Two decades of diagnostic safety research: advances, challenges, and next steps.

Since the National Academies of Sciences, Engineering, and Medicine (NASEM) report Improving Diagnosis in Health Care, various research efforts have accelerated progress to understand and improve diagnostic safety. In this opinion piece, we summarize two decades of progress in methods for identifying and learning from diagnostic errors and provide recommendations for future research. Multiple methods have been used to quantify diagnostic errors in various clinical settings, thereby facilitating a deeper understanding of the nature and magnitude of the problem and enabling studies of contributing factors. However, the use of standardized definitions of a diagnostic error and/or diagnostic safety event, a shared mental model for measurement, and more universal application of tools to measure these events across the research enterprise are still needed. We highlight progress in selected research methods and applications, such as co-development with patients, inclusion of multidisciplinary perspectives (such as those from informatics, human factors, and social and cognitive sciences), and the use of sociotechnical approaches. Specific areas where research should be prioritized include the application of cognitive science to the real-world study of diagnostic errors, understanding the costs associated with diagnostic safety, developing and implementing interventions related to patient engagement, evaluating and integrating artificial intelligence, and implementing system-related interventions to improve diagnosis. To promote broad-scale improvement in diagnostic safety over the next decade, we provide several actionable steps and recommendations for various audiences, including researchers, research funders, safety professionals, and policymakers, involved in research and implementation activities for reducing preventable diagnostic harm.