Pathway 1: factors that affect the amount of virus exhaled by an infected person

Participants were asked to give their estimate of the percentage relative difference in infectious virus produced by someone with normal respiratory rate during a range of activities in comparison to silent breathing (this was considered to be 100%). This included speaking at average volume (ie, not projecting their voice); speaking loudly such as in a classroom, theatre or lecture; coughing; singing, exercising, eating; infected with a new variant of concern (at this point, the main variant of concern being considered was the alpha variant). Seven participants contributed estimates in round 2; participants’ individual and collated responses are summarised in figure 5 and in table 1, with additional data in online supplemental materials 3. The ranges shown in each case encompass participants’ estimates of the uncertainties and the range of variability due to environmental and individual factors.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5

Participants’ estimates of the percentage increase that each of a range of different activities would cause in exhalation of infectious virus by an infected person: (A) speaking, (B) speaking loudly, (C) coughing, (D) singing, (E) exercising, (F) eating and (G) infected with a variant of concern. In each case, the top graph shows individual estimates, the middle graph the average of those estimates, and the bottom graph a cumulative probability distribution, with (in green) the median and interquartile range. Values in parentheses to the right show each participant’s estimate of the percentage likelihood of the true value lying below, within and above the range identified in round 1, shown by the orange lines. Values in parentheses in the bottom graph show 25th, 50th and 75th percentile estimates from the combined responses. The vertical dashed line shows the 100% reference case (that is, no different from silent breathing).

Most participants made the assumption that the pattern of infectious virus emission follows that of particle emissions because we cannot directly measure the amount of infectious virus. However, some specifically mentioned that studies with measurements of ‘microbial’ (not viral or SARS-CoV-2) concentrations in respiratory output may not necessarily mirror the pattern of all particle emissions.

Other factors that participants suggested might affect the amount or distribution of infectious virus produced included the individual’s age (with children potentially being lower emitters34), sex (men potentially emitting higher levels of virus), stage of infection (eg, whether they were symptomatic or not), smoking status (with smokers emitting higher levels) or use of snuff and possible other nicotine delivery systems, immune status (with those on immunosuppressive drugs or with reduced immune systems producing higher levels, as well as potentially those with respiratory allergies), and potentially co-infection with other micro-organisms or certain treatments, such as antivirals.

Participants noted that individual differences in speech such as style and language (eg, the amount of ‘plosive’ consonants), loudness and moistness of mouth were all likely to affect projection of infectious virus. Similarly wetter and more ‘productive’ coughs, with copious nasopharyngeal secretions, were thought more likely to have higher infectious components. With the estimates on emission during singing and exercise, participants often mentioned that they were particularly influenced by the studies of real-world ‘superspreader’ events which they felt were more reliable indicators of infectiousness than laboratory studies.

Participants highlighted little evidence for the amount of virus produced while eating, but mentioned reasons for believing that there might be either a potential decrease in time spent with mouth open, or that respiratory fluids were soaked into food and swallowed; or an increase in time spent with mouth open, increased salivation, mechanical movements of the jaw and tongue or people clearing their throats.

When asked about the effects of infection by any of the current variants of concern, participants commented that there was not good evidence of higher viral load or viral shedding in these variants and that their increased infectivity could be due to a greater ability of the virus to evade an antibody response as much as through increased production of viral particles. It should be noted that during round 1, this was mainly the alpha variant B117. During round 2, there were several known variants of concern with apparent higher transmissibility, but the survey was completed before the delta variant became widespread.

Pathway 2: factors that affect the amount of virus emitted to the environment in respiratory output

Size distribution of viral emissions: Respiratory output from individuals contains a spectrum of wet aerosol and droplet sizes, each of which might behave in a different way in terms of persistence in an environment and response to different mitigations. Participants were therefore first asked their opinions about how much of the virus exhaled by an infected person (taken as a baseline 100%) might be contained within each of three broad fractions: path A (aerosols <10 μm in diameter), path B (small droplets, 10–100 μm) and path C (large droplets, over 100 μm). Participants were asked this for an individual who was silent and respiring normally, but also for an individual performing each of the activities outlined in pathway 1 above. Six participants contributed data in this pathway, with data summarised in table 2 and online supplemental information 4. Again, the ranges given encompass participants’ estimates of both the uncertainty and variation due to environmental and individual factors inherent in each scenario described.

As in round 1, participants were divided on the degree to which they felt the viral particles were predominantly carried in the fraction of the respiratory output falling under the definition of ‘aerosols’. For silent breathing, six out of seven participants suggested that virus may be predominantly in aerosols, for speaking at normal volume, five out of six participants thought that the virus would likely be spread fairly evenly across the different fraction sizes. For the other activities there was further disagreement among the experts, highlighting a substantial gap in knowledge.

Similar to pathway 1, participants discussed the challenges of detecting virus particles in the air, and that a lot of studies assume that virus is distributed evenly across different respiratory particle sizes by volume. There was also discussion on whether people with symptomatic infection would produce a different distribution of particle sizes due to stress, fever and respiratory symptoms.

Participants noted that there were many individual factors that would create variation in the percentage splits, such as the proportion of the time spent breathing through the nose versus the mouth. Environmental factors such as humidity and temperature could also affect the moisture content, but the effects of these factors on viral content is unknown. During coughing, some participants thought that particles from coughing were more likely to come from the deep lung and hence be smaller, while others thought they might be larger. It was noted that there was little empirical research in patients with SARS-CoV-2 specifically.

No participants thought that any of the variants of concern were likely to behave differently on this pathway.

Impact of source control mitigation: Participants were asked to estimate the difference that four different potential face covering mitigations would have on the amount of infectious virus that would reach the environment via each of the three size fractions (A: aerosols, B: small droplets, and C: large droplets). They were asked to assume a reasonable fit (eg, covering the nose) and no resuspension of material deposited on the mask or visor.

Table 3 shows there was a good degree of consensus, with participants judging higher quality face masks to provide the greatest protection against all particle sizes. Home-made face coverings and surgical masks show the greatest variation, which may reflect assumptions relating to both fit and material effectiveness.

Pathway 3: factors that affect the amount of virus that can be inhaled or inspired from environment by a susceptible person

Seven participants contributed data on this pathway, considering loss to the environment, once exhaled, of infectious virus emitted via aerosols (0–10 micrometre diameter) or small droplets (10–100 μm) under different environmental conditions, or different mitigations that might be applied.

Dispersion of virus with location and ventilation: First, participants were asked to consider the effects of room size and ventilation on the percentage loss during transmission between a single infected person and a single susceptible person standing 2 m apart with no face coverings (note, this is loss once exhaled, while pathway 2 considered exhalation into the environment).

As seen in table 4, participants differed in their opinions of the percentage loss to the environment. Participants in round 2 had been offered low, medium or high loss of aerosols as a way of separating the ’camps’. Although participants coalesced more than in round 1, to just one or two ’camps’, their estimates still showed considerable disagreement over the likely percentage loss (see online supplemental information 5).

Participants in round 1 had already raised the possible effects of humidity and temperature, natural leakage rates of a building affecting air change rates, the physical location of the two people relative to each other and any furniture, and the potential influence of other people on the indoor setting. For the outdoor scenario, wind speed, turbulence and obstructions as well as temperature and humidity were raised as likely significant factors. Participants disagreed in their opinions on whether there would be any reduction at all of infectious virus in the air, regardless of whether the moisture component of the exhaled air evaporated (which itself was contended). Some thought that the viral elements, even in the small droplets, might settle onto surfaces over time. There was near universal agreement that the necessary experiments had not yet been done, exploring air flow patterns using suitable tracers within a number of realistic situations compared against data with actual viral particles.

Dispersion with distance: Participants were then asked to consider the effects of distance, by considering the percentage loss if the infected and susceptible person were either closer than 2 m or further apart than 2 m. Participants rated the quality of the evidence behind their estimates on the effects of distance as 3.0 for aerosols and 3.25 for small particles. A summary of their estimates is shown in table 5, and again participants’ responses showed considerable variability (see online supplemental information 5).

One participant pointed out that if participants are closer than 2 m apart then the size of an unventilated room is essentially irrelevant. In ventilated rooms, the design of the ventilation and resulting airflow patterns relative to the location of individuals may become important. Outdoors, the amount of ambient air flow in most scenarios means that beyond 2 m distance participants suggest transmission is very unlikely.

Measures to protect the susceptible person: Finally, participants were asked to estimate any additional percentage loss that might be caused by the susceptible person being behind a face covering or shield (assuming an appropriate fit and no accompanying behaviour change).

There was general agreement that exposure to larger particles would likely be reduced more than smaller ones. When it came to a plastic face visor, some thought that may even increase exposure to small droplets. See table 6.

A few participants mentioned other potential mitigation measures, such as air cleaning (although there was no clear evidence of efficacy that they could cite), or personal ventilation systems.35–41

Pathway 4: factors affecting surface transmission of the virus

In this, the largest individual section of the survey, participants were asked their opinions on a range of questions involving the transfer of infectious SARS-CoV-2 virus from large droplets (>100 μm) to surfaces and a susceptible person’s hands. Six participants contributed data to this section.

Fate of large droplets: For four scenarios, participants were asked for their estimates of the proportions of virus they thought likely to be transmitted from large droplets to the hands of an infected person onto all surrounding surfaces, or directly onto the hands of a susceptible person, compared with those likely to remain in the air (ie, going on to pathway 5).

Generally there was agreement that the amount of virus going directly onto the hands of a susceptible person was likely to be very small, and that droplets of this size would mostly tend to fall quite rapidly onto surfaces (table 7). Round 1 data did not obviously show a bimodal distribution in the responses. However, round 2 responses do indicate variations of opinion (see online supplemental information 6). For example, estimates of whether the amount of virus transmitted to surrounding surfaces would be above the indicated range (upper boundary 58.1%), in a situation where participants were not in direct contact, ranged from 1% to 64%. Similar differences are seen in many of the other estimates. We note that these values might also change if symptoms such as a runny nose are features of infection.

Participants noted that uncertainties and variability in estimates would be caused by factors such as the ambient temperature, humidity and the distance between the individuals, as well as their orientation to each other and their behaviour. Participants mentioned, for example, the amount of talking, respiratory hygiene and amount of coughing. Although these were considered in other parts of the questionnaire and here the intention was to elicit estimates regarding the fate of a given amount (considered 100%) of droplets larger than 100 μm, the proportions that might be projected different distances and in different directions by these sorts of behaviours were seen by participants as important sources of variation in the proportion that would follow each of the specified onward paths.

Participants did note that, although it stated that percentages should sum 100%, the question wording was not specific about whether ‘surrounding surfaces’ should include only those that might be touched (eg, excluding the floor), or for all surfaces, which might explain some of the diversity of views. In the scenarios involving tables or desks, most participants thought that more was likely to be deposited on these as they were closer to and more in line with people’s faces (as well as being ‘shared surfaces’). The nature of shared objects was also a cause of uncertainty in some scenarios, although again some participants’ answers suggested that they were already attempting to take into consideration aspects of the transmission that were to be considered further ‘downstream’, such as the amount of contact with those surfaces.

Some participants thought there would be no effect of duration of exposure, one that there would be an upward curve to steady state and one that it would keep increasing with time. One cited references, without comment.42–44 All agreed that it was unlikely that any variants of the virus would cause different estimates at this point in the transmission pathway.

Mitigating large droplet transfer: Participants were asked whether a range of potential mitigation measures would affect the transmission of the virus to the hands of the infected or susceptible person or surrounding surfaces; see table 8 and online supplemental information 6 for graphs of the individual and pooled responses.

Some participants thought that ventilation could change air flow paths to the extent that transmission might be reduced along all pathways to a small degree, but several others thought that ventilation indoors would make no real difference at all to large droplets. It was also speculated that ventilation might change relative humidity, which might change transmission along these pathways. Outdoors, ultraviolet levels, humidity and localised air flow patterns were all suggested to have an effect, with increased wind speed potentially carrying droplets further.

There was agreement that the infected person wearing a face mask was likely to substantially reduce large droplets being spread both through direct capture and changing the wearer’s behaviour to reduce face touching. Participants suggested that a Perspex screen between the infected and susceptible person might reduce the chance of large droplet transmission directly to the hands of the susceptible person, and reduce the extent of spread to shared surrounding surfaces.

Participants were split on respiratory hygiene, with some thinking it could almost block some of the potential transmission paths, and others being a lot more pessimistic about its effectiveness, thinking that the percussion behind a sneeze or cough is too great for a tissue or sleeve to catch the particles. One participant pointed out that contamination to the susceptible person’s hands may even increase if they were holding the tissue. When asked about gloves, participants said that absorbent materials might behave differently from skin or non-porous materials and also that gloves may affect behaviour. Some participants may have already been thinking ahead to the next stage of the transmission chain—the likelihood of virus survival or passing on to a susceptible person from hands or gloves. This may explain some diversity in answers, but it is important to note that here the diversity of opinions ranges from ‘0%’ to ‘70%–100%’ for the potential reduction in virus transmission to the infected person’s hands and the susceptible person’s hands.

When asked about halving the length of the time of the encounter, participants generally thought this would make little difference in terms of the infected person’s hands because these would be infected outside of the timeframe of the encounter. For contamination of surrounding surfaces or the susceptible person’s hands, one participant described a linear proportional increase with time, then becoming balanced by exponential natural inactivation of the viral particles (faster on skin than on most other surfaces), based on references indicated.

None believed that any of the variants of concern of the virus known about at the time were likely to behave differently in this pathway.

Transfer from hands to surfaces and people: Participants were then asked their estimates for the proportion of infectious virus present on the hands of an infected person that would likely be passed on in a viable form to surrounding surfaces compared with those that might be transferred directly to the hands of another person under a number of different scenarios —with no mitigations and with encounters long enough for virus transfer to be at steady state (but no prior contamination until the infected person enters the scenario). See online supplemental information 6 for participants’ individual estimates and table 9 for a summary.

Participants mostly based their answers on work done on other pathogens, suggesting hand-to-surface transfer efficiencies would depend on the materials involved, the area of contact, and potentially duration of contact. Inactivation rates of the virus would need to be borne in mind, but most participants assumed that in these scenarios it would be irrelevant as contact would be a short time after contamination of the hand. On the ‘shared desk’ scenario, one participant cited work showing high rates of contact with desks, but overall transfer was thought to be below 50% on this path. As direct contact increased, the probabilities increased, and one participant mentioned that there was some evidence that transfer efficiencies are log-normally distributed.45

No one knew of evidence that any variants of concern would behave differently in this pathway.

In mitigating transfer of virus through hand hygiene (table 10), participants were asked to assume it was carried out well, but raised the importance of timing (including how soon after washing their hands contact with surfaces happened, during which time there was the opportunity for recontamination) and the variable effectiveness of hand hygiene. Participants discussed uncertainties associated with contamination during donning and doffing of gloves, and the behavioural changes that gloves might encourage such as potentially less face touching, but potentially a feeling of protection leading to more sneezing into hands and more surface touching.

When asked about the effects of halving the time spent in a scenario, participants generally indicated that the question was not answerable because it was absolute time, not relative time, that was important; and far more important than time was frequency of surface touching.

Transfer from surface to hands: The final section of this pathway focused on transfer from a contaminated surface to the hands of a susceptible person. Participants were asked about likely loss during transference in a number of different scenarios, initially with no mitigations.

Participants discussed uncertainties caused by the unknown duration and frequency of touch, the humidity and roughness of the surfaces, and whether the virus was in wet or dried respiratory secretions. Some thought that metal and ceramic surfaces would potentially support longer survival of infectious virus than paper and card; others were not so sure. Some thought that transfer efficiency would likely be higher on hard, smooth surfaces and lower with paper. Others raised the differences in gripping behaviour with different surfaces and objects—with small items such as plates or condiments more likely to encourage overlapping areas of grip between hands, increasing potential transfer. See table 11 for a summary of their estimates, and online supplemental information 6 for the individual estimates.

They agreed that there was no evidence to suggest that any of the variants of concern would behave differently in this pathway.

Participants mentioned that surface cleaning (described as ‘cleaned thoroughly with an alcohol-based cleaning product’) had large uncertainties associated with the timing between cleaning and surface touching, although the cleaning itself, if done appropriately, could be highly effective. With gloves, some thought there was even enhanced possibility for transfer due to a larger surface area, and the differences between different glove materials was discussed. When it came to hand hygiene, participants mentioned that washing changes the surface properties of hands temporarily to become more hydrophilic, and some speculated that if hands are not dry they may increase surface transfer efficacy. See online supplemental information 6 for their individual estimates, and table 12 for summaries.

When asked about transfer efficacy from contaminated hands which were secondarily contaminated, rather than directly contaminated from an infected person’s respiratory secretions, participants noted that the amount of drying that would have taken place along the chain of contamination would likely reduce transfer efficacy.

Another potential mitigation that participants raised but about which opinions were not elicited was the potential role of anti-microbial surfaces such as copper.

Pathway 5: factors affecting the transmission of large particles directly to a susceptible person’s mucous membranes

In this section, participants were asked their opinions on how much infectious virus was likely to be lost to the environment from large droplets in the air (>100 μm) before reaching a susceptible person’s mucous membranes under different conditions. Four participants contributed data to this pathway. See online supplemental information 7 for details of their estimates.

Many participants in round 1 had found these estimates particularly difficult to make because of the degree of chance and variability according to specific conditions involved, such as the directions the people involved were facing, their behaviour, their speech and respiration patterns. The wording in round 2 appeared to help participants, although there was still a fair degree of difference of opinion (see figure 6). It was noted that the UK’s ‘2 m rule’ was based on the distance that these larger particles might travel but that coughing (or an exhaled ‘puff’) was not considered in this rule.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6

Participants’ estimates of the percentage of infectious virus carried in large (>100 μm) airborne particles that would NOT reach a susceptible person’s mucous membranes if the infected and susceptible person were (A) within 2 m of each other and (B) further than 2 m apart from each other in an unventilated room. In each case, the top graph shows individual estimates, the middle graph the average of those estimates, and the bottom graph a cumulative probability distribution, with (in green) the median and interquartile range. Values in parentheses to the right show percentage below, within, and above the range. Values in parentheses on the left of the bottom graph show 25th, 50th and 75th percentile estimates from the combined responses.

All participants agreed that it was unlikely that any of the variants of concern of SARS-CoV-2 would behave differently in this pathway. They also agreed that their estimates of the likelihood of transmission along this pathway would not be affected by ventilation of the room or whether the people were outdoors rather than indoors. See table 13 for a summary of their estimates.

When asked how the length of time of exposure might affect estimates, participants mentioned that this was not known but that it probably increased linearly—possibly up to a threshold at which point infection was almost certain. However, this point was not yet known.

Face coverings worn by the susceptible person were thought to reduce the large droplet transmission potential because they cover the mouth and nose (although not the eyes). One mentioned that contamination from the outside of the mask to the hands or the inside of the mask could occur, but that transfer from a porous fabric to a hand was likely to be low. Participants generally thought that FFP3 masks were likely to offer little additional reduction in these larger droplets over a surgical face mask. A large factor in variability of mitigation offered by a surgical face mask would be tightness of fit. It was mentioned that a face visor formed a physical barrier which, in other diseases and in observational data from COVID-19 outbreaks, seemed to reduce infection risk from larger droplets.

When considering Perspex screens, participants mentioned that the positioning was important, and in the real world many were located too far away. They also mentioned that a poorly positioned screen could potentially increase exposure to infected aerosols.46 47 See table 14 for a summary of their estimates.

In terms of other possible mitigations, one participant mentioned a study from China suggesting that wearing glasses may decrease transmission by protecting the eyes, but that this had not been followed up with more evidence.48

Pathway 6: how much virus is likely to be transmitted from a susceptible person’s hands to their mucous membranes under different conditions

Five participants contributed data on pathway 6. They were first asked the likely percentage loss of virus during transfer from a susceptible person’s hands to their mucous membranes with no mitigations in place, and then the loss due to different potential mitigations that the susceptible person might employ. Participants’ individual estimates and summary graphs are given in figure 7 and table 15.

• Download figure
• Open in new tab
• Download powerpoint

Figure 7

Participants’ individual estimates (top graphs) and pooled estimates (middle and bottom graphs) of the percentage reduction in infections virus transferred from a person’s contaminated hands to their mucous membranes A) under normal conditions, (B) if the person was wearing a face shield, (C) if the person was wearing gloves, (D) & (E) if the person was wearing a face covering (participants divided into two distinct camps on this question), (F) intervals elicited for the first time in round 2 from individuals (top graph, blue) and pooled (bottom graph, green) if the person were practising good hand hygiene. For graphs (A) to (E) the top graph shows individual estimates, the middle graph the average of those estimates, and the bottom graph a cumulative probability distribution, with (in green) the median and inter-quartile range. Values in parentheses to the right show percentage below, within and above the range. Values in parentheses in the bottom graph show 25th, 50th and 75th percentile estimates from the combined responses. The vertical dashed line shows the 100% reference case. For graph (F) the figures on right hand axis show individual responses in the upper graph and the 25th, 50th and 75th percentile values in the lower graph.

Participants mentioned that uncertainties around the behaviour of the individual such as the amount of face touching was important and there may be some gender differences.49 The time elapsed since hands were contaminated (‘fresh’ or dry), the environmental conditions of the hands (moisture content, pH, temperature etc) and the part of the hand used/contaminated50 were all cited as relevant factors that may influence hand-to-mucous membrane transfer. Participants pointed out that face coverings do not protect the eyes, but there is evidence that they can reduce the frequency of face touching in general. With gloves, there was not so much evidence on how much they might change face touching, but there was a view that viral transfer was still possible from contact between gloves and membranes. When asked about face visors, they mentioned that although visors may remind people not to touch their face, they were much more able to reach under it to do so than with a face mask, and that an itch could be scratched ‘though’ a fabric mask without skin-to-skin contact.

Hand sanitisation, participants said, could be very effective but depends how frequently and how well it is performed, leaving very high variability and uncertainty in any estimate.

It was pointed out that the different degrees of absorption of infectious virus through different mucous membrane routes (eg, eyes vs nose vs mouth) was unknown, and that almost all studies on this pathway (such as exist) are laboratory based and so miss out on crucial ‘real world’ behavioural aspects, and many are not on SARS-CoV-2 specifically and so their generalisability to this particular virus is unknown.

They were not aware of any evidence that would suggest that any of the variants of concern behaved differently on this pathway. However they raised other methods of transference such as touching cigarettes, inhalers and other oral devices; food and drink.