[Disclaimer]

[Extracted from larger set of notes, March 24-26.]

As far as I can tell, the news media are drawing incorrect conclusions and reporting them as fact.

I think it’s possible that the source of confusion is a letter to the editor of the New England Journal of Medicine about a research study. The correspondence was published March 17, 2020 and is available here: Aerosol and Surface Stability of SARS-CoV-2 as Compared with SARS-CoV-1.

The letter summarizes the findings of research on the viability, over time, of SARS-CoV-2 on surfaces and in the air. The letter is not the problem in and of itself, and I assume that the research study is not either. Instead, the problem is that some journalists seem to have drawn incorrect conclusions from some factually correct statements in the researchers’ letter. The journalists think these facts imply something that they clearly do not.

Background

SARS-CoV-2 virus bodies are ejected in saliva and mucus during coughing. Larger droplets in the sputum fall quickly to the ground or land on surfaces. It is unavoidable that uninfected individuals will come into contact with these contaminated surfaces. Think of packages on store shelves in grocery stores which infected people will contaminate, and which uninfected people will unintentionally infect themselves with later.

In addition to the surface droplets, there are some aerosolized droplets that have high enough surface area relative to their mass that air currents can keep them suspended in the air for hours.

Epidemiologists monitoring the spread of the pandemic, and other scientists studying the SARS-CoV-2 coronavirus itself, don’t have sufficient data yet on the rate at which SARS-CoV-2 virus bodies are being shed by the infected population during the progression of various epidemics around the world.

They do not know how many virus bodies are commonly found on surfaces in those areas currently suffering an epidemic or about to suffer one. It’s obvious that these numbers are not constant: They must be low at the start of an epidemic and must increase over time with the number of active cases.

I will show why that means there is no single “safe amount of time” to wait when evaluating whether a surface is safe to handle or that a place has safe air to breathe.

Numbers of virus bodies (virions, whole individual virus particles)

With influenza, around 50,000 virus bodies are emitted per cough (median) by a person shedding the virus, but the number is commonly hundreds of thousands of virus bodies, and changes over the course of the infection.1

In SARS-CoV-1, the virus that led to a pandemic in 2002, the concentration of virus bodies in mucus was observed to be very high. (Note: This is not the same as number of virus bodies emitted in that mucus in a single cough.)

Up to 100 million genomes per mL are found in nasopharyngeal secretions detected in 32% of patients at mean 3.2 days after onset of illness and in 68% at Day 14.

— https://www.aabb.org/tm/eid/Documents/150s.pdf

It is possible that SARS-CoV-2 shedding could happen at a much higher rate and/or at higher concentrations than with SARS-CoV-1. Research is ongoing. The letter noted this.

We found that the stability of SARS-CoV-2 was similar to that of SARS-CoV-1 under the experimental circumstances tested. This indicates that differences in the epidemiologic characteristics of these viruses probably arise from other factors, including high viral loads in the upper respiratory tract and the potential for persons infected with SARS-CoV-2 to shed and transmit the virus while asymptomatic.2

We don’t know yet how many virus bodies are likely to be present on surfaces that we are likely to bring into our homes, which is a crucial detail in knowing how long it takes surfaces to be safe after contamination by an infected person. The same goes for the viability of SARS-CoV-2 in the air.

How the news media seem to be getting it wrong

The news media seem to be using the following model of thinking:

  • The question we want to know the answer to: “Exactly long does the coronavirus live on surfaces?”
  • The conclusions we drew from the letter:
    • On hard surfaces, “Lasts 7 days”
    • Porous surfaces, “Lasts about a day”
    • Hands, fabrics, “Lasts a few hours”
    • Air, “Up to 3 hours”

This is the wrong model to use. They are not used to thinking about things that are exponential, like exponential decay. Thinking along these lines will mislead people into believing that it is safe to handle objects, or to enter spaces, when too little time has passed since the space or object was in contact with another person.

Better model: Half-lives/decay

The NEJM letter appears to me to be the eventual source of the bad conclusions being reported in many of the news articles. The letter, which was summarizing a lot of research, said the following statement of fact:

SARS-CoV-2 was more stable on plastic and stainless steel than on copper and cardboard, and viable virus was detected up to 72 hours after application to these surfaces …

In other words, no virus bodies were detectable on the surfaces after 72 hours. Other similar factual statements about detectable concentrations of virus bodies were also made in the letter. I assume the facts are accurate. However, no conclusion can be drawn from these facts alone about the safety of objects or the air after a certain amount of time.

The “not detectable after 72 hours” result is an artifact of the limited number of virus bodies used in testing

According to the NEJM letter, the observed half-life of the SARS-CoV-2 virus is approximately 6 hours on hard surfaces. The measurement of half-lives is really the only result the letter is reporting.

However, it’s trivial to see that a simple testing artifact in the research study is what is leading to the incorrect conclusion by the news media that “the virus can ‘live’ for up to 72 hours.”

Illustration of the principle of half-lives/decay:

For the purposes of illustration, we will imagine a hypothetical scenario where a person has purchased an item at the grocery store that has been contaminated with the coronavirus, and that there are 1,000,000 SARS-CoV-2 virus bodies on the surface of the item. Is that a realistic number? I am only a layman, but bearing in mind the extremely high concentration of virus bodies observed in the mucus of patients in the studies linked above, it is not hard for me to imagine this number of virus bodies being ejected in a single cough. 1 mL of phlegm contained 100,000,000 SARS-CoV-1 virus bodies.3

If left to sit on the surface, the number of viable virus bodies will begin to “decay” exponentially, as stated in the study.

For background, this happens because there are always low concentrations of molecules in the air that can react with the surface proteins that comprise the outer layer of the viral envelope (the shell of the virus particle). As they react with the envelope, they may damage it in such a way that the virus can no longer stay intact or can no longer attach to the surface of a respiratory cell, which means it is no longer viable. When there are many virus bodies on a surface (e.g., 1 million of them), there will be many of these reactions happening that will lead to many “dead” virus bodies in a given amount of time.

As the number of virus bodies decreases, there are fewer viable virus bodies remaining; therefore, the count of “deaths” of viral bodies, which is the absolute number of virus bodies made unviable per hour, decreases as well. That is why the decay is exponential: There is a very long tail of these deaths until there are zero viable virus bodies remaining. (I don’t know the data on how many virus bodies it takes on a surface to be considered extremely dangerous versus practically safe.)

In the case where we have 10,000 virus bodies (as in the study4) and a half-life of 6 hours, the decay will look approximately like this:

Half-lives elapsed Hours elapsed Viable virus bodies remaining
0 0 10000.00
1 6 5000.00
2 12 2500.00
3 18 1250.00
4 24 625.00
5 30 312.50
6 36 156.25
7 42 78.12
8 48 39.06
9 54 19.53
10 60 9.77
11 66 4.88
12 72 2.44
13 78 1.22
14 84 0.61

As you can see, after the 14th half-life has elapsed, which is after about 80 hours, there would be fewer than 1.0 viable virus bodies. In other words, the virus would not be detectable at all. This is approximately the result found in the research discussed in the NEJM letter.

However, in the case where we start with 1,000,000 virus bodies on the surface (as in our hypothetical scenario with the shopper) and the same half-life of 6 hours, the decay will look approximately like this:

Half-lives elapsed Hours elapsed Viable virus bodies remaining
0 0 1000000.00
1 6 500000.00
2 12 250000.00
3 18 125000.00
4 24 62500.00
5 30 31250.00
6 36 15625.00
7 42 7812.50
8 48 3906.25
9 54 1953.12
10 60 976.56
11 66 488.28
12 72 244.14
13 78 122.07
14 84 61.04
15 90 30.52
16 96 15.26
17 102 7.63
18 108 3.81
19 114 1.91
20 120 0.95

As you can see, it takes 20 generations, or about 120 hours, to reach the same number of viable virus bodies. If you are a member of the public who thinks an item brought into your home will be safe to touch after it has rested for 72 hours, which is what the news media are generally reporting, then you are going to be touching a surface with ~240 viable virus bodies still on it.

Here are the trivial calculations for different numbers of starting virus bodies on the hard surface.

Virus bodies Half-lives until no viable virus bodies Hours Days
10,000 14 half-lives 80 hours 3.3 days
100,000 17 half-lives 100 hours 4.2 days
1,000,000 20 half-lives 120 hours 5.0 days
10,000,000 24 half-lives 140 hours 5.8 days
100,000,000 27 half-lives 160 hours 6.7 days
1,000,000,000 30 half-lives 180 hours 7.5 days

As you can see, it all depends on how many viruses are on the surface to begin with.5

Therefore, any conclusion drawn from the NEJM letter, or any announcement to the public about how to act in the real world, clearly must take into consideration whether the concentration of virus bodies in the real world is like the concentration used in the study. I believe there is enough evidence to cast doubt on this. Therefore, I believe the news media generally are reporting a misleading conclusion.

Ad absurdum simplification

Imagine if the authors of the study had started with 1 coronavirus body on the hard surface. In that case, they might have published the factual statement, “There was no detectable level of virus bodies after 6 hours.” That would be a correct statement, but it would not be correct for the news media to report, “The virus only lasts on hard surfaces for 6 hours.”

Virus bodies in the air, the same problem

In the news, I also see the claim, “The virus stays in the air up to 3 hours.” However, the NEJM letter said only the following:

SARS-CoV-2 remained viable in aerosols throughout the duration of our experiment (3 hours), with a reduction in infectious titer from 103.5 to 102.7 TCID50 per liter of air.6 7

The statement is very different from saying the virus bodies stay viable for only 3 hours in the air. They say only that there was a reduction in the concentration of viable virus bodies in the air by a factor of 100.8, or about 6.3, so there was an 84% reduction after 3 hours. In other words, the SARS-CoV-2 virus has a shorter half-life in air than on surfaces. They did not say that they used concentrations of the virus bodies in their study that were likely to be similar to what will be encountered commonly during the pandemic.

Summary

It seams clear to me, a layman with no virtually no medical understanding, that under real-world conditions (in shops, restaurants, homes, and so on), SARS-CoV-2 virus bodies could be viable on surfaces and in the air much longer than the news media are leading people to believe.

It is obvious that as the number of virus bodies on real-world surfaces increases, the length of time that the average surface remains dangerous must also increase. I do not believe this is being represented broadly by the news media.

Update

March 28, 2020

The virus degrades outside a host because of exposure to moisture and sunlight, or from drying out. But a study published in the New England Journal of Medicine showed that in pristine laboratory conditions, some SARS-CoV-2 particles can remain potentially viable on metal or plastic for up to three days. – Washington Post

The misrepresentation of the NEJM letter is continuing to spread, with many new articles making the “viable on hard surfaces up to three days” claim every day. It is now clear that the NEJM letter is the eventual source in the majority of the reporting.

I have emailed some of the journalists producing the articles, and I have also contacted some of the academic interviewees. I used the university email address that I still have from the days of working as a programmer in a genomics lab, hoping they will at least open my email and consider looking into the matter.

Though the primary mode of transmission of SARS-CoV-2 is not via touching contaminated surfaces, it is still a known mode of transmission. Therefore, if the reasoning in this blog post is correct, I would like this information to spread in the media instead of the “hard surfaces safe after three days” claim. Slowing the rate of spread, even by a small amount, will save many lives later on. (I can imagine many households leaving their shopping in the cold garage for 3-4 days thinking the packages will be free of viable virus bodies after that amount of time. In fact, they might only have preserved a contagious dose.)

Notes

I became aware of all this by reading the NEJM letter after Dr. John Campbell discussed it. He didn’t notice the problem I’m describing in this blog post. Dr. Campbell is my favorite source for updates on the coronavirus.

Disclaimer / About

I have no formal qualifications relevant to this subject. My college job was writing supporting software and software libraries for DNA and RNA sequencing in my university’s genomics/bioinformatics lab. I aided scientists sequencing tobacco mosaic virus, an RNA virus, and other viruses. However, I have only an extremely basic and out-of-date education on viruses, chemistry, and so on. I have some experience reasoning about complex systems including pandemics, but not at a professional level.8

Contact: My email is hackeur.life at the domain ProtonMail.com.

  1. 50,000 is only the median. The max was hundreds of thousands. In the study, the standard deviation was twice the median, which indicates it’s a dataset where some of the coughs had huge numbers of virus bodies, while some had very little. The number of virus bodies shed by a person changes over the course of the infection. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4676262/ ↩

  2. https://www.nejm.org/doi/full/10.1056/NEJMc2001737. Also, the original letter that is the focus of this blog post also mentioned it. ↩

  3. It contained up to 100 million SARS-CoV-1 genomes. I am assuming this meant there were approximately the same number of virus bodies since each viable virus body contains one complete genome, but this is not necessarily the case. ↩

  4. We don’t know why they used only 10,000 virus bodies. It could be because they don’t have sufficient sample stock to replicate a real-world count. I don’t know how much supply there is of these research samples. ↩

  5. As far as the larger numbers which might seem improbable: Keep in mind that having 10 packages with an average of 1,000,000 virus bodies each, and then handling all 10 of these packages, would be equivalent to having one package with 10,000,000 virus bodies and then handling just that one package. The probability of coming into contact with an intact virus body is a function of the total starting number of virus bodies on all the surfaces to be handled. In other words, 10 packages with 1,000,000 virus bodies each would not collectively be safe to handle after only 20 half-lives or 120 hours; you would have to wait 24 half-lives or 140 hours for the same level of safety. If a grocery shopper brings back a large number of mildly contaminated items, that is equivalent to some lower number of highly contaminated items. ↩

  6. https://www.nejm.org/doi/full/10.1056/NEJMc2004973 ↩

  7. TCID50 signifies the concentration at which 50% of the cells are infected when a test tube or well plate upon which cells have been cultured is inoculated with a diluted solution of viral fluid. — https://www.zeomic.co.jp/en/glossary/virus/71 ↩

  8. I had an interest in epidemiology when I was a kid. I wrote low-fidelity computer models of viral spread as a complex system. (A complex system has feedback between the components in the model.) Models of complex systems like pandemics are not predictive, but they can educate. I learned about exponential growth, exponential decay, thinking in rates and capacities, and so on. Further education, some formal but mostly informal, increased my understanding of complex systems, risk management, and other topics relevant to reasoning about pandemics. Avocationally I have made computer models of complex systems in multiple domains for a couple of decades. I wrote this blog post because I understand that small changes in initial conditions can have a domino effect later on, under conditions of exponential growth. If a small number of people mistakenly handle unsafe surfaces now, it will lead to illness and death for many people later. ↩