Decisions in the Diagnostic Process

The medical diagnostic process is complex and involves multiple phases of gathering, integrating, and interpreting information. Physicians often perform these tasks in stressful environments, making the occurrence of diagnostic errors unsurprising. Alarmingly, these errors occur relatively often, are highly preventable, and are frequently associated with substantial patient harm. Diagnostic safety research has been underemphasized, but interest has increased since the publication of the 2015 report Improving Diagnosis in Health Care. Nevertheless, important knowledge gaps remain. This thesis aimed to address these gaps through five studies focusing on the measurement of diagnostic safety concepts and the causes and contextual factors associated with diagnostic errors. Diagnostic error is commonly defined as “the failure to 1) establish a correct and timely explanation of the patient’s health problems or 2) to communicate that explanation to the patient”. However, this definition is difficult to measure in practice, resulting in the use of other operationalizations. Several closely related concepts are often used as proxies for diagnostic error, but it is unclear whether these measure the same thing. This lack of clarity may lead to misinterpretation of findings. Therefore, this thesis investigated how different diagnostic safety concepts can be measured and how they relate to each other. In a record review study of prospectively included patients, three concepts were examined: suboptimal diagnostic decisions, diagnostic errors, and diagnostic discrepancies. The results showed little overlap between these concepts, suggesting that they capture different aspects of diagnostic safety. Building on these findings, a conceptual model was developed that also incorporated diagnostic adverse events, sentinel events, and missed opportunities for timely diagnosis. The model demonstrates that although some cases may fit multiple definitions, substantial areas exist in which these concepts do not overlap. In addition, this thesis examined in-depth when and why errors occur during the diagnostic process using multiple well-established instruments across several studies using different data sources (e.g. sentinel event reports, medical record reviews). Across studies, diagnostic errors and suboptimal decisions most frequently occurred during the assessment phase of the diagnostic process, where physicians integrate and interpret clinical information. Errors also frequently occurred during diagnostic testing, such as ordering incorrect tests or failing to follow up on test results in a timely manner. Analyses showed that both human and organizational factors often contributed to these events, and that diagnostic adverse events were more frequently associated with multiple contributing causes than other types of adverse events. Combining the different instruments provided deeper insight into the relationship between phases of the diagnostic process and specific types of errors. Errors during the assessment phase were often knowledge-based mistakes, reflecting gaps in clinical knowledge or difficulties retrieving relevant knowledge. In contrast, errors during diagnostic testing were more often related to violations, such as not following established guidelines or protocols. Contextual factors such as time pressure and poorly designed electronic health record systems were also indicated as possible contributing factors. Despite the high potential for improvement, the incidence of diagnostic adverse events has remained relatively stable. This thesis therefore recommends that policymakers and healthcare organizations prioritize diagnostic safety by improving collaboration and communication, particularly in the follow-up of diagnostic test results, and improving disease knowledge and knowledge-retrieval. The instruments used in this thesis may support healthcare organizations in systematically analysing diagnostic errors and identifying targeted improvement strategies. Furthermore, improvements to electronic health record systems are needed, as current design and usability problems may negatively affect diagnostic safety. Future research should further investigate the role of electronic health records, and explore how (AI–based) diagnostic decision support tools might support physicians in making more accurate diagnoses.