Overestimation of clinical diagnostic performance caused by low necropsy rates.
Assessing how well clinicians or diagnostic tests perform in detecting a given condition depends on knowing the number of correctly diagnosed cases and missed diagnoses. Combining their previous work with published studies of diagnostic performance for aortic dissection, pulmonary embolism, and tuberculosis, the authors show that adjusting for missed diagnoses among non-autopsied deaths lowers diagnostic performance. They estimate that 17% of patients with tuberculosis, 18% to 26% with aortic dissection, and 9% with pulmonary embolism die without clinicians considering the possibility of those diagnoses. All of these findings fall significantly below the previously reported rates. The authors point out that many missed diagnoses may represent limitations of current technology or atypical presentations of these conditions rather than errors. Nonetheless, clinically significant diagnoses escape detection at a greater rate than is generally appreciated.