The MATCH as an Institutional Clinical Reasoning Assessment for Third-Year Students

Purpose: Following an 18-week block of either family medicine, internal medicine, neurology, and psychiatry (FINP) or obstetrics–gynecology, pediatrics, and surgery (OPS), students participate in a 2-week course comprising enrichment workshops and 3 assessment experiences, the content of which is mapped to clerkship objectives. Assessments include (1) a customized National Board of Medical Examiners (NBME) multiple-choice question exam, (2) a standardized patient-based clinical skills assessment, and (3) an exam focused on analytical thinking/clinical reasoning. The focus of this discussion is on the lattermost exam which was novel to our curriculum. Based on the literature on clinical reasoning assessment practices, 1 we selected a variation of the clinical integrated puzzle 2 called Measuring Analytical Thinking in Clinical Healthcare (MATCH). 3 This prompted students to differentiate key features of common patient presentations, and it allowed us to integrate multiple disciplines within questions. We present our process of development, implementation, and exam outcomes analysis. Approach: The authors served as an oversight team throughout the development, implementation, and analysis. We met with clerkship directors to discuss common patient presentations for their discipline. We looked for overlap by more than 1 discipline and selected 4 presentations for each exam version (e.g., FINP block: palpitations, back pain, altered mental status, and fatigue). We then identified common diagnoses for each presentation and scripted key features under 4 domains: patient presentation, physical exam, work-up, and management. To construct one MATCH grid, we selected 5 diagnoses, providing corresponding key features. Two distracters within each domain provided additional but irrelevant choices limiting forced matching. Both exam versions included 4 grids. The exams were piloted and refined based on feedback from faculty, residents, and fourth-year students. One week before the assessment, students were oriented to the exam format and provided a sample MATCH grid from the literature. 3 They were instructed to select the best option out of each domain for each diagnosis, using each only once. Following individual exam administration, groups of 5 students completed the same exam. Immediately afterward, faculty debriefed each presenting problem/diagnosis and students were encouraged to explain their reasoning if there was disagreement from intended answers. Individual and group data were analyzed for mean, standard deviation, reliability, and discrimination. Each student received an individual and group score (contributing 75% and 25%, respectively) and an overall score. Outcomes: Exam development was very labor intensive. Consultation with clerkship directors, review of current literature-based guidelines, and clinical expertise were essential to ensure that each domain item adequately distinguished itself and was a “best match” for a given diagnosis. Two hundred nineteen students participated in this assessment. During group administration of the exam, faculty circulated to listen to student discussions. Students were very engaged and actively shared their reasoning. The large group debrief provided an opportunity for constructive discussions when disagreement about a particular item occurred. Group performance was superior to individual performance (e.g., FINP means 92.97, SD 2.82 vs 83.82, SD 6.68). The mean change between individual and group scores was approximately +9 points. Reliability using Cronbach’s alpha was 0.77 and 0.68 for FINP and OPS, respectively. Comparing quartile performance on the MATCH and customized NBME exam revealed differences. Significance: Despite challenges in developing a high-quality integrated assessment, this exam format offered us unique insights into students’ clinical reasoning and provided valuable information about student skill. The group examination presented an opportunity for peer teaching and cooperative problem solving. Clinical reasoning development is a challenge 4 and assessment using varied formats is essential. Following this experience, we plan to introduce MATCH assessments throughout our curriculum.