. At a personal or narrative level, identity disturbance can reflect confusion regarding career choices, social and romantic relationships etc., whereas identity disturbance at the level of the minimal or core self reflects "[…] distortions of first-person perspective, incomplete sense of substantiality-embodiment, and an ephemeral sense of selfpresence" (17). These are psychopathological phenomena often seen in patients with SSD. Similarly, feelings of emptiness at the narrative level can refer to a lack of values or direction in life. At the level of the minimal self, these feelings reflect different kinds of depersonalization linked to SSD (17). As the SCID-5-PD contains no instructions or tools for differentiating psychopathology at a structural level, it is possible for patients with SSD to be misdiagnosed with a personality disorder. This may result in a patient not getting indispensable antipsychotic medication.
The potential mistake of overlooking a condition of schizophrenia may not be a weakness inherent to the design of the SCID-5-PD. This interview is not meant to differentiate between broad categories of psychiatric illnesses, but only to assess personality disorders as described in the DSM-5. However, in clinical work the SCID-5-PD is often used as sole diagnostic instrument when a diagnosis of personality disorder is suspected. It is difficult to find quantitative data of the extent of such suboptimal clinical practices. The user guide of the SCID-5-PD itself presents an example. This is a clinical case, meant to serve as a tutorial of how the interview should be conducted. In this scenario, a young man is referred to the psychiatric department by his general practitioner, after he unexpectedly bursts into tears during a routine check. Upon entering the psychiatrist's office, he exhibits unusual behaviour, including displaying a copy of a dollar bill with the patient's face glued on instead of George Washington's (2).
If any sort of general psychiatric assessment is performed, mention of it is omitted in the user guide.
The patient is assessed with the SCID-5-PD and given multiple diagnoses of personality disorder.
A different problem may arise when a psychotic condition is established prior to the administration of the SCID-5-PD. In this case, answers are considered invalid if they exclusively relate to events occurring during psychotic or affective episodes. According to the SCID-PD guidelines the interviewer is supposed to ask directly if the answers provided occur outside these episodes (11). This question presupposes that the patient knows what is meant by the term "psychosis" and can delimit psychotic states temporally. This distinction, if at all possible, requires a high level of insight into illness (18).
Although there is no consensus on the definition of the concept of insight into illness, it is generally recognized that schizophrenic patients frequently suffer from a lack thereof (19).
The diagnostic criteria in DSM-5 and ICD-10 are meant to delimit the different mental conditions, not to be comprehensive descriptions of the psychopathology inherent to specific disorders (5,20). As the SCID-5-PD treats the criteria for personality disorders as discrete entities of equal diagnostic importance, it can be challenging for the clinician to grasp the overall clinical picture in the absence of a more comprehensive psychiatric assessment (3,7,21). Most diagnostic criteria are, in themselves, not specific for any disorder (22): The diagnostic specificity emerges from the interaction between the symptoms and signs of the disorder. The meaningful unit of parts that emerges is called the Gestalt.
Multiple interwoven sources add to the Gestalt such as the patient's history, the context, the relation to other experiences, the form and content of the experience (23,24). In the diagnostic process there are reciprocal dependencies between the whole and its single features. When the unifying Gestalt is not engaged, these single features may appear as a conglomerate of different disease entities, creating the co-morbidity that characterizes a structured, polythetic diagnostic approach (25).
According to Jaspers, the specificity of mental disorders is not graspable at the level of singular symptoms and signs, but only at the level of the Gestalt (6). Grasping the Gestalt may be particularly challenging when differentiating personality disorder from SSD, an endeavour for which a fully structured interview may not suffice.

Limitations
The major limitation for this study is the small sample size, which cannot be assumed to be representative. Furthermore, all outpatients were referred to the clinic with the purpose of getting a second opinion, as they were difficult to diagnose. It is likely that this subsample is not representative of a normal patient population.
It could also be argued that part of the discrepancy between the expert assessment and the SCID-5-PD assessment could be attributed to the use of novice raters conducting the SCID-5-PD interviews.
However, a study from Ventura et al. (26) suggests that differences in reliability and diagnostic accuracy between structured interviews performed by trained novice raters and experienced raters are not significant.

Conclusion
Even though the SCID-5-PD includes differential diagnostic notes and questions, schizophreniaspectrum disorders may be misinterpreted as personality disorder in the absence of comprehensive differential diagnosis. Furthermore, cases of schizotypal disorder may be misinterpreted or overlooked if no other psychiatric assessment is conducted. Further research with larger and more representative samples is needed to elucidate these problems. Compliance (approval no. P-2019-09). All recruited patients had to sign an informed consent statement in order to participate in the study. Patients were given a minimum of 24 hours to decide whether to participate or not and were informed of their legal rights as medical research subjects, including the right to revoke consent at any time. As the study did not involve clinical trials, no approval of ethics was required by Danish standards.

Consent for publication
Not applicable

Availability of data and materials
The datasets generated and/or analysed during the current study are not publicly available due to reasons concerning the privacy of the patients but are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no conflict of interest.

Funding
This study was funded by the Lundbeck Foundation. The Lundbeck Foundation was not involved in any stage of this study.

Authors' contributions
MB and SF performed the SCID-interviews for this study. MB wrote this article with the collaboration of SF. JN designed the study and provided ongoing quality assuring feedback regarding the SCID-5-PD interviews, as well as critique, additions and suggestions for the article. LJ assessed all outpatients at the ambulatory clinic and provided critique and suggestions for the article. All authors contributed to the final article.