COVID-19 has spread quickly throughout the world, and Cochrane is producing a series of rapid reviews to help decision makers deal with the pandemic and its impact. One of these reviews, published in June 2020, examines the accuracy of using signs and symptoms to diagnose whether someone has the disease. We asked the lead author, Thomas Struyf from the KU Leuven in Belgium, to tell us why the review is needed and what they found.
Thomas: The accurate diagnosis of COVID-19 disease is important to identify individuals who may need treatment or to isolate themselves to prevent spread of infection to others.
It’s also important not to label someone as having COVID-19 when they don’t, because this might lead to unnecessary further testing, treatment, isolation (of them and their close contacts), and
false reassurance about having immunity in the future.
Our review examines the evidence on this. We wanted to find out how accurate symptoms (such as cough, fever, or diarrhoea) and signs from medical examination (such as an increased heart rate) are for diagnosing mild COVID-19 disease and COVID-19 pneumonia.
We searched for studies published from January 2020 to April 2020 and were able to include 16, with a total of just over 7700 patients. These had investigated 27 different individual symptoms and signs, but they did not clearly distinguish mild COVID-19 disease from COVID-19 pneumonia, so we had to present the findings for both conditions combined.
The accuracy of individual symptoms and signs varied widely across studies. In all studies, the diagnosis of COVID-19 was confirmed by the most accurate laboratory test available. The studies were done in hospital settings, so the results are not representative of individuals presenting with symptoms in the community.
The two symptoms with the most studies were cough and fever. These showed that for every 1000 adults with symptoms, approximately 170 would have COVID-19. However, 500 to 700 patients would have a cough, and these would include only 70 to 120 of the patients with COVID-19. Conversely, of the 300 to 500 patients who do not have a cough, 50 to 100 would actually have COVID-19.
In the same group of 1000 adults, 100 to 300 patients would have fever, and 10 to 150 of these would have COVID-19, meaning that of the 700 to 900 patients without fever, 30 to 160 would have COVID-19.
In summary, therefore, our review of the research from the first four months of 2020 suggests that no single symptom or sign is good at detecting whether COVID-19 is present in individuals presenting to hospital. No studies were found investigating the accuracy of multiple symptoms and signs used together, which is what happens in usual practice. We will be updating this review regularly to capture any such studies and to keep this summary of the evidence as up to date as possible.