Framing diagnostic error: an epidemiological perspective

Diagnostic errors burden the United States healthcare system. Depending on how they are defined, between 40,000 and 4 million cases occur annually. Despite this striking statistic, and the potential benefits epidemiological approaches offer in identifying risk factors for sub-optimal diagnoses, diagnostic error remains an underprioritized epidemiolocal research topic. Magnifying the challenge are the array of forms and definitions of diagnostic errors, and limited sources of data documenting their occurrence. In this narrative review, we outline a framework for improving epidemiological applications in understanding risk factors for diagnostic error. This includes explicitly defining diagnostic error, specifying the hypothesis and research questions, consideration of systemic including social and economic factors, as well as the time-dependency of diagnosis relative to disease progression. Additional considerations for future epidemiological research on diagnostic errors include establishing standardized research databases, as well as identifying potential important sources of study bias.