Can CT Scans Diagnose COVID-19? Experts Argue For and Against

As COVID-19 continues to spread around the world, doctors are struggling to learn and understand more about the disease. In some cases, there are disagreements about best practices for diagnostics and treatments. Medical News Today talked to two doctors on opposing sides of one argument: whether CT scanning could, or should, serve as a method for diagnosing this disease.

According to Joseph Fraiman, MD, an emergency doctor working near New Orleans:

Early on in the COVID-19 outbreak, it became clear that RT-PCR testing is both slow and unreliable. One Chinese official estimated that swab test sensitivity was as low as 30%, a finding consistent with RT-PCR testing of fluid samples.

More optimistic estimates from the ACR suggest a range of 42–71% for the RT-PCR test. In other words, the swab misses nearly a third of cases — at best.

According to a research article in Radiology, CT is far more sensitive, commonly detecting cases the swab misses. CT’s higher sensitivity (routinely estimated at well over 90%) and real-time results come at the expense of increased false-positives.

But this downside is of far less concern to public health authorities than poor sensitivity. The World Health Organization’s (WHO) strategic objective to stop human transmission of SARS-CoV-2 prioritizes the detection and isolation of potential spreaders above all else.

According to Mark Hammer, MD, from the Department of Radiology at Brigham and Women’s Hospital, Harvard Medical School:

COVID-19 has myriad clinical manifestations. The best known and most severe is pneumonia. Experts know that CT scans of the lungs are very sensitive in the diagnosis of pneumonia, but are they useful in the context of COVID-19?

Several research studies from China have argued that CT is useful. Articles from Dr. Yicheng Fang and colleagues at Affiliated Taizhou Hospital in Zhejiang Province, and Dr. Tao Ai and colleagues at Tongji Medical College in Wuhan, published in the journal Radiology suggested that the sensitivity of CT scans for COVID-19 is 97–98%.

In other words, these articles purport to show that only 2–3% of patients with COVID-19 have normal CT scan results.

Unfortunately, these findings are misleading for several reasons.

First, the selection criteria for patients in these studies are unclear, but there appears to be a bias toward people with more severe illness, those in the hospital, or both. It is likely that the figures do not include people with milder (or no) symptoms who would probably have normal CT scans.

Second, the authors do not offer criteria for determining when a CT scan is deemed “positive.” Instead, they appear to include any CT abnormality whatsoever.

In real-world practice, claiming a person has COVID-19 based on the presence of a minor abnormality on a CT both ignores the common subclinical lung inflammation that radiologists frequently encounter and the other diseases that people may have instead of COVID-19.

Third, the results of these articles are at odds with other publications, including a paper by Dr. Shohei Inui and colleagues from the Japan Self-Defense Forces Central Hospital in Tokyo published in the journal Radiology: Cardiothoracic Imaging.

In this study, the researchers studied 104 people with COVID-19 from the infamous Diamond Princess cruise ship. They found that nearly half of asymptomatic people and one-fifth of symptomatic patients had normal CT scans.

When taken collectively, these early results suggest that CT scans produce an unacceptably high false-negative rate and thus will fail to pick up a significant fraction (up to half) of people with COVID-19.