Supporting specialized psychiatric care in remote areas: A computer adaptive diagnostic interview for personality pathology

The goal of this project was to develop an adaptive instrument for assessing personality disorders. The instruments that are currently in use are relatively cumbersome, and may be a challenge to use; especially for clinicians who do not use them on a daily basis. The new instrument that we developed is based on principles of computerized adaptive testing. This means that the instrument is governed by an algorithm and adapted real-time, so that only the questions that are relevant and contribute important information needed to make the diagnosis are selected. Using the adaptive algorithm allows for delivering a relatively short, yet precise, personalized test; while ensuring adequate content coverage. The instrument was designed to be more intuitive to use than the current generation of semi-structured interviews. This will facilitate assessment of personality disorders for clinicians who are not specialized in these disorders per se, such as clinicians working in remote areas. Since personality disorders are notoriously hard to treat, correct diagnosis is of paramount importance: using appropriate treatment designed especially for a particular personality disorder has been shown to help patients improve. Over the past year, we studied how the instruments our adaptive test is based on are used in practice. Among other things, we explored how clinicians perceive the ease of use of these instruments, as well as their views on the advantages, disadvantages and challenges associated with using these instruments in clinical settings. We used this information in the development the prototype of our adaptive test, which was shown to clinicians and researchers at the national conference organized by the Network for Personality Disorders in 2022. Through this demonstration, we obtained input from the conference participants that was used to refine the adaptive test further. The test is accompanied by a scoring module that turns the test scores into graphical displays. Based on a literature review and an experiment we did with experienced clinicians, we have developed a scoring module that promotes the correct interpretation of test results and their clinical implications.

Through feedback groups and seminars in Stjørdal, Trondheim, Namsos and Molde, the project has contributed to spreading information about PD diagnostic assessment and developments within the PD research field to multiple psychiatric clinics across the country. By actively involving our target user group throughout the development phase of our project we have also ensured that clinicians’ experiences, needs and interests were integrated into the computer adaptive test developed in our project. As this computer adaptive test is more straightforward to use and more efficient than existing pen-and-paper interviews, it will serve to improve the availability and effectiveness of PD assessment at local psychiatric clinics in Norway and thus help to ensure patients get appropriate treatment.

When personality traits are inflexible and maladaptive, and cause a clinically significant level of functional impairment and/or distress, they may be diagnostic markers for a personality disorder. Studies have shown that between 3 and 10% of the general population meet the diagnostic criteria of one or more personality disorders. Moreover, prevalence rates in psychiatric populations may be as high as 50 to 80%. Since personality disorders are notoriously hard to treat, correct diagnosis is of paramount importance: using appropriate treatment designed especially for a particular personality disorder has been shown to help patients improve. Establishing a personality diagnosis is not an easy feat; it typically requires a high amount of expertise and training. Furthermore, the current generation of semi-structured interviews are lengthy; are scored by hand; or may be perceived as highly complex. To address these issued and support mental healthcare professionals, especially those not specifically specialized in personality disorders, we aim to develop and validate a tool which is efficient, intuitive to use, and facilitates automatic scoring: a computerized adaptive test (CAT). CATs are governed by a combination of statistical and content-driven rules, and allow for delivering a relatively short yet precise personalized test while ensuring adequate validity. In this project, we will study how to best incorporate available information (theory, empirical information) to design a CAT that is as efficient and user-friendly as possible, while maintaining adequate levels of validity. Using our results, we will provide general guidelines for developing CATs supporting clinical diagnoses and CAT-based score reports for clinical purposes.