Multimethod Assessment of Psychosis

Session Abstract: This symposium is focused on the multimethod assessment of psychosis using self-report and performance measures. We start with a discussion of the symptoms of psychosis in the context of the HiTOP model followed by how R-PAS and the PAI assess different aspects of psychotic phenomena. Mark Blais presents the first paper on assessing milder psychotic symptoms with the PAI and the relationship between PAI SCZ Psychotic Experiences and the Paranoia scale with clinical ratings of psychosis. In the second presentation, using a forensic sample, Francesca Ales presents data on the R-PAS visual misperceptions variable (FQ) and its relationship to clinical ratings of hallucinations when using three different procedures to code response objects that are not in the FQ tables. The goal is to improve the FQ scale’s reliability and to save time for assessors in coding this variable as part of the upcoming short-form to assess psychosis, the Thought and Perception Assessment System (TPAS). In the third presentation, Callie Jowers presents the results of a study with male VA inpatients using nonstandard Rorschach clarification questions to determine whether more open-ended/unstructured questions result in more thought disorder and illogicality than more closed-ended/structured questions. The practical goal is to develop clarification questions for TPAS that assessors can use to administer the test even without knowing how to code responses. Finally, using a forensic sample, Angie Keene presents results of a study of the degree to which automated coding of Rorschach responses can serve as a substitute for human coding of disorganized thinking and the negative psychosis symptoms of inexpressivity. Again, the goal is to save assessors’ time in coding psychotic phenomena on TPAS. Finally, Joni Mihura will describe how these findings inform the development of TPAS and Ali Khadivi will serve as the discussant as to the applications of these findings to clinical practice.
Note: We also attempted to include MMPI-3 inpatient psychosis data but, due to the pandemic, most of these settings were not conducting assessments or not conducting research using assessments. One researcher, however, plans to start data collection in the next few months but cannot guarantee they will have data in time.

Chair Information: Joni L Mihura, PhD, ABAP | University of Toledo

Discussant Information: Ali Khadivi, PhD | Albert Einstein College of Medicine

Presentation 1 Title: Assessing Mild Psychosis with the Personality Assessment Inventory

Presentation 1 Abstract: Psychotic symptoms are associated with tremendous cost, burden, disability, and suffering. While psychotic symptoms take multiple forms, they can be conceptualized along two dimensions those that impact reality contact and that impairing thought quality. While frank psychosis is readily identified, milder presentations (e.g., residual, attenuated, prodromal) can be challenging to detect. Psychological assessment is typically requested when psychotic symptoms have “resolved” or diminished. To assess mild or residual psychosis, psychologists employ a multimethod test battery including self-report and performance-based instruments. The present paper explores the utility of the Personality Assessment Inventory (PAI, Morey, 1991) for assessing residual psychotic symptoms. The PAI is a popular broadband instrument with scales designed to measure two common psychotic conditions; schizophrenia (subscales Psychotic Experiences, Thought Disorder, & Social Detachment) and paranoia (subscales Resentment, Persecution, & Hypervigilance).

From an IRB-approved assessment database, we identified 1,359 subjects who completed the PAI as part of a standard clinical assessment. After screening for validity (see Morey, 1991), 1,047 subjects were entered into the study. The sample was 54% male, average age 41.9 (SD=15.29) and mean education was 14.7 years (SD=2.79). The sample identified as 84% white, 5% African American, 4% Asian & 6% Mixed / Other. All subjects completed a semi-structured interview (see Blais et al., 2021 for interview details) prior to the assessment. The interview captures demographics, current functioning, and life history data, including past psychotic symptoms. Previous research has shown the interview data to be sufficiently reliable. Interview psychosis data were organized into 3 composite variables: 1) Impaired Reality Contact (IRC; summed history of Auditory & Visual Hallucinations), 2) Impaired Thought Quality (ITQ; summed history of Paranoid Ideation, Ideas of Reference, Thought Insertion, Thought Blocking, & Thought Withdrawal), and 3) Total Psychotic Symptoms (summed total of all variables).

Data Analyses: Correlational analyses will explore the associations among the composite psychosis variables and the PAI Schizophrenia and Paranoia scales & subscales. Next, three subject groups will be composed: 1) subjects with only IRC symptoms, 2) only ITQ symptoms, & 3) subjects with both IRC + ITQ. ANOVAs will be conducted to explore for group differences on the target PAI scales.

The discussion will focus on the PAI’s utility for assessing residual or mild psychotic symptoms. The convergent and divergent performance of the PAI Schizophrenia and Paranoia scales & subscales will be reviewed. PAI scale / profile differences among 3 groups with different psychotic presentations examined. Potential clinical application of these findings will be highlighted. The paper will close with a discussion of limitations and possible future directions.

Author:

Mark Blais, PsyD | Massachusetts General Hospital

Presentation 2 Title: Improving the Reliability and Validity of the R-PAS Measure of Visual Misperceptions (FQ-)

Presentation 2 Abstract: ** Please note that this is a ‘Completed Study’ in the sense that the data collection is complete; only the path for coding FQ and related statistical analyses remain.

Rorschach Form Quality (FQ) describes how well a response object fits a given inkblot location and is derived from how frequently it is reported in the reference sample. Currently, in most cases, to code FQ, it is sufficient for the assessor to look up the response object in the FQ Tables. However, some percepts seen by the examinee are so rare that they are not found in the FQ Tables, in which case the examiner must make an extrapolation from similarly shaped existing responses in the database or, in the most controversial cases, make a completely subjective judgment about how well the object fits the contours of the inkblot location to which the response is given. Previous literature indicates that making the latter judgment results in a lower interrater agreement than when the response object is in the FQ tables or can be extrapolated from the available objects in those tables. Therefore, our goal was to investigate other methods of assigning FQ levels when a response object is not in the FQ tables nor can it be extrapolated from existing objects. These methods include (a) adding a separate score to FQo, FQu, and FQ- of FQ NIT (Not in Table), (b) coding the FQ NIT as FQ-, and (c) predicting the non-tabled FQ level from the existing tabled and extrapolated FQs in a given protocol. We will assess the validity of these measures by their comparative associations to PANSS ratings of Hallucinations.

In order to test these hypotheses, we are using an unpublished R-PAS dataset in which FQ-% and WD-% were positively associated with the PANSS rating of Hallucinations (i.e., r = .51 and r = .50, respectively). Data collection took place at the maximum-security Michigan Center for Forensic Psychiatry in Saline, Michigan. The sample consisted of 91 valid R-PAS protocols. The 91 respondents had an average age of 40 (SD = 13, range 19-80). The sample is 89% male and predominantly European (41%) or African American (42%). Of the 91 respondents, 53 were adjudicated not guilty by reason of insanity and 38 were incompetent to stand trial. Forensic Center staff completed a rating of the PANSS for each patient. For each protocol, the FQ level (i.e., ordinary, unusual, minus), the coding process (i.e., tabled, extrapolated, judged), and the dichotomous variable FQ NIT will be coded. Another examiner will also blindly code these FQ’s and their decision paths for 20 of the protocols to assess interrater reliability. We will report the correlations between the validity criterion variable, the PANSS Hallucinations, and the different methods of coding responses when they are not listed in the FQ tables. We will also conduct hierarchical regression analyses to determine whether any of these methods of coding provided incremental validity in predicting PANSS Hallucinations ratings over the standard R-PAS method of coding FQ.

Authors:

Francesca Ales, PhD | University of Turin- Italy

Gregory J. Meyer, PhD | University of Toledo

Joshua J. Eblin, PhD | Center for Family Psychiatry

Callie Jowers, MA | University of Detroit Mercy

Alicia W. Villanueva van den Hurk, BS | University of Dayton

Joni L. Mihura, PhD, ABAP | University of Toledo

Presentation 3 Title: Using the Rorschach to Assess Thought Disorder: The Effect of the Examiner’s Questions on Test Validity

Presentation 3 Abstract: Psychosis is perhaps one of the most functionally disruptive symptoms of many psychiatric, neurodevelopmental, neurological, and medical conditions (Arciniegas, 2015). While it is the hallmark symptom of the schizophrenia spectrum disorders, psychosis is also common in other mood disorders, as well as substance, degenerative neurological conditions, and medical ailments (Arciniegas, 2015). Formal thought disorder, or disorganized thinking, is a common feature of psychotic disorders.
One executive functioning deficit hypothesized to account for a range of cognitive deficits in schizophrenia, including disorganized thinking, is context processing deficits (CP; e.g., Cohen & Servan-Schreiber, 1992). Here, context is defined as “prior task-relevant information that is represented in such a form that it can bias selection of the appropriate behavior response” (Barch et al., 2004, p. 557). Examples of context include task instructions or a specific prior stimulus, such as Rorschach cards or Rorschach response queries. Experimental studies have found that individuals with schizophrenia display fewer instances of formal thought disorder in response to structured questions compared to unstructured questions (Barch & Berenbaum, 1997). Moreover, using only the response phase administration questions and omitting the clarification phase questions has been shown to reduce the levels of disorganized thinking on a Rorschach protocol (Ritzler & Nalesnik, 1990).
The aim of the present study is to examine whether Rorschach context variables (e.g., open/unstructured, closed/structured questions) influence the amount of disorganized thinking on Rorschach protocols using an archival dataset of 64 male inpatients from a California VA who have diagnoses of schizophrenia, schizoaffective disorder, and depression. This study used a range of different non-standard questions types. A scale was created to rate these questions along continua that included open/unstructured and closed/structured endpoints. The practical application of this study is to develop clarification questions for the short form Thought and Perception Assessment System (TPAS) that do not require examiners to know how to score the Rorschach. Authors hypothesize an association between question type and Rorschach Performance Assessment System (R-PAS) linguistic cognitive codes that tap disorganized thinking (Peculiar Logic [PEC] and Level 2 Deviant Response [DV2]), such that less structured questions will relate to greater numbers of PEC and DV2 on these protocols. Using the standard clarification, DR2 (.63), but not PEC (-.11) scores, correlated with an aggregate criterion measure of disordered thinking (the Conceptual Disorganization scale from the Brief Psychiatric Rating Scale [BPRS; Overall & Gorham, 1962] and a composite score of Loose Associations and Incoherence from the Structured Clinical Interview for DSM-III-Psychotic Disorders Version [SCID-PD; Sptizer & Williams, 1985]). After the non-standard questions are rated on the question-type scale, our goals are to determine if (a) less structured questions result in more disorganized thinking on the Rorschach, (b) if these less structured questions can increase the presence of the low base rate DR2 and PEC scores, and (c) whether question type will moderate the relationship between R-PAS DV2 and PEC codes and the criterion variables. Results from this presentation will highlight what type of examiner questions influence the type and amount of thought disorder during Rorschach administration.