Based on a broad interdisciplinary grounded theory analysis the topic of learning clinical reasoning and how it is implemented in virtual patients has been explored. The core category "multifactorial nature of learning clinical reasoning" is reflected in the following five main categories: Psychological Theories, Patient-centeredness, Context, Learner-centeredness, and Teaching/Assessment. Each category encompasses between four and six related concepts.
Based on the outcomes of grounded theory, the concept for a clinical reasoning tool and guidelines on how to embed it into VPs was developed.
This concept guided the development process which included a course concept, a technical specification, a prototype, the final software solution, and the integration into existing VP systems. Usability tests and a pilot study were implemented for quality assessment purposes. The developed clinical reasoning tool is based on a concept mapping approach and can be used as a stand-alone tool or in combination with virtual patients. Learners are asked to document relevant findings, differential diagnoses, necessary tests to rule out or confirm a diagnosis, and management options. All items can be connected with each other. Additionally, learners are pompted to compose a short summary statement about the patient. Feedback is provided based on the author's answer, peer answers and learning analytics.
Two courses have been developed consisting of virtual patients (81 in German, 63 in English) and seven short videos explaining the project, clinical reasoning as a process, cognitive errors, the purpose of a summary statement, and the handling of the software.
The researcher has implemented three studies based on the developed material:
Study 1: Analysis of the clinical reasoning process of medical students in two freely available courses from January until July 2017. Overall, the researcher analyzed 1393 completed datasets created by 317 learners. Three groups were identified: Correct final diagnosis on first try (59%), correct final diagnosis after 2-17 tried (13%), and correct diagnosis revealed by the system (28%). The clinical reasoning process differed significantly betweeh the two groups that diagnosed the virtual patient correctly (no matter how many tries) and the group that gave up. This finding has implication for the teaching and researching of clinical reasoning showing that it is more important to come to the correct diagnosis eventually than have it in the first try.
Study 2: How does the patient representation influence learners' clinical reasoning. Six virtual patients with three different patient representations (disruptive description of patient behaviour, friendly description of patient behaviour, and no patient story, just facts presented) were provided to 46 medical students in a rendomized controlled trial. The results showed that a disruptive description of a patient behaviour is more influential on the clinical reasoning process of medical students than the design variations. Telling the story of a patient increases extrinsic motivation, but does not significantly influence the clinical reasoning process. Further research is needed concerning the reliability of difficulty assessmentof virtual patients.
Study 3: A pilot study in cooperation with the Jaggelonian University in Krakow, Poland was implemented, exploring the influence of outcome- versus process-oriented feedback in virtual patients. 64 medical students at Jagellonian University, Kraków, Poland were provided with 16 virtual patients in a spaced way of four virtual patients / week. Overall, 873 clinical reasoning concept maps with 19.689 nodes and 641 connections were created. The results show no significant differences between the two feedback modalities concerning clinical reasoning process.
Overall, the project indentifed the need for a structured and longitudinal teaching of clinical reasoning for healthcare students and the implementation of a faculty development course for educators to be trained on the teaching of clinical reasoning.
