Re: Controversy in coronaViral Imaging and Diagnostics (COVID). A reply.

Sir—We thank our Moscow colleagues for their kind words regarding our scientific letter and for sharing their experience.

They highlight their practical experience of using computed tomography (CT) to triage patients with suspected COVID-19 in the context of a high prevalence of the disease and insufficient numbers of reverse transcription polymerase chain reaction (RT-PCR) tests or associated delays. We note that they describe a very different cohort of patients from the asymptomatic patients we were considering for screening. They do provide compelling data to suggest that CT is indeed as “a valuable tool for indicating the need for hospitalisation of patients with suspected COVID-19, especially in densely populated areas.“; however, we are not able in this cohort to calculate the true accuracy of CT to diagnose COVID-19 as we have no idea of the true and false positive and negative rates, so accuracy cannot be determined.

From the data we quoted, if a 50% prevalence of the disease is assumed in those tested, this would provide a post-test positive predictive value (PPV) of 56%, negative predictive value (NPV) 88%, overall test accuracy of 61%; with a 75% prevalence of the disease in those tested would give a post-test PPV of 80%, NPV 71%, accuracy of 79%, and if a 90% prevalence of the disease is assumed, the post-test PPV is 92%, NPV 45%, and the accuracy is 89%. None of these statistics provide compelling evidence for the use of CT to diagnose COVID-19 per se. Whether it helps triage for hospitalisation is, of course, another matter.

We would urge our colleagues to “close the loop” on their potentially powerful dataset by looking at their correlative results of RT-PCR versus the CT findings so that PPV, NPV, and accuracy can be determined. It is our view that too many radiological investigations are recommended based on sensitivity/specificity data, but without considering prevalence in the cohort being tested. We would be interested to see if CT added any further diagnostic value to the usual clinical and vital signs of “illness”; for example, if all those admitted to hospital were significantly hypoxic, one presumes that CT did not add very much to the “triage for hospitalisation”, other than additional (perhaps unnecessary) workload and cost, and perhaps compromising other areas of healthcare.

The Controversy in coronaViral Imaging and Diagnostics persists, and by extension, also in other areas of imaging.

Conflict of interest

The authors declare no conflict of interest.