A prospective study on 147 symptomatic Indian COVID patients demonstrates a sensitivity of merely 35% for CT. What are its implications for using CT for screening for COVID or for diagnosing COVID? Our latest Journal Watch video gives the answers. Also look at the photo below for reference.
You can read the entire article here: https://www.sciencedirect.com/science/article/pii/S0720048X20303363
PS : If the embedded video is not opening, you can view the video on YouTube by clicking here.
Welcome to the Café Roentgen Journal Watch, where we review certain important or interesting radiology articles. Should we do a screening CT for COVID patients? This is something that we discussed a couple of videos earlier as well, but now we have Indian data which categorically provides us the answer. We will be discussing a study published in the European Journal of Radiology on the spectrum of CT findings in COVID patients in India. But before we go to the study, let’s just look at what we have right now.
What literature tells us is that CT is extremely sensitive to detect COVID. In fact, many studies say it’s more sensitive than our RT-PCR. But this literature is mainly from high prevalence regions with overburdened systems. What literature also says is that CT is not very specific for COVID.
But what we have actually seen, particularly in places like Mumbai in Delhi, which are seeing a lot of cases, is that CT is actually quite specific for COVID, especially if it has the classic CO-RADS 4 or 5 picture. In fact, it is so specific that hospitals will accept patients having a positive CT and admit them directly into the COVID ward if the RT-PCR report is delayed or not available.
What we still don’t know is how sensitive CT is, particularly for the mild cases or the asymptomatic cases. We are also not clear on whether CT has any utility for screening asymptomatic patients before hospital admissions.
This is where we see CT being used right now in India – 1. for evaluating any complications of COVID, 2. for diagnosis of symptomatic patients, and 3. as a screening study for patients before hospital admissions or procedures. Nowadays, many doctors and hospitals are asking for routine CT screening, along with RT-PCR, with the patient getting the admission or undergoing the procedure only if negative on both.
Let’s dive into the study. This was a prospective observational study of 147 consecutive patients with RT-PCR proven COVID-19. Only symptomatic patients were included in the study, having fever, cough, sore throat or dyspnea. Asymptomatic patients were excluded. CT was performed an average of six days after the onset of symptoms.
The results were quite surprising. The mean age of the patients was 41 years, which is about what we’ll see in a country like India with a predominantly young population. Very few people (<10%) had comorbidities. Most patients are mild symptoms, while very few patients had severe symptoms like this dyspnea or reduced oxygen saturation.
Overall only 35% patients had a CT abnormality, while 65% had a normal CT. So CT was actually not very sensitive to catch COVID patients. The CT features are pretty much the same as other studies, so I am not describing them. One interesting feature, though, was that there was segmental or sub-segmental pulmonary vessel enlargement, which was seen in around 70 percent patients. And this was deemed to be quite specific for COVID as it is not described in other infections as per the authors.
I must confess that when I first saw the results, I was quite surprised, because most international studies say that CT is extremely sensitive to detect COVID; between 60-100% sensitive in different studies. But, as we know, India is a little different. We have less case fatality. We have a large young population. So perhaps the Indian data might not really match the data in the rest of the world, especially when the rest of the world data mainly comes from overburdened health systems. Also important to remember that the authors imaged mainly mild cases, but also certain moderate to severe cases were included. Asymptomatic cases were excluded.
So then does this data actually make more sense? It is quite possible that in India many patients have only upper respiratory symptoms and the infection doesn’t really reach the lower respiratory tract, and hence we do not see any CT chest findings. Most of us get very excited nowadays when we see a CT chest with COVID-like features. But what is important is that we really we do not know the denominator. Perhaps we saw two CTs which looked like COVID today, but there might have been 8 other unknown COVID patients who we also imaged but who did not have who did not have any CT features. And that is where this study is really excellent, because it included consecutive patients, making this data quite strong.
So let’s try to apply the findings of the study. Many hospitals or doctors are demanding doing CT screening along with an RT-PCR before any procedure/admission. Let’s assume we are in a place like Bombay or Delhi, which is around 10 percent COVID prevalence. Other areas will have even less prevalence. And let’s assume that this study is accurate. So the CT sensitivity is 35%. Remember that this sensitivity is actually for symptomatic patients, while we are talking about asymptomatic hospital admissions. So actually, the sensitivity may even be lower than 35%.
Now, out of 100 patients on which we do these screening studies, only 10 will actually have COVID, because that’s the prevalence of the disease in this region. Because RT-PCR has a 70% sensitivity, we will catch seven of these 10 patients on RT-PCR. And because CT has a sensitivity only around 35%, amongst the remaining three patients, we’ll be able to catch only one patient. Thus, out of 100 patients who have to undergo CT screening, we will catch only one additional COVID patient. Let’s say that CT sensitivity is actually not 35%, but closer to 70%. Even then, we’ll catch only 2 additional COVID patients out of 100 screening CTs.
I don’t think that a screening with 1-2% positive result is good enough to support CT screening. And this is in high prevalence areas; its utility will be even lower in the low prevalence zones. And that is what ACR and Fleischner’s society guidelines also recommend. We should not be doing CT screening in asymptomatic patients. That’s a practice that hospitals should just stop doing.
What about symptomatic patients? So here the situation is slightly different. Again, if we assume the study to be true, CT is not very sensitive. But whether we do CT or any other imaging or not depends on the availability and turnaround time of the RT-PCR tests or the antigen assays andn the prevalence of the disease in the region. So it’s a slightly more complex situation here. I talked to many people who are doing a lot of COVID studies in India, mainly from Bombay and Delhi, and they all believe that the 35% figure is slightly lower than what they believe. Maybe the truth lies in between the 35% this study demonstrates and the 60-100% that other studies claim. But what is clear is that a normal CT really means nothing if the patient has fever. We certainly need more data from India and hopefully one of the hospitals which perform screening studies or one of the imaging centers we do a lot of CTs for symptomatic patients can publish their data. That will certainly help us. And we certainly guidelines on what to do for areas with limiting testing ability. That is perhaps an area where imaging may play a more important role.
But what is clear right now is that we should not be performing CT screening on asymptomatic individuals. It is not recommended by any guideline and not useful based on the Indian data just published.
Thanks a lot for watching this video. Stay safe and we’ll see you again soon.
Post script: The large vessel ‘sign’ described in the paper is not all that specific, as this paper in AJR demonstrates that it may be seen even in H1N1 patients.