Suboptimal diagnostic decisions in hospitalized patients with fever: a prospective record review with physician Interviews.

Background
The medical diagnostic process is vulnerable to suboptimal decision-making (that is, decisions with any deviation from an optimal diagnostic process) due to its complex nature. It is unknown how these suboptimal diagnostic decisions and other measures of diagnostic safety (diagnostic error, diagnostic discrepancy) relate to each other.

Methods
The authors prospectively included a convenience sample of 53 hospitalized patients with fever between February and May 2023. After discharge, independent internal medicine physicians reviewed their medical records to identify suboptimal diagnostic decisions and diagnostic errors. When such suboptimal decisions were observed, involved physicians were interviewed about the thought processes behind these decisions. Established tools and taxonomies were used to identify and categorize suboptimal diagnostic decisions, diagnostic errors, and diagnostic discrepancies.

Results
The authors initially identified a total of 110 suboptimal diagnostic decisions in 38 of 53 cases (71.7%). After the physician interviews, 29 cases with suboptimal decisions remained (54.7%), with a total of 72 suboptimal diagnostic decisions across those cases (median of 2 suboptimal decisions per case; interquartile range 1-4). Cases with a higher number of suboptimal diagnostic decisions had significantly higher rates of diagnostic error and diagnostic discrepancy. No significant association between diagnostic error and diagnostic discrepancy were found. Almost all suboptimal decisions were human, and most took place during assessment of the patient and diagnostic testing.

Conclusion
Cases with more suboptimal diagnostic decisions were associated with higher rates of diagnostic error and diagnostic discrepancy, but the level of overlap between the three was relatively low, suggesting that these reflect different concepts of diagnostic safety and should be treated as such. Future research should incorporate physician interviews to enrich understanding and account for contextual factors.