What is corona virus or
COVID-19 ?
It's very important to know when a person is infectious with COVID-19. Individuals are infectious when they're shedding the virus, which means that the virus replicates in the cells of their upper or lower respiratory tract. And this viral shedding can be measured by obtaining swabs and subsequently doing PCR on the swabs to detect viral RNA. Let's look at the temporal trend of this viral shedding. This is what viral shedding in patients with COVID-19 looks like. This graph is taken from a paper by Gabriel Leung and coworkers published in Nature Medicine. On the X axis are days since symptom onset, shown on the Y axis are cycle threshold values. The fewer PCR cycles are necessary to detect viral RNA, the lower the CT value. Meaning more viral shedding. In other words, the Y axis shows the amount of viral RNA that's detected. And we can clearly see that viral shedding decreases after symptom onset. What we can tell from this graph is that patients are already infectious when symptoms develop and infectiousness decreases thereafter. But what does the curve look like before symptom onset over here? It's very unlikely that it looks anything like this, right? It's more likely and biologically plausible that it looks something like this, this, this, or this. Essentially what we want to know is when do they start to become infectious? When is the peak of infectiousness and when are they no longer spreading the virus? When are they no longer infectious? The problem is once patients become known to us because they develop disease, we can measure viral shedding and thereby infectiousness. But we usually don't know who will become infectious, so we can't measure that part of the curve. So how can we determine that left part of the curve? Well, we can do that by studying transmission pairs. That's a very common practice in epidemiology. When someone gets sick with, let's say, COVID-19 it's very often possible to determine whom they got it from.
So the person who infected them was the infector in that transmission pair, and they themselves are the infectee. Sometimes, for example, in families, it's hard to tell who was the infector because three people might have already had the disease. So one of them must have been the infector. In this case, the name transmission pairs might be a bit of a misnomer. So that's what the authors looked at. They had a group of 77 transmission pairs. Red dots indicate the onset of symptoms in the infector or primary case. Blue dots indicate onset of symptoms in the infectee or a secondary case, and green bars indicate the exposure window of transmission during which infection must've occurred. The authors basically used two key concepts in epidemiology, the incubation period, which is the time from infection to symptom onset and the serial interval, which is the time from symptom onset in the infector to symptom onset in the infectee. Now, we can't infer the incubation period from looking at this graph, but we can definitely determine the serial intervals. That's the duration from the red spots to the blue spots for each transmission pair. And when they did that, they found that the serial interval in their transmission pairs was distributed like that. It looks a little bit like a bell curve with the right skew.
They found that the mean serial interval was 5.8 days, and that the median serial interval was 5.2 days. And here's something that's very interesting. In 7.6% of pairs, the serial interval was negative, which means that in these pairs, the infectee got sick before the infector. Here are two such examples where the blue dot occurs before the red dot. And we said that they also needed to know the incubation period, which they didn't calculate from their own data. They use the results from this New England Journal publication by Li et al, who found that the mean incubation period was 5.2 days. Now, let's look at a couple of examples to understand what they did. Let's say the incubation period was always five days, followed by a symptomatic period. This is the infector at patient A.
Now, let's look what happens in a disease where patients only become infectious after they develop symptoms, as was the case in SARS or Ebola. So let's say that this infectee, patient B, gets infected here, and after five days he developed symptoms. So what's the serial interval? Well, patient A developed symptoms here. Patient B developed symptoms here. So this is the serial interval. Let's look at patient C, who gets infected here, and after five days become symptomatic. Again, this is what his serial interval looks like. And we see that in these patients. The serial interval is longer than the incubation period. Now, let's look what happens in a fictitious disease where patients are infectious in the pre-symptomatic period. Let's use patient A again, as the infector. So this patient D is infected here and after five days he developed symptoms. And this is his serial interval.
From symptom onset in patient A to symptom onset in patient D. Let's look at another patient, patient E, who gets infected here, develops symptoms here. And this is his serial interval. You see that in patients D and E, the serial intervals are shorter than the incubation period. So when the serial interval is longer than the incubation period, infection occurs primarily after symptom onset. And when the serial interval is shorter than the incubation period, infection occurs primarily before symptom onset. So just to reiterate, this is the infectiousness we know and can measure.
What does the left part look like? That's the question. So we know that the mean incubation period is 5.2 days and that the median and means serial intervals are 5.2 and 5.8 days. So based on what we already know, without doing much calculation, we can already guess that there must be just as much transmission going on before symptom onset as thereafter. Obviously the authors did a much more sophisticated analysis to find out which distribution of infectiousness best fit their data, and this is what they found. They found that infectiousness starts 2.3 days before symptom onset. That patients are most infectious 0.7 days or around one day before symptom onset, and that's seven days after symptom onset there was very little infectiousness going on anymore.
They also found that 44% of infectiousness or infections occurred in the pre-symptomatic phase, and that's really bad because it makes containment very hard. Let's say that a patient develops very mild symptoms on day zero and on day one the symptoms get worse and she seeks medical attention and is diagnosed with COVID-19. Well, at that point, most of the secondary cases have already been infected. Let's look at our transmission pairs again to see what that means. Let's focus in on these guys. So this infector developed symptoms here, and we know from looking at this graph that the infectee will develop symptoms two days later. But since infectiousness starts two days before symptom onset, he's already infectious and should be isolated immediately as indicated by this line. Similar situation in this pair. The infector developed symptoms and we only have one day until infectiousness starts in the infectee. In some pairs we'll have a little more time. Whereas in others, time to act has already lapsed. This is a situation where technology could be of great help. If we had an app that tracks the contacts of infectors, it would be much easier to break the chain of transmission. That's what these authors have also found. They found that with the help of a good contact tracing app, we could even prevent lockdowns and all the economic damage that occurs in its aftermath. Obviously it's a social discussion we need to have about it. But if we can agree on a strict standard for these apps, that ensures that our data is protected as much as possible, these apps could be a true game changer in the fight against COVID-19.