There are broadly three reasons why we perform tests in clinical medicine: diagnosis (what is the disease?), etiognosis (what caused a disease?), and prognosis (how will the disease evolve?). It is also important that the outcome of a test should guide treatment in some way, especially when it is being touted as being a monitoring test that provides unique information that cannot be obtained by easier means. Considering how widespread the use of computerised tomography (CT) scans of the thorax during the novel coronavirus pandemic has been, one would assume that the test would satisfy one, if not all the above criteria, for an accurate diagnostic test.
Data from studies
The Cochrane (previously known as the Cochrane Collaboration) pooled together all the available data from studies conducted over the last year (https://bit.ly/33ftvVt) to try and test the accuracy of CT scans in diagnosing COVID-19. It included 41 studies with a total of 16,133 participants. It was found that a CT scan accurately diagnosed COVID-19 in about 88% of individuals with a positive RT-PCR. Since an RT-PCR itself misses 30% of people who have COVID-19, a chest CT is likely to diagnose only 62% of all individuals having COVID-19, making it a relatively inaccurate test for diagnosis. In these difficult times in obtaining RT-PCRs due to overworked laboratory services, the use of a CT chest as a surrogate needs to come with a caveat: a normal CT chest does not exclude COVID-19, and, therefore, should not be a reason to come out of isolation, especially when the CT is done very early in the disease.
Mislabelling the cause
An accurate test for etiognosis would be one in which a result would make the cause almost certain. The same Cochrane review mentioned above found that when radiologists convincingly labelled a CT pattern as being consistent with COVID-19 disease, they mislabelled 20% of those who did not have the disease as having COVID-19, getting the etiognosis wrong in a significant proportion of individuals. Telling someone who does not have COVID-19 that they do have the disease has serious implications, leaving the real diagnosis undetected, and subjecting the individual to the psychosocial consequences of the knowledge that she/he has the disease.
The third reason that is often cited as being a reason to do a CT is for prognostication: a CT that appears worse is likely to lead to worse outcomes than a CT that appears better. Two comments need to be made in this context: the severity of lung involvement as seen on a CT is reflective of the status of the lungs at that point of time, and we know that this is a dynamic process, i.e., a limited involvement at an early stage could progress with time to a severe involvement; and a CT scan revealing severely affected lungs while oxygen levels remain high and unchanged is an extremely improbable event, suggesting that a CT is unlikely to give a treating physician more information than a simple tool such as an oximeter. It needs to be mentioned that in research settings, certain patterns of lung involvement (and not the mere quantum as reported by a score) have been associated with worse outcomes (https://bit.ly/3xInyi5), but unfortunately, these have not been widely validated, and are not the reason why CT scans are presently being performed.
The risks
“What is the harm in getting a CT of the chest done?” is another argument one hears often. A study published in The New England Journal of Medicine in 2007 (https://bit.ly/3eSuIru) postulated that “0.4% of all cancers in the United States may be attributable to the radiation from CT studies”, and further speculated that the current estimate could be in the range of 1.5%-2%. This potential harm would have been clearly acceptable had this been a highly accurate and useful test. In addition to this risk to the individual undergoing the scan, there are risks to radiology technicians, staff and doctors that need to be accounted for. Moreover, considering the fact that CT scanners need to be kept in closed air-conditioned spaces, the risk of transmission of the virus at such centres cannot be ruled out.
As a physician treating COVID-19 disease over the past year, I have ordered CT scans for less than 1% of the patients whom I have treated. I have ordered them to evaluate the possibility of other lung diseases when two RT-PCR swabs were negative in patients whose symptoms were consistent with COVID-19 disease, in patients in whom there was a possibility of blood clots in the lungs when hospitalised, and to look for secondary infections in individuals who have been in hospital for a long time and can sometimes have new infections after being admitted for COVID-19. Intensivists have on occasion used CT scans to optimise ventilator strategies for individuals with severe COVID-19 disease. Indications outside of these should be the exception, not the norm.
Raise queries
So, if a physician asks that a CT scan be done, ask her/him a few questions. If it is being done for diagnosis, why not do an RT-PCR instead (or two RT-PCRs), considering the higher accuracy of the test? If it is being done despite COVID-19 being proven, ask whether a minimal involvement on the scan guarantees an uneventful clinical course, or whether a more than minimal involvement (when the oxygen levels are high, and the patient seems to be getting better) is a sign of impending deterioration. Ask whether treatment strategies have been proven to work better when guided by chest CTs (rather than clinical findings such as oxygen levels). If the answer to none of these satisfies you, consider the potential risks involved in getting that CT done, and feel free to make an informed decision.
Lancelot Pinto is Consultant Respirologist, P.D. Hinduja National Hospital and Medical Research Centre, Mumbai