Improving High-Stakes Obstetrics & Gynaecology Exams: A 3-year Analysis of SBAs

Introduction: This study evaluates the quality of Single Best Answers (SBAs) in the summative assessments of Year 5 medical students at Defense Service Medical Academy (DSMA) in Yangon, Myanmar. It examines the Difficulty and Discrimination Indices of SBAs in the Obstetrics and Gynecology curriculum across three cohorts (2022–2024), aiming to improve assessment quality through insights into refining the SBAs and informing broader quality assurance strategies. 
Methods: A quantitative analysis was conducted, focusing on 120 SBAs drawn from the Obstetrics and Gynecology curriculum across three cohorts. Statistical methods, including item analysis, Cronbach's Alpha for internal consistency, and cohort performance comparison, were used to assess the psychometric properties of the SBAs. 
Results: The Difficulty Index showed that most questions were classified as "easy," while there were notable variations in the Discrimination Index across cohorts, suggesting that some questions did not effectively differentiate between high and low-performing students. Although the exams maintained high internal reliability (Cronbach's Alpha 0.78–0.82), the weak correlation between the difficulty and discrimination indices highlighted the need for improvements in question formulation. 
Conclusion: The findings suggest that the exit exam for the medical program should include a higher proportion of average and difficult-level questions. Continuous refinement of SBAs, informed by item analysis and integrated within a comprehensive quality assurance framework that includes robust exam blueprinting, item writing and review protocols, and systematic item banking, is needed to improve difficulty levels and balance discriminatory power. By addressing these issues, DSMA can enhance its assessments, ensuring they more accurately reflect student performance, align with outcome-based medical education, and better prepare students for their future roles in healthcare.