The rapid emergence of COVID-19 has created tremendous uncertainty in medicine. We don’t know where this pandemic is headed. We don’t know the ideal management strategy. Every day brings conflicting information. Emergency medicine is a field that embraces (or at least tolerates) uncertainty, but knowledge is an important pillar of our sense of control in medicine, and COVID-19 is doing a good job highlighting massive gaps in our knowledge. One of those gaps is the precise mechanisms through which infectious diseases spread and how best to protect ourselves. We hear terms like “aerosol generating” and “droplets”, but their precise meaning can be unclear, and so it is hard to know how to adjust our practice. In this post, I will review everything I have been able to learn about aerosols and droplets, how they spread, and how they should impact our practice.
I will start with a major caveat: despite reading hundreds of papers on this topic, I still have a lot of uncertainty. I think that uncertainty is born from uncertainty in the literature. There was debate and conflicting information with every new paper I found. However, it is also important to recognize that I am an emergency physician attempting to distill in a few weeks topics that people have dedicated entire careers to. If you think I missed something, or want to add to the discussion, please do so below.
I also want to acknowledge that these are incredibly trying times. We are all anxious, and that anxiety is made worse by the conflicting information that we are receiving. There is a risk that by adding even more potentially conflicting information I might add to that anxiety. I think science is fundamentally important. I think this information is important. How we act on this information is equally important. Remember that nothing here is definitive. In already trying times, we don’t want to create conflict with our colleagues. Try to use any information available to work collaboratively, focusing not on the negatives of uncertainty and disagreement, but on the positives of growth and a common goal of safety for all healthcare workers and our patients. For the most part, I am reassured by what I read, and will continue to work hard to use this information to keep my entire team safe.
What exactly is an aerosol?
I have to say, I didn’t expect this to be such a complicated question to answer. There is actually a pretty heated academic debate, centering around desiccation rates and the formulas for turbulent flow, such that it seems that no one really agrees on an exact definition. You will see some pretty definitive definitions in some sources, but that definition will invariably be refuted in the next paper you encounter. In general, aerosols are liquid or solid particles suspended in air. (Tellier 2009; Judson 2019) They can be visible, like fog, but are most often invisible, like dust or pollen.
They are often divided into small droplets (and many, but not all, people reserve the term “aerosol” to refer only to these small droplets) and large droplets. Large droplets drop to the ground before they evaporate, causing local contamination. Disease transmission through these large droplets is what we often refer to as “droplet/contact spread”, where disease transmission occurs because you touch a surface contaminated by these droplets, or get caught within the spray zone when the patient is coughing. Aerosols are so small that buoyant forces overcome gravity, allowing them to say suspended in the air for long periods, or they evaporate before they hit the floor, leaving the solid particulate (“droplet nuclei”) free to float very long distances, causing what we often refer to as “airborne” transmission. (Nicas 2005; Judson 2019)
Respiratory aerosols are created when air passes over a layer of fluid. (Fiegel 2006; Morawska 2006) There are a large number of factors that can alter this process. The viscosity of the fluid layer is an important determinant of aerosol generation, and could be a very important practical consideration in medicine. Increases in surfactant increase overall droplet formation, and produce smaller droplets (which will travel farther). (Fiegel 2006) This could be an important consideration, as some people are discussing the use of surfactant to manage COVID-19 lung disease. Conversely, nebulized saline has been shown to decrease the number of bio-aerosols produced, and has been suggested as a possible (but unproven) infection control strategy. (Fiegel 2006)
In the world of aerosols, there seems to be two main points of contention. The first is the size cutoff between large and small droplets. Various sources will put the cutoff at 2 µm, 5 µm, 10 µm, 20 µm, or even 100 µm. (Judson 2019; Morawska 2006; Fiegel 2006; Xie 2007; Chen 2010; Nicas 2005; Tellier 2009) This is a key distinction, because it is the difference between airborne and droplet precautions. Many papers make definitive statements based on one of the cutoffs that would be incorrect if a different cutoff was used. (For example, Morawska 2006 states that droplets smaller than 100 µm, which is almost all droplets, will evaporate before hitting the floor, meaning that they can transmit disease through the airborne route, while other documents will use 5 µm as the cutoff.) There is probably a grey area in which droplets can behave either way, depending on how quickly they evaporate compared to how quickly they fall to the ground based on the atmospheric conditions of the room.
The second main point of contention is exactly how clean the distinction between airborne and droplet transmission is. Some sources treat this as black and white, but others point out that large droplets evaporate and become smaller, and most activities create a very large variety of sizes, so it is more like a spectrum than a dichotomous distinction. A lot of epidemiologic studies will make strong claims that a disease is only spread by close contact, but we have to remember, those studies cannot possibly distinguish between short distance aerosol transmission (I caught it while breathing a few feet away from you) and contact transmission (I touched the door handle and then rubbed my eye.) Too often, if you were close together, studies will just assume it was contact instead of aerosol spread, biasing the literature in that direction.
What are aerosol generating procedures?
An aerosol generating procedure is a medical procedure that creates aerosols in addition to those that the patient creates regularly from breathing, coughing, sneezing, and talking. (Judson 2019) In other words, it is important to remember that patients will create their own aerosols even when we are not performing these procedures. Aerosol generating procedures can produce both large and small droplets. Each procedure will be unique, so they really need to be considered independently. (Judson 2019) Importantly, aerosol generating procedures can cause transmission through pathways that microbes don’t usually use (a virus normally spread through contact or droplets can become airborne). Procedures can either generate aerosols directly or by inducing the patient to cough or sneeze, a distinction that may be important when trying to mitigate risk. (Judson 2019)
Although respiratory infections are the primary source of aerosols, they are created in other ways as well. Surgery can aerosolize pathogens found in the blood or tissues. (For example, HIV was found in aerosols created by surgical power tools.) (Judson 2019) Aerosols can also be produced by seemingly mundane things, such as fast running tap water and flushing toilets. (Morawska 2006)
Because the individual risks (and benefits) of each procedure is likely unique, I will consider them each independently. For the sake of space, I have done so in a second post that accompanies this one.
Aerosols and normal activities
Throughout our preparations for COVID-19, we have all been incredibly focused on aerosol generating procedures, but it is important to understand that aerosols are also produced through normal human activities, including simply breathing. (Tellier 2009; Asadi 2019) Essentially any air passing through the respiratory tract will create droplets. The clinical significance will depend on the number of droplets produced, their size, the concentration of infectious agents, the frequency with which the activity is performed, and the PPE used by staff. (Morawska 2006) For example, although a single cough produces far more droplets (of all sizes) than a single breath, breathing occurs much more frequently, and so may be responsible for more droplet production overall. (Morawska 2006; Fiegel 2006) It is also important to understand that although the majority of the droplets produced by a cough may be small enough to stay airborne, their small size means that collectively they add up to only a tiny fraction of the volume produced (perhaps less than 0.1%), and therefore only a tiny fraction of the total virus spread. (Nicas 2005) However, despite carrying smaller numbers of microorganisms, there is evidence that smaller droplets don’t need to contain as many microorganisms as larger droplets to cause a clinical infection (by several orders of magnitude). (Nicas 2005; Tellier 2009) Furthermore, we must remember that not every droplet will contain virus, and even if it does, it may not be enough to effectively transmit disease.
Table adapted from Morawska 2006, with similar numbers reported in the Fiegel 2006 review:
|Activity||Number of droplets produced||Small (1-2 um) aerosols?|
|Normal breathing (5 min)||A few||Some|
|Single strong nasal exhalation||Few to a few hundred||Some|
|Counting out loud (talking)||Few dozen to few hundred. Some sources say a few thousand (Xie 2007)||Mostly|
|Cough||Few hundred to many thousand||Mostly|
|Sneeze||Few hundred thousand to a few million||Mostly|
If you want a more specific breakdown, you can look at table 2 from Nicas 2005, but these numbers are estimates, and you will see different numbers even in this same paper:
Older studies concluded that humans primarily produce large droplets, but they were significantly limited because their instruments were insensitive to smaller sizes. (Morawska 2006) Recent research has indicated that as many as 80-90% of the particles generated by human exhalation are smaller than 1 µm in size. (Papineni 1997) Although the exact size of droplets produced is still debated, most sources agree that speaking, coughing, and sneezing produce droplets that are sufficiently small to remain airborne. (Fiegel 2006; Chen 2010)
Interestingly, the total amount of bioaerosols produced varies tremendously among individuals, with some people creating very few, and others acting as “super producers”. (Fiegel 2006) I wonder whether this explains why we have observed super-spreaders of SARS and COVID-19, as “it appears that a minor percentage of the population will be responsible for disseminating the majority of exhaled bioaerosol”. (Fiegel 2006)
Vomiting, in which humans can shed up to a million virus particles per milliliter of vomit, can also produce aerosols. (Morawska 2006) A vomiting SARs patient was associated with nosocomial spread in a hospital in Hong Kong, although it isn’t clear by what route (contact, droplet, or airborne) the transmission occurred. (Morawska 2006) Similarly, there can be as many as a hundred million virus particles in every gram of feces, and flush toilets are known to result in aerosolization. As is discussed below, this form of aerosolization is thought to have spread SARS in the Amoy Garden apartment complex in Hong Kong. (Morawska 2006)
However, whether these aerosols are capable of transmitting disease still depends heavily on the number produced, the concentration of the infectious agent, the virulence of the microbe, environmental factors (the virus needs to be able to survive, whether in the air or on a surface, until it enters a host), and the health and immunity of the host. (Morawska 2006) Although it is clear that aerosols are commonly produced, it is also clear that the vast majority of disease transmission occurs among people who are in very close contact and therefore exposed to the largest of the droplets.
The fact that humans constantly produce aerosols is really important when assessing studies of aerosol generating procedures. The result sections of these papers will often only present a change in aerosols from baseline, and frequently our procedures won’t produce more droplets. However, if you look closely, we are already producing a ton of aerosols, and even if the procedures don’t produce more, their ability to spread those aerosols further is a big concern. (Simmonds 2010; Rule 2018)
Update: In one of the more entertaining and yet still scientific tweetorials of all time, Dr. Andy Tagg asks the question, “Is farting an aerosol-generating procedure?”:
What happens to the aerosol after it is expelled?
The only constant I could find in this data was a general lamentation of the lack of experimental data. (Nicas 2005; Xie 2007; Tellier 2009; Judson 2019) Most of the numbers we use clinically are based on mathematical models making large numbers of (potentially faulty) assumptions. Where droplets end up is regulated by a huge number of factors. The primary factor is probably the size of the droplet. A 1000 µm droplet will fall 1 meter in 0.3 seconds. A 100 µm droplet will take 3 second to fall 1 meter. A 10 µm droplet will take 300 seconds, and a 1 µm droplet will take 30,000 seconds. (Morawska 2006) How long a droplet remains in the air is clearly a huge factor in how far it is able to travel, and how likely healthcare workers are to be exposed.
As I said, exact size cutoffs are controversial, but Chen (2010) suggests that the distribution of all droplets between 0.1 and 200 µm will primarily be influenced by ventilation patterns and the initial velocity of the droplet, rather than gravity. In other words, these droplets do not just drop to the ground within 1-2 meters of the patient, as many infection control practices assume. However, the distribution of droplets is also influenced by a very large number of factors, including relative humidity, temperature, ventilation pattern and rate, initial velocity, shape of the human body, and droplet nuclei size and composition. (Xie 2007; Chen2010) Most of these factors are dynamic (droplet size changes as it evaporates and temperature changes as you move away from a febrile patient), making simplified calculations difficult. At smaller sizes, Brownian motion, electrical forces, thermal gradients, and turbulent diffusion have much bigger impacts. (Morawska 2006) Overall, it is complicated, there are lots of formulas, and reading these papers generally left me with a headache.
Many calculations regarding droplet distribution have significant assumptions embedded. For example, initial studies that estimated droplet dispersion made the assumption that the droplets were introduced into the air without any velocity, which is a bad assumption when coughing and sneezing can create tremendous initial particle velocities. (Morawska 2006) As general estimates, particles produced by normal breathing have a velocity of approximately 1 m/sec, talking 5m/sec, coughing 10 m/sec, and sneezing 20-50 m/sec. (Xie 2007) Thus, even though large particles are often assumed to land close to the patient, that assumption is frequently incorrect. (Bourouiba 2020) Think about walking by the sea on a windy day. Large droplets that usually only travel a very short distance can easily reach you a long way from the shore. (Judson 2019)
There are some mathematical models and experimental data that support the 2 meter rule for normal breathing and talking, but most suggest that coughing and sneezing spread droplets far further. (Xie 2007; Hui 2014; Bourouiba 2020) However, this rule only applies to large droplets. Smaller droplets remain trapped in the air and therefore can travel much greater distances. Unfortunately, most of these models ignore the impact of the patient covering their mouth and nose when they sneeze. Hopefully all of these patients are wearing masks while sneezing in the hospital, which will clearly change the distribution of droplets, and makes the 7-8 meter number less likely to be true. (Bourouiba 2020)
So how far do these droplets travel? I think this science makes it clear that there is no simple answer. Small droplets will remain in the air for very long periods of time (become airborne), but the exact cutoff is unknown, and can change significantly based on factors like temperature and humidity. With normal breathing, large droplets mostly fall to the ground within a 2 meter radius, but they can evaporate and become small droplets. (Nicas 2005) Coughing and sneezing can propel these large droplets much further – at least 6 meters or 18 feet. (Bourouiba 2020)
Perhaps the most important thing to remember is that this distribution is probabilistic. (Morawska 2006) There is nothing that guarantees a droplet will stop before a certain distance. When we make statements like, droplets over size X will fall to the floor within distance Y, what we really mean is that most droplets will. Luckily, for most diseases, knowing where most of the droplets end up is probably good enough.
As a quick aside, some sources will state that very small particles are not dangerous, because although you might breath them in, they remain in the air and are not retained in the alveoli. However, it appears this is not true, with 50% of particles smaller than 1 µm being retained in the respiratory tract. (Morawska 2006)
Influenza: droplet or airborne?
We need to think about the practical implications of all this basic science. Nobody is all that worried about influenza right now, but it is important to know a little bit about the transmission of influenza, because almost all of our PPE recommendations are based on extrapolations from this more common disease. Although the science is not definitive, it appears that influenza is transmitted by both large and small droplets (ie, transmission occurs through both droplet and airborne routes). (Judson 2019) However, although airborne transmission is possible, large droplet or contact transmission is probably responsible for the vast majority of disease transmission.
Although some experts seem to doubt that influenza can be spread through small droplets or airborne droplet nuclei, (Brankston 2007) there are multiple lines of evidence that support this hypothesis. Studies in ferrets showed that influenza spread even when the animals were separated by “S” and “U” shaped ducts that would not allow for the passage of larger droplets, indicating “that infection was conveyed either by droplet-nuclei or very fine dust particles.” (Andrewes 1941) Likewise, in studies of mice, lower ventilation rates led to higher transmission rates, leading the authors to conclude spread must be by airborne droplet nuclei. (Shulman 1962) There are numerous other animal studies demonstrating the spread of influenza uphill (against gravity) over distances longer than droplets are supposed to travel, strongly supporting the conclusion of airborne spread. (Tellier 2009)
Influenza RNA has been detected in aerosol particles from infected patients while just breathing comfortably. (Fabian 2008) Influenza has been found in aerosols in random samples of air around an emergency department during flu season. (Blachere 2009) In one study that tested the air in an emergency department during flu season, influenza was identified in 43% of the samples taken. The total amount of virus found on samplers worn by healthcare providers was about twice that in the air, suggesting that the risk is still highest from close contact with patients, but that airborne spread is clearly possible. (Rule 2018) In another study, total viral load in air samples was higher outside patients’ rooms than right next to the patients. (Cummings 2014) On the other hand, Bischoff (2013) did find higher viral loads closer to the patient, but virus was still detected 6 feet or 2 meters away. It has also been proven that humans can develop influenza after breathing air artificially contaminated by the virus. (Francis 1944) (And I just have to point out that Jonas Salk is one of the authors on this paper!) The question is how often this occurs in a real world setting.
The TCID50, or the concentration of virus particles at which 50% of cells become infected, has been estimated to be as low as 3 for some strains of influenza A. (Thompson 2013) Assuming a respiratory rate of 10 L/min, during a 10 minute visit to a patient room a healthcare worker will breath 100 L of air. At baseline, one study found 7,690 viral particles per L of air in an influenza patient’s room. Therefore, in a 10 minute visit, the health care provider would have inhaled 769,000 viral particles, which is clearly above the TCID50. (Thompson 2013) Even studies that find lower viral counts in the air (approx 300 in Rule 2018) will still be well above this threshold.
There is indirect evidence that airborne transmission occurs, as good ventilation and UV irradiation limits influenza transmission. (McLean 1961; Shulman 1962; Drink 1996; Fiegel 2006; Tellier 2009) There are also multiple influenza outbreaks where airborne spread is hypothesized as the best explanation, however it is impossible to know retrospectively. (Moser 1979; Klontz 1989; Davis 2009) There is the interesting case report of an outbreak on a plane in Alaska in 1977. Because of a mechanical failure, there was a 4.5 hour delay with the ventilation turned off. 1 person was sick at the time of the flight, but over the next 3 days 38 others (72% of the people on the plane) developed influenza A. The size of the plane seems to make airborne transmission much more likely than large droplets, which we are told should not spread beyond 1-2 meters. That being said, if the index patient was the first person to use the washroom, the outbreak could easily be explained by droplet spread. (Moser 1979)
I think the data is pretty clear that influenza can spread through airborne aerosols, and I find that fact reassuring in the era of COVID-19. We generally treat influenza in “droplet precautions”, and transmission to healthcare workers is generally low. (Although the analogy fails because vaccination and prior exposure to influenza provide a level of immunity that does not exist with COVID-19). Even when a virus can be spread through airborne transmission, you are still much more likely to become ill as the result of close contact. If COVID-19 is transmitted similarly to influenza, we can be somewhat reassured by our current practices.
Some faulty epidemiologic reasoning
Many people dismiss airborne spread for any disease that doesn’t behave like measles or tuberculosis. They suggest that airborne must equate to long distance transmission. They suggest that a low R0 and the lack of large scale outbreaks proves that airborne spread is impossible. I think that is a mistake.
I think the animal studies above make it pretty clear that influenza can spread by the airborne route. This would provide direct evidence against those epidemiologic arguments. Influenza has a low R0. It occasionally causes large scale outbreaks (some of which have been blamed on airborne transmission), but those outbreaks are rare. It is hard to know exactly how people get sick, but we don’t routinely see influenza spread of long distances. Thus, the epidemiologic argument that “the R0 of this disease is too low for it to be airborne” doesn’t hold water. Any infection that behaves like influenza could easily be spread through aerosols.
Aerosols are tiny and their concentration drops off exponentially as you get farther from the source (especially with good ventilation). Of course we don’t frequently see transmission over large distances or large scale outbreaks. The chances of encountering viable virus across the room are just too lw.
However, that line of reasoning completely discounts close range aerosol spread. Aerosols will be most concentrated within a few meters of the patient. At that distance, aerosol spread is almost indistinguishable from droplet spread. In fact, it is generally completely ignored, because many people just assume that short range spread is due to droplets. However, short range aerosol spread would have significant implications for our PPE choices.
This short range aerosol spread is exactly what we are talking about when we discuss aerosol generating procedures. It isn’t the nurse 3 rooms down that gets sick (usually). It is the people in the room that are exposed to aerosols. However, as we discussed above, normal human activities like talking and coughing produce just as many aerosols as most of our aerosol generating procedures. Therefore, it is important to consider short range aerosol spread in order to appropriate protect ourselves.
Coronaviruses: droplets or airborne?
I don’t think we can say definitively, but it is highly likely that COVID-19 can be spread through airborne aerosols. Both SARS and MERS were thought to spread primarily by large droplets, but there were also outbreaks that were best explained by airborne spread of the disease (and others in which aerosols were thought possible, even if they weren’t highly likely). (Wong 2004; Li 2004; Morawska 2006; Xie 2007; Judson 2019) SARS was contracted by a nurse that never entered a patient’s room, which could be explained by airborne transmission, but could also have been through fomites. (Scales 2003) In a retrospective analysis out of Singapore, insufficient ventilation on hospital wards was one of 5 major factors that increased the risk of transmission of SARS. (Chen 2009) Although the evidence isn’t definitive, the pattern of nosocomial spread followed hospital ventilation patterns, rather than the random distribution you would expect if the infection was spread via surfaces.
An outbreak of SARS that affected more than 300 people across 150 apartments in the Amoy Garden apartment complex in Hong Kong is thought to have resulted from airborne spread of aerosols through the sewer system. Rather than the random distribution that you might expect from contact or droplets spread through common areas, residents on higher floors were more likely to be infected, “consistent with a plume of contaminated warm air”. (Morawska 2006)
It is also pretty clear that these viruses have been spread as the result of aerosol generating procedures. There was a strong association between multiple aerosol generating procedures and transmission of SARS to healthcare workers. (Tran 2012)
For COVID-19, the virus has been found in the air more than 6 feet away from the patient, in ventilation systems, and even in the air in hallways outside patients’ rooms, indicating the potential for airborne spread. (Santarpia 2020 preprint data; Ong 2020, Liu 2020) Guo (2020) found virus RNA in the air up to 4 meters from the patient. However, the presence of RNA doesn’t mean that there is viable virus, nor that it was present in large enough numbers to cause clinical infection. That being said, if COVID is aerosolized in large volumes, it is likely that the virus remains viable for at least 3-5 hours. (Van Doremalen 2020)
There is definitely debate on this point. Some organizations have stated very strongly that SARS-CoV-2 is not spread by the airborne route. However, others say the opposite. For example, the CDC guidance for coronaviruses has always been to treat them as airborne, although that is based more on the precautionary principle than hard science (see here and here). Furthermore, the National Academies of Sciences, Engineering, and Medicine states that “while the current SARS-CoV-2 specific research is limited, the results of available studies are consistent with aerosolization of virus from normal breathing.” (Fineberg 2020)
At this point, I think the only safe conclusion is that airborne transmission is possible. However, that doesn’t make it likely. Because of their larger size, large droplets contain as much as 99.9% of viral particles exhaled. Although aerosols may carry small amounts of virus, they become very diffuse the further you are from the patient and are effectively managed by modern ventilation systems. I don’t think we should be making black and white statements. We need to consider the potential for aerosol spread, and how that might impact our PPE practices, while simultaneously recognizing that droplets and close contact with patients represent a far greater risk.
Update: It is becoming increasingly clear that there is significant pre-symptomatic spread of COVID-19. When discussing this issue on the Saint Emlyn’s podcast, professor of medical virology Pam Vallely made the interesting point that presymptomatic spread is further evidence of aerosol spread, as large droplets aren’t formed in any volume until you are symptomatic.
A few infection control notes
One of the most important aspects of managing bioaerosols is good ventilation. (Fiegel 2006) In ideal circumstances, 65% of all airborne droplets can be removed with each air exchange, although because air doesn’t mix perfectly, the number is probably in the 20-60% range in real life. (Fiegel 2006) In medicine, we are used to thinking in half lives. Each air exchange might take away half of the aerosols in a room, and therefore, if you can determine the air exchange rate for your facility, you can estimate the half life of aerosols, and use that to make PPE and clinical decisions.
You can also disinfect air using a number of different systems, such as HEPA filters and UV light. (Fiegel 2006) I wonder if anyone is using portable HEPA air purifiers in hallways to limit airborne spread of COVID? And of course, the most important mechanism for managing aerosols is almost certainly PPE, with a properly fit N95 mask being the medical standard, which I will discuss further below. (Fiegel 2006)
Is there evidence for the “2 meter rule”?
There is a widely spread infection control concept that as long as you are 2 meters away from the patient, you are safe from droplets. This claim is usually made without citation, and there is plenty of data to say that it is wrong, at least as a definitive cut off. The idea that all large droplets will fall to the floor within 2 meters seems to have been initially proposed by Wells, based on a very simplistic calculation, with assumptions that have since been questioned, and limited empirical data. (Xie 2007) Unfortunately, as is reviewed above, most of the existing data seems to refute that hypothesis. For example, one recent study had 5 volunteers cough after gargling with food colouring, and there was visible macroscopic contamination beyond 2 meters with 4 of 5 participants. (Loh 2020) Simple pictures of sneezes show a droplet cloud out to 8 meters. (Bourouiba 2020) In another study, in the absence of any aerosol generating procedures, influenza viral loads were actually higher further away from the patient, and they were highest outside the patient’s room. (Cummings 2014)
We should not rely on the 2 meter rule to keep us perfectly safe. That being said, because droplets spread through 3 dimensional space, the concentration of droplets decreases exponentially as you get further from the patient, and there is data that the majority of droplets created from normal breathing fall within 1 meter, although coughing and sneezing increase that distribution significantly.
Overall, the further you are from the patient the safer you are. You are more likely to become contaminated at 50 cm than you are at 1 meter. Your risk is lower again at 2 meters, but it doesn’t drop to zero. You are even safer 4 or 8 meters away from the patient (or even better, behind a closed door).
Practically speaking, this means that you should take off your PPE as far from the patient as possible. In an ideal world, you would always take your PPE off behind a screen or door to completely limit droplet contamination. However, although increasing the distance will decrease your risk from droplets, it actually increases the risk of contact spread. Clearly, you don’t want to bring dirty PPE into clean hallways. The risk of spread through contact with fomites is almost certainly higher than the risk from droplets once you are further than 2 meters away from the patient, which is why the 2 meter rule often works practically, even though it is not scientifically accurate.
What this means clinically will depend a great deal on the layout of your own space. If there is an empty anteroom to doff in, that makes the most sense. If you can doff behind a curtain, that would be great. Otherwise, realize that your risk is very small beyond that 2 meter mark as long as the patient is wearing a mask, not actively coughing or sneezing, and there is not an ongoing aerosol generating procedure.
Update: A new systematic review examined this topic, and 8 of the 10 studies included demonstrated droplet spread beyond 2 meters. The authors state, “although the studies employed very different methodologies and should be interpreted cautiously, they still confirm that the spatial separation limit of 1 m (≈3 ft) prescribed for droplet precautions, and associated recommendations for staff at ports of entry , are not based on current scientific evidence.” (Bahl 2020)
Comparing N95 and surgical masks
We actually use masks for 2 different purposes and it results in a common misunderstanding among the lay public. We can use masks to keep droplets out of the respiratory tract (to keep ourselves safe), but we can also use masks to keep droplets in (to keep others safe). The surgical mask is significantly less effective than the N95 at keeping particles out, but it is very good at keeping particles in. (It’s primary purpose is to protect patients from the surgeon during surgery). With coughing and sneezing patients in the hospital, one of the most important infection control strategies is placing surgical masks on the patients to limit the number of droplets then make it to the environment. (This is also one of the reasons that “aerosol generating procedures” are higher risk – because we generally need to remove that primary layer of protection.) Similarly, when hospitals are asking employees to wear surgical masks at all times right now, it is not a strategy designed primarily to protect employees, but to limit unintentional spread from asymptomatic or minimally symptomatic individuals.
There are a few studies that compare N95 and surgical masks in healthcare workers. (Loeb 2009; MacIntyre 2013; Smith 2016; Radonovich 2019) All were looking at influenza, so will only extrapolate to COVID-19 if the mechanism of spread is the same, which might not be a good assumption. Similarly, all studies will be impacted by the rate of compliance with the mask (as well as things like hand hygiene). In general, compliance is lower with the less comfortable N95 masks, so the studies may be biased to show no difference, even if there is a difference with perfect mask use. Studies done in an outpatient setting may not extrapolate well to critical care. Finally, although the studies contain what look like large numbers of people, the power of the studies comes from the event rate (or the number of people who get sick despite wearing a mask), which is much lower, and so the confidence intervals are very large. A systematic review and meta-analysis on this topic found no statistical differences, but the point estimates are all on the side of N95s being better, and the confidence intervals are huge. (Long 2020) Therefore, I don’t think it is fair to conclude that the two masks are equivalent, but just that we don’t know. It is hard to know what to do with that information. In an ideal world, I think using N95s as the standard until surgical masks were proven to be non-inferior makes the most sense, but that only works if we have an adequate supply of N95s to use for all COVID-19 encounters.
This is a lot of information, and unfortunately it does not allow for any black and white conclusions. There is pretty wide consensus that the science surrounding aerosol transmission of disease is severely lacking. (Morawska 2006; Chen 2010; Judson 2019) There are more questions than there are answers. We should avoid making definitive statements, and instead discuss the uncertainties and the trade-offs between alternating risks. Overall, given the lack of solid evidence, it is generally recommended that we rely on the “precautionary principle”. (Judson 2019) In other words, we should not be looking for evidence that a practice is harmful before avoiding it, but should instead be looking for evidence that a practice is safe before adopting it.
Based on this data, it doesn’t make sense to dichotomize into just airborne aerosols and localized droplets. It is clearly far more complicated than that, with larger droplets becoming smaller as they evaporate, and plenty of evidence that virus can be found further from patients than our current models predict. This literature also makes it clear that almost every activity, including normal breathing, can create aerosols. However, the risk from those aerosols is far lower than the risk of droplets and close contact with the patient.
Perhaps most importantly, we need to move beyond black and white statements and think in terms of probabilities. This data disproves definitive rules, such as “you are perfectly safe when 2 meters away from the patient”. Instead, it tells us that the chance of infection is much higher when close to the patient and much lower as we get further away.
We need to distinguish between possible and likely. Airborne transmission of influenza and COVID-19 is clearly possible, and probably does occur occasionally, but we also have to acknowledge that it is very rare. How can we know that it is rare? If airborne transmission was highly likely, we would see much bigger outbreaks. There have been thousands of COVID patients managed in normal hospital rooms, or behind curtains, and not every person in that department gets sick. Similarly, there have been many sick COVID patients on planes, and although there is some transmission, most people are fine. I think that message is key. Airborne spread is possible, but if there was a COVID patient coughing without a mask on the other side of the department, it is still very unlikely you will actually catch the disease, even if you aren’t wearing PPE. You are much more likely to catch it from droplets or close contact, which is why infection control practices are so focused on those activities.
If you are a numbers person, people have done some calculations. There are a lot of assumptions, but the best estimates are that if you spend 15 minutes in a room with a coughing patient, your chance of catching influenza from large droplets and self inoculation (contact) is about the same (and not very high). Transmission through airborne aerosols is about 100 to 1000 times less likely than the other two routes (Telllier 2009)
That is reassuring. However, it is a mistake to take that line of reasoning too far. Just because droplet spread is more likely, doesn’t mean that airborne spread should be ignored. It is wrong to categorically say that COVID does not spread through an airborne route.
The focus should be on droplet and contact spread. At the same time, we should not ignore the airborne route. The above science review gives hints at activities that could increase airborne transmission: sicker patients with higher viral loads, more coughing and sneezing, higher respiratory rates, longer times spent with the patient, and of course aerosol generating procedures. In these cases, the risk of airborne transmission increases, and we should consider adding airborne precautions to our standard of contact and droplet PPE.
Unfortunately, at this point I am not sure we understand the concept of “super spreaders” well enough. It is clear they exist. I think the above review explains their existence: some people produce far more aerosols than others, and if those same patients have high viral loads, you have an infection control problem. The issue is we don’t know how to identify these patients, nor do we know how to appropriately account for them in our current infection control practices. In my mind, their existence is probably the best argument for universal mask use during an outbreak.
I have said many times that the goal of every hospital during COVID-19 should be to ensure that ZERO healthcare workers become infected in the course of their normal duties, while still providing exemplary care to all of our patients. However, that goal may not be perfectly attainable. There are always tradeoffs between risks. Removing your PPE further from the patient may limit your exposure to droplets, but increases potential exposure to fomites as we carry dirty PPE further from the source. The small risk of airborne transmission might suggest increased use of N95 masks, but if we use our equipment in low risk scenarios it might not be available for us in higher risk encounters. You might suggest wearing an N95 at all times, but we have already seen providers suffering from skin breakdown and other complications.
There are no easy answers. If N95s were plentiful, I think if you make sense to wear them for all encounters with suspected COVID patients. However, that is not the world we live in, and although airborne spread is technically possible, it is incredibly unlikely. I am happy to wear a surgical mask when assessing the average patient with respiratory complaints. I think that will keep me close to 100% protected. But as the patient gets sicker and the risk of aerosols increases (either through procedures or natural activities like coughing), I will switch to an N95.
Other FOAMed and Reporting
You can watch aerosol being created by talking in this Japanese news program:
Andrewes CH and Glover RE. Spread of Infection from the Respiratory Tract of the Ferret. I. Transmission of Influenza A Virus. Br J Exp Pathol. 1941 Apr; 22(2): 91–97 PMC2065394
Asadi S, Wexler AS, Cappa CD, Barreda S, Bouvier NM, Ristenpart WD. Aerosol emission and superemission during human speech increase with voice loudness. Sci Rep. 2019;9(1):2348. Published 2019 Feb 20. doi:10.1038/s41598-019-38808-z PMID: 30787335
Bahl P, Doolan C, de Silva C, Chughtai AA, Bourouiba L, MacIntyre CR. Airborne or droplet precautions for health workers treating COVID-19? [published online ahead of print, 2020 Apr 16]. J Infect Dis. 2020;jiaa189. doi:10.1093/infdis/jiaa189 PMID: 32301491 [article]
Bischoff WE, Swett K, Leng I, Peters TR. Exposure to influenza virus aerosols during routine patient care. J Infect Dis. 2013;207(7):1037–1046. doi:10.1093/infdis/jis773 PMID: 23372182
Blachere FM, Lindsley WG, Pearce TA, et al. Measurement of airborne influenza virus in a hospital emergency department. Clin Infect Dis. 2009;48(4):438–440. doi:10.1086/596478 PMID: 19133798
Bourouiba L. Turbulent Gas Clouds and Respiratory Pathogen Emissions: Potential Implications for Reducing Transmission of COVID-19 [published online ahead of print, 2020 Mar 26]. JAMA. 2020;10.1001/jama.2020.4756. doi:10.1001/jama.2020.4756 PMID: 32215590
Brankston G, Gitterman L, Hirji Z, Lemieux C, Gardam M. Transmission of influenza A in human beings. Lancet Infect Dis. 2007;7(4):257–265. doi:10.1016/S1473-3099(07)70029-4 PMID: 17376383
Chen WQ, Ling WH, Lu CY, et al. Which preventive measures might protect health care workers from SARS?. BMC Public Health. 2009;9:81. Published 2009 Mar 13. doi:10.1186/1471-2458-9-81 PMID: 19284644
Chen C, Zhao B. Some questions on dispersion of human exhaled droplets in ventilation room: answers from numerical investigation. Indoor Air. 2010;20(2):95–111. doi:10.1111/j.1600-0668.2009.00626.x PMID: 20002792
Cummings KJ, Martin SB Jr, Lindsley WG, et al. Exposure to influenza virus aerosols in the hospital setting: is routine patient care an aerosol generating procedure?. J Infect Dis. 2014;210(3):504–505. doi:10.1093/infdis/jiu127 PMID: 24596280
Davis J, Garner MG, East IJ. Analysis of local spread of equine influenza in the Park Ridge region of Queensland. Transbound Emerg Dis. 2009;56(1-2):31–38. doi:10.1111/j.1865-1682.2008.01060.x PMID: 19200296
Drinka PJ, Krause P, Schilling M, Miller BA, Shult P, Gravenstein S. Report of an outbreak: nursing home architecture and influenza-A attack rates. J Am Geriatr Soc. 1996;44(8):910–913. doi:10.1111/j.1532-5415.1996.tb01859.x PMID: 8708299
Fabian P, McDevitt JJ, DeHaan WH, et al. Influenza virus in human exhaled breath: an observational study. PLoS One. 2008;3(7):e2691. Published 2008 Jul 16. doi:10.1371/journal.pone.0002691 PMID: 18628983
Fiegel J, Clarke R, Edwards DA. Airborne infectious disease and the suppression of pulmonary bioaerosols. Drug Discov Today. 2006;11(1-2):51–57. doi:10.1016/S1359-6446(05)03687-1 PMID: 16478691
Fineberg HV. Rapid Expert Consultation on the Possibility of Bioaerosol Spread of SARS-CoV-2 for the COVID-19 Pandemic (April 1, 2020) Washington, D.C.. National Academies Press; 2020. https://doi.org/10.17226/25769
Francis T, Pearson HE, Salk JE, Brown PN. Immunity in Human Subjects Artificially Infected with Influenza Virus, Type B. American journal of public health and the nation’s health. 1944; 34(4):317-34. [pubmed]
Guo ZD, Wang ZY, Zhang SF, et al. Aerosol and Surface Distribution of Severe Acute Respiratory Syndrome Coronavirus 2 in Hospital Wards, Wuhan, China, 2020 [published online ahead of print, 2020 Apr 10]. Emerg Infect Dis. 2020;26(7):10.3201/eid2607.200885. doi:10.3201/eid2607.200885 PMID: 32275497
Hui DS, Chan MT, Chow B. Aerosol dispersion during various respiratory therapies: a risk assessment model of nosocomial infection to health care workers. Hong Kong Med J. 2014;20 Suppl 4:9–13. PMID: 25224111
Judson SD, Munster VJ. Nosocomial Transmission of Emerging Viruses via Aerosol-Generating Medical Procedures. Viruses. 2019;11(10):940. Published 2019 Oct 12. doi:10.3390/v11100940 PMID: 31614743
Li Y, Huang X, Yu ITS, Wong TW, Qian H. Role of air distribution in SARS transmission during the largest nosocomial outbreak in Hong Kong Indoor Air. 2005; 15(2):83-95.
Liu Y, Yu ZN, et al. Aerodynamic Characteristics and RNA Concentration of SARS-CoV-2 Aerosol in Wuhan Hospitals during COVID-19 Outbreak. 2020. Preprint, not peer reviewed, here.
Loeb M, Dafoe N, Mahony J, et al. Surgical mask vs N95 respirator for preventing influenza among health care workers: a randomized trial. JAMA. 2009;302(17):1865–1871. doi:10.1001/jama.2009.1466 PMID: 19797474
Loh NW, Tan Y, Taculod J, et al. The impact of high-flow nasal cannula (HFNC) on coughing distance: implications on its use during the novel coronavirus disease outbreak [published online ahead of print, 2020 Mar 18]. Can J Anaesth. 2020;1–2. doi:10.1007/s12630-020-01634-3 PMID: 32189218
Long Y, Hu T, Liu L, et al. Effectiveness of N95 respirators versus surgical masks against influenza: A systematic review and meta‐analysis J Evid Based Med.. 2020
MacIntyre CR, Wang Q, Seale H, et al. A randomized clinical trial of three options for N95 respirators and medical masks in health workers. Am J Respir Crit CareMed. 2013;187(9):960-966.
McLean RL. The eff ect of ultraviolet radiation upon the transmission of epidemic infl uenza in long-term hospital patients. Am Rev Respir Dis 1961; 83: 36.
Morawska L. Droplet fate in indoor environments, or can we prevent the spread of infection?. Indoor Air. 2006;16(5):335–347. doi:10.1111/j.1600-0668.2006.00432.x PMID: 16948710
Moser MR, Bender TR, Margolis HS, Noble GR, Kendal AP, Ritter DG. An outbreak of influenza aboard a commercial airliner. Am J Epidemiol. 1979;110(1):1–6. doi:10.1093/oxfordjournals.aje.a112781 PMID: 463858
Nicas M, Nazaroff WW, Hubbard A. Toward understanding the risk of secondary airborne infection: emission of respirable pathogens. J Occup Environ Hyg. 2005;2(3):143–154. doi:10.1080/15459620590918466 PMID: 15764538
Ong SWX, Tan YK, Chia PY, et al. Air, Surface Environmental, and Personal Protective Equipment Contamination by Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) From a Symptomatic Patient JAMA. 2020
Papineni RS, Rosenthal FS. The size distribution of droplets in the exhaled breath of healthy human subjects. J Aerosol Med. 1997;10(2):105–116. doi:10.1089/jam.1997.10.105 PMID: 10168531
Radonovich LJ Jr, Simberkoff MS, Bessesen MT, et al. N95 Respirators vs Medical Masks for Preventing Influenza Among Health Care Personnel: A Randomized Clinical Trial. JAMA. 2019;322(9):824–833. doi:10.1001/jama.2019.11645 PMID: 31479137
Rule AM, Apau O, Ahrenholz SH, et al. Healthcare personnel exposure in an emergency department during influenza season PLoS ONE. 2018; 13(8):e0203223-. [article]
Santarpia JL, Rivera DN, et al. Transmission Potential of SARS-CoV-2 in Viral Shedding Observed at the University of Nebraska Medical Center. 2020. Preprint here.
Scales DC, Green K, Chan AK, et al. Illness in intensive care staff after brief exposure to severe acute respiratory syndrome. Emerg Infect Dis. 2003;9(10):1205–1210. doi:10.3201/eid0910.030525 PMID: 14609453
SCHULMAN JL, KILBOURNE ED. Airborne transmission of influenza virus infection in mice. Nature. 1962;195:1129–1130. doi:10.1038/1951129a0 PMID: 13909471
Simonds AK, Hanak A, Chatwin M, et al. Evaluation of droplet dispersion during non-invasive ventilation, oxygen therapy, nebuliser treatment and chest physiotherapy in clinical practice: implications for management of pandemic influenza and other airborne infections. Health Technol Assess. 2010;14(46):131–172. doi:10.3310/hta14460-02 PMID: 20923611
Smith JD, MacDougall CC, Johnstone J, Copes RA, Schwartz B, Garber GE. Effectiveness of N95 respirators versus surgical masks in protecting health care workers from acute respiratory infection: a systematic review and meta-analysis. CMAJ. 2016;188(8):567–574. doi:10.1503/cmaj.150835 PMID: 26952529
Tellier R. Aerosol transmission of influenza A virus: a review of new studies. J R Soc Interface. 2009;6 Suppl 6(Suppl 6):S783–S790. doi:10.1098/rsif.2009.0302.focus PMID: 19773292
Thompson KA, Pappachan JV, Bennett AM, et al. Influenza aerosols in UK hospitals during the H1N1 (2009) pandemic–the risk of aerosol generation during medical procedures. PLoS One. 2013;8(2):e56278. doi:10.1371/journal.pone.0056278 PMID: 23418548
Tran K, Cimon K, Severn M, Pessoa-Silva CL, Conly J. Aerosol generating procedures and risk of transmission of acute respiratory infections to healthcare workers: a systematic review. PLoS One. 2012;7(4):e35797. PMID: 22563403
Wong TW, Lee CK, Tam W, et al. Cluster of SARS among medical students exposed to single patient, Hong Kong. Emerg Infect Dis. 2004;10(2):269–276. doi:10.3201/eid1002.030452 PMID: 15030696
Xie X, Li Y, Chwang AT, Ho PL, Seto WH. How far droplets can move in indoor environments–revisiting the Wells evaporation-falling curve. Indoor Air. 2007;17(3):211–225. doi:10.1111/j.1600-0668.2007.00469.x PMID: 17542834