Posted on

CoronaVirus – FAQ, misconceptions, information, from a statistical perspective

Hi, I am in the statistics field and have been working directly on the nCoV-2019 outbreak with local and international teams for the last 2 weeks. I’m based in the US but speak to local doctors, administrators, WHO advisory teams, and academics all around the world on the virus. I haven’t had time to really do this post until now since it’s been pretty much nonstop 18 hour days for most of us since the outbreak started (also because of the time difference).

First the disclaimer: This is not medical advice. I am not a medical doctor or virologist (though I work side by side with teams of both). I will not reveal any non-public information, both for privacy and legal reasons. I am not acting in any official capacity. Any views I may present are my own, based on my work in the space, and may not be peer-reviewed or condoned by official bodies. I will not engage in any political discussions.

Now I’ve seen a lot of very common misconceptions about nCoV. Partially this is due to the media distorting, misinterpreting, and cherry-picking data to fit a narrative. Partially this is due to polarization of the “doomsday” crowd and the “it’s ok” crowd. Mostly it is due to the general public having not enough understanding of medicine and statistics, and lacking the tools to interpret the data/news. I want to clear some of these common questions up and provide some good resources and charts.

Final Edit: I didn’t know this excellent thread was going on while I was writing this. Please consult that as well, as it contains excellent responses from many, many more experts!

Common questions/concerns/misconceptions FAQ:

1) What is the incubation period? Why do I keep hearing 14 days? Is this scary?

The incubation period so far shows a period of 2-7 days with a 95% confidence interval, with median cases at 4.8 days[1] The 14 day limit is the current maximum theorized incubation period from a Zhejiang case study. The exact maximum is difficult to know because this is based on patient survey and contact reconstruction and prone to error, but 14 days is the “safe” upper bound so far. This figure is similar to the ~5 day incubation for SARS. [2] There is no need to panic about this as it’s very normal viral behavior.

2) But what about asymptomatic transmission? Is this worth worrying over?

So to be clear, so far over 95% of patients in most studies do eventually display symptoms. [3] However, transmission during the asymptomatic incubation stage above has also been confirmed by local and international studies. I believe the US decision to vastly heighten travel restrictions on China last night was largely due to this German confirmation. Ironically US CDC previously did not believe Chinese warnings this was happening.

While confirming asymptomatic transmission is important, it is not rare viral behavior, especially in the latter stages of incubation where viral load is high. Currently, we have no statistical evidence that there is a major risk from asymptomatic spreading. The incubation period is short enough that if this were a major dynamic, the end patients would have already shown up in the statistics.

3) What about super-spreaders? Why do I hear this has spread to 14 people from one infected?

Actually this is one of the positives about this virus so far. Unlike SARS, we have had no evidence of super-spreading occurring rapidly. What has been confirmed so far is 1 case of a “super spreader” which in epidemiology means a carrier that has infected at least 8 people. [4]

Now let’s study this one case so far. It was honestly a VERY special case. Several rare factors all compounded to create the conditions for him to “superspread” nCoV to 14 healthcare professionals:

  1. He lied about having had lots of exposure to the Wuhan Seafood market
  2. He was admitted to the hospital because of pre-existing conditions requiring neurosurgery, before the danger and extent of the nCoV outbreak was known to the staff there. So proper quarantine procedures weren’t followed
  3. He required sputum suction, tracheotomy and tracheal intubation, which all unfortunately expose medical staff to a LOT of his body fluids.

So in the current opinion of the epidemiology community looking at nCoV cases, this is a fairly rare instance and unlikely to be repeated outside of a very specialized setting. There is no need to be worried about this vector yet.

4) What is the R0? Is it 2? 5? 12? What does this mean for the viral evolution?

Since popular media (Contagion, Pandemic) really brought the concept of R0 into public focus, there’s a lot of confusion about this simplification of statistical methods. Put simply, R0 is a variable used in theoretical epidemiology analysis, derived from the data through various mathematical methods. It is not an intrinsic property of the virus, nor is it set in stone – R0 will change as properties of the outbreak, and our containment efforts, adjust it. There’s a good further discussion of R0 here, but generally, without understanding the underlying methods that led to the calculation of a specific R0, you shouldn’t overly focus on this number, nor compare it or make conclusions based purely on it.

As best as our models can tell, the R0 of the virus was well above 2-3 in the beginning, where it was infecting people in Wuhan through the Seafood market and across many vectors before broad awareness. This was from Dec of last year to maybe early January. Since increasing awareness and containment factors, the R(t) has likely declined to below 2, and optimistically will head below 1. We are awaiting data from Chinese New Year containment to see the lagged reporting data, but current extreme measure will have a major effect on the outbreak, but is unrealistic to maintain for long. The plan is to identify, treat, and isolate the vast majority of cases before life and travel normalizes.

Edit: to be clear here, I am not suggesting that R0 is currently 1 or anything like that. I am trying to communicate the point that R(t) is not fixed over time, but a function of our response to the virus. I am hoping that current containment measures will be enough to bring the R(t) to 1 or below, as is the case with any epidemic once it’s under control and declining.

5) Why is the official case count so low? Why do I keep hearing larger numbers of infected? Is there a government cover-up?

The official “confirmed cases” number is not meant to be a “live” count of the # of infected or even identified infected individuals, and the professional community understands this. This number is exactly what it says on the tin, eg, this is the official number we have been able to test and confirm to our satisfaction. In our current fast-response information-driven society, we are used to having access to immediate, live data, and we expect such. The fact we have any confirmation at all at this point is actually a miracle. Back in the days of SARS, no accurate testing existed for many months after the outbreak, so ALL numbers were estimates!

Now due to Chinese bureaucracy and how the confirmations work in China, lack of supplies and personnel when Wuhan hospitals were overwhelmed last week, and difficulty producing the test kits, there is a lag time of up to 12 days to someone being suspected and able to be tested in Wuhan. I think this week they’re working hard on bringing that lag down, and the lag is a lot shorter in other provinces due to still-functioning logistics, but it’s still about 5 days at least in almost all of China, due to the multiple bureaucratic checks they force it to go through before it’s deemed “confirmed enough”. There’s a trade-off between accuracy (yes, they wouldn’t want to make an embarrassing mistake misdiagnosing or mistaking identity) and speed.

In the rest of the world, the delay can be very fast, ~1 day response to 3 or 4 days as well, depending on the country’s infrastructure and availability of test kits/proximity to CDC center that’s stocking it.

So really the way to think about the number of confirmed cases in China is, this is the number of cases that we can confirm from about 7-10 days ago. This is how we’re roughly working with the data. I think most laypeople are just assuming this is a “live” number which is just not the case, it takes time from patient intake to screening to testing to confirmation to double checking.

6) What about deaths? Have a lot of people died? Why is the official death rate so low? Is there a cover-up?

It is true that the death rate reported by China is heavily misleading. But this is NOT due to an active cover-up. There are 2 main structural reasons:

  1. This is primarily due to the structural method of how China records deaths on their certificate. It is established policy/practice in China to record the final cause of death, rather than all existing conditions and overlapping factors.

For example, if a (say 85 yo) patient in the US with diabetes and an existing heart condition gets nCoV, is admitted in the hospital, is confirmed with nCoV, then dies of heart failure, he is recorded as dying of nCoV AND heart failure with other complications. However if the same patient dies in China, he would only be recorded of dying by heart failure.

This is a well-known issue with China and co-morbid diseases. I don’t agree with it, I wouldn’t do it, but I don’t run China. But this is not a new method they made up to try to hide deaths here, it’s just the way it’s done. This has led to jokes in the epidemiology community that “it’s impossible to die of flu in China”, because they basically don’t record any deaths where the patient has flu. See here this recent article from the Global Times, which is one of China’s state-sponsored newspapers.

This is not something even China is really trying to hide. They just tell us, sorry, our doctors just do things this way, we have no interest in changing it.

2) The other reason is, right now if a patient is awaiting test results (turnaround can be 3-5 days in China still), and passes away in the meantime, they are not recorded as nCoV. I guess this I can understand, I think similar policies in US, we don’t like to go back and edit death certificates because it’s a huge hassle.

Ok so – definitely, the death count is too low. We all agree there. But before you freak out, there’s a bright spot. We CAN also put an upper bound with a fair amount of certainty on the general death rate. How? Because there have been enough cases reported globally already, and enough data from the patients OUTSIDE of China, that we can tell the death rate is NOT anywhere near 10% with a strong degree of certainty (many patients have recovered, and are just awaiting the viral test all-clear before they can be discharged. Most other patients are in stable and recovering condition).

Edit: I’m going to take out the actual back of the envelope illustration I was using here, because it’s been rightfully criticized as being over-simplistic to the point of misleading. I still believe that the fact that global death rates remain very low is encouraging and can be used to remove extremely high death rate arguments, however, even adjusted for quality of care and health of the traveling population.

7) Great, so we don’t know the number infected or the number of fatalities. Why am I refreshing the number repeatedly?

Well, it’s ok that we don’t know all the exact specifics of a virus while we’re fighting it. It’s the same as every past pandemic. However as long as we can keep making good approximations, we can get closer and closer to the truth with each iteration and develop the best methods for fighting it. It’s important for professionals to understand the limitations, systematic errors, and other adjustments in the data so we can best utilize it. Laypeople shouldn’t pay too much attention to the data releases, but if you are still curious, there are some cool novel ways researchers are using to get to the number approximations.

8) <Removed>

Edit: I’m taking this out under good advisement. I was clearly going for an optimistic skew by this point in the writing, but better to provide no data than provide flimsy data that could be misleading.

9) I’m still not convinced, I hear there’s a huge government cover-up, mass graves, people dropping dead on the street, invisible super-carriers and we are days away from complete anarchy!

That’s not a question, but if you are still worried, just remember the basic law of conspiracies: The more people involved, the less likely it is to keep secret. Currently the outbreak is being carefully scrutinized by thousands of professionals across the world, as well as about a billion very worried Chinese citizens. The simple fact is that extreme assumptions about deaths and coverups just don’t fit with the most basic math of the distributed data we have seen in the international population. By now, if the apocalyptic assumptions were true, we would be either seeing a LOT more international infections, and/or a LOT more deaths. Unless you believe that the entirety of global response efforts are “in” on the deception and trying to kill the world.

10) Fine, I’m not going to buy a fallout shelter yet, but what can I do?

If you are not in China, there’s not much to do. Keep an eye on the news, but don’t panic or make drastic decisions. This and this are nice articles about how to keep safe. If you’re unsure, seek help from a healthcare professional. Overall, how much preventive care depends on what level of risk you are personally comfortable with. If you’re most comfortable doing a little more prevention, that’s ok too. There’s no one-size fits all answer for how much you should react.

11) This is all well and good, but surely something worries you and other professionals too? There’s more draconian responses announced every day, surely it’s in response to a real risk?

While I can’t speak to the policy response choices of every country, generally it’s become politically difficult to resist a harsher response, because of the fear and attention the virus has generated. While the economic damage is real, the tail risks from a perceived lack of response is too politically damaging, so most countries are responding with forceful measures. From a disease control viewpoint this is great, because it means the virus is that much more likely to be contained.

What I’m most worried about now is still whether self-sustaining infection locales are being propagated in Chinese cities outside of Wuhan. This data is still inconclusive as of now, and bears a lot of attention. Most CDC policy is watching this, because if the virus was not contained in Hubei, then the next easiest border is to contain it in China, but doing so is an order of magnitude harder.

If you’re still with me after all those links and math – take a breather. From an epidemiological data standpoint, the virus is still in its infancy days. The fast information and news flow has allowed the coverage to ramp up much faster than any other outbreak, which is a double-edged sword for the public. There are thousands and thousands of professionals around the globe working on the dangers around the clock, often risking life and infection. Rest assured they do have your health interests in mind.

I will try to be around to answer questions as my schedule permits.