AJR: 'Old-fashioned' autopsy could serve as useful corrective for high-tech radiology

July 27, 2012
by Brendon Nafziger, DOTmed News Associate Editor
Autopsies are on the decline in U.S. hospitals, but they could serve as a quality control measure for medical imaging procedures, according to a new study.

The autopsy rate fell from 19 percent in the 1970s to about 8 percent today, according to recent Centers for Disease Control and Prevention data, partly as a result of the procedure's costs, provider worries about liability, and family reluctance to have deceased relatives go under the knife.

But researchers say that the postmortem examinations could act as a useful corrective for radiology, and vice versa.

In a study published in the current issue of the American Journal of Roentgenology, researchers compared autopsy results with imaging scans of patients before they died. They said they found more than a dozen clinically significant misses or misinterpretations.

"The results of this study suggest that even in 2008 patients sometimes died with undiagnosed or misdiagnosed diseases," wrote the authors, led by Douglas R. Murken, with the department of radiology at the University of Pittsburgh School of Medicine.

Looking at autopsies of 179 patients who had undergone a CT, MRI, ultrasound, X-ray, angiography or molecular imaging scans within one month of their deaths in 2008, the researchers found about 300 discrepancies between the imaging exams and postmortem studies. Of these 300 discrepancies, 201 were of a condition believed to be present at the time the imaging study was actually performed, the researchers said.

Most of these 201 discrepancies were fairly inconsequential or were caused by known limitations in the sensitivity of the modalities or the type of study ordered. However, the researchers did find 11 missed diagnoses and seven misinterpretations deemed clinically significant, resulting in a radiologic error rate of about 3.3 percent.

Researchers defined as a missed diagnosis a condition that was not found on the imaging exam but which contributed to the patient's death or would likely have required treatment. Examples included a liver tumor bigger than 1 centimeter.

Misinterpretations included lung metastases having been reported as pneumonia, and vice versa.

The researchers also found seven "descriptions without a diagnosis," where the radiologist clearly mentioned a finding -- such as a liver cyst -- in the radiology report but never mentioned a specific diagnosis.

Interestingly, the researchers also turned up 32 autopsy findings that were discrepant with accurate radiologic scans, such as a goiter described in the postmortem findings on the right side that was really on the left, suggesting that radiological procedures and autopsies can both act as quality control measures for each other.

"[B]y retrospectively reviewing radiologic studies equipped with diagnostic information gained at autopsy, clinicians and radiologists can potentially gain valuable insight into the nature of their diagnostic errors or uncertainties that might help them to avoid similar difficulties in the future," the researchers said.

The study is "Autopsy as a Quality Control Measure for Radiology, and Vice Versa."