Monaz: Hello, I'm Monaz Mehta, editor in the Cochrane Editorial and Methods department. The COVID-19 pandemic is creating challenges for diagnosis of the disease, as well as for its treatment. The rapid reviews being prepared and updated by Cochrane are covering both areas. We asked the lead author of the November 2020 versions of our review of the evidence on using imaging tests to diagnose the condition, Nayaar Islam from The Ottawa Hospital Research Institute in Canada, to describe the latest findings in this podcast.

Nayaar: People suspected of having COVID-19 need accurate diagnostic tests to determine whether they are infected, so that they can receive treatment and self-isolate to prevent the spread of the infection. However, accurately diagnosing COVID-19 is challenging, and the value of imaging tests, such as X-rays, CT scans and ultrasounds, remain undefined. In the previous version of our review from September, the evidence showed that chest CT scans were not capable of making a distinction between people with and without COVID-19 as the scans gave an approximately equal proportion of positive test results in those with and without the disease. However, as the research is rapidly evolving, we updated our review to further investigate these findings.
Our most recent search for research published up to late June 2020 allowed us to review a total of 34 studies in people suspected of having COVID-19. In these studies, participants underwent the imaging test as well as the current best available test to diagnose an infection, the RT-PCR test, which determined if the imaging tests were as accurate at identifying if someone had COVID-19.Most of the studies tested chest CT scans and, based on the results of these 31 studies, we found that the scans gave a correct diagnosis of COVID-19 in 90% of people who had the disease, but an incorrect diagnosis of COVID-19 in 39% of people who did not have the disease.
There were only three studies on chest X-rays, including one that also tested chest CT scans. These studies showed that X-rays may give a correct diagnosis of COVID-19 in 57 to 89% of people with the disease and an incorrect diagnosis of COVID-19 in 11 to 89% of people without the disease. In the one study evaluating ultrasound, this correctly diagnosed COVID-19 in 97% of people with the disease, and incorrectly diagnosed COVID-19 in 38% of people without the disease. However, because of the low number of included studies for X-rays and ultrasound, this evidence is weak and should be carefully considered.
Looking at the studies we reviewed as a whole, there were large differences across their results and most had a high or unclear risk of bias. For example, many studies either recruited participants that did not represent the general population of people with suspected COVID-19 or did not clearly state how their participants were recruited. These risks of bias mean that our findings may not be representative of typical clinical situations or generalizable to all patient populations, making it difficult to judge the implications of the current evidence at this time. However, we believe that one important conclusion from this update is that CT scans perform well in identifying COVID-19 but have limited capability in differentiating COVID-19 from other causes of lung infection.
In summary, the diagnostic accuracy of CT scans has improved since our previous review, perhaps because radiologists reading the scans now use better definitions of a positive CT result. Also, as we are now in a later stage of the pandemic, more recent studies have likely built on knowledge from previous experience and the studies published earlier in the pandemic. We will continue updating this review as more evidence becomes available.

Monaz: If you would like to read more about the current evidence, and watch for future updates of this review as the evidence moves forward, it's available free to view at Cochrane Library dot com. Just go to the website and type in a search for 'thoracic imaging and COVID-19' to find it.