New guidelines for reporting diagnostic accuracy studies

An international panel has released new guidelines for researchers and scientists on how to report accuracy studies of new and existing diagnostic tests. The guidelines list 30 items that should be included in every such report.

“This is really the question of how do you know the test to diagnose disease really does diagnose disease?” said co-author David E. Bruns, MD, of the University of Virginia School of Medicine’s Department of Pathology, in Charlottesville, VA. “It sounds like such a simple question to ask, but like most simple questions, it turns out it’s not so easy.” 

The new guidelines update and streamline the STARD (STAndards for the Reporting of Diagnostic accuracy) guidelines, first issued in 2003. “After about 10 years of experience with this, it’s been used thousands of times in the literature—we felt it was time to bring it up to date,” said Dr. Bruns, founding architect of STARD.

Diagnostic accuracy studies are those in which a test is evaluated against a clinical reference standard (a “gold standard”). The results are typically reported as estimates of the test’s sensitivity and specificity, which express how good the test is in correctly identifying patients as having the target condition.

Despite their apparent simplicity, such studies are at risk of bias, the panel noted. So, the updated guidelines incorporate recent evidence about sources of bias and variability in diagnostic accuracy.

When a study fails to describe useful details, it can have serious consequences for doctors and their patients, Dr. Bruns noted.

“How would you do the study to see if your new cancer test works? The first thought might be, ‘Well, we’ll get samples from patients who have cancer and we’ll get samples from some lab technicians and we will compare the two,’” Dr. Bruns said. “If you do that, you’ll greatly overestimate how good that test is. We know that for a fact."

He explained, "Part of the reason is that patients who have disease that has been clearly diagnosed are relatively easier to identify by a test than the patients that your test is designed to detect—people who have very early disease that hasn’t created symptoms yet. That’s much harder to detect.”

To that end, the guidelines enumerate clear details on important study aspects such as:

• specifics of the study hypothesis
• limitations of the study
• sample size assessed by the study
• intended use and clinical role of the test

A journal article, which explains and includes the guidelines, was simultaneously published in October 2015 by The BMJ, Radiology, and Clinical Chemistry. The guidelines are also available online.